DIDACTICAL PHENOMENOLOGY OF MATHEMATICAL STRUCTURES

DIDACTICAL P H E N O M E N O L O G Y OF MATHEMATICAL STRUCTURES

HANS FREUDENTHAL

DIDACTICAL
PHENOMENOLOGY
OF MATHEMATICAL
STRUCTURES

KLUWER ACADEMIC PUBLISHERS

NEW YORK, BOSTON, DORDRECHT, LONDON, MOSCOW

eBook ISBN:
Print ISBN:

0-306-47235-X
9-027-71535-1

2002 Kluwer Academic Publishers

New York, Boston, Dordrecht, London, Moscow
Print 1999 Kluwer Academic / Plenum Publishers
New York
All rights reserved
No part of this eBook may be reproduced or transmitted in any form or by any means, electronic,
mechanical, recording, or otherwise, without written consent from the Publisher
Created in the United States of America
Visit Kluwer Online at:
and Kluwer's eBookstore at:

http://kluweronline.com
http://ebooks.kluweronline.com

TABLE OF CONTENTS

EDITORIAL PREFACE

vii

A LOOK BACKWARD AND A LOOK FORWARD

ix

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.

As an Example: Length
The Method
Sets
Natural Numbers
Fractions
Ratio and Proportionality
Structures: in particular, Geometrical Structures
Putting into Geometrical Contexts
Topology as a Geometrical Context
The Topographical Context

Figures and Configurations

Geometrical Mappings
Measuring by means of Geometry
Topography with Geometry
Negative Numbers and Directed Magnitudes
The Algebraic Language
Functions

1
28
34
73
133
178
210
223
250
281
296
330
351
407
432
461
491

LIST OF THE AUTHORS PUBLICATIONS ON MATHEMATICS

EDUCATION

579

INDEX

586

This page intentionally left blank.

EDITORIAL PREFACE

The launch of a new book series is always a challenging event -- not only for the
Editorial Board and the Publisher, but also, and more particularly, for the first

author. Both the Editorial Board and the Publisher are delighted that the first
author in this series is well able to meet the challenge. Professor Freudenthal
needs no introduction to anyone in the Mathematics Education field and it is
particularly fitting that his book should be the first in this new series because it
was in 1968 that he, and Reidel, produced the first issue of the journal Educational Studies in Mathematics. Breaking fresh ground is therefore nothing new
to Professor Freudenthal and this book illustrates well his pleasure at such a
task. To be strictly correct the ground which he has broken here is not new,
but as with Mathematics as an Educational Task and Weeding and Sowing,
it is rather the novelty of the manner in which he has carried out his analysis

which provides us with so many fresh perspectives. It is our intention that this
new book series should provide those who work in the emerging discipline of
mathematics education with an essential resource, and at a time of considerable
concern about the whole mathematics curriculum this book represents just such
resource.
ALAN J. BISHOP

Managing Editor

vii

This page intentionally left blank.

A LOOK BACKWARD AND A LOOK FORWARD

Men die, systems last. Immortality is assured to those who build their name into
a system. Although even immortality is not what it used to be any more, and I

did not crave for it, I once a day set my mind on writing my first systematic
work, after a few that can rightly be called chaotic. The result has been the
most chaotic of all so chaotic that when the reader expects a preface he has
to wait for Chapter 6. Moreover the work is incomplete. When unexpectedly
Chapter 18, Text and Context, showed the first symptoms of elephantiasis,
I cut it off, appointed Chapter 17 to be the last, and exiled the remainder to
a separate book, which is very likely to become still-born. Let me add that my
secretary and collaborator for almost 25 years, Mrs. Breughel read and wrote
the last line of the illegible Dutch manuscript of this book the day before she
retired.

But a manuscript like the present deserves a look backward by its author,
which at the same time should be to its reader a look forward.
A common theme of the greater part of my publications in mathematics

education has been: change of perspective; in particular what I called inversion

and conversion, a mathematical virtue, practised and cherished from olden times.
No mathematical idea has ever been published in the way it was discovered.
Techniques nave been developed and are used, if a problem has been solved,
to turn the solution procedure upside down, or if it is a larger complex of
statements and theories, to turn definitions into propositions, and propositions
into definitions, the hot invention into icy beauty. This then if it has affected
teaching matter, is the didactical inversion, which as it happens may be antididactical. Rather than behaving antididactically, one should recognise that
the young learner is entitled to recapitulate in a fashion the learning process
of mankind. Not in the trivial manner of an abridged version, but equally we
cannot require the new generation to start just at the point where their predecessors left off.
Our mathematical concepts, structures, ideas have been invented as tools
to organise the phenomena of the physical, social and mental world. Phenomenology of a mathematical concept, structure, or idea means describing it in
its relation to the phenomena for which it was created, and to which it has been
extended in the learning process of mankind, and, as far as this description is
concerned with the learning process of the young generation, it is didactical
phenomenology, a way to show the teacher the places where the learner might
step into the learning process of mankind. Not in its history but in its learning
process that still continues, which means dead ends must be cut and living
roots spared and reinforced.
ix

A LOOK BACKWARD AND A LOOK FORWARD

Though I did not use the term explicitly, didactical phenomenology already
played a part in my former work. In the present book I stress one feature more

explicitly: mental objects versus concept attainment. Concepts are the backbone
of our cognitive structures. But in everyday matters, concepts are not considered
as a teaching subject. Though children learn what is a chair, what is food, what is
health, they are not taught the concepts of chair, food, health. Mathematics is
no different. Children learn what is number, what are circles, what is adding,
what is plotting a graph. They grasp them as mental objects and carry them out
as mental activities. It is a fact that the concepts of number and circle, of adding
and graphing are susceptible to more precision and clarity than those of chair,
food, and health. Is this the reason why the protagonists of concept attainment
prefer to teach the number concept rather than number, and, in general, concepts
rather than mental objects and activities? Whatever the reason may be, it is an
example of what I called the anti-didactical inversion.
The didactical scope of mental objects and activities and of onset of conscious
conceptualisation, if didactically possible, is the main theme of this phenomenology. It was written in the stimulating working atmosphere of the IOWO.* So
it is dedicated to the memory of this institution that has been assassinated, and
to all its collaborators, who continue to act and work in its spirit.

* The Netherlands Institute for Development of Mathematics Education.

CHAPTER 1

AS AN EXAMPLE: LENGTH

1.11.11. P H E N O M E N O L O G I C A L

1.13. What is Length?

1.1. Length has more than one meaning. At length, going to the utmost
length, length and width include in their context length in different
meanings. The one I am concerned with becomes clear if along side the question
what is length?
I put a few other questions:

what is weight?
what is duration?
what is content?
Length, weight, duration, content are magnitudes, among which
length has its special status.
If I use the word length in the sense, made more precise here, I mean length
of something, of a long object. Length then is synonymous with width,
height, thickness, distance, latitude, depth, which are related to
other dimensions or situations. For the sides of a lying rectangle one prefers
length and width, for a standing one, width and height.

1.2. Without stressing it, I have turned my question what is length? towards
an answer such as length of ... is .... This is a typically mathematical turn:

transforming apparently isolated terms into symbols of functions. The question

what is mother?
what is brother?
what is neighbor?
are more easily answered according to the pattern

mother o f . . . is ...
brother o f . . . is ...
neighbor o f . . . is ....
More precisely:

mother of x is she who has born x,

1

CHAPTER 1

brother of x is every y such that y is a male and x and y have the same
parents,
neighbor of x is every y such that x and y live beside each other.
Afterwards mother can also be defined in an isolated way:
x is mother if there is a y such that x is mother of y.
Linguistically man, stone, house belong to the same category as
mother, brother, neighbor as nouns they enjoy a substantiality, though
that of mother, brother, neighbor differs from that of man, stone,
house. Being mother, being brother, being neighbor get a meaning
only by the explicit or implicit addition of whom. In they are brothers,
they are neighbors the additional of ... seems unnecessary but is not: they
are brothers or neighbors of each other.
1.3. Back to length, interpreted in length of ... as a functional symbol:
a function that talks about long objects how long they are, though not necessarily numerically specified, as in
the length of this bed is 1.90 m.
Functional value may be vague: long, very long, short, very short, and so on.
The reason why I neglect these values now is that I will start by focusing on a
phenomenology of mathematical structures. Are long, very long, short,
very short not mathematical concepts? Such questions will be answered later
on; in order not to complicate things, I delay the answer.
1.4. Magnitudes*
Before continuing let me consider the terms mentioned earlier. All of them aim
at functions:
weight: weight of (a heavy object),
duration: duration of (a time interval),
content: content of (a part of space).
Let me introduce abbreviations:
l(x):
p(x):
d(x):
v(x):

length of x,
weight of x,
duration of x,
content of x,

* Directed magnitudes will incidentally be considered in Sections 15.912. Otherwise

magnitude is always understood in the classical way. In this context rational and real
always means positive rational and positive real.

AS AN E X A M P L E : LENGTH

where x is something that can properly be said to have a length, weight, duration,
content.
We again pose the question of the possible values of the function l (and of
p, d, v). Not long short, heavy light, big small, respectively, but since
we speak mathematics, more precise values. This does not oblige us to state
something like 1.90 m, 75 kg, 7 sec,
, expressed in the metric system, or
in any a priori system of measures. This is a liberty we can profit from to get
deeper insight. Indeed, it appears that we can go rather far without accepting
any special system of measures.
Let us call the
values of l
values of p
values of d
values of v

lengths,
weights,
durations,
contents,

and the
system of lenths
system of weights
system of durations
system of contents

L,
W,
D,
V,

and look for their properties.

1.5. Adding Lengths
The first thing we notice is that we can add lengths even before conceiving them
numerically. How is it done? Given two lengths and we provide ourselves
with two long objects x and y with lengths

respectively, and compose them (in a way that asks for detailed explanation)
into a new long object
This object has a certain length, consequentially
named
It was our intention to define the sum of lengths and and by
definition we put

that is to say
1.5.1

In other words
the length of the composite equals the sum of the composing parts.

CHAPTER

As regards this kind of definition one has to pay attention to one point:
For the lengths x and y we have chosen representative long objects x and y,

respectively, with lengths as prescribed. Instead we could have chosen other

representatives, say x' and y', thus such that again

which would lead to a composite

be meaningful, we must be sure that

In order for the definition 1.5.1. to

1.5.2

in other words, that

the length of the composite does not depend on the choice of the
representatives.
I have to take care that my way of combining long objects fulfills this
condition.
In a similar way this ought to be true of weights. Given two weights and
the sum of which I propose to define, I take two heavy objects x and y
with weights and
respectively, compose them into a new heavy object

and define

Again, replacing x and y by x' and y' with the same weights, respectively, must

not change the weight of the composite. This requirement looks self-evident, and
it is so for a good reason, indeed: we would never have focused on length,
weight, and so on if this condition were not fulfilled.
A second remark: If composing is meant to lead to defining the sum, it must
be carried out in such a way that the components do not overlap. Suppose I
want to add a length to itself in order to define the length
Then for
each of the summands I need another representative, thus

in order to get

So I cannot manage with one representative for each length. Fortunately with
lengths it is rather easy to provide oneself with two, three, or more representatives of the same length; instruments like a ruler can repeatedly be applied. In
the case of weights and so on, the difficulty of obtaining enough representatives
looks greater, but this is a point we are not concerned with here.
In carrying out the operation as imagined in the various cases, order does
not play a role and, as a consequence, addition of lengths, weights, etc., obeys
the laws of commutativity and associativity:

AS A N E X A M P L E : L E N G T H

The first property stated in the systems L, P, D, and V of lengths, etc., therefore
is:
I.

A commutative and associative operation (+) of addition in L, and

so on.

1.6. Order of Lengths

Adding will later be joined by subtracting; that is, the smaller from the bigger.
But smaller and bigger are ideas we have not yet come across. They will
now be considered.

Relations like smaller bigger belong to the so-called order relations: any
pair of elements

of L is in exactly one of the situations

1.6.1

and for three of them,

1.6.2

if

and

then

holds (the so-called transitivity).

Such an order relation can now be defined in L by means of the addition.
We express the property that

by adding, something can become only larger,

in a formula
1.6.3
for any lengths
if
and

and
This immediately ensures transitivity 1.6.2. Indeed,
then there is a k and a such that

so

and thus
The first requirement, 1.6.1, on an order relation is a bit trying. It means

1.6.4

If
then there is
either a k with
or a with

though not both together.

CHAPTER 1

For a moment I call two long objects x, y directly comparable if either x

can be considered as a composing part of y or y as a composing part of x. Then
1.6.4 can be translated as follows into the language of long objects:
1.6.5

Given two long objects x, y, then I can find directly comparable

long objects x', y' such that
and however

I choose x', y' in this way, one thing is true:

either x' is a composing part of y',
or y' is a composing part of x' .
The second property we have stated for the systems L, W, etc. is:
II.

The definition
in L, etc.

for all lengths

determines a total order

1.7. Multiplying Lengths

If we repeatedly add the same length, then the resulting lengths can be denoted
as

and so on; in general

with n summands.
Laws like

are obvious.
From adding we have now derived multiplying elements of L, etc., by positive
integers, that is, elements of
As an inverse of this operation one has dividing,
which means:
Given a length and an
then the equation
1.7.2

has a solution

then

There is only one such

since if

AS AN E X A M P L E : LENGTH

In the first and third case this would result in

respectively, which is impossible, and leaves us with

The solution of 1.7.2 gets the name

Thus

is defined by

1.7.3
This then is our third property of lengths, etc.:
III.

For every
such that

etc., and every

there is one

etc.,

The following laws for dividing are easily verified:

1.7.4

1.8. Rational Multiples of Lengths

Multiplying and dividing elements of L, etc., by elements of
One puts

can be combined.

1.8.1

which results in multiplying of lengths by positive rational numbers. Since,

however, a rational number can be denoted in various ways,

we have to make sure that the definition 1.8.1 is valid; that is, we have to prove
that

CHAPTER 1

This is indeed true. According to 1.7.4

thus

Thus we can multiply every length, etc., by every positive rational number
We easily find the laws, for
etc.:

1.8.2

1.9. Real Multiples of Lengths

Starting from one length, etc. say
we can form all its rational multiples.
They constitute a set
In
two arbitrary elements are rational multiples
of each other. So
cannot possibly exhaust what we imagine the system of
lengths to be. Indeed, the diagonal and side of a square are not rational multiples
of each other. However,
does exhaust the system of lengths, etc. aproximately. One knows about a property, the so-called Archimedean axiom:

IV.

Given an
than all of

etc., then there is no element of L, etc., bigger

and no element of L, etc., smaller than all of

I now take an arbitrary

etc. It does not necessarily belong to
but according to IV it must lie in between. For each
1.9.1
holds. I now want to represent as a real multiple of

in such a way that the order fits, that is

In particular for
1.9.2
How to find such a u? Well, 1.9.1 causes a partition of

into three

AS AN E X A M P L E : LENGTH

classes (the second can be empty, or can consist of one element if

multiple of Such a partition is called a Dedekind cut:

the lower class:

the upper class:

the
the

is a rational

with
with

where at most one

can escape this division. Now there is a real
that causes the cut, that is to say
if
if

then
then

If now we put
we fulfill the requirements of 1.9.2.
It has been shown that

of two given elements of L, etc., each is a positive real multiple of

the other.

We can now conclude with the property

V.

For each

etc., and each

there is an
one can formulate laws for

Similarly to those of

etc.
and

etc.:
1.9.3

if

then

1.10. Length Measure

Let us break off the exposition and not insist on a systematic approach to
magnitudes.
For instance we could continue with a numerical treatment of magnitudes:
A measuring unit (m, kg, sec, , or suchlike) is chosen in order to express each
length, and so on, as a positive real multiple of the unit. Then each length, etc.,
is represented by a measuring number and according to its generation we find
the fundamental rule

under the composition the measuring numbers are added,

from which follows among others

the longer, heavier,... object has the bigger measuring number.

1.11. What is Lacking Here

The preceding was an example of phenomenology; namely, for the mathematical

10

CHAPTER 1

structure magnitude. Or rather, it was a fragment of such a phenomenology.

No attention was paid to measuring; connections between different magnitudes
should have been considered; and finally, what has not been mentioned at all
is that length is ascribed not only to long objects but also to broken and
curved lines. How broken lines, say the perimeter of a triangle, should be dealt
with is easy to guess. Curved lines are a different case. The classical way is approximation by broken lines but I shall skip it here in order to resume it later on.
1.121.29. DIDACTICALLY PHENOMENOLOGICAL

The preceding was not didactical phenomenology. In order to stress the difference I started with phenomenology as such. But also in the sequel didactical
phenomenology will often be preceded by phenomenology as such, to create a
frame of concepts and terms on which the didactical phenomenology can rest.
The difference between phenomenology and didactical phenomenology
will soon become apparent. In the first case a mathematical structure will
be dealt with as a cognitive product in the way it describes its possibly nonmathematical objects; in the second case, it will be dealt with as a learning
and teaching matter, that is as a cognitive process. One could think about one
step backwards: towards a genetic phenomenology of mathematical structures,
which studies them in the cognitive process of mental growth.
One might think that a didactical phenomenology should be based on a
genetic one. Indeed I would have been happy if, while developing the present
didactical phenomenology, I could have leaned upon a genetic one. This, however, was not the case, and the longer I think about the question, the more I
become convinced that the inverse order is more promising. In the sequence
phenomenology, didactical phenomenology, genetic phenomenology each
member serves as a basis for the next. In order to write a phenomenology of
mathematical structures, a knowledge of mathematics and its applications
suffices; a didactical phenomenology asks in addition for a knowledge of instruc-

tion; a genetic phenomenology is a piece of psychology.

All the psychological investigations of this kind which I know about suffer
from one fundamental deficiency: investigations on mathematical acquisitions
(at certain ages) have involved the related mathematical structures in a naive
way that is, they lack any preceding phenomenological analysis and as a
consequence, are full of superficial and even wrong interpretations. The lack of
a preceding didactical phenomenology, on the other hand, is the reason why
such investigations are designed in almost all cases as isolated snapshots rather

than as stages in a developmental process.

1.131.25. COMPARISON OF LENGTHS

1.1314. Length Expressed by Adjectives

1.13. Many mathematical concepts are announced by adjectives. Adjectives

AS AN EXAMPLE: LENGTH

11

belonging to length are: long, short, but also broad, tight, thick, thin,

high, low, deep, shallow, far, near, wide, narrow, and finally also
tall, sturdy, diminutive, insignificant. Of course the ability to distinguish
such properties precedes the ability to express them linguistically. For the adult
it is at least unconsciously clear how these expressions are related to the
same magnitude, length, and he often presupposes children to be well acquainted
with this relation. Researchers in this field are often not aware of this difficulty.
It is not farfetched to ask oneself how the child manages to develop a knowledge
of these connections. A disturbing factor is the overarching of this complex of
adjectives by big and little, which can serve so many aims (up to big boy
and little girl).
Bastiaan (5; 3) asks how big is a mole. When I show with my hands a moles length, he insists
no, I mean how high. He is compelled to differentiate big. Clearly he is conscious of
the fact that both cases mean a length.

The insight that both expressions mean a length is not at all trivial, for instance,
that a high tree, if cut, is long. As a matter of fact, even adults may have

problems with the equivalence of distances in the horizontal and the vertical
dimensions, at least with regard to quantitative specification.
How does the connection within this complex of adjectives come into being?
How is the common element constituted? If I may guess, I would attribute a
decisive role to the hand and finger movements that accompany such statements
as that long, that wide, that thick, and so on (likewise that short, and so on)
movements that can turn in different directions and possess different intensities
but always show the same linear character. (Compare this with the mimic
expressions of embracing, which may accompany that much, but also that
thick, and with the mimic and acoustic expression of lifting belonging to
that heavy.)
The common element in this complex of adjectives for length is possibly not
yet operational in all young school children; as a matter of consciousness it may
even be absent in many older ones. Acquiring it and becoming conscious of it are
an indispensable condition for mathematical activities.
1.14. Around such adjectives as long there is a complex of relational expressions like:
longer, longest, as long as, less long, not as long as, too long, very long.

Here again the ability of distinguishing precedes that of linguistic expression

(for instance, something cannot pass through a hole because it is too thick;
the smaller cube is placed upon the bigger). Inhibitions work against using
comparatives and superlatives big is used where bigger and biggest
are meant.
The adjectives of the last list aim at comparing objects with respect to length.
This activity develops long before what mathematicians call the order relation

12

CHAPTER 1

of lengths is constituted, not to mention becoming conscious of the order

relation. The constitution of an order relation in whatever system includes at
least the operational functioning of transitivity, that is, drawing factual conclusions according to patterns like
a as long as b,
b as long as c,
thus a as long as c
and

a shorter than b
b shorter than c
thus a shorter than c,
which of course does not mean the ability of verbalising or even formalising
transitivity.
Contradictory Piaget, P. Bryant* showed that young children (from the age of four onwards)
possess an operational knowledge of transitivity. On the other hand, I reported** on third
graders who could apply the transitivity of weights in seesaw contexts but were not able
to understand a formulation of transitivity.

Little if any information on the development of the concept of length can be

drawn from traditional research. Thought on this subject is obscured by such
terms as conservation and reversibility, which are supposed to cover the

most divergent ideas, and by a faulty phenomenology.

1.15. Congruence Mappings

One of the mathematical notions that have been absorbed by conservation in
order to be mixed together with quite different ones is
in variance under a set of transformations.
As an aside I will illustrate this notion by a number of examples:

The number of elements of a set (cardinality) is invariant under one-toone mappings.

Convexity of a plane figure is invariant under affine mappings.***

Parallelism of lines is invariant under affine mappings.***

* P. Bryant, Perception and Understanding in Young Children, London, 1974, Chapter
3.
** Weeding and Sowing, p. 255.
*** A reader not acquainted with the concept of affine mapping, may read instead:

parallel projection.

AS AN EXAMPLE: LENGTH

13

The difference between the surfaces of a sphere and of a ring is invariant

under arbitrary deformations (the surface of a ring cannot be deformed
into that of a sphere).
Lengths of line segments and measures of angles of pairs of lines are invariant under congruence mappings of the plane or the space (movements
and reflections, with glide reflections included).

The length ratio of line segments and measures of angles of pairs of lines
are invariant under similarities.
The property of being a regular pentagon is invariant under similarities.
The property of being a cube with side 1 is invariant under congruence
mappings.
The property of a plane figure to represent the digit 2 is invariant under
movements (though not under reflections).

The shape of a figure is invariant under similarities.

Both the shape and size of a figure are invariant under congruence mappings.

The expression congruent is well-known; congruent figures are, as it were, the

same figure laid down in various places. In mathematics this concept is made
more precise by that of a congruence mapping, which extends to the whole
plane or space; then figures are called congruent if they can be carried into

each other by congruence mappings.

The simplest figure is the line segment. Equal rather than congruent
line segments is an older terminology. The now prevailing terminology reserves
equality to coincidence; that is, actual identity. Yet congruent line segments
are equal in a sense; that is, with respect to length. And conversely: line segments
with the same length can be carried into each other by congruence mappings.
1.16. Rigid Bodies

Line segments are mathematical abstractions. They are connected with the
former long objects via the phenomenon of the rigid body. A rigid body can
be displaced, and provided it is not badly belaboured, it remains congruent with
itself under this operation. Rigidity is the physical realisation of the property we
called in variance of shape and size under movements. The fact that in geometry
we consider by preference properties that are invariant under movements is
related to the dominance of rigid bodies in our own environment molluscs
would prefer another kind of geometry.
I am pretty sure that rigidity is experienced at an earlier stage of development
than length and that length and invariance of length are constituted from rigidity
rather than the other way round. Rigidity is a property that covers all dimensions
while length requires objects where one dimension is privileged or stressed.
However, stressing this one dimension may not lead to restricting the preserving

14

CHAPTER 1

transformations. If lengths are to be compared, the free mobility of rigid bodies

must play its part. The mobility must be fully exploited; all movements must
be allowed, not only the most conspicuous translations, but also rotations, in
order to compare long objects in all positions. The shape of a body or the
stressing of one dimension as its length may not result in restricting the mappings
under which rigidity expresses itself as invariance. Adjectives like high, low
within the complex of terms that indicate length can exert an influence to restrict
the set of transformations; high, low, stressing one direction in space, can
lead to restricting the set of transformations to those that leave invariant the

vertical direction displacements along and rotations around the vertical

a restriction that would impede the overall comparison of line segments and
long objects.
1.17. Similarities

Side by side with the congruence mappings I repeatedly mentioned the similarities. The latter play a part in interpreting visual perception. What is farther
away, looks smaller (at least at big distances); this is a feature unconsciously

taken into account by a perceiver and sometimes made conscious to himself a

curious interplay which has been studied many times. If a rigid body moves

away, its shape as understood by us remains invariant; visually conceived the

rigid bodies are invariant even under similarities, while the similarity ratio

depends on the distance between object and perceiver.

Nevertheless just this fact can contribute a great deal to the mental constitution of rigidity. The invariance suggested by the continuous behaviour of some
striking characteristics might provoke the attribution of more in variances, in
particular those of size and length.

1.1820. Flexions
1.18. The rigidity of rigid bodies has to be understood with a grain of salt.

Though its wheels and doors can turn independently, a car can globally and
under certain circumstances be considered as a rigid body. Another extreme
case is clay, which by mild force can be kneaded and deformed. In defining
rigidity all depends on what you call not badly belaboured. A liquid or
a gas can be given some other shape without using any force, but according to
the degree of rigidity more or less strong forces are needed to deform a rigid

body. More or less rigid parts can be movable with respect to each other, such
as in the case of animal bodies, while certain arrangements of the parts with
respect to each other may be privileged, such as the state of rest, which can
be congruently copied ad lib. It is that privileged state in which length measures
of animal bodies are defined. The heights of, say, two people are compared
while they are standing; we are convinced that they do not change when they
sit down, and we know that they will show anew the former relation if they

AS AN EXAMPLE: LENGTH

15

rise again. We also judge that if they sit down and the taller person looks smaller,

the difference must be ascribed to longer legs something we can reconsider

under the viewpoint of addition of lengths.
1.19. What comes about here is another principle of invariance of length, that
is to say, invariance under a kind of transformations other than planar or spatial
congruence mappings. It is transforming long objects by plying or bending
them with a negligible effort: two objects to be compared are laid side by side
or one on top of the other while certain deformations are allowed. Typical

examples of this are measuring instruments other than the ruler and the measuring stick for instance, the measuring tape, the folding or coiled pocket-rule
but a more primitive device used to measure lengths, the piece of string,

should not be forgotten. It shows marvellously two ways of comparing lengths:

in the tight state it measures a straight length, and fitted around a curvilinear
shape it measures a circumference.

As opposed to the rigid bodies considered earlier, I will call these objects
flexible the admissible deformations of these objects being called flexions.
Flexions are reversible this is an important feature. Moreover, flexible objects
possess one or more privileged states. Among the privileged states there might
be one in which the object is straightened and used as a measuring instrument:
the measuring tape, the folding pocket-rule, the coiled-rule, and again the

piece of string that can be stretched with a little force and that in this state
resists further stretching. Ones own body is of the same kind; in order to have
it measured, one jumps to ones feet (though not to ones toes). Similarly, one
measures the length of a stalk or reed or a stair-carpet: by stretching. Or of a
car antenna: by pulling it out. A sheet of paper is flexible, though there is a
well-defined state of maximal stretching. Plastically deformable substances such
as clay are again different, a long object made of clay, if carefully handled,
can be considered as flexible, though a kneading transformation is no flexion.

1.20. Where can we put the flexions mathematically? The mathematical counterparts of the rigid bodies (which may be moved without being badly belaboured)

were the geometrical figures subjected to movements in the plane or in space,

transformations that map everything congruently; in particular every line segment
whatever its length or direction might be. If our objective is measuring lengths,
this requirement is exaggerated; in order to serve for measuring, the long
objects need display this invariance in the length direction only. Only in the
length direction should the object be rigid; there is no need for rigidity in the
other dimensions. This kind of object is mathematically idealised by what is
called curves curves which are described by a moving point or appear as
boundaries of a plane figure. Of course curves which are entirely or partially

straight are also admitted: straight lines and broken lines. It is these mathematical curves that are subjected to mathematical flexions. What does this
term mean? If it refers to curves, I am concerned with one dimension only

16

CHAPTER 1

no width and no thickness and in this one dimension they shall be rigid.
The arc length, which as a measure replaces straight length, should be invariant
under flexions. Mathematically, flexions are defined as mappings of curves that
leave the arc length invariant.
But what do we mean by the arc length of a curve? The answer looks obvious:
straighten the curve while not stretching it and read the arc length on the

final straight line segment. Well, isnt it a vicious circle? What do we mean
by straighten without stretching? No stretch this just means that the arc
length must be preserved, but arc length still has to be defined. As a matter of
fact, it is curious that I prohibited stretching only, and kept silent about shrinking, but of course the mistake you can make when straightening the object,
is pulling too hard and stretching. This shows once more that the alleged clarity
of the straighten-out definition of arc length rests not on visual but on kinesthetic
intuition.
Yet another definition of arc length deserves to be considered. In order to be
compared, curves are rolled upon each other. In particular, in order to measure
the length of a curve, it is rolled upon a straight line. Rolling yes, but of course
skidding is forbidden. But what does it mean mathematically: no skidding? That
the pieces rolling along each other have the same (arc) length. This again closes
the vicious circle.
There is no escape: In order to define flexions mathematically, we must first
know what arc length is, and arc length must be defined independently with
no appeal to mechanics.
How this is to be done, I have already mentioned. First, one defines the
length of a polygon that is, a curve composed of straight pieces as the sum
of the lengths of those pieces. Given a curve, it is approximated by inscribed
polygons, that is, with their vertices on the curve. The smaller the composing
straight pieces, the better the curve is approached. In this approximation process
one pays attention to the respective lengths: as the curve is approached by the
polygons, the lengths converge to what is considered as the length of the given
curve. Not only should the total curve get an arc length by this definition,
but also each partial curve, and it is plain (though the proof requires some attention) that these lengths behave additively: if a curve is split into two partial
curves, the length of the whole equals the sum of the lengths of the parts. It is
now clear what we have to understand by a mapping that preserves the arc
length (by a flexion): not only should the total arc length be left invariant,
but also that of each part.
It is strange that an intuitive idea like invariance of arc length and straightening without stretching requires such a cumbersome procedure in order to be
explained mathematically. The reason is now obvious: when trying attempts
at explaining arc length mathematically, we are compelled to renounce our
mechanical experiences. It is particularly intriguing that physically I can compare
two flexible objects by flexion or the borders of two plane shapes by rolling the
one upon the other, before I start measuring length, whereas our mathematical

AS AN EXAMPLE: LENGTH

17

definition of flexion presupposes arc length, which includes the whole measuring
procedure and even the addition of lengths.

1.21. Rigidity and Flexibility

We have been concerned with two kinds of mappings:
congruence mappings in plane of space, and
flexions of curves.
Both are mathematically defined by the invariance of length, though the first
requirement cuts deeper than the second if the view is fixed on curves and arc
length.
The fact that congruence mappings and flexions leave length invariant is
implicit in their definition. In physics the counterpart of mathematical congruence mappings and flexions is the movement of rigid bodies and the bending
of flexible bodies, but whether in physical practice something is (at least approximately) a rigid body or a flexible body and which physical operations are

allowed if length should (at least approximately) be preserved are physical facts,
depending on experiences we have somehow acquired. This acquisition of
experience starts rather early, certainly as early as in the cradle. It is empirical
and experimental, and though this experimenting starts, as Bruner asserts, in an
enactive way, in the course of development it is supported more and more
by representative images of what is recollected or pursued (the ikonic phase),
and it becomes more and more conscious in order to be verbalised (the symbolic
phase). In the context of the phenomenon of length a phenomenological
analysis is required to state and to distinguish invariance under congruence
mappings and flexions, but anyway it is clear that the related learning process
starts in the enactive phase (with no representative images and unconsciously,
that is in the most effective way) and that pieces of it can be made conscious
in the learning process.
Bastiaan (3; 9) finds a glass marble on the foot path: If I push hard, it would roll into the
street. It does happen. The marble rolls under the tyre of a car parked at the curb. Bastiaan

cannot reach it. I show him a little stick. By sight he judges: It is not hard enough. It
is a soft stick, but nevertheless he succeeds.

This example does not concern using rigid or flexible objects to compare lengths.
What matters here is experiences with mechanical properties of things. At a

certain moment in his development a child judges at sight whether something

is hard enough to be applied as a lever to exert a certain power (ikonic phase),
and he even finds words hard enough to express this fact (symbolic phase).
I do not have the slightest idea how this complex of mechanical properties

becomes mentally constituted; an able physicist, observing children, could discover a lot of things in this field. There is one conjecture which I dare pronounce:
that rigidity precedes flexibility. The environment strongly suggests the rigid

18

CHAPTER 1

body as a model. Surprising experiments show that under conditions of incomplete information about kinematical phenomena there is a strong tendency
to interpret them as movements of rigid bodies.
As a consequence I think that length is first constituted in the invariance
context of congruence mappings that is, connected to rigid bodies and only
at a later stage gets into that of flexions that is, of flexible objects. This can
happen if the child sees lengths compared or even measured by flexible instruments fitting (is the sleeve long enough?) and measuring with a tape.
In any case it is crucial to pay attention to the double invariance context
of length.
1.22. Make and Break

I hesitated unjustly as it will shortly appear as to whether I should augment

the two kinds of transformations that show invariance of length (that is, congruence transformations and flexions) with a third, which I would call
breakmake transformations:
a long object is broken into pieces and remade.
The long object may be a stick that is factually broken, or a string that is
cut, or a train of blocks that is split into two or more partial trains. In the first
two examples remaking will not yield a complete restoration of length even if
carried out carefully, with some loss in the second case if the partial strings are

tied together. In the third case the restoration can be complete though it need
not be: the parts can be put together in another order, and this can even be
visible if the particular blocks are distinguished by length, colour, or other
characteristics.
It is a meaningful and non-trivial statement that under breakmake transformations length is invariant. It is meaningful if it is the original and final state
that are compared, disregarding the intermediate ones. Indeed, how should we
formulate the question if the intermediate states are to be admitted? Do they
remain as long together? If together means adding lengths, this question
is premature at the stage of simply comparing lengths, and if together means
taken together the question aims at comparing the initial with the now
also mental final state, which is no news.
Whenever the breakmake transformation reproduces the initial state, the
question are they of the same length? is trivial. Or rather, the answer reveals
only whether the child that was questioned has remembered the initial state and
is able to compare an actual and a mentally realised state with each other. If
the final state is not wholly identical with the initial one, the answer also reveals
whether the child knows which characteristic matters if length is meant. These
two abilities will be reconsidered later.
Insight into length invariance under breakmake transformations can be split
into two components:

AS AN EXAMPLE: LENGTH

19

first, that under breaking (partitioning) and making (composing) long

objects are transformed into long objects, and
second, that in composing long objects length is not influenced by the
order of the composing parts.
Actually, these two insights form the basis for measuring lengths and will reappear in that context. If the second insight is to be placed into the context of
invariance of length with respect to certain transformations on long objects,
instead of breakmake transformations we could better use the term

permutation of composing parts.

I can now explain the hesitation I felt before writing this section. Break
make transformations or permutations of composing parts as a third kind of

transformations look logically and phenomenologically superfluous. Within

a phenomenology of magnitudes, and particularly length, as sketched in the
beginning, the breakmake transformations (permutations of composing parts)
and the associated invariance of length can be derived from the congruence
mappings, flexions and their invariance properties. But this derivability is a
consequence of coupling the comparison of lengths with measuring, which is
genetically and didactically premature. It is true that composing long objects
occurs in that phenomenology as a special operation, indicated by but the
context in which it occurs is length rather than comparing length; namely

the formula

occurs there as a logical rather than geometrical and mechanical operation,

appears as something that is uniquely determined by x and y, whereas for
breakmake transformations it is essential that x and y can be put together
in various ways and however composed, yield objects of the same length.
1.2324. Distance
1.23. Up to now in our didactical phenomenological analysis we have considered
length as a function of concrete objects (possibly replaced by their mental
images). This, however, does not cover all cases of length. Length as distance
between A and B answers the question how far is B from A? In a purely

formal sense how far? is quite another interrogative than how long? In
how far ... ? two points occur as variables, whereas in how long is this
object?, the object is the only variable. Length is a function of whole objects,
whereas distance is a function of two points here and there. We are so
accustomed to the procedure which connects both of them that we can hardly
imagine the early stage where we must have acquired it by a learning process
and ask ourselves whether this connection is as obvious for children as it is for
us.

20

CHAPTER 1

If how far? is to be reduced to how long?, a long object must be

placed between A and B, between here and there. So, if A and B are railway
stations or stopping places along a highway, the rail connection or the stretch
of highway may be considered as concrete long objects whose distance is
asked for. In general, if there exists a concrete path between A and B, their
distance is the length of the path; if there are more such paths, it should be
stipulated which one is meant. But how far is it from the front room of my
ground floor to the rear room on the first floor? From here to across the canal,
if no bridge is visible? From here to the sky? Only from the context can it
be understood what is meant. In the context of geometry, mechanics, and
optics the distance is measured along a straight line; in the context of spherical
trigonometry and in the context of (surface of air) navigation, along arcs of great
circles, geodesics or shortest paths as determined by straightened strings on
curved surfaces. Of course, with this remark I do not mean spherical trigonometry
or navigation should have been studied or exercised in order to decide that
lengths should be measured along geodesics; contexts like this develop long
before they are made conscious. The value of rectilinearity is suggested to the
young child, enactively, if he is called to come straight in your open arms, the
ikonically by all the horizontal and vertical straight lines in his environment,
and symbolically by straight lines in schemas and by the word straight line.
The part played by rectilinearity in the constitution of length remains unconscious until it is explicitly discussed. Straightening flexible objects if lengths
are to be compared may still be an automatic act for instance, automatic
imitation and there might be children who as automatically put between two
unrelated points a mental long object, an imagined ruler, or a string in order
to interpret distance as length. Well-known experiments where children get
disoriented as soon as a screen is placed between the two points may prove
how important this act of inserting a long object can be for reducing how
far? to how long?. But whatever these experiments mean, if some judgment
about the distance of unrelated points must be motivated, one cannot but
make explicit the necessity of rectilinear connections. From this moment
onwards the significance of rectilinearity for the concept of length becomes
more and more conscious another connection between length and rectilinearity
will be indicated later on.
1.24. How does a child learn what matters if lengths are to be compared? Sets
of objects of the same kind but of different length may play an important part:
big and little spoons (and equal ones), long and short trains (and equal ones),

high and low trees (and trees of equal height). The objects are compared at sight
if they are lying parallel and side by side; in order to be compared they are
brought into such a position, physically or mentally, as rigid bodies, by congruence mappings. This requires comparing physical with mental objects, and
mental ones with each other. Memory for length initially functions in a rather
rough way, it seems. Remembering length during long periods remains a difficult

AS AN E X A M P L E : L E N G T H

21

task. As for myself, I am often surprised that relations of length differ greatly
from what I remember they should be. Comparing objects side by side gains
precision in the course of development: the ruler is laid close to the line to be
measured, while observing the prescription to aim perpendicularly to the line.
The connection between length and distance is stressed, and the weight is
shifted to distance if one of the objects to be compared, or both of them,
bear marks by which the ends of the objects to be compared can be indicated.
Comparing can be done indirectly, using the transitivity of the order relation;
for instance, by taking distances between fingers of one hand, between two
hands, between the points of a pair of compasses, or between two extant or
intentionally placed marks on a long object, and carrying them from one place
to another. With all these methods length as a function of long objects is replaced
by distance as a function of a pair of points. It already starts with showing
that big or that small with fingers or hands, although in its exaggerated
appearance this gesture is more an emotional expression of awfully big or
miserably small than a true means to compare lengths. More refined methods
of comparing lengths are based on geometry and will be dealt with in that
context.
1.2 5. Conservation and Reversibility
Before extending the analysis of measuring lengths I tackle the question already
touched in Sections 1.12 and 1.15: how psychologists interested in the development of mathematical concepts deal with such concepts, in particular length.
The investigations, started by Piaget, show the following pattern. The general
problem is to acquire knowledge about the genesis of such fundamental concepts
as number, length, area, shape, mass, weight, and volume. Subjects are shown
groups of objects which agree with respect to one or more of these magnitudes
(the same number of chips in a row, reeds of the same length, and so on) and
are asked to state that they agree with respect to the characteristic A (number,
length, or suchlike). Then one of the objects of the group is subjected to a
transformation that according to adult insight does not change the characteristic
A while other characteristics may be changed (for instance, changing the mutual
distances of the chips in the row or bending the reed). After this operation
the subject is asked whether the characteristic A has remained unchanged;
if this is affirmed, one speaks of conservation, and the subject is classified as a
conserver. Psychologists are reasonably unanimous about the average age of
conservation of the various characteristics, whereas people who have some
didactical experience with children usually judge these ages absurdly high.
The large percentages of non-conservers in psychological experiments are achieved
by a particular strategy: The transformation that should be ascertained to
conserve A is intentionally chosen so that it changes another characteristic
B so drastically that the attention is diverted to B (for instance, if A is cardinal
number or mass, a striking change of length, or if A is length, a striking change

22

CHAPTER 1

of position or shape). What is actually being investigated is whether the subject

is able to separate these characteristics sharply from each other and how strongly
he can resist attempts at misleading him. Built-in misleading is in general characteristic of the psychological, as opposite to the didactical, approach.
By no means should the question be rejected as to the stage of development
at which children master invariances of certain magnitudes. On the contrary, it
is a merit of Piagets to have been the first to have formulated such problems.
The problem, however, is obscured by the use of such terms as conservation;
often the researchers themselves have no clear idea of the kind of transformations
with respect to which the so-called conservation should be established. For
each experiment designed to take in young children, one can contrive a more
sophisticated version to embarrass adults. For instance, show a person two
congruent paper clips and ask him whether they are equally long; the question
is of course affirmatively answered. Then unfold one of them, straighten it out,
and repeat the question. Whatever he answers can be wrong. It depends on what
the experimenter meant. An adult subject would react to the question by asking,
What do you mean? (In our terminology, length in variance under congruence
mappings? Or under flexions?) Young children in the laboratory are not likely
to ask counter questions. The fact that they do not ask proves that they are
intimidated (in the terminology of the psychologist, put at their ease) their
critical behaviour being eliminated by situational means.
For a good experimental design it is indispensable that experimenter and
subject have a clear idea of the kind of transformations with respect to which
invariance is to be established. Perhaps psychologists would answer that then the
fun goes out of it, as the chance of getting wrong answers would be minimised.
So much the better, I would say. Such a result would better agree with the
opinions of childrens capacities held by didacticians.
Of course this does not mean that all problems are disposed of. I could
enumerate a lot of developmental problems that from the viewpoint of a sound
phenomenology are interesting enough. For instance I would like to know
whether constituting rigidity mentally precedes length, whether length invariance
under congruence mapping and length invariance under flexions help or impede
each other, what role is played by similarities in the mental constitution of
length, and how the equivalence of long and far is acquired. So there are
many more questions I would like to have answered. The most urgent question,
I think, is about the significance of the breakmake transformations for socalled conservation (not only of length). If I may trust my own unsystematic
experience, I would consider them as crucial. Yet in order to answer such
questions, a quite different design of experiments is required than that of
snapshots, registering which percentage of subjects at a certain age do conserve.
Also required is a more positive mentality than that of tricking children into

making mistakes.
Another vague term that is often used in that kind of research is reversibility. Originally it was related to answers given by subjects when they motivate

AS AN E X A M P L E : L E N G T H

23

pronouncements on conservation. For instance, one of two strings of equal

length is made crooked while the other remains straight; the subject is asked
whether they are still equally long. If it is affirmed, the subject is asked to give
reasons. If he answers If straightened out, it is again the same, he shows
reversibility; that is, the capacity to mentally reverse the transformation,
which is considered a good argument for equality of length. Of course, it is no

argument at all, and though it is interpreted by the experimenter as such, it was

probably not meant that way by the subject. From the equality of initial and
final states nothing can be derived about intermediate ones. If the subject had
said, They are equal because I got the one from the other by mere crooking,
the answer would have been as good as, or even more to the point than, the
argument of reversal. The subject, however, would not have been counted
among the true conservers, because he lacked reversibility.
This reversibility as a proof for conservation is the original meaning,
but subsequently it has been used in many other and mutually unrelated senses.
There are, however, also researchers who reject the reversibility argument. They
postulate standard answers that have to be given in order to establish conservation. Then the question why is this as long as that? must not be answered by
a material argument but by a formal one, if the subject is to be classified as a
conserver; he should answer something like because they have the same length.
To the question why do they have the same content?, it must be because
they include equal parts of space. It goes without saying that such investigators
are even farther away from meaningful mathematics.
The lack of insight into the difficulties with the equivalence between long
and far has already been mentioned. Often they are increased by a stress on
intentionally misleading connecting paths a pattern in the plane that suggests
a system of paths or two points on the rim of a round table that invite marching
along the edge where the experimenter had, of course, meant straight paths.
These details may suffice. I would certainly not judge that all the investigations I have in view are worthless, but many of them suffer from wrongly placed
stresses. The method of snapshots need not be rejected but in order to be
applied it requires a background theory or at least ideas about the intermediate development. Such theories do exist, but they are so vague and general
that anything can be fitted to them and they do not provide criteria for attributing relevance to certain questions or complexes of questions.
What is lacking here can be made clear parabolically. Let us assume somebody
is investigating the development of flora during the year by snapshots. On trees
and brushes he notices various kinds of buds at various stages. In the next
snapshots he identifies leaves and petals at the same places. Later on the former
have remained whereas the latter have been replaced by fruit. Then the fruit,
and finally the leaves too, have disappeared. He has not paid attention to
stamens, pistils and insects and does not know where the fruit and leaves went.
Perhaps he does not even know that the leaves and flowers were locked up in
the buds. His phenomenology was utterly fragmentary, he did not know what

24

CHAPTER 1

he had to look for, and there is a good chance that he will wrongly interpret

what he has seen. Perhaps terms like growing, blooming, bearing fruit are lacking
in his vocabulary, or they mean states rather than processes. Ideas about development would have given him a greater chance of noticing essentials.
1.2629. MEASURING LENGTHS

1.26. Yardsticks

Measuring length requires instruments measuring sticks or rules. At first the

measuring instrument will be smaller than the thing to be measured. Remarks
to the contrary in the psychological literature rest on misapprehensions about
measuring, or on artificial experiments.
The first yardstick I see used by children is the step. For a long time they
do not care whether all steps are equally long. Almost always they count one
step too many (the zero step as one). From the beginning it is clear that fewer
steps mean a shorter interval, though it is not as clear that composition of

intervals goes along with addition of numbers of steps. At about the same time
as measuring distances by steps, or somewhat earlier, one notices the activity of
jumping over a certain number of pieces in patterns of tiles in order to see how
far one can jump. I do not claim that this is really a measuring activity, though

this kind of jumping may influence measuring by steps.

Bastiaan (4; 10) spontaneously measured the width of a path by steps. This is six further.
I show him I can do it in one step. He does the same with two steps. He continues measuring
by pacing.
Bastiaan (6; 5) has made a large construction of roads, bridges, walls and tunnels in a
sandpit. In order to make a drawing of the construction he measures distances with his
two forefingers parallel at a fixed distance (about a decimeter), proceeding with the left
forefinger in the hole made by the right one.
Bastiaan (almost 7; 6) measures distances with a span between thumb and little finger
which he knows is one decimeter.

Measuring with a measuring instrument means laying down the instrument

congruently a number of times until the length to be measured is exhausted.

If the object to be measured is a distance between two points, rectilinearity

of continuation must be practised as the measuring instrument is repeatedly laid

down. It is surprising that even 12 years olds may neglect this. If the straight
line between the two points is blocked, the path is partially replaced by a
parallel one. It is a remarkable fact that usually parallelism is better observed
than is the rectilinearity of the continuation in the non-blocked case. Indeed,
the latter is more difficult. To do this reasonably, one has to develop a certain
technique, which requires more geometrical insight than unfortunately is
being taught in the primary school.
There is a rich variety of yardsticks. Most of the traditional length units are
taken from the human body: inch (which means thumb), finger, palm, foot,

AS AN EXAMPLE: LENGTH

25

short and long ell, yard, step, double step, fathom; for larger distances the
stadium (= 100 fathoms = 600 feet), the Roman mile (= 1000 double steps),
an hours walk. The so-called metric measures are related by powers of 10:
metre, kilometre, centimetre, millimetre, micrometre, picometre. At variance
with them: light year, parsec.
1.27. Change of Yardstick
If the object to be measured is not exhausted by applying the yardstick congruently a number of say n times, the problem arises of what to do with
the remainder. In many cases one will resign oneself to the fact that a bit is left
or is lacking, which means that the object is a bit longer or a bit shorter than n

times the unit. Likewise the case where the remainder looks to be about half,
one-third, or two-thirds the unit is not problematic. For greater precision a more
systematic procedure is required. Two systems are familiar: common and decimal
fractions. A less usual variation is binary fractions (or fractions with another
base). A most natural system, now obsolete because of its complexity, is continued fractions, as I have explained elsewhere*. If is the measuring unit and
the object to be measured,
then the remainder

is used as a new unit,

and so one goes on, expecting that eventually the division will terminate, that is
Then is a common measure of
will find, say

and

and by reckoning backwards, one

which implies

It is an advantage of this procedure that it involves a systematic search for a

a denominator, provided
is truly a rational multiple of
that is, if the
procedure indeed stops. But this need not happen. Then the procedure has
to be stopped at a certain stage, the remainder is neglected, and the length
of is expressed approximately in terms of
With the methods of decimal fractions one is saved the trouble of finding
* Mathematics as an Educational Task, p. 203.

26

CHAPTER 1

a suitable denominator. The measuring unit is again and again divided into ten
equal parts (even if such a partitioning is not yet marked on the measuring
instrument), and one has only to see how often the subdivided unit goes into
the remainder. It is a disadvantage of the decimal method that even simple
fractional lengths such as of the unit can only be indicated approximately.
Length is one of the concepts by which common and decimal fractions can
operationally be introduced. This subject will be resumed in the chapter on
fractions.

1.2829. Measuring Length at an Early Stage

1.28. Terms that should occur early in measuring length are double, three
times, half, and a third. It struck me that 56 year olds who reasonably
understood length did not know these terms, or at least, not as related to length;
the dominance of the adjective big seems to block applying double and
half to the linear dimension.
Bastiaan (5; 3), at a certain moment during a straight walk at the other side of our canal
between two bridges at a large distance from each other, does not understand the question
Are we half-way?, but later spontaneously indicates the point where the middle starts
(that is, the second half).

Terms like half full and so on (of a glass) function earlier and better.

Additivity of length is still a problem at this age. A long object is paced off
anew after it has been lengthened by a second object. It is not noticed that the
second pacing gives another length for the first piece.
One should realise that these are not trivial things knowing
how lengths are composed,
that the results are again lengths,
that pieces of lengths are again lengths,
that the length measure of a part is smaller than that of the whole, and
that length measures behave additively under composing.
1.29. The length of flexible objects is measured after straightening. The circumference of curved figures is measured by means of a flexible object a string

laid along side. It can also be done by rolling the curve upon a straight line.
It is not at all trivial that this yields the same result. The length arising from
rolling a circle is grossly underestimated by children, and even by adults.
Conversely, rolling a wheel can be used to measure linear distances (expressed
by the number of revolutions of a bicycle wheel or a measuring wheel).
Geometrical knowledge can lead to more sophisticated methods of measuring
distances. Some of them are possible at an early age. We will reconsider this
question later.
Reading and designing maps with distance data does not necessarily presuppose acquaintance with ratio.

AS AN EXAMPLE: LENGTH

27

The relation between distances and the times needed to cover them does not
necessarily presuppose an acquaintance with velocity.

Climbing stairs can be put into relation with distance.

The distances in a network of streets are accessible early.

CHAPTER 2

THE METHOD

2.1. Aspects ofPhenomenology

I started with an example to be used as a subject matter which I can appeal to
when I explain my method. I chose length because it is both a rich and
relatively easy subject.
First of all, what of the terms phenomenology and didactical phenomenology? Of course I do not mean phenomenology in the sense that might
be extracted from the works of Hegel, Husserl, and Heidegger*. Though the
clearest interpretation I can imagine is that by means of the example of chapter
I, which is to be continued in the following chapters, nevertheless it is worthwhile trying something like a definition.
I start with the antithesis if it really is an antithesis between nooumenon
(thought object) and phainomenon. The mathematical objects are nooumena,
but a piece of mathematics can be experienced as a phainomenon; numbers
are nooumena, but working with numbers can be a phainomenon.
Mathematical concepts, structures, and ideas serve to organise phenomena
phenomena from the concrete world as well as from mathematics and in the
past I have illustrated this by many examples**. By means of geometrical
figures like triangle, parallelogram, rhombus, or square, one succeeds in organising the world of contour phenomena; numbers organise the phenomenon of
quantity. On a higher level the phenomenon of geometrical figure is organised
by means of geometrical constructions and proofs, the phenomenon number
is organised by means of the decimal system. So it goes in mathematics up to
the highest levels: continuing abstraction brings similar looking mathematical
phenomena under one concept group, field, topological space, deduction,
induction, and so on.
Phenomenology of a mathematical concept, a mathematical structure, or a
mathematical idea means, in my terminology, describing this nooumenon in
its relation to the phainomena of which it is the means of organising, indicating
which phenomena it is created to organise, and to which it can be extended,
how it acts upon these phenomena as a means of organising, and with what
power over these phenomena it endows us. If in this relation of nooumenon
and phainomenon I stress the didactical element, that is, if I pay attention to
how the relation is acquired in a learningteaching process, I speak of didactical
* Is it by accident that with Habermas included the names of the most pretentious
producers of unintelligible talk in the German philosophy start with an H?
** Mathematics as an Educational Task, in particular, Chapters II and XVII.

28

THE METHOD

29

phenomenology of this nooumenon. If I would replace learningteaching

process by cognitive growth, it would be genetic phenomenology and if
is ... in a learningteaching process is replaced by was ... in history,

it is historical phenomenology. I am always concerned with phenomenology

of mathematical nooumena, although the terminology could be extended to
other kinds of nooumena.

2.2. The Part Played by Examples

The piece of phenomenology with which Chapter I began was clearly an a
posteriori constructed relation between the mathematical concept of length and
the world of long objects structured by an operation of composing, Length was
interpreted as a function on this world. I did not analyse how I arrived at this

function. Although this was indispensable, I omitted it because I had to tackle

this question in the didactical phenomenological section and I wanted to avoid
duplication. But as a consequence the didactical phenomenological section contains pieces of pure phenomenology, such as Section 1.15 about the congruence
mappings and Sections 1.1819 about the flexions. Likewise in the sequel I will
not clearly separate phenomenology and didactical phenomenology from each

other. As promised in the preface I would not sacrifice readability to systematics.

Where did I look for the material required for my didactical phenomenology
of mathematical structures? I could hardly lean on the work of others. I have
profited from my knowledge of mathematics, its applications, and its history.
I know how mathematical ideas have come or could have come into being. From
an analysis of textbooks I know how didacticians judge that they can support
the development of such ideas in the minds of learners. Finally, by observing
learning processes I have succeeded in understanding a bit about the actual
processes of the constitution of mathematical structures and the attainment of

mathematical concepts. A bit this does not promise much, and with regard to
quantity it is not much, indeed, that I can offer. I have already reported a few
examples of such observations, and I will continue in the same way. I do not
pretend that at this or that age this or that idea is acquired in this or that way.
The examples are rather to show that learning processes are required for things
which we would not expect would need such processes. In the first chapter I

showed a child suddenly confronted with the necessity to differentiate big

according to various dimensions, a child placing far into the context of long
and learning about the connection between half and middle. I am going

to add another story, which happened a few hours after the event where half
and middle were tied to each other:
Bastiaans (5; 3) sister (3; 3) breaks foam plastic plates into little pieces, which she calls
food. He joins her, takes a rectangular piece, breaks it in about two halves, lays the two

halves on each other, breaks them together and repeats the same with a three-layered
combination the fourth piece was already small enough.

30

CHAPTER 2

I do not know where I should place this observation, whether I should classify
it as mathematics, say geometry, or whether it belongs to general cognitive
behaviour. I report this observation because I think it is one of the most im-

portant I ever made because it taught me a lesson on observing. I do not know

whether the age of 5; 3 is an early or a late date for this kind of economic
breaking; I do not know either whether Bastiaan imitated or adapted something

he had observed before. I know only one thing for sure: that what he did is
important and worth being learned. For myself it is fresh material to witness
that in no way do we realise all the things children must learn. If I look at what

people contrive to teach children, I feel inclined to call out to them: do not
exert yourself, simply look, it is at your hand.
Why do people not look for such simple things, which are so worth being

learned? Because one half of them do not bother about what they think are
silly things, whereas those who do bother are afraid to look silly themselves if
they show it. Weeding and Sowing is full of such simple stories. I told them

in lectures. I do not care whether a large part of the audience interprets my

reporting as senility, provided that by my example a small part of the audience
is encouraged to follow suit this, indeed, requires courage.

2.3. Enactive, Ikonic, Symbolic

Above I used Bruners triad enactive, ikonic, symbolic. Bruner* suggested

three ways of transforming experiences into a model of the world: the enactive,

the ikonic, and the symbolic representation. Corresponding to the dominance of

one of these, he distinguishes phases of cognitive growth.
Bruners schema can be useful. It has been taken over by others, and its
domain of application has been extended, in particular towards the attainment

of concepts in learning processes, where similar phases are distinguished. Later

I will explain my objections to the idea of concept attainment as such, although

I would not oppose the extension of Bruners triad to concept attainment.

As a matter of fact, in Bruners work there is an example that shows how the
three ways of representation can be extended to concept attainment: enactively
the clover leaf knot is a thing that is knotted, ikonically it is a picture to be
looked at, and symbolically it is something represented by the word knot,
whether or not it is accompanied by a more or less stringent definition.
There is a well-known pleasantry: ask people what spiral stairs are. All
react the same way: they make their forefinger mount imaginary spiral stairs.
Of course, if need be, they would be able to draw them. Does this mean that
they are in the enactive or in the ikonic phase? Of course not. For the concept
in question they possess a symbol, the words spiral stairs, though if a definition
is to be produced, one would have more or less difficulty in passing from the

enactive or ikonic to the symbolic representation.

* Studies in Cognitive Growth (Edited by J. S. Bruner), Toward a Theory of Instruction,
1966, pp. 1011.

THE METHOD

31

Consider the number concept three and the geometrical concept straight.
Before the child masters these words, he can be familiar with what they mean:
clapping his hands thrice and running straight to a goal if it is suggested to him
(the enactive phase); sorting out cards with three objects or straight lines pictured
on them (the ikonic phase). Mastering the word three (or straight) means he is
in the symbolic phase, since three as a word is a symbol for the concept
three (or straight is for straight). But likewise the three dots on dice can be
a symbol; for instance, in playing the game of goose. A child that counts intelligently is in the symbolic phase even if this counting is accompanied by moving
counters on the abacus. Adding on the abacus is enactive only for a moment.

After the first experience it has become symbolic, though the symbolism differs
from that of the written digits. The Roman numerals are as symbolic as the
Arabic ones. Notches and tallies to indicate numbers belonged to the symbolic
phase, even before people invented numerals they are as symbolic as Roman
and Arabic numerals. The cashier in the supermarket who prints amounts of
money is neither enactively nor ikonically busy. A little child who claps his
hands in joyfulness expresses his feelings symbolically even if he cannot yet
pronounce the word joy. As early as kindergarten, children accept a drawing of a
dance position where dancers are represented by strokes rather than manikins.
If the doors of the mens and ladies rooms are distinguished by plates of figures
in trousers and skirts it does not mean that the decorator imagined the users to
be in the ikonic phase; he did so because this difference is differently symbolised
in the hundreds of languages that mankind speaks and writes moreover the
plates themselves are already symbols.
With these examples I intend to say that in learningteaching situations,
which are our main interest, Bruners triad does not yield much. Bruners
domain of application is the psychology of the very young child, and in this
period the phases can meaningfully be filled out.
2.45. Concept Attainment and the Constitution of Mental Objects
2.4. I would like to stress another idea, already stressed in my earlier publications. Let me start with a semantic analysis of the term concept. If I discuss,
say, the number concept of Euclid, Frege, or Bourbaki, I set out to understand what these authors had in mind when they used the word number. If I
investigate the number concept of a tribe of Papuans, I try to find out what
the members of this tribe know about and do with numbers; for instance, how
far they can count.
It seems to me that this double meaning of concept is of German origin.
The German word for concept is Begriff, which etymologically is a translation of
Latin conceptus as well as comprehensio and which for this reason can
mean both concept and (sympathetic) understanding. Zahlbegriff
can thus mean two things, number concept and understanding of number;
Raumbegriff, concept of space and geometrical insight; Kunstbegriff,
concept of art and artistic competence.

32

CHAPTER 2

Actually, in other languages too concept is derived from a word that means
understanding (English, to conceive; French, concevoir) which, however, does
not have the misleading force that the German begreifen has. I cannot say
whether it has been the influence of German philosophy in particular, philosophy of mathematics that created the double meaning of number concept,
of space concept, and in their train as it were, of group concept, field concept,
set concept, and so on. At any rate the confusion has been operational for
a long time and has been greatly reinforced by the New Math and by a rationalistic* philosophy of teaching mathematics (and other subjects) which
in no way is justified by any phenomenology. It is the philosophy and didactics
of concept attainment, which, of old standing and renown, has gained new
weight and authority in our century thanks to new formulations. In the socratic
method as exercised by Socrates himself, the sharp edges of concept attainment had been polished, because in his view attainment was a re-attainment,
recalling lost concepts. But in general practice the double meaning of concept
has been operational for a long time. Various systems of structural learning
have only added a theoretical basis and sharp formulations. In order to have
some X conceived, one teaches, or tries to teach, the concept of X. In order
to have numbers, groups, linear spaces, relations conceived, one instills the
concepts of number, group, linear space, relation, or rather one tries to. It
is quite obvious, indeed, that at the target ages where this is tried, it is not
feasible. For this reason, then, one tries to materialise the bare concepts (in
an embodiment). These concretisations, however, are usually false; they
are much too rough to reflect the essentials of the concepts that are to be
embodied, even if by a variety of embodiments one wishes to account for more
than one facet. Their level is too low, far below that of the target concept.
Didactically, it means the cart before the horse: teaching abstractions by concretising them.
What a didactical phenomenology can do is to prepare the converse approach:
starting from those phenomena that beg to be organised and from that starting
point teaching the learner to manipulate these means of organising. Didactical
phenomenology is to be called in to develop plans to realise such an approach.
In the didactical phenomenology of length, number, and so on, the phenomena
organised by length, number, and so on, are displayed as broadly as possible.
In order to teach groups, rather than starting from the group concept and
looking around for material that concretises this concept, one shall look first
for phenomena that might compel the learner to constitute the mental object
that is being mathematised by the group concept. If at a given age such phenomena are not available, one gives up the useless attempts to instill the
group concept.

For this converse approach I have avoided the term concept attainment
* In the 18th century sense of a priori concepts epistemology.

THE METHOD

33

intentionally. Instead I speak of the constitution of mental objects,* which

in my view precedes concept attainment and which can be highly effective even
if it is not followed by concept attainment. With respect to geometrically
realisable mental objects (square, sphere, parallels) it is obvious that the constitution of the mental object does not depend at all on that of the corresponding

concept, but this is equally true for those that are not (or less easily) geometrically realisable (number, induction, deduction). The reader of this didactical
phenomenology should keep in mind that we view the nooumena primarily
as mental objects and only secondarily as concepts, and that it is the material
for the constitution of mental objects that will be displayed. The fact that
manipulating mental objects precedes making concepts explicit seems to me
more important than the division of representations into enactive, ikonic, and
symbolic. In each particular case one should try to establish criteria that ought
to be fulfilled if an object is to be considered as mentally constituted. As to
length such conditions might be
integrating and mutually differentiating the adjectives that indicate length,
with long, short,
comparing lengths by congruence mappings and flexions,
measuring lengths by multiples and simple fractions of a measuring unit,
applying order and additivity of measuring results, and
applying the transitivity of comparing lengths.

2.5. In opposition to concept attainment by concrete embodiments I have

placed the constitution of mental objects based on phenomenology. In the first
approach the concretisations have a transitory significance. Cake dividing may
be forgotten as soon as the learner masters the fractions algorithmically. In
contradistinction to this approach, the material that serves to mentally constitute
fractions has a lasting and definitive value. First concepts and applications
afterwards as it happens in the approach of concept attainment is a strategy
that is virtually inverted in the approach by constitution of mental objects.

* Fischbein calls them intuitions, a word I try to avoid because it can mean inner vision

as well as illuminations.

CHAPTER 3

SETS

3.19. Sets in Advanced Mathematics

3.1. Sets as an Aim in Itself
Sets as an aim in itself is a restricted, almost sterile, and hardly popular domain
of mathematical research. Terms characterising it are cardinality, continuum
hypothesis, well-ordering, transfinite ordinals, alephs. Up to the middle of the
present century there was a certain need in algebra and analysis for transfinite
ordinals, which subsequently were eliminated by means of the so-called Zorns
lemma.
For a while, general set theory was a subject of profound axiomatic research;
in particular, the continuum hypothesis was in the focus of attention.

3.2. Sets as Substrata of Structures

Wherever sets in advanced mathematics are not an aim in themselves, they

fulfill various tasks. For instance, they serve as substrata for structures a
metric space, a group, a category is a set with a number of properties. (Sometimes
the term set is verbally replaced by class in order to avoid certain paradoxes

of set theory.) Structuring the substratum set can happen in various ways, for
instance:
a set becomes a metric space by putting a distance function upon its pairs
of elements,
a set becomes a group by prescribing a certain operation between its
elements, and
a set becomes a category by imposing certain mappings between the
member sets.
If sets serve as substrata for certain structures, they are in general not subjected
to drastic set theory operations. Subsets are formed, mainly to introduce substructures; substrata of structures are mapped on each other to define mappings
of the imposed structures; set theory products are formed to get structure
products of the imposed structures; and the substratum set is partitioned, for
instance, as a set of equivalence classes, in order to derive new structures.

3.3. Sets as a Linguistic Tool

In an even weaker way, sets occur virtually as only a linguistic tool where some
34

SETS

35

predicates are replaced by their extension. For instance, to avoid repeating the
clumsy predicate n times differentiable, one introduces the set
of n times
differentiable functions and expresses f is n times differentiable by

A more recent acquisition is the conventional set symbols N, Z, Q, R, C, which

replace the predicates ... is a natural, integral, rational, real, complex number,
respectively.
A certain operation on sets (though important as such namely, forming
the power set
of A (that is, the set of subsets of A) functions purely as
a linguistic tool if the fact
(C is a subset of A) is expressed by
or the fact that is a set of subsets of A is expressed by
or even
by
Sometimes the empty set occurs as a purely linguistic tool;
for instance, if the non-existence of solutions of a certain equation is expressed
by the emptiness of the set of solutions. Interpreting a mapping from A to B
or a relation from A to B as a subset of the product
is also a purely
linguistic use of a set theory operation (the set theory product). A more essential
use of set theory language will be dealt with later.
3.4. Sets in Topology

Operating genuinely and explicitly with sets happened first in topological

contexts. Not with the substrata of limits and convergence that are sequences
and series rather than sets; nor even in the case of the maximum of a function
was the set of its values originally made explicit. This changed when the need
was felt to use and to define the upper bound of a function (or functional) that
has no maximum a phenomenon important in history because of Dirichlets
principle. In this case the formulation is indeed easier if the set of values is made
explicit. A similar case: in order to prove that a continuous function vanishes
somewhere in the interval between a negative and a positive value, one conceives
the set of points where the function is not positive, and then takes the upper
bound of this set. Upper bounds, lower bounds, upper limits, lower limits, and
accumulation points are indeed an opportunity to explicitly introduce sets and
to operate with them. In real and complex function theory the need is felt to
consider open and closed sets, interior points and boundaries of sets the
drawings illustrating such concepts reinforce the set theory context. In set
theory topology the adjective expressively indicates this context, although in
algebraic topology it is no less influential.
3.5. Measures

Area, volume, and measure are functions on sets which, as it happens, can call
for explicit sets on which they are defined. As long as it is simple figures to
which an area or volume is to be ascribed, there is no urgent need for making the

underlying set explicit; even the part of the plane delimited by the horizontal
axis, a function graph, and two ordinates, whose area is expressed by an integral,

36

CHAPTER 3

need not be explicitly described as a set. The need really arises as soon as areas
and volumes are to be attached to more or less arbitrary sets, that is in measure
theory. Here algebra of sets becomes operational: unions and intersections (not
only finite ones) and complements. Measures, that is, functions on sets with
certain additivity properties, are an important organising tool for many mathematical phenomena. Probability is one of them.
3.67. Solution Sets for Conditions
3.6. Traditional geometry knew an explicit procedure for introducing sets as

solution sets for conditions. Instead of sets one called them loci the circle
with centre M and radius r is the locus of the points at distance r from M.

Among the set theory operations the intersection played an important part,
arising from the combination of two or more conditions.
3.7. Considering within some structure the sets of elements maintaining a
certain relation with a given element is an important principle in algebra, too.
The multiples of 4 in Z form a subset: the set of numbers divisible by 4. The

intersection of multiples of 4 and multiples of 6 is the set of multiples of 12

a connection between intersection and least common multiple.
In algebraic structures one often focuses on subsets closed with respect to
some operations. A set theory pattern, which is typical, may be illustrated by an
example taken from group theory: Let G be a group and A a subset of G; one
asks for the subgroup of G generated by A, that is, the group obtained constructively by starting from A and applying over and over the operations of product

and inverse. However, the subgroup of G generated by A can also be obtained

in one blow: defining it as the smallest subgroup of G containing A, or in still
another way as the intersection of all subgroups of G containing A. The same
pattern works similarly in rings, fields and similar algebraic structures.
We meet with sets when considering divisibility properties. The multiples
of a given element a in a commutative ring R form the (principal) ideal generated
by a. Starting with a set A in R and forming all linear combinations of elements
of A with coefficients from R, one gets the ideal generated by A. Ideals I in R
are characterised as non-empty sets of R with the property:
The ideal generated by A in R is the smallest ideal of R
containing A, or otherwise, the intersection of all ideals of R containing A. In
Z every ideal is a principal ideal; this is equivalent to the existence of a greatest
common divisor of a subset A of Z. For instance, the ideal of Z generated by
by 4 and 6 is also generated by their greatest common divisor 2. Ideals have
been invented and the theory of ideals has been developed to master divisibility
in such rings where the existence of the greatest common divisor is not assured,
that is, where not every ideal is a principal ideal. Ideals were in history the first
explicit occurrence of sets and set theory methods in algebra.

37

SETS

3.8. Zorns Lemma

Applying set theory in advanced mathematics, as mustered up to this point, is
light or even very light guns. The heaviest and then downright heavy is
Zorns Lemma. Its use is in eliminating the transfinite ordinals of olden times,
and the way to prove it is similar to that of formerly providing a set with a wellordering. Zorns Lemma runs as follows:
Let be a non-empty set of sets (partially) ordered in the natural way, that
is, by means of inclusion. A chain is a subset of such that for each pair
we have either
or
(thus a chain is totally ordered).
Suppose that the union of each of its chains also belongs to
( is closed with
respect to forming unions of chains.) Then Zorns Lemma asserts
possesses
maximal elements, that is elements
such that if
then
for
every
By means of Zorns Lemma one can prove, for instance, that each group G
possesses maximal proper subgroups: Let
not be the unit element of G;
take for the set of all subgroups of G that exclude a. Let
be a chain in
then the union of is obviously a subgroup* that excludes a, and thus belongs
to
which shows that the condition of Zorns Lemma is fulfilled. According to
Zorns Lemma
possesses a maximal element, which is a maximal proper
subgroup of G.
This proof is characteristic of applications of Zorns Lemma. A concept from
this complex of ideas that is crucial in modern algebra and analysis is that of
filter:
Let R be a set. A filter F of R is a set of subsets of R with the properties:
the empty set is not a member of F,
and

By means of Zorns Lemma one shows that each filter of R is contained in a

maximal filter, different from R (also named an ultrafilter).
If
then there is one maximal filter of R containing the one-element
set {a}, namely the set of all sets of which a is an element. But otherwise as
soon as R is infinite, maximal filters are almost pathological objects, beyond
intuition and construction: each maximal filter F of R possesses the perplexing
property:
for each

either

* Indeed: If
then
or
and in the fust case

or

thus

for certain
, thus

now
and

similarly in the second case. This line of reasoning is characteristic of applications of Zorns
Lemma.

38

CHAPTER 3

3.9. Cardinality
All sets occurring in the applications shown hitherto had some structure, which
was at stake in the set theory operations performed on them. This is also true
of Zorns Lemma. Indeed, was a set of sets and by this very fact bears the
structure of pairs of member sets being part of each other.
The absolutely unstructured set still possesses one characteristic: cardinality.
Sets that can be mapped one-to-one upon each other have by definition the
same cardinality, and if A can be mapped one-to-one into but not onto B, A is
by definition of smaller cardinality than B. This then operationally defines
cardinality and order in cardinalities. After this definition, it does not matter
much whether cardinalities themselves are introduced as mathematical objects.
In fact, it can be done by putting sets having the same cardinality into one
class, which is named their cardinality.
First of all, cardinality should be appreciated as a historical-philosophical
phenomenon: the courage to extend an elementary concept like number, which
apparently needs no analysis, to infinite sets and to defy the seeming paradox
that an infinite set can have the same cardinality as some true subset. Technically
viewed, cardinality appears to be important for five reasons:
Firstly, equal cardinality of sets and subsets can be exploited in a positive
sense.
Secondly, the countability of the sets of integer, rational, and even algebraic
numbers at first sight unexpected allows unexpected constructions in this
field.
Thirdly, the uncountability of R guarantees in a simple way the existence of
non-algebraic numbers and in a more general way that by the difference of
cardinality one set can be distinguished as a true extension of another.
Fourthly, the unexpected phenomenon that forming the product of an
infinite set by itself does not increase its cardinality and, as a consequence,
that line segment, square, cube, and so on have the same cardinality, is the
source of the problem of how to distinguish dimensions, which has been solved
by paying attention to more structure, namely topological structure.
Fifthly, the well-known drawer principle: if a set A is being mapped in a
set of lower cardinality, at least two elements of A have the same image.

3.10. Sets as a Purely Linguistic Phenomenon

The reader may ask why I have elaborated so much on the phenomenon of sets
in advanced mathematics, at the risk of supplying no more than verbal information. In fact what I did was not much more than to lift a tip of the magic veil

spread over set theory by innovators in the past, who claimed to have shown
that set theory, hitherto a privilege of advanced mathematics, could be successfully taught in primary school and even kindergarten. After the back to basics
reaction it is still or even more necessary to analyse those pretentions because

39

SETS

they are characteristic of an approach to designing mathematics education that

lacks any phenomenological background.
Set theory as meant by those innovators was quite another subject than is
meant in advanced mathematics. In todays school mathematics sets are not an
organising device for mathematical (or non-mathematical) phenomena, but an
aim in themselves. Intersection, union, complement, and power set are not
introduced when and where the subject matter asks for these organising devices.
Instead a subject matter has been created to exercise and train these operations.
It should be noticed that this is not an unusual way of creating school teaching
matter. Rather than developing set theory schemas as organising tools from
subject matter that asked for such organising and schematising, empty boxes
are taught and, in order to appease ones didactical remorses, filled with (false)
concretisations. Sets originating this way always remain within the palpably
concrete sphere or are purely linguistic phenomena. Collecting a finite number
of objects in a set, which as a mental object nobody has asked for, only in order
to apply set operations on it, is one aspect of this false concretisation. Another
is the so-called Venn diagram; a third, logical blocks. Later we will stress that
the proper problem with sets is to grasp and recognise them abstractly.
Set as a purely linguistic phenomenon expresses itself in this kind of school
mathematics in particular by the cult of creating sets in extension. Every
predicate can be transformed into a set indeed, the set of things sharing
the predicate. From ... is red one can form the set of red things; from ...
has long hair and wears spectacles the set of long-haired spectacle-wearers,
which in turn is the intersection of the long-haired people and the spectaclewearers. Or closer to mathematics: from ... is more than 7 or less than 3
one forms the set of the numbers more than 7 and less than 3, to be represented
as the union of the set of numbers
and the set of numbers
Or divisor
can be transformed into the set of pairs
such that x is a divisor of y. In
advanced mathematics sets are created in extension where they are needed. In
the kind of school mathematics we have in view, sets are training matter, and
what is trained are things to which only side attention is paid in advanced
mathematics. Consider the parallelism between logical and set theory operations:

to
to
to

(and) corresponds
(or) corresponds
(not) corresponds

(intersection),
(union),
(complement).

The last line exhibits what is wrong with this parallelism: in order to have
not correspond to complementation, one must know with respect to which
domain the complement has to be taken. Here I indicated it by R. In general it
is known under various names: basic set, universe, choice set, reference set.
It is a conception occurring in school mathematics only;in normal mathematics
it is entirely unknown and this for more profound reasons than the difficulty
with the negation.
True mathematics is a meaningful activity in an open domain, rather than

40

CHAPTER 3

a haphazard one in an a priori fixed reference set. I admit that in set theory
axiomatics, it is postulated that from every object one can make a one-member
set, that for every set of sets the union exists, that every set has its power set,
and that every predicate has its extension. In the school mathematics I sketched,
one feels obliged to exert oneself to realise all these possibilities. But in true
mathematics there is not the slightest need for, say, the union of the set of
natural numbers that are divisible by three and not the power of a prime number,
with the set of finite subsets of a three-dimensional vector space, with the set
of all sunsets. Sets are formed and used where they are needed.
With regard to the parallelism of logical and set theory operations, I would
not exclude the need at a certain moment to illustrate an equivalence like

or perhaps even
by its set theory counterpart. As soon as these logical equivalences provide for

real wants, nobody will raise objections. Even then, exaggerations and wrong
concretisations should be avoided. By wrong concretisations I mean concretising
the corresponding set theory equalities

dogmatically by means of Venn diagrams rather than by rectangular partitions

(Figures 1 and 2). It is pedantry to go further if there is no reason for it, and
in particular trying to extend this parallelism to
and
such is the
whole ritual of sets as a purely linguistic phenomenon.

SETS

41

3.11. The Good Reference Set

A few lines above I stated: True mathematics is a meaningful activity in an
open domain, rather than a haphazard one in an a priori fixed reference set.
Curriculum developers and textbook authors apparently believe they offer

teachers and pupils firm ground under their feet as soon as they restrict mathematical activities to a fixed reference set. This, they think, prevents possible
surprises that might be disappointing. Fixed reference sets have been a source of
confusion in particular in probability (which we are obliged to anticipate
here) and have provided rather than firm ground under the feet, a swamp of

contradictions. In traditional probability, one can also learn what should really
be done if reference sets are tried, and this I will explain here.
Take a die. The six possible results of a throw form such a reference set R.
Related propositions are of this kind: the result x of a throw is even, or is 3,
or is 6. Such propositions determine subsets of R; these subsets form a Borel
system, where one can play set theory algebra and form unions, intersections,

complements.
Yet this reference set is not of any real importance. In fact, one promptly
passes to considering throws with 2,3, ... , n dice, or rolls of one die rolled

2,3, ... , n times, or together with the die a coin showing head or tails, and so
on. This requires forming ever and ever new reference sets, arising from the old
ones by forming products; that is, products of different sets, powers of the

same, and products with a finite (or even an infinite) number of factors. One
lives, as it were, with an open reference set which at any moment, if need be,
can be enriched by adding new factors or impoverished by neglecting factors.
A colloquium lecture on a probability theme may start with the announcement
that the lecturer supposes the reference set (more precisely the probability
field) so large that every stochastic variable he will introduce is meaningful if

conceived as a function on this set. It is just by means of stochastic variables that

the probabilist can bring any reference set he means to the fore. For instance,
the result of the nth throw of a certain die is considered a stochastic variable
defined on a probability field, which has among its factors at least the nth throw
of that die though it can have many more factors.
It is a point of view that works very well wherever reference sets are at stake.
Let us illustrate this. Consider the following two reference sets:
the set of flowers with predicates like

... is red,... is multicolored,... is tall,

the set of animals with predicates like

... is a mammal,... lives in Africa,... is extinct.

As soon as both are to be placed within the same context, the correct reference
is not obtained by throwing
and
on one heap but by forming their
product

42

CHAPTER 3

In this way any proposition p on

can be considered as one on R, which is not
influenced by the second component; similarly any proposition q on
can be
considered as one on R, which is not influenced by the first component. By this
interpretation any proposition

about

is jacked up to a proposition

and any proposition

about

to a proposition

The proposition
the flower is red

on flowers is jacked up to a proposition on

by stating that
in the pair

the flower is red,

which of course is equivalent to

the flower is red,

and this means that we can dispense with the more complicated form and even
need not introduce it.
In order to make the connection between prepositional logic and set algebra,
indicated in Section 3.10, we have only to ensure that all propositions under
future consideration (or all predicates) deal with the same set of objects that
is, for the predicates, have the same domain. This is achieved by means of the
product of the domains of all propositions occurring in the context, jacking
up at least in theory each proposition to one about the new domain, the
reference set of the given context. Extending the context involves enriching the
reference set by new factors. Restricting the context means omitting factors
of the reference set; that is, projecting it on a poorer one, with a restricted
system of factors.
Rather than an organisation of mathematics, the present explanation organises
a system of propositions on reality by mathematical means, in particular, by set
theory. It is less relevant for mathematics than it is for relating mathematics

SETS

43

to its applications. The most embracing reference set which, starting from reality,
can come under consideration might be dubbed: world. In this sense world
is neither a set of space and time points, nor a set of objects in space-time,
nor any physical datum whatever, but the reference set of the context dealing
with reality. World is the product of an awful number of factors such as
flowers, animals, colours, throws of coins, throws of dice, moments, points, persons, feelings, thoughts, and whatever one may
think of.
If I pronounce some proposition such as
the flower is red,
the animal lives in Africa,
the die shows a six,

I occupy myself with one factor, one facet of the world, while disregarding
the others. If I pronounce the three of them together, I am grasping more of the
world. Yet this is still a small reference set. Realistic contexts require a much
broader reference set actually but also potentially by their openness, by the

potentiality of extending.
The reason why this phenomenological analysis is less relevant for mathematics than for applications (for instance, probability, where it originated) is
to be found in the peculiarities that distinguish mathematical language from
the vernacular, in which applications are usually formulated. I shall return to
these peculiarities, but I will anticipate the most essential one: The variables
of mathematical language are omnivalent in principle; letters can indicate anything, whereas any restriction of domain must be made explicit. Such variables
are rare in the vernacular the variable something is akin to mathematical
variables. Domains of variables in the vernacular are most often extremely
restricted; flower can be applied only to flowers, now only to moments.
Yet this also offers us an advantage: the linguistic symbol promptly displays the
reference set, the facet of the world, to which it is restricted, and by this
fact doubts about the reference set of a context are virtually excluded. It is the
same situation which we meet with stochastic variables, which are not variables
in the usual mathematical sense. If I denote the result of the next throw with a
die by , then the proposition

has a different meaning from

in a mathematical context. In a mathematical context the variable x must be

bound in some way in order to occur in a proposition, but regarding the proposition
can directly pronounce judgments such as ... is true or ...
has the probability .

44

CHAPTER 3

3.12. Elementary Structures

From a didactical phenomenological point of view implicit and explicit uses of
sets are to be distinguished. As a matter of fact for mental objects to function
well it is not required and often not even beneficial that they are made explicit.
This is particularly important for the didactical phenomenology of sets as mental
objects because, unlike numbers and geometrical objects, the vernacular knows
no terms to indicate sets in general. Nevertheless sets as substrata of structure
must play a role early in cognitive development even though they are neither
recognised nor isolated.
Sets, as they emerge in spontaneous development, are heavily structured.
This is not surprising. Attention is more strongly attached to structures than to
what lacks structure, and the inclination towards structuring that which lacks
structure dominates that of destructuring structure, at least as far as building
rather than destroying activities are concerned.
As examples I shall indicate such structures. First, the structure of succession
or file, placing objects besides or behind each other, for instance to make trains
for riding or putting blocks upon each other, perhaps according to size. Or
the same in a rhythmic-acoustic or rhythmic-gymnastic way, combined or not
with files of solid objects. Counting in the early meaningless manner or in
a wrong sequence shows the same pattern. Creating from a given file a new
one by regular skipping (for instance when jumping in a tile pattern) is a more
sophisticated way of structuring, as is the regular inserting of new elements
into a file.
By being provided with periodicities (have beads of different colour regularly
alternating in a string), the file is transformed into a repetition structure. With
repetition structures one can didactically distinguish: passive recognising, imitating, and describing. Repetition structures are developed not only linearly but
also in the plane and the space it would be of interest to know in which stage
the potential infinity of the repetition structure becomes conscious. Akin to
the repetition structure is the cyclic structure, arranging objects cyclically
persons around a table or a centre or the mental structures of the day, week and
year cycle.
More topological structures are: the detour structure (the child makes a
detour after which it meets the adult guide again), the border structure (walking
on borders), the enclosure structure (a true or symbolic enclosure of oneself
or somebody else by means of drawn or otherwise marked enclosures), the
barrier structure (blocking the path with the arms stretched), the hide-and-seek

structure, the look-around-the-corner structure, the labyrinth structure, the

island structure (objects are declared to be islands within a surrounding sea).
This list of general structures can be complemented by more or less specialised
ones: the box with blocks that cannot be packed arbitrarily, the puzzle that
must be composed in a definite manner (that is, where the parts are mutually
closely related); the family in general, or in varying realisations, structured

SETS

45

according to generations, sexes, kinship, the content of the toy-cupboard,

structured with respect to a lot of relations, the seating accommodation in a
car or bus, a pack of cards.
In appropriate places I will reconsider these structures, which can be processed
into a wealth or didactical material. The enumeration was meant only to evoke
an idea of how strongly the early cognitive possession is dominated by structures.
Sets as substrata of these structures are in a few cases easily recognised, in other
cases it is harder or even difficult. I mean: for the author and his readers. Does
the child notice the substratum set? One should ask rather whether it would
be relevant to him.

3.12a. Order Structures

As far as elementary structures are concerned, those that lend themselves at the
earliest stage to mathematising seem to be the order structures linear and
cyclical order (with partial orders as a rather deranging phenomenon). They,
too, are obtained from richer structures by impoverishing in a particularly
effective way. At the start this richer material is of a quite concrete character,
that is, felt by hands, feet, senses. In a developmental line the first order relations
might be

spatial inclusion,
spatial succession.
Where

temporal succession
is to be placed is not easy to say. Everybody knows the difficulties that many
though not all little children have with yesterday and tomorrow: they
use the same word for both of them, and as by preference, it seems, tomorrow.
What is wrong here? The mental grasp of the time direction, the idea of something like a future that in fact is (still) non-existent? Is the preference for
tomorrow to be explained by the greater frequency of the word?
Daphne (3; 8) gets a number of sweets. Mother says: This we will keep for tomorrow.
Daphne: And this for yesterday, and that for Thursday. Mother: Thursday is today,
isnt it? Daphne: Then for Friday.

Order relations are expressed in

comparing spatial and temporal dimensions and quantities,

characterised by comparatives such as

longer, shorter, thicker, higher, older, more, fuller, and so on,
by prepositions such as
in front of, behind, after, between, and so on,

46

CHAPTER 3

by adverbs such as
first, then, more, less, and so on.
I would not be able to say in which succession. Farther away, though perhaps
only partially, is

comparing intensities
by means of
heavier*, hotter, sweeter, dearer, nicer, and so on.

Order relations are

perceived,
stated,
actively imitated,
created,
for instance, if the child follows a prescribed path, from one stone or step to
the next, or if, more or less consciously he chooses his path, jumping from one
stone to another in the neighborhood, climbing from one branch to the next.
Much concrete material paths, stairs, strings, straight and crooked linear
objects, nests of objects suggest, a

global order that is, comparing totally,

though under less concrete circumstances with temporal or intensive criteria
this global suggestion can be lacking. Whereas in the first case the global order

is
accepted and at most analysed afterwards,

it happens in the other cases that the global order is

synthesised from the local one by pairwise comparison.

The link between local and global order then is

transitivity
two local mutual situations when combined yield a third, which is mathematised by

Transitivity, if not imposed by definition, is an empirical feature. Cognitively

viewed, the complex of order relations as it develops itself is a complex of
justified and sometimes unjustified analogies, linguistically supported by

comparatives and superlatives such as

* Developmentally weight starts as an intension.

SETS

47

big, bigger, biggest,

sweet, sweeter, sweetest,
much, more, most,
and so on. Putting moments, weights, intensities, and so on, on a number line
may arouse the suggestion of global linear order.
Necessary and perhaps even sufficient for the constitution of
one mental object ordered set

seems to me the
operational mastery of transitivity*

and the

operational inverting of order.

Necessary for the constitution of

the mental object linear order

seems to me the
operational mastery of endo- and isomorphisms of orders
and the

rejection of only partial, as it were, total order.

Our exposition has increasingly focused on linear order. A priori
cyclic orders

are

globally
given;
spatially

around the block, around the tree, the closed string of beads
temporally

the day cycle, the week cycle, the merry-go-round, continued repetitions. A
restricted grip on the global character is within the scope of the mental object
possible by means of
local analysis
* Possibly as early as the age of four, as shown by P. Bryant, contradicting Piaget. See
Chapter 1.14.

48

CHAPTER 3

an ordered triple determines the difference between a cyclic order and its
converse
neighborhood relations determine the cyclic order,
but formalising
what corresponds to transitivity
is far away in the conceptual domain of cyclic order.* However, for constituting
cyclic order as a mental object, the

operational mastery of endo- and isomorphisms, and

operationally inverting
look even more important than in the case of the linear order.
In the planar geometrical context the cyclic order is enriched by
right and left turn,
which will be considered at appropriate places.**

3.13. The Elimination of Structure

Recognising the substratum set requires thinking away the structure based on
it. This can be quite difficult if the structure has arisen from a natural order or
topology as is the case with most of the examples of general structures. It
may be easier, though still difficult, with structures generated by classification.
The colours in a supply of beads are more easily thought away than the structure
of the human family. It is easiest when the elements are not, or hardly, distinguished from each other, and unrelated (chips of the same kind, a flock of
sheep, the reverse side of a pack of cards) or another extreme if the whole
looks like a hotchpotch, which does not invite structuring.
Didacticians who undertook to make children familiar with explicit sets have,
as it appears, struggled with the problem of how to eliminate structure in order
to evoke the suggestion of an unstructured set. The most natural method
indistinguishable objects has been the least common, probably because of a
wrong concretisation of the requirement that the elements of a set should be
different. The most usual paradigm of the bare unstructured set is the hotchpotch the elements are being put into a Venn diagram or between braces in
such a way as to kill any suggestion of order, higgledy-piggledy, and if they
are numbers or letters, criss-crossed in the diagram, and in a chaotic order
between the braces. All that could be experienced by the child as order, is set
at defiance the order in the family where every member knows his place
at table, has his own bed to sleep in, the order in the tidy toy-cupboard, in the
flower garden so as to evoke the set deprived of any structure. It is suggested
* Cp. Mathematics as an Educational Task, p. 472 sq.
** Sections 10.3 and 14.1216.

SETS

49

that as soon as objects figure as elements of a set, they are freely movable in
space an ontology based on a wrong phenomenology. Indeed, in the great

majority of cases the elements of sets are mental objects, and as soon as point
sets are discussed, the pupil is asked to relearn what he has been taught to
unlearn: that for the individual elements of point sets the only thing that matters
is the place.

In spite of the chaotic order in the Venn diagram or between the brackets,
the suggestion that the place does not matter is still not strong enough. On the
contrary, teacher and pupil must explicitly be informed about it, as they must
about the possibility that contrary to the verbal suggestion a set is allowed
to consist of one or even of no element. Why is it allowed? By prescription,
since this is the only way to introduce such objects as one-element sets and the
empty set as long as there is no need for them.
The didactical means by which children are forced to constitute structureless
sets are kill or cure medicine, like the horse-drench. It is impossible to check
whether and how they function, because in fact the constitution of the mental
object is skipped and replaced by a verbalism that does not cover any mental
object. These verbal products are afterwards subjected to purely linguistic
operations, as demonstrated earlier.
3.1416. Equality

3.14. The approach through blinkers is the most disturbing feature of these
horse cures. If sets are to be constituted close to reality a broader context is
required. Close to reality does not mean Venn diagrams of pictures, but
a living context.
School texts often explain: A set is determined if for each object it is known
whether it is element of the set or not. This is misleading. If it were literally
true, I would not be able to speak of sets of natural numbers without precisely
enumerating all their elements. It is as misleading as the suggestion that in order
to become elements of a set, objects must be stripped of their position.
The decisive feature, however, is what equality means in the case of sets. In
order to speak about sets, one must know criteria of equality. Do two pictures
represent the same family? Does this picture represent the family across the
street? Are these the same marbles that were here last week and that are now in
the bag? Do A and B have the same sorts of trees in their gardens? (Notice that
rather than concrete trees, I said sorts.) Do A and B have the same school
marks?
Of course, in order to answer such questions a broader context is required.
We should first know something about the pairwise equality of the elements
belonging to such sets, that is, know what equality means with simple objects.
Afterwards we can look for the composition of the sets in question.
By broader context I mean formally the ontological question of what is
equality, that is, criteria in general, not necessarily restricted to sets. Not absolute
criteria, but criteria determined by the context itself. As a context I now choose,

50

CHAPTER 3

as regards subject matter, closeness to reality, and as regards didactics, young

children. I am going to sketch a broad theme.
3.15. Imagine a number of pictures of the same church building, taken from
different directions, under different angles, perhaps one before the restoration,
one with scaffolding, one after the restoration. Is it the same church? a
question which can be discussed, affirmed, and denied. The result does not
matter. What really matters is the insight that criteria are required to answer
such questions and that there is nobody to offer you the criteria.
In such a discussion analogies will be cited as arguments. A man in different
clothing, with one tooth more or less, a woman who got a different name by

marriage, a mother-to-be with a child in her belly are they different persons?
I feel a new man, Saul became Paul, the saltpillar called Lots wife
themes of endless discussions.
The most likely final result is: The church building is continually the same,
but the church building previous to restoration, the church building under
restoration, and the church building after restoration are different things.
Mathematically formalised: From C (the church building) and t (time) we get
the pair
thus besides the object C we have new objects
all different. In a similar way the other examples can be dealt with:
the church building according to different aspects, SaulPaul before and after
Damascus, and so on.
If the church building, or Paul, is considered to be a simple (not composite)
object, then the church building previous to, during, and after restoration,
Paul before and after Damascus, are composite (mental) objects. The way of
composing, however, by mental pair formation, of course differs from that of
composing a family, a construction box, a jigsaw puzzle, the seating accommodation in a bus or car, a pack of cards. Composing in the latter sense is how sets
are composed of elements. Of course, as was stressed earlier, in all such examples,
in order to get to the substratum sets, the structure has to be eliminated, which
is not always easy. Sometimes this can be done by means of criteria of equality.
The family remains unchanged when its members move, the construction box
and the jigsaw puzzle are the same however their parts are arranged (though
the orderly packed state can be considered as a special object), the pack of
cards is not changed by shuffling; the seating accommodation in the bus is the
same whether occupied or not.
Yet all these examples show much more structure, which cannot be done
away with by these operations. Family, construction box, puzzle, quartet game,
seating accommodation keep their individuality not only as sets but also as
structures (unless parts get lost or are changed).
Let me illustrate this equality of structure by another example. Take a
ludo board* such as you can buy in a shop a quite simple structure with a
* A cyclic arrangement of squares with four entrances on which four gamblers move their
men according to certain rules.

SETS

51

set of squares as set substratum. Take a second specimen, exactly congruent. Is

it the same object? No, not as a concrete ludo board. But you can consider both
of them as pictures of the same abstract ludo board. As a matter of fact,
the boards may even be of different manufacture, of different size, in other
colours and nevertheless serve as images of the same object. (If by chance they
have different numbers of squares, one should consequently distinguish ludo
games with different numbers of squares, as with the game of draughts, where
there is the international and the PolishGerman variant.) The concrete boards
appear as pictures of the mental board in the same way as you can make various
snapshots of a family or consider a construction box, a puzzle, a pack of cards,
the seating accommodation in a bus as an abstract pattern, which can be realised
in various congruent or non-congruent ways. The term picture which I
used, evokes the idea of a two-dimensional image, while other examples suggest
visual or instrumental realisations, but reciting the same poem in various ways or
singing a song can also be interpreted as a picture of a more abstract object. A
more comprehensive term than picture would be model* the concretisations
are models of the same abstract structure.
The structure of the ludo board is what is called in mathematics a directed
graph. The square fields can be thought of as being replaced by their centres
(the nodes of the graph) and neighboring ones connected by arrows in the
direction they are passed through in the course of the game. This is a simplified
topological image of the usual ludo board. In the background there is an even
more abstract structure, lacking any geometric appeal: the set P of nodes and
on it the special relation R (x, y) (there is an arrow from x to y). However, the
set substratum can also be chosen differently: along with the set P one considers
the set Q of arrows, while adding the relations
(x, y) (node x is the tail of
arrow y) and
(x, y) (node x is the point of arrow y). Choosing the set substratum of a structure is a matter of taste and practice.
I should add that I have analysed the structure of the ludo board only, not
that of the game, which is much more involved. Then a set of pawns should be
added, a concept of position which is a relation between nodes and pawns
, a rule as to how a position can change into another, data on the starting
position and a rule on the consequences of a final position. Here I have restricted
myself to the ludo board.
I will now reconsider the other examples of structures. First of all, the
family, or rather a particular family, say, of two adults, father and mother, three
children, boy, girl, boy. This structure looks more variegated than that of the
ludo board. There is a linear order on it according to age, there are classifications
according to generations and sex, and there is a genealogy. Another structure
is the family, of which the particular families are models. It must allow for more
* As explained in Weeding and Sowing (p. 130 sq.) this is one of the diametrically opposed
meanings of model, that is, as after-image. Model as pre-image would just be the abstract
ludo board, the abstract construction box, after which the concrete ones have been manufactured, or the poem or song on paper as conceived in the authors mind.

52

CHAPTER 3

than two generations, for more complex and disconnected genealogies. The
structure of the construction box is determined by classes of geometric shapes,
perhaps also by functional classes. The jigsaw puzzle reminds one of a graph
representing the particular pieces by means of nodes and neighborhood by
means of joining line segments although such a structure would be too rough
to allow one to reconstruct the puzzle from the graph, which does not determine
the jigsaw puzzle structure sharply enough. The true structure is the geometric
division of the rectangular point set according to the jigsaw pieces. If I focus on
this structure, I only neglect what is stuck or painted on the pieces.
3.16. When the set theory frenzy reached the Netherlands, it was the stamp
collection that was raised to the rank of a paradigm for sets indeed, the plain
meaning of the Dutch word for set is collection. So one could even argue why
all elements of a set should be different: a serious philatelist does not insert
more than one specimen of a postage stamp into his collection. Following this
argument the 50-cent stamps in stock at midnight on 1 August, 1975 at the
Utrecht Central Post Office would not constitute a set.

One can distinguish at least two different concepts of postage stamp: the
concrete printed and gummed piece of paper, and the sort of which this piece
of paper is a specimen. (Different philatelists may adhere to different notions
of sort; some of them know more sophisticated sorts than do others, but this
does not matter here.)
What then is a stamp collection? The set of printed pieces of paper? Yes and
no. The collection of philatelist X grows, is transformed by exchange, but does
not change its identity. Well, this is not a new point of view; we can account for
it by speaking of Xs collection at time t. Is this really the set I mean? If in this
set I replace the specimen of a certain sort actually or mentally by another of
the same sort, it remains the same stamp collection although the set has changed.
The philatelist does not collect pieces of paper but sorts. He does not collect
so as to own a sort. The stamp collection is not a set of sorts. In a stamp collection sorts are represented by specimens not all sorts are represented and
perhaps some sorts are represented by more than one specimen.
In the gardens of A and B are the same sorts of trees we met this example
earlier. In all YZ bookshops you can buy the same books, at all MN ice cream
stands you can buy the same ice creams, the countries P and Q have the same
fauna. Of course, a book as a concrete object can sit in one bookshop only,
two ice creams are never the same, and one lion is not another.
The concept that fits these situations is assortment rather than set, the same
assortment of stamps, trees, books, ice creams, animals, respectively. One
considers the set of all stamps, trees, books, ice creams, animals, respectively.
This set is divided into sorts, which form a set An assortment is a set of sorts,
a subset of
It seems a rather weak concept: such an assortment could be empty, indeed.
In itself, such an assortment is poor: a set with a bit of structure its elements

SETS

53

being sorts. In the stamp collection case it is no more than a list of the respective
sorts. However, this set bears a strong external structure by its relatedness to the
set
of all stamps and the various stamp collections of the other philatelists.
The assortment gets its significance from the fact that it is an assortment of the
set
partitioned into sorts. In this way it becomes meaningful to ask whether
two philatelists share the same assortment of stamps (although their collections
are separate), or whether the one is larger than the other, or whether they
overlap, or have nothing in common (different countries or ages).
Consequently, a stamp collection is a set (of pieces of paper) plus an assortment; the set is, as it were, externally structured. A group or a linear space can
be studied in itself. On the other hand, one can stress the relations of the group,
or the linear space with others. Then one acts in what is called the category of
groups or linear spaces; there the external structure of the particular group or
space expresses itself by the so-called morphisms of the category.

3.17. Structures as Mental Objects

Except in artificial examples and exercises, sets are usually endowed with, and
are dependent on, structures and can be grasped through these structures only.
As a substratum a set becomes explicit if the structure is recognised and consciously eliminated. This has been the meaning of the preceding discussion. The
examples of structures we chose were taken from common experience. Artificial
ones need not be rejected provided one realises their deficiencies. Logic blocks,
for instance, suffer from four ills: All is prestructured, structure is restricted to
classification, a closed (even finite) mathematical universe is mirrored in order
to exercise set algebra, and finally, all relevant predicates always determine one
unique element together. These four ills are the expression of a systematism
that, like unfortunately all systematisms, attracts didacticians of mathematics.
(Of course, reasonably applied, logic blocks may be a useful tool.)
Recognition of structures and through the structures of substratum sets
is a schematising abstraction, which in the didactical process deserves concrete
support, though not so strong a support as is supplied by logic blocks. One is
better advised to draw pictures that reflect structures and substrata schematically,
for instance, along with the ludo board the schematising graph. For classifying
according to one characteristic, the most suggestive picture is the set of bags
(Figure 3) and in the case of double classifications two sets of crossing bags
(Figure 4). A pack of cards is schematised as the product of
{2,3,4,5,6,7,8,9,10,J,O,K,A}
and

{C,H,S,D}.

54

CHAPTER 3

There are many ways to represent assortments schematically: by means of a

check list, by a schematic page of a stamp album where some places are filled
in and others are not, by a page of a stamp catalogue with some items checked.
Yet with all these representations the set remains in limbo. In order to remedy
this defect one can combine these pictures with that of the classification.

We will encounter more of this kind of schema. Again we stress that they
do not concretise concepts as do Venn diagrams but are derived from real
structures by schematising abstractions and have to serve for constituting mental
objects. Whether and how fast and by which stages one passes to constituting
concepts is another question. Stages would be characterised by their verbal
machinery. One can restrict oneself to simply introducing words like set,
group, ordered set, graph, directed graph for this or that, in the way
one speaks about numbers and addition, one can denote particular structures
by letters, and in formalising the notation of a structure one can even go so far
as to distinguish formally the substratum and the structuring relation (such as
with an ordered set
It depends on the level of formalising whether these
are pure names used to call up objects or whether one intends to describe mental
operations with the objects by formal operations with their names. From traditional arithmetic instruction one knows that formalising is possible at an early
age the first formalised activity a child is taught is column arithmetic. It
is also known that premature formalisms can be pernicious. One should profit
from these experiences in the didactical subject matter that is dealt with here;
introducing formal machinery only where it covers mental objects and where
it is required to describe and facilitate operations with these objects. With the
view on this issue fixed, I will review a few set theory concepts.
3.183.20. N U M E R O S I T Y AS A SET T H E O R Y OBJECT

3.18. Constitution of Cardinal Numbers

At present I will not deal with the constitution of the cardinal numbers; this
will be done in Chapter 4. Yet as far as constitution of sets and operations

SETS

55

on sets are concerned, something must be anticipated. In the constitution of

numerosity (or cardinal number) I distinguish the following activities:
Eliminating from structures with the same substratum the structuring
component in order to arrive at substratum sets,
and in this connection transforming the inclusion relation into an order
relation (less instead of contained in).
Using isomorphisms of structures with different substratum to compare
different sets.
Using transitivity of equality and order (of numerosity number).
Intentionally I did not add:
structure-free comparison of sets,

which by set theory is, as it were, ordained for comparing cardinalities, but
as important as it might be at the level of concept formation it is irrelevant
and ineffective for the constitution of mental objects which will be shown
immediately. Meanwhile I shall soften this strong assertion. There is one exception: whether one set has many more elements than the other can be decided
with the naked eye (or ear), without structuring the sets. There are early developmental stages in which a short sequence of constituted small numbers leads to

an indefinite many (which can be expressed verbally by a definite numeral).

Then the development progresses by differentiating many. The mental constitution of many could, at least partially, be performed by a procedure of
structure-free comparison, and in this way structure-free comparing would
influence, via the motor of many, the development of numerosity as a mental
object.
I review the three activities which at the start of the present subsection I
pointed out as being essential for the constitution of numerosity. In order
to clarify what I said, I review some well-known experiments: Two rows of
objects are laid down in such a structure (Figure 5) that by means of the greater

length of one row the larger number of objects in that row is made visible: the
isomorphism of one of the structures with a substructure of the other and

transitivity are ostensively used. Then the longer one is compressed to make
both of them equally long. Thus in order to maintain the order of numerosity,
one has to eliminate the change of structure and again apply transitivity. Such
experiments are considered by psychologists as tests of conservation, they do
not, however, unveil the rock on which the so-called non-conservers foundered.

56

CHAPTER 3

What these three activities mean for the constitution of numerosity number
will be discussed later. At present we are concerned with the part played by sets
implicitly or explicitly.
Of course, one may first ask: do they really play a part? As I have stressed
repeatedly,* numerals as cardinal numbers indefinite ones included are
a different linguistic category than adjectives: if I want to say about a football
team not that each of its members but the team itself is good, it is not enough
to say eleven good football players; I shall rather attach the adjective good
to the object eleven: a good eleven, or a good team. Numerals, however, are
by their very essence related to collectives of things, the particular football

player cannot be many or eleven, but it should be: eleven football players.
This collective, however, is not explicitly indicated as it is in the case of the
football team: the numerals stamp by their mere use the object they are attached
to as a collective.
Well, it is possible that in the first developmental stages the numerals as
well as the non-verbalised mental numerosities are concerned with structures
rather than with sets. At least with regards to the number 2, this is a quite

natural assumption. At an early stage the childs attention is drawn to natural

pairs: two eyes, two ears, two hands, the pair of parents, sun and moon, twins.
Is the two at this moment already mentally constituted or is it a new step

that the two is recognised in unstructured sets of two? Indeed, the pairs we
just cited are more or less strongly structured, the first three, in particular,
by right and left; but also externally structured because a union of pairs of
eyes, and so on, displays a natural classification structure which extends to the
pair of parents; sun and moon show a clear internal structure; in the case of
the (identical) twins it is weak or absent.
Natural triples are less conspicuous, natural quadruples, however, are, for
example the legs of many animals, and tables, the wheels of cars strong
internal structures in the first and third cases, and external ones in all of them.
Natural quintuples, such as the fingers of the hand, are imposed explicitly on
the child as means of constitution of a number. This imposition may not be
too successful, as a consequence of the strong structure of the system of fingers,
stressed by nursery rhymes, and impeding rather than favouring the constitution
of number: it is always the same triple or quadruple of fingers that is lifted if
children accompany the answer to the question of their age by this obligatory
gesture.
Structures may impede the constitution of number but they are also, as
seems to me, indispensable in instigating this process. I repeat the red currants
story:
At the rectangular table Bastiaan (4 ; 3) is seated opposite his younger sister, father opposite
mother, grandpa opposite grandma. At the dessert of red currants he suddenly lifts his little
spoon in the greatest agitation and ejaculates So many are we! Indeed they were six.

* For instance, Weeding and Sowing, p. 215 sq.

SETS

57

I asked him: Why? and he answered I see it so, and then two children, two adults,
two grandpa and grandma. Possibly the six currants lay on the spoon in the same configuration of six as we occupied around the table, but this I could not see. At that time
Bastiaan was still quite unsure with numbers and obstinately refused to count. This event
distinctly marks in his development the constitution of numerosity numbers.

This is one of the observations that taught me how important for the development of mental numerosities is the structured comparison of sets. Adults impose
a special kind of structured comparison: by means of the ordered set of natural
numbers, N, used to count out a given set S. This asks for two remarks: On the
one hand, for comparing sets, N is not indispensable. Often, yet not always,
N is useful as a mediator between sets that are to be compared. There are cases,
examples of which we shall deal with later, where thanks to mutually related
structures on the sets that are to be compared, the intervention of N is inappropriate. On the other hand, the intervention of N is in general insufficient to
settle the cardinal number of a set. Unless this number is very small, settling
the cardinal number of a set requires the use of an extant structure or the
creation of a new one, in order to be sure that by the counting process S is
one-to-one exhausted. The elements of V might be given in a spatial or temporal
linear order, as a string of beads, as a passing train, as strokes of a clock. If this
is lacking, an order must be imposed on V. How it should be imposed is not
trivial and is a matter of learning. It can be done by a kind of coordinate system:
a line from the upper left to the upper right is mentally constructed, a second
line below it, and so one continues to the lower right corner. A variation is the
so-called ploughing scheme; first line from left to right, second from right to
left, and so on, alternating. In a horizontal and in a three-dimensional field,
one may start in the rear and progress to the front, or follow the inverse way.
The use of polar coordinates also occurs spiralling from the centre to the
exterior, or the other way round.
Structuring by analysing according to coordinates can be promoted in the
learning process by means of examples in which such structures are conspicuous.
The learner is seduced into strengthening them where they are weaker, and
to imposing them when they are absent. Another way of structuring is by means
of classification: if the elements of V are different in shape, colour, or size,
these characteristics can be used for classification. Neighborhood as a classifier
can lead to local groupings in S, mentally or physically marked by enclosures.
It is my intention to stress the significance of structures in the mental development of numerosities and numerosity and for the factual determination of
cardinal numbers. Though the cardinal number is a characteristic of the structureless set, its development and application can be effective only through structures
(at least with regard to definite numbers, rather than indefinite ones, which are
indicated by words like many and few). The didactical consequences are
clear. One should not try to have the constitution of numerosities depend on
structureless sets, as system fanatics prefer nor on the counting-out structure of
N adhered to by parents, when teaching their children the number sequence.

58

CHAPTER 3

I restrict myself to this remark. I did not aim at a phenomenology of number.

This will be developed later. Then methods of determining numerosities in the
reality will be systematically reviewed.

3.19. Cardinals of Sets Connected by a Structure

At this point I will discuss structures that may serve to determine cardinals
and relations between cardinals.
First of all, comparing cardinals of sets that are mutually connected by a
structure, which may be realised by a more or less explicit mapping between the
sets and
Examples:

the nose of x;
a set of people,
their noses by counting noses
the number of people in
is determined;
ballot paper assigned to x; instead of the votes cast for a person, one
counts the ballots;
hooks and eyes;

ticket for seat x in the theatre; the unoccupied seats are counted by
means of the unsold tickets;
counting out a set by numbering its elements by means of numerals or
written numbers.
A more sophisticated relation between two or more sets that can lead to a
comparison according to cardinality is alternating arrangement. The cards of
a pack are equally distributed among four persons A, B, C, D by dealing one
card (or any number of cards) at a time while alternating cyclically, starting with

A. In a directed string of beads where colours A, B, C, D follow each other

cyclically, there are equally many of each colour provided the string starts with
A and finishes with D. Such structures determine mappings where each element
(except the last) is mapped upon the next, and these mappings settle the equality
of numbers.
With planar structures similar phenomena can be perceived: A chess board
has as many black as white squares. The same holds for any like structure of
m by n squares if m or n is even. For
(Figure 6), the squares in each row

are mapped upon those in the other to shows that there are as many white
squares as black. From this the general assertion follows for all even m by
splitting the structure into strips of two lines. The same method shows that for
odd m and n the numbers of black and white squares differ by one.

A similar phenomenon is the planar tesselation into equilateral triangles of

59

SETS

alternating colour or other regular and regularly coloured tesselations (Figure

7 and 8). At a glance one states that globally each colour is equally frequent,

though it requires caution to cut out of the infinite pattern a finite piece where
indeed all colours occur equally often.
There are cases where the cardinal equivalence of sets is shown by a one-toone mapping rather than by counting out. The mappings are natural ones inspired
by geometry rather than arbitrary ones like those used with the structureless

sets pictured in Venn diagrams. According to my experience pre-school children

are able to recognise the equality globally albeit without motivating it; they are
also able to construct equipartitions of stocks of similar objects by alternately
dealing them out to persons or laying them on separate piles. Young school
children are even able to reason about it.
3.20. Cardinalities of Sets in a Union Structure

The formulas
for

empty

and

and similar ones with more terms describe in a formalised way facts that are
basic to certain methods of counting and comparing sets. These methods are
not at all self-evident or trivial.
When second graders in groups of two were asked to count large quantities of objects (sticks,
chips, thumb-tacks, paper clips, and so on), those who had not followed the experiment
program in the first grade did not hit spontaneously on the idea of sharing the work, whereas
those who had followed it, immediately shared the work.

60

CHAPTER 3

It is one thing, if a set is divided into subsets, to act according to the addition
formula; it is another thing, even if the formula is mastered, to divide a set
intentionally in order to use the formula. For all the structures we are going to
deal with, it is the same experience: using given structures or introducing new
ones can require abilities of different levels. Everything depends on how distinct
or how vague the structure is.
3.21. Cardinalities of Sets in a Rectangle Structure
The abstract set of pairs from A and B is geometrically visualised by the
rectangle structure (Figure 9): A horizontal and a vertical order structure is

imposed on A and B, respectively. Almost as effective with small sets is the

rhythmic model of pairs, such as those known, for instance, from reciting the
number sequence (twenty-one, twenty-two, twenty-three, ... , thirty-one,
thirty-two, . . . , forty-one, with strong stresses at the end) or from reading a
matrix (a-one-one, a-one-two, a-one-three, ... , a-two-one, a-two-two, ... ,
with stresses on the first number).

But even if the product structure is visualised or rhythmically scanned, its

recognisability can be vague. In a meeting room with m rows of n chairs each,
arranged in straight columns, the product structure is clearly visible. If the
chairs of subsequent rows shift the width of half a chair (as do the stars in the
American flag), the two intertwining product structures must be unravelled
before applying the product formula. Or the product structure can display
holes or other irregularities, which are first filled up or mended to make the
product structure clear. The number of type-writer touches on a sheet of paper
is structured according to lines, but the column order structure of touches,
though indicated on a rail, is rather weak.
In the case of equicardinality of the two factors the product structure is
reinforced; in its geometric visualisation as a square the equality of cardinals
is easily recognised. The square structure invites one to conceive other structures
as parts of it. For instance the border structure (Figure 10), which can be
understood as the difference of two square structures, and the triangle structure
(Figure 11), completed to a square structure, which in turn can be interpreted
as the union of two congruent triangle structures intersecting in the diagonal,
leading to the numerical evaluation

SETS

61

All these geometrical visualisations rectangle, square, triangle need not

be realised in the euclidean geometrical way; the pattern may be deformed

affinely or even topologically, provided the structure according to rows and

columns remains more or less clear. Instead of a linear order the columns can be
endowed with a cyclic one; then the product is borne, as it were, by a cylinder;
or if the rows are also cyclically ordered, by a torus. The column elements can
be the faces of a die; the product set then consists of sequences of faces of
dice, placed in a row.
On the other hand the euclidean structure can play an essential part if sets
are to be compared. With a row or a tower of congruent objects the length or
the height, respectively, can be a criterion of the cardinality. A large number
of such objects can be counted by arranging them in rows or towers of 10, where
the number 10 is verified only once by counting, and assured subsequently by
congruence.
If the rectangular structure of a set of pairs is not given a priori, its mental
image can be supported by visual images. If from m boys and n girls all pairs
boy-girl are to be formed, the actual pairs cannot be present at the same time.

62

CHAPTER 3

However, with n pictures of each boy and m pictures of each girl all pictures
of pairs can be formed (Figure 12); the rectangular pattern then serves to

arrange the m n pictures of pairs, in order to make sure that none is missing

and none is counted twice. Once this has been understood, the pictorial component can be dropped, and a diagrammatic rectangular schema, stripped of
pictorial features can take over, which in turn is replaced by a mental schema.

3.22. Cardinalities of Sets in an Equipartition Structure m Baskets with n

Eggs Each
I stressed elsewhere* that in many cases the arithmetical product operation
cannot or not without reinterpretation be justified by the set theory

product. In m baskets with n eggs each the resulting m n eggs suggest a set
product, which in fact is absent. The structure involved is a mapping structure,
which I will name equipartition. The set C of eggs is mapped into the set A of
baskets egg x lying in basket y and in this mapping f each element of A
has the same number of originals
is

The corresponding formula

In the present case the equipartition can somewhat artificially be restructured

as a product: In the original set of each element of A (that is, in each basket)
the elements (that is, the eggs) are arbitrarily numbered from 1 to
by
this procedure the particular sets
are mapped one-to-one upon each
other, such that C is represented as the product of

A by
In other cases the particular
might be naturally related. The set
of legs of n cats is structured on one hand by the cats and on the other by the
set {left foreleg, right foreleg, left hindleg, right hindleg}.
* Mathematics as an Educational Task, pp. 189, 250.

SETS

63

If on each of its n pages a book contains two pictures, one above and one
below, the number of pictures can be obtained as well by n times two pictures,
as by two times n pictures, namely, n upper and n lower pictures. Many examples

in the early phase of learning multiplication are of the kind n times m where
n is large and m is small. As long as the tables of multiplication are not yet

sufficiently memorised and multiplications are performed as (n times) repeated

additions, such multiplications can be arduous. Teachers are inclined to impose
the commutativity of multiplication and its application. Instead, one can proceed
conceptually by strengthening the structure to such a degree that it is seen as

m times n in the present case, n times two pictures as n upper and n lower
pictures, thus twice n pictures.
In other cases it can be utterly useless to strengthen the equipartition as to
form a product structure. If in a club A of n persons a president and a secretary
are to be chosen and the number of possibilities is asked for, the set of boards
is structured according to president and to secretary. Let f be the mapping that
maps each board on its president; then the original sets
(for various a)
cannot meaningfully be mapped upon each other with the aim that the equipartition is strengthened to become a product structure; attempts at doing this
can only cause confusion. Didactically the equipartion is needed as a structure
of its own.
The example of the last paragraph still admits of another structuring: the
ordered pairs formed by different elements of A (nobody can be both president
and secretary at one time), which leads to the number
However, if more factors are at stake (president, secretary, treasurer) this filling
up to full products becomes laborious.

3.23. Roads Model of the Product

There is no use in imposing the rectangle image of the set theory product by
every means, and certainly not if in the long run this model would push aside
or even suppress more effective models. In the example I have often cited of
three roads from A to B, two roads from B to C, how many ways from A
through B to C? an image is suggested that fundamentally differs from the
rectangular one. Of course, one can help the pupil recast it into the rectangular
pattern, but the rhythmical structure of the description of ways is certainly
more effective. Anyway, it is desirable that the pupil encounters the road
structure once disentangled as a structure of its own that embodies the
set theory product. Whereas visualising set products by rectangles can be continued to at most three factors (visualised by boxes instead of rectangles),
visualising by means of road structures knows of no restrictions on the number
of factors. The road structure, initially a problem situation that is elucidated
by the set product, is raised to the rank of a model of its own, the road model

64

CHAPTER 3

of the set product. (Figure 16) It is an extremely flexible model; it admits of an

arbitrary number of factors. It invites the embodying of products with equal
factors by running to and fro between A and B (n equal factors by running n
times) (Figure 13). It also invites problems like that of president and secretary

(from Section 3.22) to be embodied by running to and fro while not repeating
roads already used. The set of sequences of length k out of a set with n elements
is embodied by running k times while roads used earlier are prohibited.
3.24. Tree Model of the Product
Consider the task of colouring flags, the upper bar by three colours, the middle
by two others, the lowest again by two others, all prescribed. After a number of
failures, the task is structured by the pupils according to upper bar, middle
bar, lower bar. This can be an action structure (first colouring all flags with
an upper bar black, and so on) if the flag patterns to be coloured are given. In
the process of describing, a verbal structure, possibly rhythmically accentuated,
can develop. Schematising produces the tree model (Figure 14).

A variant of the tree model is the substitution model, which can produce
beautiful patterns (Figure 15).
The tree model, which is useful in many applications, does not account
adequately for the set product, since the equality of branches of higher order
is not visualised. By identifying the nodes on the same level one can pass from
the tree model to the road model (Figure 16).
3.25. Equipartitioning Relation

A relation between two sets A and B can be visualised as a subset of the rectangular image of
although this is hardly illuminating. A more effective
illustration is by means of the set of connections between the related elements of
A and B (Figure 17), and in the case of
by directed connections (arrows).

SETS

65

In the case of the full relation each element of A related to each of B this
gives an image of the set product of A and B. Although this picture is less

appropriate as a model of the set product, it is desirable that pupils recognise

the set product therein.
In combinatorics an important part is played by what I would call an equipartitioning relation: A relation between A and B such that each element of A
is related to k elements of B, and each of B to l elements of A. Then in the
model
connections leave from A, and
from B, which leads to the
equality

66

CHAPTER 3

This equality plays a part in counting problems.

Examples:

(1)

A: the corners of a cube,

B: the edges,
relation: element of A lying on element of B,

(2)

X has N members,
A: subsets of X with u members,

B: subsets of X with v members

relation: inclusion.

It is worthwhile signaling the different character of these examples. The sets

A and B of the first example (and similar examples) as well as the relation
between both of them can be given visually. In the second example even if
u and v are numerically specified the sets A and B are given purely verbally
(though the description can be simplified by speaking of u-tuples and v-tuples
from X rather than of subsets with u and v elements, respectively). Notwithstanding this verbal definition, the sets A and B might be mentally constituted
in this context, thanks to the visual power of the relation of inclusion between
elements of A and B.

3.26. Structure of the Power Set

A similar phenomenon is shown by
the set of subsets of X. Again it is a
verbal definition, which can become operational only in rather strongly formalised
mathematics. If on a lower level
is to be constituted as a mental object,
the rather strong and visually active structure of
must not be neglected.

This structure arises from the inclusion between the elements. For small X the
power set
can be considered as a graph, directed by inclusion (Figure 18

SETS

with
transition from

67

which visualises the structure of

schematically. By the
to
an induction can be prepared,

leading to the formula

if
3.27. Model of the Set of Subsets with u Elements
Let us reconsider the set A of u-tuples from a set X with N elements.* To
constitute A mentally, in particular with a view to determining #A, there are
other methods than that indicated at the end of Section 3.25. I will enumerate
some of them.
(1) One introduces the set C of (u l)-tuples from X and the relation of
inclusion between A and C, which is equipartitioning.
(2) One introduces the set D of sequences consisting of u different elements
of X (earlier embodied, for instance, by means of the tree or the road model
of Figures 14 and 16), and the forget mapping of D on A (forgetting the
order) that maps a sequence into its substratum set.
(3) One introduces an order in X, interprets an element of A as a choice
sequence of length N consisting of u choices of yes (belongs to A) and N u
choices of no (does not belong to A), and visualises this by means of a Galton
board (Figure 19), with N + 1 stories, where no means a fall to the left, and

yes a fall to the right. All choice sequences with u yes choices finish at the
uth place of the lowest story (counted from the left, starting with 0). So the set
A is embodied by the set of all paths on the Galton board finishing at the uth
place of the lowest story.

3.28. Drawer Principle

The last examples especially show the importance of an extant or imposed

structure for the constitution of sets as mental objects. Elsewhere** I have
investigated the difficulties pupils can have with the problem we generally
describe as the drawer principle. It appears that pupils who have no difficulties
with problems of the type
* N and u should be taken as definitely fixed numbers.
* * Weeding and Sowing, p. 210 sq.

68

CHAPTER 3

Are there in any school class children that have their birthdays in the same
month?
can be greatly embarrassed by the type

Are there two people in the world with the same numbers of hairs on their
heads?
or

Are there two match boxes with the same number of matches?

The analysis showed as the source of the difficulty not the cardinal aspect of
number but the constitution of the decisive sets. In the first example the relevant
sets are the children in a class and the months of the year, which are immediately
recognised. This immediateness is lacking in the other examples, although they
look isomorphic to the first one. Of course, the set of people and the set of
matchboxes are not problematic, whereas the other superficially viewed, the
set N has at first sight not enough relevant structure to be helpful. But in
order to be helpful an external structure must be imposed on N as the set of
possible numbers of hairs on heads or of possible numbers of matches in matchboxes if the drawer principle is expected to work. In problems where the
cardinal aspect of numbers plays a part, it can easily happen that by the data
of the problem N is externally structured and that recognising this structure is
the very thing that matters.

3.29. Formal Machinery for the Constitution of Relevant Structure

With the last example I have ventured beyond the frame set out in the beginning
of 3.19 to discuss structures that play a part in determining cardinalities and
relations between cardinalities. The drawer principle is instead a method for
gaining information about certain mappings from the inequality of cardinal
numbers. I now take up the thread from the end of Section 3.17, where I
promised to define the minimal formal machinery needed to constitute as
mental objects such sets as appeared in the paradigmatic examples I displayed.
Obviously this machinery will comprise much less than the means by which
I presented my examples, although even this representation could have been
formalised much more heavily. I could also have opted for less formalising.
Instead of N and u I could have said 20 and 4. It would have been extremely
difficult to avoid the letters X, A, B for sets; the presentation would probably
have become incomprehensible or cumbersome if I had tried it. I could have
avoided the symbol # though not the word cardinal number or something
like it. I could no more easily have renounced words like set, union,
product, and so on, because it was just their didactical place that was being
discussed. In teaching, however, and particularly in teaching young children,
one can at most aim at doing something with sets, unions, products, and so on,
rather than describing their status.

SETS

69

In Section 3.18 I gave and illustrated criteria that aimed at the cardinal
comparison of more or less clearly structured sets. The terminology used when

exercising this ability at pre-school age, is characterised by such words as as

many as, more, less. As many black as white, as many stars as rounds,
or more, or less. If not in kindergarten, then in the first grades, one can afford
to follow such a statement by the question Why?. After each black a white,
after each star a round this could be the answer, and the pupil could make
such statements more precise by the language of arrows. This, indeed, would
be a new symbolic means, enriching the pupils language paradigmatically.
It should be used to describe or delineate a visible mapping rather than as
happens in modern methods to construct an arbitrary, artificial, unnecessary
mapping. Is there any reason or need to introduce verbal constructions like sets
of whites, blacks, stars, rounds and to speak of mappings of such sets? I do not
think so. The set of all whites has the same number of elements (or the same
cardinal, or cardinality) as that of the blacks is a weighty expression by which
mathematical language is intentionally dissociated from the vernacular a
dissociation of language that at least suggests, if not instigates, or even puts the
seal on, a dissociation of realities.
The additivity of cardinal number in the union of disjoint sets is mathematically the principle that defines addition of cardinals, and is a posteriori a fact
to be stated and an expedient in counting. x chips and y chips together is
chips (with of course, definite numbers instead of x and y) is the minimal

linguistic means to verbalise this fact. In set language it is done by the formula

At the level of, say, primary school it is the only opportunity to use the symbol
(there are somewhat more opportunities for
Between the formulation in
the vernacular and its formalisation by means of the above formula many
intermediate stages can be distinguished, like if I form the union of disjoint

sets, their numbers are added. What good can be gained for understanding
addition by teaching a formalisation that transgresses the vernacular? Nothing,
as far as I can see. The didactical problem of the arithmetic operations is
besides teaching algorithms a problem of application, which I shall reconsider.
The pupil must learn, for instance, that an addition is required to know how
much money he had yesterday if he has this much now and spent that much
in the meantime, and that he should add 7 to his own age to know the age of
somebody 7 years his senior. Sets do not help much in this case. Even though
I can make a set of seven marbles, with seven florins or seven years this is an
artificial if not meaningless concern.
Compared with the addition formula there are better reasons for explicitation
of the multiplication formula

70

CHAPTER 3

This formula does not spring from a mere definition at least if multiplying
is generated as repeated adding. There is no doubt that sets of pairs can and

should be constituted as mental objects at an early age and that the product
formula can and must function. Should the formula be made explicit and, if
so, how? I say yes, and add in a paradigmatic way, though one paradigm

is not very likely to suffice.

From 4 boys and 3 girls I can form 4 3 boygirl pairs.
4 roads from A to B and 3 from B to C is 4 3 from A via B to C.
Should one go further? Replace 4 and 3 by letters? Replace this much ... by
the number of elements of the set of ...? Anyway, it would be less artificial
than with the addition formula since the members of the pair in the multiplication formula can indeed naturally be interpreted to form a set, at least, up to a
certain limit. The product formula leads via the rectangle model of the set of
pairs to the area formula for rectangles. As far as I know, even the most fanatical

advocates of sets did not dare to interpret length and width as sets of the number
of times a measuring unit could be laid down along the one side and the other,

which indeed is didactically absurd. Formalising the area expression stays, as

far as I can ascertain, always in the sphere

4 horizontal by 3 vertical pieces is 4 3 squares,

which is the same vernacular expression as in the earlier examples, or more

tightly formalised,
area = length times width,
which is a simplifying rather than complicating way and clearly lacks any association with sets.
In the case of equipartition mappings is there any need to go beyond vernacular formulations like
5 boxes with each 12 eggs is 5 12 eggs

or in the case of equipartition relations beyond

3 edges from each of 8 corners and 2 corners on each of 12 edges is 24
pairs

(of course with schematic illustrations)? Yes. In the case of equipartition relations it can pay one not to be satisfied with the sloppy motivation of thrice
counted but to understand and to have understood that behind this multiple
counting is the counting of related pairs.
3.30. Need for More Formalising
The sets most often discussed in the present chapter have the peculiarity that

they can be visualised (or otherwise embodied) in a way that is both more

SETS

71

honest and more subtle than Venn diagrams. It is exactly for this reason that
premature formalising can be dispensed with, at least if the examples where
the sets occurred had sufficient paradigmatic force.
When and in what measure can a need to formalise arise? I must anticipate
this question, which will be asked and answered later in a broader context. If
I ask it, I mean not only formalising but also transgressing the limits of the
vernacular by adaptations of it that, as seen from the vernacular, look like
jargon. The sets considered in the present chapter bear, or can be indicated by,
meaningful names, such as ludo board, paths finishing at the uth node.
Sometimes they were denoted by letters, but this happened because I had to deal
with a variety of such examples side by side and to distinguish them efficiently
and economically.
Formalising is a linguistic activity, and the need for formalising is in the first
instance a need for a better means of communication (communicating with
both others and oneself). In particular, far-reaching requirements are to be
fulfilled by the formalism as soon as oral, or often even ostensive, communication should give way to written and other kinds of communication that lack
ostensive means. However, not only do languages have a communicative function,
but they also serve as closed, more or less automatically functioning, systems.
In a more advanced stage of formalising, the need is felt to guarantee this automatism: formalising is made subservient to developing algorithms.
As far as the communicative aspect of formalising is concerned, one can
distinguish: formalisations of vocabulary, some of a simplifying, some of an
elucidating character, and formalisations of syntax, mostly of a systematising
character. An example of the first sort: the reinterpretation of the vernacular

quadrilateral and introduction of terms like square, rectangle, rhombus,

parallelogram, trapezoid. An example of the second sort: the convention that
squares are to be counted both among the rectangles and among the rhombuses.
Another example of the first sort: indicating definite objects by letters, and of
the second, using letters to indicate variables.
Historically letters for variables were first used in euclidean geometry, where
they were obviously needed for communication: as soon as geometrical constructions and reasonings are to be laid down in writing, ostensive means of
expression like this point and that point do not work any more; they are
replaced by conventional variables, indicated by letters. This situation becomes
more involved if the exposition is illustrated by a drawing: if the same letters
are placed at certain points on the paper they also have the appearance of
constant proper names of these points. On the other hand, this formalisation
in geometry has an algorithmic offshoot: a straight line through A and B does
not get a brand-new name but an algorithmic one, AB. But this does not lead
to developing any even weakly automatised algorithm.
A formalisation of great value and momentous consequences, which we are
so well acquainted with that we do not experience it as such any more, is the
creation of our familiar system of number notation. In contradistinction to the

72

CHAPTER 3

ancient Egyptian system and the so-called Roman numerals, which, as it were,
literally translate the numerals of the vernacular, the IndianArabic system is
a translation of the abacus language and was distinctly motivated historically
by the need to transfer the advantages of the abacus expression onto the powder
table or writing paper. The IndianArabian figures have become an inseparable
piece of our written language (beyond the frontiers of any particular written
language); nevertheless they have remained a foreign element in it, as appears
from the fact that they are taught in the arithmetic lesson rather than in the
language lesson. This is so for a good reason, because their flawless syntax is
very unlike that taught in the language lesson. Formalising the number language
by means of the IndianArabic notation is highly motivated by its algorithmic
sequel: IndianArabic arithmetic. Even the notation of operations has then
been formalised, both within the vernacular and by special symbols beyond
the vernacular.
Fixation of a spoken language in writing is quite another sort of formalising
it is a more primitive, less specialised activity and does not invite algorithmisation. It is certainly not by accident that teaching both formalisms starts at the

same age and that they are very much alike regarding their fertility in education.
This is a question which will be reconsidered later.
The function of formalising the number system is primarily communicative,
and only by the algorithms of column arithmetic is it made subservient to
guaranteeing the reliability of algorithms. Too often, this component is identified
with mathematics. Modern mathematics in the primary school is readily interpreted as formalising for the sake of algorithms. It is the traditional view that
at the age of obligatory instruction the child becomes susceptible to learning
formalisms (though not to learning formalising), but it is a mistake to believe
that this is also the moment of formalising for the benefit of automatising.
Innovators were not satisfied with the formalisms taught according to tradition
at the lower levels of the primary school. A large number of new formalisms
were invented for this stage of development, but their communicative function
is often doubtful, and the algorithmic one remains within the narrow limits
of little games. Up to now, only formalisms with a strongly communicative
character have proved successful at low levels. An example: the arrow language
for arithmetic operations and their inversions and for symbolising operations
on the number line. No one ever tried to go beyond this limit to teach the
denoting of mappings by letters at an early age. In general, to decide at which
developmental stage this would be feasible would depend on the total state of
formalising. The most natural, and experimentally tested, formalism are letters
for variables representing numbers or magnitudes, which possibly now starts
too late. With respect to sets, which were the subject of the present chapter,
I can discover a need for formalising and formalisms only at a quite advanced
stage.

CHAPTER 4

NATURAL NUMBERS

4.14.9. PHENOMENOLOGICAL

4.1. According to Peano

A great deal of what would be expected here, has been presented elsewhere
earlier.* In order to avoid useless repetition, I am going to summarise it within
the context of what has to be added.
The most usual introduction of natural numbers at an advanced level is
Peanos axiomatics:
A set N (natural numbers) with
(1)

an element

and

(2)

a one-to-one mapping f (successor of...) of N into N,

such that
(3)

(thus: each element has precisely one successor and up to 0 precisely one
predecessor by means of f) and

(4)
(thus: with the same f the set N is minimal with respect to (1) and (3)).
Instead of using subsets, (4) may also be formulated with predicates as the
principle of complete induction:

(4)

Let E be a property of natural numbers such that

E(0) and
for all

then

E(n) for all

If we let
then under the suppositions of (4)

* Mathematics as an Educational Task, Chapter XI.

73

74

CHAPTER 4

and because of

it follows that

and because of (4)

so
E(n) for all

Thus

and conversely it can be argued that

In N, operations can be introduced inductively, the addition by means of

for all n,
which defines generally the sum

(though the proof is not as obvious as it looks). Commutativity and associativity

of the addition are inductively proved. Similarly, the multiplication:

The order relation in N is no problem at all.

From here a bridge can be built to the finite cardinals: putting

one has to prove that indeed all these cardinals are differnt. These then are the
finite cardinals, and any set with such a cardinal is called finite. Moreover, one

can prove that for finite sets

which means that the inductively defined and the set-theoretically defined
operations coincide.

In Peanos axiomatics N accounts for the counting sequence, f for the counting
act, and (4) for the idea, not easily verbalised, that the counting act successively
exhausts the counting sequence. By the axiomatisation the counting sequence
has been frozen as it were; N is static; time and action look as though they are
eliminated.
On the algorithmic plane the counting sequence is, as it were, concretised by

means of the decimal notation: in

NATURAL NUMBERS

75

each may be one of the symbols 0, 1, 2, 3, 4, 5, 6, 7,8,9. The counting act

changes these symbols in a well-defined way. Infinity seems to be encapsulated,

but it again appears in the subscript k (the number of digits minus 1 if

which fortunately is of a lower order than n itself.

),

4.2. According to Euclid

The phenomenology at the background of Peanos axiomatics is the counting
process in the time flow; the numerals one, two, three, ... should properly be
read as first, second, third,.... Another phenomenology, which is more original,
is accounted for in Euclid VII. Before Peano, and even years later, it has influenced theoretical expositions on the number concept.
Euclid says
Unit is that after which a thing is called one.
Number is a set composed of units,

and still in the Encyklopdie der mathematischen Wissenschaften, founded and

edited by Felix Klein, we can read*
Counting things means considering them as of the same kind, taking them together and

assigning to them individually other things, which are also considered as of the same kind.

Each of these things to which by counting other things are assigned is called a unit, and
each of the things which by counting are assigned to the others, is called one. The result
of counting is called number. Because of their being of the same kind, as regards the units
and the ones, respectively, the number does not depend on the order according to which

the ones are assigned to the units. (Translation by the author.)

Hilbert in 1904** still tried to base number theory on a phenomenology of this

sort, and his later formalist approaches are to be understood from the same
phenomenological viewpoint.
From Euclid to Schubert a number consists of units (ones according to
Schubert) a mental object composed of simple objects of the same kind.
Number as a cardinal is explicitly mentioned by Euclid only in the case of the
one (though in fact the one is no number): by means of the unit one
can say of a thing that it is one thing. Euclid could have added that by means
of the two-ness, three-ness, ... two, three, ... things can be named. Euclid
* I A 1 (1898), H. Schubert, Grundlagen der Arithmetik. In his criticism of H. Schuberts
contribution (ber die Zahlen des Herrn H. Schubert, Jena, 1899), G. Frege stands on
somewhat firmer ground than in that of Huberts Grundlagen der Geometrie (Jahresbericht
DMV 12 (1903), pp. 319324, 368375), but as to insipidity both criticisms are well

matched. Freges lack of understanding both of phenomenology and of axiomatics has

proven symptomatic of the logistic attitude in general.
** Verhandlungen des III. Intemationalen Mathematikerkongresses in Heidelberg, 1904.
From 3rd to 7th ed., Appendix VI of Grundlagen der Geometrie.

76

CHAPTER 4

kept silent about the way this designation is to be established. Various authors

have filled in this gap with various formulations.* In order to count a heap of
things, one has to see them as a set of things of the same kind cows and
horses as a herd of animals, because after the numeral a noun is expected and
to these objects other objects, likewise all of the same kind, are assigned
counters on the beadframe, or chips, or tallies, or mental equivalents of them,
which together form numbers.
At the time that Schubert wrote his article, such a consideration was still the
normal one, though it should have been obsolete since in the course of the
nineteenth century complete induction had become the most striking property
of N and had been formulated as such by Peano. This then is the ordinal approach
to natural numbers, preferred today, which leads in the straightest way and with
a stress on what is essential for mathematics to the mathematical use of numbers
as arithmetical objects. It is a disadvantage of this approach that it does not
match a phenomenology where a number consists of units, and an even greater
disadvantage that numbers serving to count something and being (cardinal)
numbers of something come at the end of a long and arduous course of reasoning.

4.38. According to Cantor, Frege, Russell & Whitehead

4.3. Number as (cardinal) number of something is Cantors approach. Cantor
says:**
Potency or cardinal number of M we call the general concept which by means of our
active thinking power is created from a set M in the way that we abstract from the quality
of its different elements m and the order of their presentation.
Since each single element m, if its quality is disregarded, becomes a one, the cardinal
number
(of M) is itself a definite set composed of nothing but ones, which exists in our
mind as the mental image or projection of the given set M. (Translation by the author.)

This is followed by the definition of equivalence:

Two sets M and N are called equivalent, denoted by
or

if it is possible by a law to put them in a relation with each other, where to each element
of one of them corresponds one and only one of the other.

And later on:

that two sets M and N have the same cardinal number if and only if they are equivalent.

The if is motivated by the fact that whenever the cardinal number of M is

at stake, the position and the character of the elements of M does not matter
* For instance E. Schrder, Lehrbuch der Arithmetik und Algebra, Leipzig, 1873.

** Math. Annalen 46 (1897), pp. 481483 = Gesammelte Abhandlungen, pp. 282284.

77

NATURAL NUMBERS

and consequently the elements of M can be replaced with the corresponding

ones of N according to the correspondence. The only if is motivated in

an even more curious way,
that between the elements of M and the different ones of their cardinal number there
exists a mutually one-valued correspondence ... Thus we can say that
Likewise
So if
then

We see the mathematical construction entwined with a phenomenology that

continues Euclids: the cardinal number of M consists of units, albeit of more
than a finite set.

This cannot do any harm, one should say, because the only thing that matters
is the definition of equivalence. It suffices to know when sets are equipotent;
it does not matter what the cardinal number of a set is. This is intentional
concept building specialised on cardinal number. I assign the same cardinal
number to two sets as soon as they are equivalent.
4.4. Cardinals are also numbers in the sense that you can perform operations
on them defined as follows: the addition by means of the union, the multiplication by means of the set product:

This is for finite sets simpler than in Peanos approach, and the various

laws for these operations are much more easily proved.

Powers is another feature obtained by set operations

where

is defined as the set of mappings of

into

4.5. Notwithstanding the concession made in the last paragraph of Section

4.3, the question of what the cardinal number of a set is has been answered up

to a point. The answer, which stems from G. Frege and Russell & Whitehead,
is a paradigm for what is called extensional concept formation. All sets equivalent
with M must possess the same cardinal, thus cardinality of M is defined as the
thing common to them. But what is the thing common to them? Set theoretically
viewed it is the class of all sets equivalent with M. (A remark: the class of all
sets equivalent with M is an object that must be handled carefully, as appears
from various paradoxes; the term class instead of set is meant as a warning
sign.) At FregeRussells standpoint one is in a world far away from Eulid
Cantors phenomenology. For non-mathematicians this definition is so extraordinary that, for instance, Piaget read the FregeRussell definition as though
a cardinal number were a class of equivalent units.

78

CHAPTER 4

4.6. Here we are not interested in cardinals in general, but especially in number,
that is, finite cardinals.
Cantor* starts with a set
with one element and assigns to it the cardinal
1. He adds another element to get a set
to which he assigns the cardinal 2:
By adding fresh elements we get the series of sets
which in an unlimited sequence deliver successively the other, so-called finite, cardinals,
denoted by 3, 4, 5, . . . . The auxiliary use we made of the same numbers as subscripts is
justified by the fact that a number is not used in this sense until it has been defined as a
cardinal.

The last sentence shows that the finite numbers are not presupposed, but truly
defined.
All cardinals obtained by this process are called finite. The set of the finite
cardinals (N\{0}) has the (countably) infinite cardinal
Each infinite set
M (that is, whose cardinal is not finite) contains a countably infinite subset,
which arises as follows:rTake an element
of M; since M is infinite,
Hence there is an
with
, since M is infinite,
So it
goes on, to produce an infinite sequence
of different elements of M.
4.7. Cantors presentation of natural numbers is as naive as Euclids. The only
difference is that the numbers are not statically present as in Euclid. They are
produced in the course of time, and counting out a countably infinite subset

from an infinite set is also done in time. Cantors new idea is not his presentation
of natural number but the first clear explanation of what it means for a number
to be the number of a set of things.
On the first point, Frege has explored a less naive approach: He took all sets
of cardinals such that
contains the 1
with each cardinal

contains the one that is 1 larger,

and of all these sets he took the intersection thus the smallest and this was

the set of finite cardinals. This method to describe in one blow infinite constructions previously projected in time has since become paradigmatic. At the end
of his work Frege confesses that his plan became unsettled by paradoxes that
had been discovered in the meantime.
Russell & Whitehead proceeded more carefully: At this point they do not
consider sets of cardinals. They define sets according to Freges prescription,
but their elements are finite sets rather than cardinals:
contains the empty set,
contains with any set X the set
* Math. Annalen 46 (1897), pp. 214215 = Gesammelte Abhandlungen, pp. 289290.

NATURAL NUMBERS

79

and among all these sets they determine the smallest. The example sets for the
finite cardinals then are

It looks, as a satirist observed, like a library with

in the first volume, nothing

in the second, the table of contents of the first,
in the third, the table of contents of the second,
and so on .

These are the units from which the Russell & Whitehead numbers are composed.
The existence of this or some infinite set must explicitly be postulated
according to Russell & Whitehead.
4.8. Another point where Cantor proceeded naively was in counting out a
countably infinite subset from an infinite set. In order to describe this process in
one blow, one imagines that assigned to any non-empty subset X of M is one of

its element
The desired countably infinite subset A of M consists of
consecutively the elements
More precisely, by induction

and finally

Or, without explicit induction,

and
if

then

and among these sets take the smallest (the intersection of all of them).
The indispensable instrument for this definition is the function
which

to every non-empty part X of M assigns an element of X a function whose

existence is obtained by the so-called axiom of choice.
4.9. According to Dedekind

A countably infinite set can be mapped one-to-one on a proper subset; for

instance, the A is mapped on
just obtained by an f such that
Thus every infinite set can be mapped one-to-one upon a proper subset

80

CHAPTER 4

the M which we have just considered, by mapping A according to f and the

remainder identically.
This characteristic property of infinite sets, that of being equivalent with
certain proper subsets, was R. Dedekinds* starting point for a quite original
theory of finite sets.
In the theories we have summarised up to now the finite cardinalities were
inductively generated, though finally the inductivity became somehow encapsulated. In Dedekinds method the induction is self-generated. Dedekind
defines
a set is infinite if it is equivalent with a proper subset,
and of course

a set is finite if it is not infinite.

He starts with some infinite set M his proof that such a set exists is not
convicing.
Since M is infinite, there is a one-to-one mapping f of M on a proper subset.
Let
A subset N of M is defined by

(1)
(2)

(3)

N minimal with respect to (1) and (2).

Substituting N for N, a for 0, and f for the successor mapping, one gets exactly
the system of properties that defines N. Moreoever without the axiom of
choice one has succeeded in finding in any infinite M a subset equivalent
with N, thus a countably infinite subset.

4.1011. Tested DidacticallyPhenomenologically

4.10. If all or part of the preceding analysis is considered too high level and
consequently out of place, I would object that it is subject matter found in
some modern textbooks, for instance, for the 6th grade. There it looks like a
closed system, though in fact it is, as it were, tiptoeing in wooden shoes, displaying subtleties in a rudely defective context. I have discussed this elsewhere.**
It will have become clear from my exposition that with each profundity we
get further away from the phenomenology of number as it is naively experienced.
The definitions quoted from Euclid interpret a directly-from-the-abacus abstracted
consciousness about numbers. They are, however, not operational in the number
theory to which they are meant to be the introduction. Complete induction
was repeatedly applied from antiquity onwards, sometimes even in a profound
way, but not until Pascal was it formulated. Complete induction turned up in
* Was sind und was sollen die Zahlen? 1887, 1893.
** Mathematics as an Educational Task, Chapter XI.

NATURAL NUMBERS

81

ever more examples, and was ever more consciously used, but not until two and

a half centuries after Pascal did the key position of complete induction for the
concept of natural number become clear. Likewise it became clear that complete
induction could serve to define the arithmetical operations and prove their
properties. The most modest interpretation of Peanos axiomatics is to consider
it as descriptive. Anyway, it is operational, wherever induction plays a part in
mathematics. A step further is to ascribe a definitory character to it, that is,
defining what natural numbers are; then Peanos axiomatics is only operational
as long as one moves within the foundations of mathematics.
As far as Cantors definition of cardinal numbers describes number as numerosity and draws attention to its invariance under one-to-one mappings, it is

a phenomenology of the naively experienced number; if it is meant as defining

number, it belongs to the foundations of mathematics; if, moreover, Freges
and Russell & Whiteheads formulations are used, it is research in foundations
by methods that have arisen from a profound criticism in the foundations of
mathematics. Dedekinds definition of infinite set, produced by a clever change
of perspective, presupposes a well-stuffed arsenal of mathematical strategies.
The definitions of addition and multiplication by complete induction can
be seen as an adequate phenomenological description of the naive process of
adding by counting forth and of the naive process of multiplication by repeated

adding (producing multiplication tables); interpreted as definitions, they belong

to the foundations of mathematics, and the same holds for the laws of addition
and multiplication, whereas, when pronounced for numerosity rather than
for counting numbers, they are obvious the need for proofs arises only in
systematising the counting number approach by Peanos axiomatics.

In the numerosity approach the definition of addition describes adequately

what happens if quantities are taken together, though it is expressed at a linguistically high level; in order to be practically operational, it must be completed
with the counting-forth definition. The definition of multiplication in the

numerosity approach adequately accounts for the rectangle model of multiplying, which is an indispensable complement of multiplication defined by repeated
addition. For the power definition in the numerosity approach (cf. Section
4.4) traditional school mathematics does not possess an analogue, though
creating one would be feasible and worth paying attention to.
4.11. Numbers, counting, and arithmetical operations are first of all a means

to organise phenomena where quantities play a part. All theories of natural

numbers are rooted in these means of organisation. But all theories go beyond
that. Mathematics is characterised by a tendency which I have called anontologisation: cutting the bonds with reality. This tendency is entirely justified. It is,
however, the result of historical and individual development and cannot be
supposed to be innate to the learners mind. Even less can the learner be presumed
to be susceptible to anontologised mathematics. Attempts at instilling it lead
to false concretisations.

82

CHAPTER 4

Auxiliary tools in the process of anontologising are: change of perspective

from description to definition, where observed properties of mathematical
objects are used to define the objects in order to detach them from their origin;
and the replacement of intentional by extensional concept formation. Neither
tool is simply present in the mind. They require development and exercise.
If a pupil who can identify rhombuses at glance isolates from the multitude
of properties of the rhombus a few characteristic ones, which bring about
a formal definition of rhombus, he performs a local change of perspective and
a first step on the way to anontologising mathematics. With number, the same
step is much longer because number is more familiar to him and the necessary
change of perspective would be more global in character. Even more difficult
is extensional concept formation by equivalence, which I paid attention to
elsewhere*. This restricts to narrow domains teaching theories of natural number.
4.12. Phenomenologically Too Low a Level
One current in the nineteenth century, which I have neglected up to now, was
characterised by a superficial kind of formalism: the natural numbers considered
as symbols, sequences of digits, with operations performed according to merely
conventional rules. In a sense it is not as mad a phenomenology as one would

think. Rather it is to the point as far as it describes a kind of instruction in

arithmetic and the attitude fostered by this instruction: an attitude of viewing
numbers as symbols and doing arithmetic according to conventional rules.
Nevertheless, this phenomenology is unsatisfactory. It does not account
for the relatedness of numbers and arithmetic with reality, where numbers
and numerical operations have a meaningful counterpart. Moreover, this
phenomenology does not carry farther than understanding the most primitive
arithmetic. The activity described by this phenomenology is that of a mechanical
(or electronic) calculator rather than that of a human calculator who can handle
the instrument and is obliged to understand at a higher level the aim of the
numbers and operations. It cannot be denied that in the nineteenth, and even
for many years into the twentieth century there was a need for human calculators
who experienced the numbers and the operations at this primitive level, and
this need justified socially a corresponding kind of instruction. With the rise
of better perfected calculating machinery, and its mass production, the need
vanished for people educated according to those principles. If numbers and
numerical operations are taught today, the instruction can no longer amount
to programming a computer, to which numbers and operations are indeed
meaningless and conventional.
The primitive formalism of the nineteenth century has given away in the
twentieth century to a more profoundly understood formalism, by which not
only elementary arithmetic but the whole of mathematics is interpreted as a
* Mathematics as an Educational Task, pp. 3132, 213214.

NATURAL NUMBERS

83

language with a conventional syntax. However, the mathematician does not

then move within this language but reflects on it reflections that can be
syntactical or semantical. The same holds for all kinds of formal languages which
are extensively studied today; they are spoken only by and with computers.

4.13. Developmental Phenomena

From the concept of natural number, which has occupied us up to the previous
section I now turn to the mental object or objects that can play a part in acquir-

ing the concept. I use the plural mental objects because I must distinguish
at least between particular natural numbers and the natural number as mental
objects, but certainly also between numerosity numbers and counting numbers
as well as between numerosity number and counting number. For acquisition
of the concept natural number even more is required: constituting certain
relational patterns between natural numbers.
Developmental facts about the constitution of numbers and number are, as
far as I know, scanty. Small cardinals seem to be constituted early as people
assert, 2, 3, 4 at the ages 2, 3, 4, respectively. In Section 3.18, I asked what
part structures play in this. One could add many more questions: to what degree
the numbers aimed at are bound to certain objects or kinds of objects or to
certain representations, whether and how the constituted numbers are expressed
verbally or enactively, what in a certain stage of restricted constitution impedes
or stimulates the constitution. According to my experience the continuation,
at least as far as verbalising is concerned, has the character of differentiation:
a numeral used in the sense of many starts playing a more specific part, in
order perhaps to be replaced by another that takes over the meaning of many.
The majority of children learn counting before constituting cardinal number;
counting is reciting numerals, initially in an arbitrary order, later in the right
order though with gaps, finally without gaps. In this period they also learn
counting something; it is arbitrarily pointing to the objects, before it becomes
a systematic procedure; first graders can still have difficulties with one-to-one
exhausting even if they know that this is the way to do it. The connection
between counting and cardinal number can be lacking in spite of verbalising.
I know of only one case of a child that started counting only after the complete
constitution of the numerosity number (cf. 3.18).

4.14. Conservation
With a view to this lack of factual data on the development there is no way out
except the phenomenological method. In fact, Piaget took it and, influenced
by mathematical ideas, chose conservation as a criterion of what I have called
constitution of number. For the cardinal number as defined by Cantor, conservation or in mathematical terms invariance with respect to certain transformations is indeed an important characteristic.

84

CHAPTER 4

I have already discussed conservation critically (Section 1.25). The term

itself is confusing because it does not indicate the transformation with respect
to which in variance is postulated. In the case of length it could be, according
to my enumeration, congruence transformations, flexions, and break-and-make
transformations or permutations of parts. Numerosity looks simpler: everything

that has to do with transformations can be brought under the concept of oneto-one mapping. It can, indeed, but this does not mean that it must. Many ideas
reduced to one when logically possible can refer to ideas that belong to different
abilities and to abilities that are being learned under different circumstances.
Moreover, we noticed already in Sections 3.1827 that the constitution of
cardinals may depend on structures and isomorphisms, rather than on sets and
one-to-one mappings.
Up to the present day nobody can say whether the stress Piaget put on
conservation was justified, whether indeed certain invariances characterise
the constitution of certain mathematical objects, whether this might be true
of all mathematical objects for which Piaget developed conservation criteria,
for some of them, or for none. I guess that in principle Piaget chose the right
way, but I believe he deserved to be followed more critically on this way in
every detail than he has been in fact.
Another doubtful element in Piagets method, and a matter of principle, is
the use of temporal cross-sections instead of longitudinal observation (which
he rarely performed) The unavoidable drawback of that method is that the
observed conservation phenomena are possibly only criteria of the constitution
of mathematical objects rather than developmental phenomena on the way to
it.
4.15. Invariances With a Single Set
With all these provisos I am going to follow Piaget in the phenomenological

search for invariances that might matter in the process of constituting the
numerosity number. I first mention four of them, all related to one single set
rather than to comparing two sets:
In variance
over time,
under change of standpoint,
under shake transformations,
under break-and-make transformations.

Some comments on these terms, in particular the first:

A child (4; 8) counts her fingers. Five. And how many did you have yesterday? I have
forgotten. Another child (4; 6) laughs at this question.

This example shows that the invariance over time is not at all self-evident. This
insight is acquired earlier or later by different children. It is no logical fact.

NATURAL NUMBERS

85

Indeed, the invariance depends on conditions. The child knows circumstances

under which something becomes more or, less plants that germinate, flowers
that drop, families that increase or decrease, drops that split and unite.
If something is added, it becomes more,
if something is taken away, it becomes less,
if nothing happens, it remains equal,

these are principles for all kinds of magnitudes, discrete and continuous. The
only problem is to know whether nothing happened. This knowledge depends
on a vast body of factual experience. Adults are hardly able to imagine that
and how they acquired this knowledge in their own development.
Invariance under change of standpoint is a similar experience. A set moved to
another place and observed there, or observed in the same place by another
observer, must possess the same number. This too is not self-evident. It can
mean a discovery to look for a cause for why under certain circumstances a
set observed from another standpoint can be more or less.
The shake transformations again are concerned with one set: cookies on a
plate can be moved, flowers in a vase rearranged, sheep in a flock run and
at least mental continuity ties the initial to the final stage and guarantees the
invariance. Like the invariance over time and under change of standpoint,
invariance under shake transformations has never properly been tested with
children in Piaget-like experiments as far as I know. First of all, in the classical
test for conservation of number the relevant shake transformation is obscured
by a built-in misleading cue: the discrete set of objects is presented in a such
way that by the very fact of presentation not only the number is defined, but
also the length, which is intentionally not kept invariant. Moreover, the shaken
set is compared with an unshaken one, which produces a new difficulty.
In the cases of duration, change of standpoint, and shaking, one could maintain that they concern conservation of set rather than of number: the set of
fingers, cookies, animals remains the same over time, under change of standpoint,
after shaking, and so does of course its number as a property of the set.* I
have no objection against seeing it this way. I do not believe it matters much.
The case of the break-make transformation, however, is different. It is breaking
and rebuilding a set. The reason why I consider these transformations separately
is similar to the reason I gave in the case of length. I should add that it was
in considering length that I first became aware of this kind of transformations
and their importance.
Breaking divides a set into two or more whether the pieces are taken apart or
merely separated by true or symbolic walls. Remaking brings them together or
removes the separation. Meanwhile the objects do not cease to belong together,
* I could have considered these invariances also with respect to lengths; then the invariance
of set would have had as its analogue the invariance of long object.

86

CHAPTER 4

which can be expressed in the intermediate question, Are there still as many
together?
Breakmake transformations seem to me to be the most effective means
of developing in variance as a general feature, for number as well as for length.
The learning process could take the following course: A small number of objects
is removed from the set in order to be added back again possibly at another
place. This procedure is continually repeated, while the question is asked, Are
there now more, or less, or the same?
If information on the development of conservation of number is wanted,
these invariance phenomena should be tested separately: How many were there
yesterday? How many are they if you look that way? How many are they
now? (after shaking) How many are they now together? (after breaking).
A child that is not familiar with these invariances will not be able to answer
immediately; he will rather try counting anew.
A child that cannot yet count sets could be asked the same questions in the
version Are there as many? instead of How many are there? Or one should
ask him to estimate the number before and after.

4.16. The Counting Number

It has often been noticed that many children count before having constituted
number as a mental object. Let us distinguish
counting, that is, reciting the number sequence;
counting something, that is, in the counting process connecting the
numerals with the set that is counted out or produced;
interpreting after counting the counting result as number of the counted
or produced set.
Counting may or may not be accompanied by insight into the structure of the
counting system.
Various aspects of this insight are tested with questions like

what follows 3?
what follows 23?
what follows 9?
what follows 29?
what follows 99?
what precedes 4?
what precedes 24?
what precedes 10?
what precedes 30?
what precedes 100?
which is earlier, 6 or 9?
which is earlier, 26 or 29?
which is earlier, 29 or 36?

NATURAL NUMBERS

87

Children who lack this insight count silently anew from 1 up to the desired
place, as do many adults, if the order in the less structured alphabet is at
stake.

Insight into the structure of the system of numerals is made easier or acquired
by

reading and writing numbers,

producing the number sequence by writing.
This ability does not depend on counting something and on the relation to
numerosity.
Counting by tens, by hundreds
allows one to get on quicker it plays a part in some games. House numbers
suggest
counting by two.

A difficult job is

counting back
if it is not supported by acquaintance with the written image. Likewise
counting back or forth a given number of steps from a given number

does not in principle depend on counting something and on the relation with
numerosity, though this counting activity will usually be motivated, or be
accompanied by counting something.

The objects to be counted can be

visible, palpable, audible, kinesthetic, rhythmic,
all of these combined,
movable, fixed,
demonstrable, observable, mental.

The counting activity can be accompanied by movements of

fingers, eyes, hands, feet, or other limbs.

The counted objects can be

indicated, marked, separated.

The numbers can be firmly attached to the objects to be counted by

mapping the number sequence on the number line
and
operations on the number line.

Then it means

88

CHAPTER 4

counting steps on the number line,

and

combining sets of steps into jumps on the number line.

Illustrative are the

simultaneous* counting of steps taken

simultaneous* counting of steps on stairs,
which can be

real or symbolised,
in general,

simultaneous* counting of rhythmic events.

A usual mistake in counting steps (taken, or of stairs) is taking the zeroeth
for one. Possibly
in counting rhythmic events, counting something
arises or gets experienced: rhythmic movements of the forefinger. Whereas

correct simultaneous* counting of a rhythmic event (for instance the

rhythmic counting process) is a matter of coordination,
correct counting out is a matter of organisation,
that is,

organisation by setting apart or marking the counted objects,

or if the objects to be counted cannot be manipulated,
organisation by means of present structures,
or by structuring.

By this organisation
skipping and multiple counting

must be prevented.
It is a well-known experience with children at a certain stage that, when
asked how many? they count, say, 1, 2, 3,4, 5, without answering the question properly, and count anew if the question is repeated. So they do not
consider the counting result as that characteristic of the set which we call its
number. The step to this can perhaps be provoked by replacing the question
how many are they?

* That is, simultaneously with the acts performed.

NATURAL NUMBERS

89

with the task

give me ... beads
from a set of beads. If now the child reacts by counting, he is compelled to
interpret the counting result as the number of beads. Continuing this line, one
can charge the child with
picking m beads out of a set of n beads, with
and in this way
according to the feasibility try to have the cardinal order relation constituted.

The strongest stimulus to constituting numerosity number in this context is

structural isomorphisms: if I have successfully counted the eyes of six persons,
I do not need to count anew their ears. The insight is created that
isomorphically structured sets possess isomorphic counting structures,
hence
isomorphically structured sets lead to the same counting result,
counting a part leads to a smaller counting result.

This insight can be generalised while abstracting from counting: without counting, the child draws the conclusion that a group of people have as many ears
as eyes. This leads to the insight that

isomorphic structures have already potentially isomorphic counting

structures
and to the possibility of asserting that

on account of structural isomorphism of sets, they have as many elements,

on account of inclusion one set has fewer elements.
If three sets are considered where the first is isomorphic with the second by
another kind of structure than with the third, it can still be concluded that the
second and the third have equally many elements. It can by replacing structures
of a different kind by counting structures, thus of the same kind. From here the
use of transitivity of equally many and less than
can develop.
By this course one can imagine the numerosity number constituted as a
mental object: starting from the counting number and eliminating it while
preserving its possibility.

90

CHAPTER 4

4.17. The Structure of the Counting System

The mathematical structure of the counting system is the successor relation with
well-known properties that allow induction and, consequently, the defintion of
N. But in genetic and didactic phenomenology the counting system is endowed
with a richer structure, from which N is derived by a theoretic impoverishment.
First of all, genetically: Acquiring numbers is, in general, preceded by acquiring numerals, which as a system is with a few exceptions at the start
well-structured by tens, hundreds, thousands, and so on, and most often in its
acquisition is integrated with the total vocabulary. But however this process of
acquisition may run, there can be no doubt that acquiring the mathematical
structure of N is possible only by way of acquiring the stronger structure with
which N is endowed, the decimal system from counting forwards and backwards to the numerical operations. In true mathematics the decimal structure
of N does not count and, as is easily understood, for good reason. There are
no good reasons, however, why no attention is being paid in developmental
research, as far as I know, to the decimal structure. There is a strong tendency
to read the developmental and sometimes even the didactical phases from
the structure of mathematical science, and by preference from one particular
structure of science, Bourbakis system. Some developmental psychologists,
such as Piaget, have even raised this parallelism to the rank of a principle. With
regard to the organisation of mathematical subject matter, this view has been
expressed, if not earlier, then anyway with new aplomb, at Royaumont (1959)
and Dubrovnik (1960). For very good reasons the decimal structure of N does
not appear in any scientific structure of mathematics except as a curiosity that
is just mentioned. In such genetic or didactic research as is steered or even
dominated by the idea of structure of science, there is no inducement to pay
attention to the stronger structure of N. This explains the lack of interest I
alluded to earlier. Yet whoever is influenced by genetical or didactical phenomenology feels the stronger structure of N as a conditio sine qua non and as
a new argument against the idea that spontaneous and guided development are
determined by the structure of science.
One additional remark: Even if the decimal structure is rejected as irrelevant
within the scientific structure of mathematics, one cannot bypass the fact that
the principle of building fixed numbers of units to introduce higher units should
certainly fit somewhere into the scientific structure (cf. equipartition by mappings and relations in 3.22 and 3.25). It is rather the restriction to one system
of bundling that looks unmathematical. Yet even structuring by bundles is in
general neglected in developmental research.
In the decimal structure of N two components can be distinguished,

the (decimal) bundling,

the positional arrangement of the bundles.
Both have been known from olden times to all civilisations, the oldest included

NATURAL NUMBERS

91

the SumerianAkkadian and the Egyptian. The sexagesimality of cuneiform

texts was a sophistication of professional arithmetic, artificially imposed upon
an originally decimal system, but anyway it included the positional idea. In
Egyptian arithmetic we find the decimal bundling, but the positional idea
seems lacking. There can be no doubt, however, that operations on numbers,
in particular on big ones, were performed from olden times on some kind of
abacus; and however an abacus is constructed, it certainly embodies the positional idea. On cuneiform tablets the numbers were also positionally written.
It is the decimal digit system that is of more recent date, but it was preceded by
a decimal abacus, which was gradually superseded by the decimal digit system
when writing material became widely available. Unfortunately one would add
from a didactical point of view.
For the mathematical layman the decimal counting structure is an indispen
sable element of the counting system. I doubt whether the man in the street
ever disregards the decimal structure, being an inessential element of the counting system not until true mathematics is at issue is there any need for such an
elimination. Quite a few know that other counting systems will also do and that,
for instance, computers work in the binary system. This knowledge, however,
is most often not deeply rooted or it is directed in the wrong way: the impression
can arise that some number system in whatever base is required; that such a
supplementary structure as decimality is indispensable. The large part decimality
plays in arithmetic puzzles indicates another mentality than is appreciated in
mathematics.
From a practical rather than theoretic perspective the extra structure of N,
decimality, is valuable and even indispensable. This holds for the didactical
phenomenological perspective too. The extra structure is a powerful means to
actually master the bleak successor structure of N. One comes to grips with
the counting system by decimalising it, or at least one imagines that one does;
and with regard to mere routine, it is certainly true. Performing the arithmetical
operations and estimating orders of magnitudes is organised by means of
the decimality. Not until mathematics is carried on to a more advanced level,
do the idea and the need arise to eliminate the particular number system from
the phenomenal structure of the counting system in order to get the truly
mathematical N.
It is well known how in counting large quantities of objects of the same
kind, decimality is concretised: objects are laid down in rows of ten, which are
combined in squares of ten by ten, to systems of tens of such squares, and so
on. Or towers of ten or one hundred such objects are built, for instance, when
counting money. This principle is also systematised by such material as blocks,
rods of ten blocks, plates of ten rods, blocks of ten plates.
Such material concretises the bundling only it neglects the positional order.
This latter component is fully embodied by the abacus, an old instrument
rediscovered in Western Europe. Whereas this kind of instrument maintained
itself in Russia and the Far East, it disappeared in the Western world after the

92

CHAPTER 4

rise of IndianArabian arithmeticstrangely enough also in instruction, which

relied on the counting frame with 100 beads, each of the same value. As a
didactic tool the abacus enjoys an increasing popularity. If I am not mistaken,
its reintroduction was due to Maria Montessori.
Unlike the Eastern abacus with its intermediate units of 5, 50, and so on,
our instructional abacus is purely decimal, as is the Russian one. In the Russian
abacus the units, tens, hundreds, ... follow up each other in the vertical
direction. Our instructional abacus, however, follows the order of the written
positional system. It is fabricated with 10 or 20 beads on a line or a bow; if
it is 20, then 10 are of one colour and 10 are of another. Another variant is the
pictorial abacus which allows for an arbitrary number of chips between vertical
lines. On such abaci numbers admit of a multiplicity of representations; in the
standard representation, corresponding to the digital numeral, the number of
beads or chips used is minimal.
A variant of the decimal abacus is the so-called minicomputer, where in each
position the numbers from 0 to 10 are represented according to the binary
system.

After these technical details on the abacus and before dealing with its use
in arithmetical practice, I will briefly tackle a question of principle. When
I mentioned how calculating on the abacus was superseded by written column
arithmetic, I heaved the sigh unfortunately. Why unfortunately? Is it not

a blessing we would be silly to waive writing numbers as one likes it, beside and
below each other when compared with working in the shackles of the abacus,
however flexibly it might be constructed and used?
Yes, it is a blessing for those who can afford it. Writing the digits neatly
below each other is the precondition for the functioning of the positional
system, and by neatly I mean not only a calligraphic fact but mentally observing the positional idea. And this holds for decimal fractions, too. A pupil who
calculates excellently in N and gets into difficulties with decimal fractions proves
that he has not yet grasped the essentials of the positional system.
The abacus, in whatever form, compels the learner to reflect again and
again not only on what units, tens, hundreds, ... are, but also on what tenths,
hundredths, ... are, and how the ones arise from the others. It looks like a
shackle calculating on the abacus. Indeed it is for those who have outgrown
it. But the chance to outgrow it should be granted to anyone who has no other
chance, who cannot do without it, who is otherwise left with the choice: number
chaos.

4.18. Invariances With Two or More Sets

In Section 4.15 aspects of conservation were discussed with regard to a single
set. Numerosity, however, does not become an essential characteristic and
instrument until more sets are compared.
Recall the enumeration in 3.18 of activities which together might represent
the constitution of numerosity number:

NATURAL NUMBERS

93

Eliminating the structuring component from structures with the same

substratum in order to arrive at substratum sets,
and in this connection transforming the inclusion relation into an order
relation (less instead of contained in).
Using isomorphisms of structures with different substratum to compare
different sets.
Using transitivity of equality and order (of numerosity number).
The structures meant here can be present in the material in order to be
discovered and used afterwards, or they can also be imposed intentionally by
the user. The structure most usually imposed in this context on unstructured
or insufficiently structured sets is the counting structure, though as the examples

of Sections 3.1830 show it is certainly not the only one, and often it is operational only through a one-, two-, or three-dimensional order structure or some
other structure which is needed to make systematic counting feasible.
In Section 4.16 I sketched how the elimination of the counting structure
might develop; the elimination could be triggered by other, more striking structures that supersede the counting structure as an example I took the relation
between numbers of eyes and ears when determining them for a group of people.
My phenomenologjcal sketch of the constitution of numerosity number via
the counting number allows for variants and short cuts. In principle it is possible,

and though it seems exceptional, in fact it happens, that numerosity number

is not constituted via counting number but precedes it.
In both cases the invariance of the numerosity number under one-to-one

mappings, though mathematically the most prominent and even constituting

property of the concept of cardinality, is an a posteriori phenomenon if compared
with the invariance under isomorphism of structures, and the same holds for
invariance of the order as established by the cardinals. Constitution of those
invariances requires that structures be impoverished to sets and isomorphisms
to one-to-one mappings. The invariance principles to be acquired are: if the
mapping f of A in B is

one-to-one, onto
one-to-one, not onto,
onto, not one-to-one,

then
then
then

and conversely,
if
if

if
if

and f one-to-one, then f upon,

and f upon, then f one-to-one,

then f not onto

then f not one-to-one.

Modern textbooks often teach and test such properties by means of pairs of
Venn diagrams. Such exercises are false concretisations which are likely to

block the mental constitution of these properties, or at least to thwart their

94

CHAPTER 4

application. Recognising relevant sets and mappings is the very thing that matters
in applications. I have concluded this from experiments with the drawer (or
Dirichlet) principle, which, indeed, is an efficient test of the last property I
mentioned above.

4.19. Comparing Cardinals Numbers by Estimate

Earlier I indicated the part played in the constitution of cardinals by indefinite

numerals such as many and few. As far as they express rough estimates,
more and less are of the same character, if they mean many more and
much less.
The comparison by estimate is based on a most often unconsciously handled
pattern, which shall be made explicit with a view to didactics: Two sets of
objects, A and B, are compared

while A, at least as densely distributed as B,

covers a larger space than B,
or

while A, more densely distributed than B,

covers the same space as B,

A conflict arises if
A, more densely distributed than B,

covers less space than B.

When applying this pattern,
the spaces covered by the sets are globally, at sight, compared
whereas

the densities are locally compared

and

the homogeneity of the distribution is globally judged.

A special case is the following:
A and B are densely packed within equal spaces
whereas the particular objects of A cover less space than those of B.
A variant:

The spaces covered by A and B are replaced sets A' and B' ,
respectively, which are known to have the same number of elements
or the one is known to have more than the other.

Another variant:

N A T U R A L NUMBERS

95

The spaces are replaced with time intervals

and the elements of A and B are events.

In general, the principle explained here can be formulated as follows:

Consider a finite set X with all its subsets. X is internally or externally structured in such a way that subsets apart from their number
possess a certain

character k which can be described extensively; that is, by means of a magnitude

that behaves additively under composition:

This is called a k-homogeneous set.

Indeed if k is thought to be a

volume (area, length),

weight,
duration,
thus k(Y) the
volume (area, length) covered by Y,
weight of the combined members of Y,

duration of the combined events of Y,

then
is the number of members of Y per unit of

volume (area, length),

weight,
time,

independently of

which justifies the expression k-homogeneous set.

is what was called

density
in the introductory examples. Here, too, we can speak of a distribution density
with respect to k:

Let A and B now be two k-homogeneous sets. The principle we wish to

formulate, is

96

CHAPTER 4

and the conflict is expressed by

and

The k-inequalities in the premisses can globally be tested, the

locally and perhaps by samples.
The assertions based on applying this principle can be strengthened, such as
more than the double,
more than half,

and so on, if the estimations of

allow it an extension we shall consider
when dealing with ratio and proportion (Chapter VI).
4.2028. The Constitution of Addition as a Mental Act
4.20. For many years it was a habit to indicate the operation of set union by
a plus-sign thus
instead of
and even now there are authors, in
particular in measure theory, who stick to that notation. It happened first in

topological algebra that one abandoned the old notation: in additive structures
one needed
to indicate the set of
with
and
The plus sign for the union formation was of course inspired by the close
connection between union and sum, witnessed by
and in fact used to define addition. If m and n are to be added, one provides
oneself with two disjunct sets A and B of which m and n are the cardinal numbers

forms their union and puts

It is easily shown that the result does not depend on the choice of A and B,
that is, if
then
At the lowest level this then is the way to perform addition; the learner who has
to add m and n creates the required sets of fingers, beads, strokes or whatever.
But it can happen as well that the sets are given in advance: he is asked to
add not m and n, but the numbers of objects presented in some way, say
5 cars and 3 cars together,

where the cars can be real cars, five on this side of the road and three on the

NATURAL NUMBERS

97

other side, or drawn, or suggested by a story, or some of the cars may be realised
in one way and the others in another. Well, if their reality is not palpable enough,
the calculator can provide himself with substitutes A' and B', equivalent to the
sets A and B, which he is able to take together; A' and B' can be sets of fingers,
or beads, or strokes. In all these cases the numbers to be added are clearly
recognisable as cardinals of sets, and their addition as reflecting the union
operation, even though the sets themselves may be inaccessible, and uniting

them cannot actually be performed.

4.21. Problems do not arise until

the terms to be added are not plainly recognisable as cardinal numbers of

sets
or

the addition is not plainly recognisable as reflecting the union operation of

sets.
In

5 marbles and 3 marbles

the related sets and the operation are clearly recognisable. They are less so in
the case

John has 5 marbles, while Pete has 3 more, how many does he have?

Petes set is not obtained by taking two given sets together. One should rather
consider the imaginary set of Petes marbles split into two sets, one set of
marbles equivalent with Johns and another set of the 3 marbles he has more
than John, which can be done in many ways. The formula
is again being applied, though with
prescribed and to be split into A and
B.
Another type of less clear recognisability of the related sets is
John has 5 marbles, yesterday he had 3 more, how many did he have?
Here a lost set must be mentally reconstructed in order to be united with a
present one.
In
John won three marbles today, how many more does he have than he had
yesterday?

nothing is really to be added, though an addition is suggested by won and

more, and the situation is even more troubled by the foggy set of Johns
marbles yesterday. Though looking more complicated,

98

CHAPTER 4

Yesterday John won 5 marbles and today 3 marbles, how many more

does he have than he had yesterday?

is probably an easier case, since one has really things to add. But what about
replacing won with lost and more with fewer?
This kind of example could be extended with many more where the sets
are somehow recognisable although the operation is harder to identify in
particular, as a consequence of misleading linguistic cues, such as in
John has lost three marbles, he still has 5, how many did he have originally?
4.22. This stock of examples can be multiplied ad lib., but we restrict ourself

to this choice in order to signal another type of addition problems in which it

can be properly said that no sets are distinguishable, and constructing sets from
the objects occurring in the problem is perhaps less desirable. In
5 steps (of stairs) and 3 steps,

5 days and 3 days,

5 km and 3 km,

5 florins and 3 florins,

5 times and 3 times,

one can hardly speak of sets consisting of 5 and 3 elements, respectively. Whereas
in the former group of examples the sets are still constructible in the domain of
the objects, in the latter it is hard, if not impossible. Images that are to represent
5 or 3 marbles can be quite realistic, for instance, in a Venn diagram, but an
explanation that 5 or 3 chips should be understood as being 5 or 3 days is

nothing more than a verbalism.

In such examples new problems of recognition can be created by the way in
which the addition is suggested:

John is 5 years, how old will he be 3 years from now?

Today is 5 January, what date will it be 3 days from now?

It is 5 oclock, what will the time be 3 hours from now?

with all kind of variations which can arise, for instance, from a change of
perspective.
4.23. How is the arithmetical knowledge about addition that was acquired
by uniting sets transferred didactically to this kind of problem? Possibly the
numerical performance of the act of addition is simply suggested on the basis

of the genuine set operation via the abstracted arithmetical problem

This

is certainly the case with such patterns as 5 times and 3 times, where every
indication is missing concerning what should be present, happening, done,
taken three times. Though originally the numbers were cardinals of sets, and

operations with numbers were operations with cardinals based on operations

with sets, in these problems they are stripped of all substance in order to become
numerals and algorithmic operations on numerals (computing number).

NATURAL NUMBERS

99

It is not astonishing that such didactical procedures, if applied again and again,
create hosts of underachievers: children that fail on any kind of word problem.

According to this didactics the transition from

5 marbles and 3 marbles

to

5 days and 3 days

is performed via

thus the transition occurs algorithmically and without any experience as to what
adding days can really mean. How can and should this be improved?
4.24. Earlier I drew attention to how closely numerosity and counting number

are interwoven.
is defined cardinally, but from olden times it has been
calculated ordinarily. Set theory fanatics tried to fight this didactically with
Venn diagrams, but fortunately with little success thanks to the natural inclination to operate by counting. The result of
is obtained by counting 3 steps
forth from 5 onwards: to the 5 beads on the abacus one adds 3 one, two,
three both terms as well as the sum being defined by concrete sets. Or: starting
with the mental 5, one counts on 6, 7, 8 while thumb, forefinger,middlefinger, one after the other, are raised in order to steer the deployment of the
second summand. Or: the same, without the fingers, while the activity of adding
is steered by rhythmic or visual images.
One could teach children the elementary additions in the same way as the
tables of multiplication are often taught, by intentional memorising. Perhaps
here and there it really happens this way, but certainly not as a rule. Memorising
addition is unintentional; the addition tables are learned by performing additions
again and again. Yet memorising is not the sole aim of performing the operations.
By the process of adding the addition is experienced as meaningful and understood in such a way that, if needed, it can be made explicit. By this meaningful
activity on summands more complex additions are prepared where the summands
and the sum are not represented by sets. In brief, by the activity of adding the
addition is constituted as a mental activity.
Let us review our examples. 5 steps and 3 steps with feet on the fifth step,
it goes on one, two, three mentally to the sixth, seventh, eighth counted
aloud. 5 days and 3 days turn three days (pages) further in the calendar.
5 km and 3 km the signpost shows 5 km to the point you came from and
3 km to the point you are going to, and tells you to count from point 5 further
to point 8. To the 5 florins, represented by a banknote or by 2 coins of 2
florins, three florins are added 6,7,8 which are mentally counted as 1,2,
3.

100

CHAPTER 4

4.25. These examples are distinguished from the former ones not only with
regard to the materialisation but also structurally. One is no longer taking the
union of two unstructured sets. We are not faced, as in the case of the marbles,
with five and three disconnected steps, days, km, florins, times, which are

raked together. They are objects that follow each other in space or time, not as
elements of a set but as paces on a road, in a process that in turn suggests and
perhaps elicits a counting process.

I already stated that counting processes are embedded in time and possibly
also in space. By pronouncing the number, by indicating objects, points in
time and space are isolated in the continuous medium, and this can be done
more or less sharply a string of beads the size of a dove egg or a mustard
seed, densely threaded or loosely spread. Counting queuing cars they follow
each other at a distance and are counted one by one. Counting the wagons of
a freight train, a continuous stream articulated by the buffers the rear buffers,
of course, because with the rear buffer, the ordinal number of the wagon is
accomplished, as the odometer of a car accomplishes the kilometers. Counting
when playing at hide-and-seek numerals are stretched over an interval, and
after 100 it is: I am coming.
One counts marbles, but likewise intervals: pronouncing 21 such that it takes
a second. It depends on the intention: counting spatial or temporal phenomena
or articulating the spatial or temporal stream. Of course, one can maintain that
mathematically it is all the same: the disconnected marbles, the beads loosely
or tightly on the string, the sequences of spatial or temporal intervals all of
them are sets, all have cardinal numbers, and adding them is based on taking the
union. This is true and it proves how universal mathematical concepts are. If
you assign to each kilometer of a trip its end, to each year the stroke of the
clock that completes it, to each
of water passing the bridge of the Meuse
its last drop, to each ton of flour streaming out of the mill its last particle, you
obtain a mapping where the successive intervals are mapped on a sequence of
points a one-to-one mapping of something that looks continuous but by this
mapping is cut into a discrete looking sequence of slices. This then is the mathematical justification of a procedure by which natural numbers are used to count
in a continuous medium; but though it is mathematics it is nothing but the
unconscious background made explicit.
How old are you? Four the child says, raising four fingers and knowing
that at some precise moment a fifth will be added. Before the natural number
is constituted as a characteristic of sets in the discrete realm, it is already applied
in the continuous one that is, to magnitudes. The examples I mentioned
length, time, monetary value are paradigmatic. Continuous climbing is articulated by steps, the stream of time by the torn-off pages of the calendar, the
road by km-posts, money by coins and notes, so many times means so many
times the same it is these articulations that are counted. In traditional teaching,
magnitudes are delayed until the children are ready to learn common and
decimal fractions. This reservation is justified by nothing but a pseudo-didactical

NATURAL NUMBERS

101

systematism. The first step in analysing a magnitude, where measuring the magnitude is articulated by the natural multiples of a unit, is possible and desirable at
an early age; counting can and must immediately be transferred from discrete
quantities, represented by sets, to magnitudes. Modern textbooks start measuring
much earlier than tradition allows, but unfortunately this kind of measuring is
not yet sufficiently integrated with the operations on natural numbers.

4.26. The device beyond praise that visualises magnitudes and at the same time
the natural numbers articulating them is the number line, where initially only
the natural numbers are individualised and named. In the didactics of secondary
instruction the number line has been accepted, though it is often still imperfectly
and inexpertly exploited; in primary education it makes progress little by little.
The progress is slowed down on the one hand by Venn diagrams, on the other
hand by rudimentary material like Cuisinaire rods. It seems to me a disadvantage
of the number line that it is so easily drawn and that it cannot be sold together
with the textbook as teaching material.
The Cuisinaire rods which in fact have come down from Frbel were,
once introduced, a large step forward: a translation of natural numbers into
lengths, and of operations on natural numbers into operations on lengths.
The articulations in the continuous stream are being concretised, the intervals
coagulated and embodied in coloured rods. The lengths are torn from their
context length; numbers is the peak of what the rods can represent no
other magnitudes.
The number line eclipses the Cuisinaire rods in many respects: The virtual
infinity is better expressed by the number line. The number line knows no
compulsory scale; number lines on different scales on the blackboard and
on paper are immediately identified, notwithstanding their incongruency. And
what is most important: in manupulating Cuisinaire rods the route from the
visual to the mental realm is diverted by an irrelevant motorism. Later we shall
consider what relevant motorism is and how it applies to working on the number
line.
Does it look strange that I deal with these matters under the title The
constitution of addition as a mental act? Counting and adding are closely
knit in the constitution of number. Counting is again and again adding one,
and additions are performed by counting. This then characterises the arithmetic
of the number line and its didactics which we are engaged in.
The number x stands on the number line where x is accomplished. What x?
x cm of the ruler,
x km of the road,
x cars in a row,
x books on a stack,
x pages of a book (tens only)
x ticks of a clock,

102

CHAPTER 4

x hours, days, years (on the time axis),

x grams or kilograms,
x cc of the measuring glass.

Yes, accomplished. The ruler or road is accomplished by scanning or pacing it

with real or imaginary fingers or feet, the queue runs out, the books are fetched
from and put into the stacks, the book is leafed through to page x, the time is
accomplished and the hand of the clock or the spring balance accomplishes the
interval x. The number x stands where it is full. I can move or jump to the
point. And if it is a jump it is again pictured in a continuous manner by an
arrow, perhaps with a
on it. The jump can be composed or thought to
be composed from little jumps. The jump can be done from every starting
point, pictured by the same arrow
and this then pictures the addition,
in one blow, or dissolved in little jumps. The intermediate jumps have been made
explicit or just blurred. The number line can bear numbers at the tens only
like a ruler with intermediate strokes for the units, or at the fifties like the
measuring glass with intermediate strokes for the tens, or only at the hundreds
and thousands, like the kilometers along the road. In between, interpolation
takes over: the 175 should be at a certain spot between 100 and 200 second
graders are able to localise it. And finally there is the line (or path) with no
number or marks at all, except an origin.
How far is it from here to there on the number line? The little steps are
counted. But you can also take the from here to there between your thumb
and forefinger, carry it back to 0, and read it off. Adding n to m can be
performed by counting but it can also be done in one blow: the piece that
is accomplished at n is taken between the fingertips and carried over to m
between the fingertips or on a ruler or a strip. Addition is being performed
geometrically, rather than by counting. But that means that you can dispense
with numbers on the number line. The number line is mirroring a magnitude,
and the geometrical shift on the number line mirrors addition for this magnitude,
as it does subtraction.
This then is another way to have addition begin, with magnitudes rather than
with marbles, but with magnitudes that are visualised and bound to the number
line the length on a ruler, the volume on the scale of the measuring glass, the
weight on the scale of the spring balance, and (somewhat harder) the time on
the time axis. Davydov* has shown how adding and subtracting with magnitudes
length, volume can didactically precede the numerical operations in order
to develop and support them. I tried it with a boy (5; 6) whose arithmetic
abilities were negligible; he was able to perform meaningful operations on
length, volume, and, with the spring balance, on weight, up to fair understanding
of Archimedes principle.**
* Cp. Hans Freudenthal, Soviet Research on Teaching Algebra at the Lower Grades of

Elementary School, Educ. Stud. in Math. 5 (1974), pp. 391412.

** Bastiaans Experiments on Archimedes Principle, Educ. Stud. in Math. 8 (1977), pp.
316.

NATURAL NUMBERS

103

Well, let it be true that the title of the present chapter does not seem to
justify the present subject, which is magnitudes; then, however, the title of the
present section, which is about addition, does justify it. From the beginning
addition should be of a broader range than operating on natural numbers. The
natural numbers are characteristics of discrete sets, adding them is rooted in
throwing together such sets. The set theory medium of magnitudes is continuity,
and adding there means composing. The universal model of magnitudes is the
ray, primarily structured not by numbers but by congruent displacement,
corresponding to addition of magnitudes. Not until a unit is chosen for the
magnitude and correspondingly a 1 is placed at the ray, are addition of
magnitudes and addition of numbers related to each other, then the ray is
articulated by points corresponding to the natural numbers to become the
number ray, and cardinal addition is translated into set theory addition.

4.27. With the idea of natural numbers on the number line we are as far from
their cardinal origin as geometrical addition is from uniting sets. In this analysis
the connection between numerosity and measuring number has been made
didactically via the counting number and magnitudes. Earlier I explained a short
cut: dividing the ray into succeeding congruent intervals and interpreting number
as the cardinal of a set, to wit, of subsequent intervals. Anyway from the start
onwards the natural number obtrudes itself on the learner in all its aspects. Only
a system fanatic could be offended by this challenge. It is a fact that natural
number has many aspects, one of which is its use and its indispensability as a
measuring tool, not only in applications of arithmetic but also on behalf of its
didactics.
Problems arise around addition (I have already elaborated on this theme) as
soon as the summands to be added and the operation of addition are not plainly
recognisable as cardinals of sets and their union, and attempts at a set interpretation are artificial or obnoxious. For this reason, as I said, the constitution of
addition and learning to add should be more broadly oriented towards adding

magnitudes represented on a number line, bearing initially the natural numbers

only. Neither a recourse to sets nor algorithms that exploit verbal cues can be
of any help if such problems are to be solved. The Venn diagram visualisation
of adding is too narrow, and algorithmising prematurely involves the risk of
a wrong perspective that may influence the mind in a way obnoxious to mathematics. It might be wholesome to first understand addition broadly and to
visualise it in a way that can be considered as definitive from a mathematical
viewpoint.
This suggests that not only adding but even counting should be accompanied
by activities on the number line. One more marble from the bag to be counted
is accompanied by one block joining the train of blocks and one segment on the
number line. Counting by estimate is accompanied by vague indications on the
number line. More and less mean directions on the number line; this
more and this less make more precise how much to proceed in one direction
or the other. Two parallel number lines, used for John and Pete: the marbles

104

CHAPTER 4

John wins from Pete are transferred as a line segment from the Pete line to the
John line. Transferring contents and weights from one receptacle to another is
similarly represented. Concretely or symbolically adding numbers given by
objects or at the fingers or the abacus beads, or by tallies is one of the accesses,
which certainly should not be neglected, but it is not the only one. The other,
broader one is via magnitudes and the number line.
4.28. A few lines above I stamped the visualisation of the addition on the
number line as definitive. This is true as long as I restrict myself to N. The
extension to Z seems artificial at an elementary level. A more natural extension
is towards two dimensions, vector addition in
or in the lattice
which then
includes the extension to Z. This addition would be introduced geometrically,
by parallel shifts of vectors, in order to be expressed arithmetically afterwards.
This subject can lead into a field of free activities which as motivations may
influence the more regimented, properly arithmetical, activities. Later (in
Chapter 11) we will reconsider these questions.
4.29. The Additive Structure of N
The additive structure of N includes more than the act of adding. It is, as it were,
the whole complex of relations

possibly also expressed as

and supplemented by

and all other relations one would like to consider in this context.
On a higher level it includes experience, and on a still higher one formulated
knowledge, of such properties as are
commutativity,
associativity,
equivalence of

and

and many more properties of this kind.

This structure grows as the learner explores N, but fundamental properties
such as the ones just mentioned can be experienced fully and clearly and even
formulated within a quite restricted part of N.
The structure of N is partly accessible to memorising; beyond this, it is
obtained and analysed by means of the algorithms of the decimal system, which
will be dealt with later in the present chapter.

N A T U R A L NUMBERS

105

However, the relations considered so far yield an insufficient grasp of the

structure of N. For instance, the relation
can be structured by prescribing c and asking for the totality of solutions
the list of splittings

which exhibits a striking structure of increasing and descreasing sequences and

a central symmetry. Of course, splittings are also useful for the algorithm of
passing over the tens when adding, but there is more to it. A structure like the
one exhibited by this list invites questions of why?, which with their answers
help one to understand the additive structure of N more profoundly. One list
like this is not enough; they are available for each c. Interpreted in two dimensions, in the lattice, these lists become point sets with a remarkable interior and
exterior structure.
Other lists are created if

is viewed with b fixed as a condition for a, c, such as in

again with a striking structure, which asks for explanation. What characterises
these pairs of points, when viewed on the number line?
The order relation, viewed in the context of addition, also belongs to the
structure of N. It is obvious that adding more yields more, and subtracting
more yields less, but it is not so obvious what this means for solving inequalities
and for other applications; for instance, that to solve

one is advised start with the largest solution. Much insight into the structure of
N is required to solve
Other additive structures in N are arithmetical sequences
1,4,7,10,...
corresponding to jump sequences on the number line. Where do two sequences
of this kind, the preceding and
2,6,10,14,...,

106

CHAPTER 4

meet each other? The question is easily answered but to understand it requires
the multiplicative structure of N.
The field of hundred is a structure in N, and in this field the arithmetic
sequences show special structures. The table of addition is also a structure in
N, symmetric with respect to the diagonal, and a chessboard distribution of even
and odd numbers why?
Properties of such structures can be traced, understood, explained. Properties
of N can be known and applied; for instance commutativity to add the smaller
to the larger number conveniently, and associativity to simplify additions
algorithmically for instance,by completing tens.
4.30. The Constitution of Subtraction as a Mental Act
Discussing subtraction after addition does not aim at a didactical separation, and
certainly not at a succession in the genetic and didactic process. In all contexts
where addition is didactically offered, subtraction is implicitly present in order
to be made equally explicit.
Formally,

subtraction results as the converse of addition,

and in fact this aspect of subtraction should not be neglected,

because of

is reasoning by which each subtraction can be reduced to an addition, which

might be known in advance by memorising. In the case of division as the converse
of multiplication, this is indeed the way to solve elementary problems after
the tables of multiplication have been memorised. Subtraction is dealt with
differently, and there are reasons why it is so. The inference pattern of inverting
an operation is probably less familiar to a six-year old than to an eight-year old;
but the main reason seems to be that subtraction is as concrete as addition,
whereas division is much less so than multiplication.
In the domain of objects, subtraction means taking away, as addition means
annexing. The older arithmetic books had great difficulty when picturing subtraction. Venn diagrammatici, at a loss what to do with subtraction, invented
the strangest aberrations. Meanwhile new inventions bear witness to the fact
that a fresh spirit has befallen developers: rigidity and unifying dogmatics have
given away to a creative imagination of situations that suggest mental activities.
With regard to subtraction: withering flowers, birds flying up, dwarfs running
away, walls and towers breaking, and many more of this kind. The problem
suggested by the picture is often not uniquely determined. If a story is to be
told about the picture, it may be the style of
there were c,
b went away,
thus a were left;

NATURAL NUMBERS

107

or

there were c,
a were left,
thus b had gone;
but also
a are left
b runs away
that is c together.

Yet this lack of uniqueness is not at all a didactic shortcoming.

The objects can be pictured in disorder or in a row; the objects to be taken
away can be on one side of the row, on the other, or criss-crossed in between.
But explicit taking away suffices as little for the mental constitution of subtraction as uniting explicitly given sets suffices for addition. Which ones are more,
and how many more? In order to answer this question about a picture with
two kinds of objects, indirect taking away is required: one set is diminished by
a subset equivalent to the other.
The numerosity aspect suffices as little for subtraction as for addition to
mentally constitute the operation. I desist from repeating for subtraction the
arguments I displayed in the case of addition. Subtractions too should be constituted more broadly, with magnitudes geometrically interpreted on the number

line.
In numerical subtractions on the abacus, with tallies, on the number line,
or mentally one observes two methods,
taking away at the start,
taking away at the end,
and counting the remainder. Among 8 children, 3 girls, how many boys? can
be answered by from 4 up to 8 and counting these numbers one, two, three,
four, five on the fingers. Or the three are taken away at the end: one away
is 7, another away is 6, another one away is 5, and meanwhile thumb, forefinger,
and middle finger have been raised to control the process. Children learn quickly
which method is more useful in each particular case: if the subtrahend is smaller
than half the minuend it is taken away at the end, otherwise at the start.
With geometrical subtractions on the number line both methods can apply:
the subtrahend can be congruently cut away at the start or at the end of the
minuend, and the remainder is measured again.
The geometrical concreteness of the number line is particularly useful in
understanding problems like

108

CHAPTER 4

4.31. The Constitution of Multiplication as a Mental Act

Multiplicative terms like double and times precede multiplication as an
arithmetical operation.

I told you three times,

walk three times up and down,
the clock struck three times,
when the (minute-) hand has gone around three times,
you must sleep three more times, then it is your birthday.
It is

three times doing, undergoing, experiencing, awaiting

something. And then in simple additions, before multiplication is on the program,

it can be
seizing three times four marbles

or simply

three times four marbles,

and this means
four marbles and four marbles and four marbles.
No operation not even addition and subtraction offers itself as naturally
and is understood as spontaneously as multiplication.
Daphne (5; 1) is asked: How many prongs does this fork have? Two forks, three forks,
four forks? Almost imperceptibly her fingers (the thumb excluded) tap on the table while
she is counting on. When, at five forks she hesitates a bit, her elder sister whispers 20
to her grandmother, which makes Daphne angry.

Though this is not an explicit multiplication problem the term times

does not occur it shows the inductive origin of multiplication as repeated
addition. This then is the way products are calculated and how tables are built.
Remembering our exposition on addition one should apply so many times
in magnitudes early. Indeed, this is one of the functions of multiplication
3 kg, 3 m, 3 km (travelling)
are 3 times as much and cost 3 times as much as

1 kg, 1 m, 1 km;
3 km cycling

is 3 times as much and takes 3 times as long as

1 km cycling;

NATURAL NUMBERS

109

three of the same

weigh three times as much as

one;
and so on. Especially on the number line multiplication as repeated addition can
be effective:
3 logs reach 3 times as far as 1 log,
with 3 steps (jumps) you go 3 times as far,
with 3 turns a wheel covers 3 times the path.
This implies that
3 times 4

can be realised by
3 rods, steps, jumps of 4,

subsequently performed on the number line.

So far multiplication was present as nothing but repeated addition. 3 rows of

4 marbles each may be placed
after each other;
with a change of perspective, they can be placed
in order below each other,

a two-dimensional arrangement, the

rectangle model
of multiplying. From two set A and B the

set of pairs or product set

is formed. This restructuring reveals new aspects of the product. To calculate

mn
one provides oneself with two sets A, B such that
forms the set of pairs

and puts

where the result does not depend on the special choice of A, B.

Only by means of the rectangle model of the product do properties of multiplication become visible: commutativity, the rectangle is rotated a quarter of
a turns and distributivity, two rectangles of equal height (or width) moved side
by side. (Figures 20 and 21.)

110

CHAPTER 4

In the preceding
set of pairs and the rectangle model were used to
restructure the set that is basic to the arithmetical product,
among other reasons in order to create insight into properties of multiplication.
The converse function is accomplished by the

set of pairs and the rectangle model as structures discovered or introduced

in sets in order to calculate their numbers as products.

This function has been discussed extensively in Chapter 3.

The rectangle model leads in a natural way to
the area of the rectangle:
When one passes from the discrete number sequence to the number line, the
discrete point -like beads are, as it were, condensed into unit rods; in two dimen-

sions they form unit square tablets, which together cover the rectangle. Its
area is expressed by unit squares after multiplying length by width as the
implicit or explicit arithmetical operation. The part played by the rectangle

model in multiplying magnitudes in general shall be discussed at another place.

All that has been said about the set of pairs and the rectangle model can be
extended to the
set of triples and plank model.
The latter shows

associativity
and leads to the

volume of the plank

expressed by unit cubes. Sets are

NATURAL NUMBERS

111

structured as triples
on behalf of calculation of their cardinals.
In Chapter 3 we displayed other product sets with two or more factors, like
the
tree model, the roads model,
as showing product structures.

4.13a. The Role of Equipartition Structures in Learning Processes on Behalf of

Multiplication

Among the models just mentioned, I forgot the one illustrated in Section 3.22
by
m baskets with n eggs each.
Not intentionally but as an oversight, which in Section 3.22 I imputed to others

as a mistake. I myself had overlooked what I had earlier signaled as a serious gap
in the whole didactical literature. Was I so overcome by set theory dogmatism

as to forget about my own ideas? Anyway, by observing learning processes I

recalled them.
Multiplication is, first of all, repeated addition and this operation can efficiently be structured by the set of pairs in the rectangle model the set theory

product in order, among other things, to calculate cardinalities as products.

This model, however, is insufficient. Not mathematically, since the eggs in each
basket can be numbered in order to impose on the set of eggs the stronger

structure of set of pairs with the elements

number of egg within its
It is didactically insufficient since the sophisticated mathematical restructuring

can hardly be expected to arise spontaneously or perforce in learning

processes or, if it arises at all, be made conscious in order to be made explicit

and, if need be, available.

Let us consider the following sequence of problems:

(a)

A picture representing seven baskets, each with six eggs. Question:

how many eggs?

(b)

A picture representing seven baskets, with the legend: in each basket

there are six eggs. Question: how many eggs?

(c)

The text: I have seven baskets; in each of them there are six eggs.
Question: how many eggs?

(c')

Or more streamlined: 7 baskets, 6 eggs each;how many eggs?

It is a sequence leading from material and pictorial to mental and

112

CHAPTER 4

verbal, which, for that matter, can be lengthened and refined. The factual
situation which I observed in a traditional third grade, treating more modern
material, was that of problem (b).*
Pupil A counted very fast 1, 2, 3, 4, 5, 6, with a finger at the first basket, 7,8,9,10,
11, 12, with the finger at the second basket, and so on, up to 37, 38, 39, 40, 41,42 with
the finger at the seventh basket.

Pupil B counted, while his finger glided from one basket to the next: 6, 12, 18, 24, 30,
36, 42.
Pupil C did the same as B, without using his finger.
Pupil D said: 7 times

(Between pupil A and B one could interpolate a variant A' who does the same as
A but without using his finger as a marker.)
How would these pupils have reacted in the situations (a) and (c)? In particular would pupil D in situation (a) have used multiplication, and would pupil
A have interpreted situation (c) as a multiplication or would he have failed

completely? What background part did multiplication play with pupils B and C
in situation (b)? Rather than using
it is
easier to recite the sixes table in the singsong 6, 12, 18, . . . , and meanwhile,
with the finger or an eye on the basket or while counting on ones fingers,
control the number of steps to be taken in the multiplier. A bunch of questions.
A long term observer could have said more about them.
I can say just as little about whether some pupils saw the structure of set
of the pairs in the data one needs to know much more about their former
learning processes. The structuring articulation after each six may be presumed
to be operational in all pupils, A marked it with the displacement of his finger,
and B and C with counting by six. The solely mental presence of the eggs was
certainly a factor in favouring this structuring. It may be guessed that in situation (a) it would have been less favoured.
How would these pupils have reacted to a counting problem in a rectangle
pattern (seven rows of six eggs)? Probably A would also have counted; B and C
would perhaps learn by such examples to interpret a rectangle model directly
as a multiplication.
More questions. I have never felt so frustrated by the lack of continuity in
the learning processes I have had the chance to observe.
4.32. The Multiplicative Structure of N

The multiplicative structure of N is the whole of the relations

possibly also expressed as
* I have adapted the problem to my model terminology. Originally it was about house
trailers and people.

N A T U R A L NUMBERS

113

complemented by

and all one can think about in this context.

At a higher level it includes experience, and at a still higher level formulated

knowledge, of such properties as
commutativity,

associativity,
distributivity,
equivalence of

and

and many more properties of this kind.

The structure grows as the learner explores N, but fundamental properties

such as the ones just mentioned can be experienced fully and clearly and even
formulated within a quite restricted part of N.
The structure of N is partly accessible by memorising tables; beyond this, it
is obtained and analysed by means of the algorithms of the decimal system,
which will be dealt with later in this chapter.
However, the relations considered so far, yield an insufficient grasp of the
structure of N. The relation
for instance, can be structured by fixing a and having b run through the number
sequence; this yields for c an arithmetical sequence visualised by jumps on
the number line.
Or I can fix c and find systematically its splittings into two factors, which
are illustrated by rectangles with a given area.

I can also split into more factors, factorising into prime factors. Divisors,
multiples, remainder classes are other means of structuring. Divisibility properties
can serve to simplify multiplication.
As in the case of the additive structure of N, the order structure deserves
attention if tied to the multiplicative structure. It is obvious that larger factors
yield larger products, but this does not necessarily include the change of
perspective:
given the product, increasing one factor and decreasing the other go
together

is an order reversing behaviour, which plays a part in division:

smaller portions larger numbers,
bigger portion smaller numbers.

114

CHAPTER 4

In another way the order structure of N is operational in division: the number

sequence is divided by an arithmetical sequence into intervals;
which numbers are in such an interval?
in which interval does a given number lie?
are relevant questions.

4.33. The Constitution of Division as a Mental Act

The relation of dividing to multiplying is much more involved than that of sub-

tracting to adding. It is much less symmetric in the case of multiplication and

division than it is for addition and subtraction. This discrepancy extends from
the most simple activity to the fully developed concepts. At a glance one can
tell whether a number can be subtracted from another: the smaller from the
bigger. Whether a division terminates is a surprise. Nothing like the remainder
in division exists in subtraction, or should it be
remainder 5 (that is,
if the problem is about money, the amount by which one is another's debtor)?
On the other hand, at an early stage children know, besides the indivisible
object, other continuous objects that suggest unrestricted divisibility and
invite the extending of N to
whereas suggestions of restricted subtraction
and extension of N to Z are scarcer and weaker.
Division does not occur as universally in the function of inverting multiplication as does subtraction with respect to addition. There is no counterpart in
division to performing an act three times as shown by the former examples.

Even half is no good as a counterpart of double, as witnessed by bigger

half in the vernacular.
Phenomenologically viewed, dividing arises in three ways: as
continually taking away,
distributing in equal parts,

inverting a multiplication.
Dividing by repeated subtractions is the counterpart of multiplying by
repeated additions:
how often can you take away a set of three from this pile;

how many jumps of three do you need to go from here to there;

how many times does the three-metre rod fit along the corridor;
how many times can you scoop 3 litres from this vessel?

Being able to subtract q times d from a number or magnitude a means that the
remainder must be smaller than d:
with
The ancient mechanical calculator, which did not know multiplication tables,
performed divisions as repeated subtractions; that is to say, rather than starting

NATURAL NUMBERS

115

with d, starting with a 10md in order to apply the same procedure with 10mld
to the remainder, and so on. Likewise, in the usual long division algorithm the
subtractive structure of the process is recognisable, though perhaps shortened by
the knowledge of multiplication tables.
This kind of division should be contrasted with that of dispensing equal
shares of equivalent objects to, say, q persons. Each gets one of the equal shares,
a qth part a strange terminology, if you stop to think about it, but so familiar
that it looks natural.* In every language I know, the ordinal number is used to
indicate how many shares it is of one share leaving aside division by two, for
which there are special expressions like one half. This terminology looks even
stranger if instead of a quantity of objects divided among q persons, a number,
say 12, is divided in, say, three parts. Then 4 is the, rather than a, third part of
12. It is so easy to pass from concrete sharing to abstract dividing and at the
same time from a qth part to the qth part, but whether it is as easy for the
learner, we simply do not know.
Distributing a small quantity in a small number of equal parts is most often
an intuitive procedure, in particular if done with magnitudes, which in principle
can be divided with no remainder left. It is exercised and understood early; in
particular, as meting out fair shares to a number of persons. It can be done by
giving them cyclically the same share until nothing is left or something is left
that does not admit of dividing.
These two kinds of division were formerly distinguished as
ratio division
and

distributive division,

and separately learned and trained as such.** The question of the ratio division
is
how many times does d go into a?,
that of the distributive division is

what is the qth part of a?

The difference is particularly striking if a and d are concrete numbers of the
same kind:
how many persons are dealt with if each gets d out of a florins?
how many florins does each of q persons get if a florins are distributed?

Under both aspects the remainder has the same function: a remainder too small
to fit what is taken away, or too small to be fairly shared.
* This question will be tackled once more in Chapter 5, on fractions.
** Cp. Mathematics as an Educational Task, pp. 252254.

116

CHAPTER 4

With division as the converse of multiplication it is different. The arithmetical

sequence of multiples of d determines intervals in N; the number a is placed in
one of them. If it is between qd and (q + l)d, then q is the quotient of a
divided by d . After q steps of d on the number line, there are still r simple
steps left to reach a, and that is the remainder.
This is how elementary divisions are learned: one takes ones bearings in the
frame of the table of d and situates a with respect to it. To achieve this, one
has to know the multiplication tables well.
From the viewpoint of inverting multiplication the two kinds of division
can be distinguished as follows: If I read
as a equals q times d, the factors q and d are not exactly the same thing: d is
the thing that is taken q times. This is strongly felt if d is not simply a number
but a concrete number or a magnitude:

3 notes of 25 florins equal 3 25 florins.

The set whose product is the cardinal bears a certain structure, which means
that the factors are not a priori commutants.
In the equation

I can ask for q if a and d are given ratio division , or for d if a and q are given
distributive division. Thus, also as a converse of multiplication, division shows
two facets. Well, one could do the same with the addition, distinguishing the
summands and asking two questions:
how much added to a to get c?
to what is b added to get c?
Yet the asymmetry is not so striking in addition as it is in multiplication, where
one of the factors can be concrete and the other abstract.
More profoundly viewed, two aspects of division are not yet enough. Concrete

numbers are multiplied by concrete ones too, magnitudes by magnitudes, such

as

and though such equations can also be solved with respect to different factors,
it is not usual to distinguish here various kinds of division. For good reason
indeed: Mathematics is powerful thanks to its universality. One can count all

NATURAL NUMBERS

117

sets by the same sort of numbers, as one can measure all magnitudes by the
same sort of numbers. Numbers do have a numerosity and a counting aspect;
addition has a cardinal, an ordinal, and a measure aspect; multiplication has the
aspect of repeated addition and of pair formation, and likewise division has its
own variety of aspects. But in spite of this wealth of aspects, it is always the
same operation a fact that expresses itself by algorithmics. As a calculator

one may forget about the origin of ones numbers and the origin of ones
arithmetical problem in some word problem. But at the same time one must
be able to return from the algorithmic simplicity to the phenomenal variety in
order to discover the simplicity in the variety. It is the secret of aspects that they
are discovered at one time and neglected at another time, and the knowledge of
it is part of the constitution of the mental object or act in question.

4.3443. Algorithmics in N
4.34. Algorithmics was touched on in the last section and earlier in Section 3.30.
In the case of division I alluded to a very special algorithm, whose definitive
version is long division. Algorithmics can, however, be understood more broadly
as an organisation by which one is advised to follow prescribed rules, where each
particular step requires a decision bound by certain criteria, under which the
step is easily performed. For the computer such rules are more narrowly formulated than for man; the human calculator should have the liberty of replacing a

multiplication by 98 with one by 1002 in order to apply distributivity, but

in a computer program the advantage of less computer time does not outweigh
the disadvantage of a more involved program. It is as though a bank whose
customers may have several accounts would think about saving postage by
sending messages regarding the various accounts in one envelope. The savings
might not outweigh the costs of a foolproof program for such a complicated
system. For crossing a road pedestrians follow rules such as first look left,
then look right (not useful on one-way streets), but this can hardly be called
an algorithm. A crossroads with traffic lights is different cross when the light
is green looks more like an algorithmic rule.
Though algorithms may sometimes look like an aim in itself, they are not.
They serve to simplify complex activities complex up to impossibility. Insight
is superseded by automatisms; that is, automatisms that are dependable even
though they are controlled by little or no insight. This is what you teach
computers by means of good computer programs. Man is less easily programmed.
Programs are inculcated by numerous repetitions; whether people are slow or
fast learners, flawless functioning is quite exceptional.
This seriously restricts the usefulness of learning algorithms, and in particular
if it happens with no appeal to insight. How much time and trouble should be
spent on teaching some algorithm which might be expected to be applied only
rarely by the learner? In a period of non-activity the algorithmic ability might
fade away or be lost. If some opportunity would then occur to apply it, it

118

CHAPTER 4

would be forgotten or shrouded in fog. But if one has learned the algorithm by
insight, there is some chance that it can be reconstructed. I can remember there
was a trick, but whether the reconstruction succeeds or not may depend on
how that trick has been learned.
Algorithmics, as considered here, springs from radical formalising. Mental
objects are fixed, or even replaced by, linguistic symbols in the language of
the abacus or the digital system , operations on mental objects are supported,
or even replaced, by strictly regimented syntactic operations on linguistic
symbols the column arithmetic relations between mental objects are translated into, or replaced by, regimented syntactic propositions, properties of the
operations and relations are expressed, or replaced, by formulas but this is a
new stage of algorithmising: algebraisation.

4.35. The usual algorithmics of the operations in N rests on the decimal structuring of N, followed by a formalisation by means of the positional system. A
number
is dissected into a sum

where
and for the the values 0, 1, ... , 9 are allowed. If
is prescribed, this presentation of
is unique. Positionally n is written as
that is with beads, chips, strokes in the jth column (starting at the right with
the 0th column) of the abacus, or with the digital symbol for in the jth
position (starting at the right with the 0th position) of the writing material.
(I neglect abaci with intermediate units of 5, as used in the Far East).
The school abaci now available realise the columns by means of metal rods
with mostly 20 beads each. Representing numbers requires no more than
9 beads to a rod; the surplus is meant as a reserve for adding numbers or dissolving a higher unit into 10 lower ones. Consequently, number representation
is not unique on the school abacus; there is a most economical, the reduced
one, which reflects the digital representation with 9 beads at most on each rod.
For initial learning of the algorithms of adding and subtracting, this nonuniqueness, compared to the uniqueness of the reduced presentation, is an
advantage. This is one of the reasons why in initial learning the abacus deserves
to be preferred above the digital representation. This does not mean neglecting
the digital presentation; on the contrary, the two tasks
to represent a written number in various ways on the abacus,
to note down in digits a number arbitrarily given on the abacus,
program the learner in a natural way to
transfer 10 units as one higher unit to the left,
transfer one unit as 10 lower units to the right

NATURAL NUMBERS

119

in the operations of
adding

and
subtracting.
For the addition

8+5

first 8 then 5 beads are made visible from the hidden supply on the units rod,
from which 10 are taken away and replaced with a bead in the tens rod, after
which the result is read as 13.
For the subtraction

13 8
one bead is placed on the tens rod and three beads on the units rod, then
the one on the tens is replaced by 10 on the units, after which 8 beads are
pushed to the rear, leaving 5 visible.
This procedure is usually maintained for a short time only. Soon, in performing the addition, one observes pupils structuring the 5 as 2 + 3 when they push

the 5 to the fore; then comes a stage where they more or less mark pushing the
string of 5 beads, and finally one where they immediately unite 5 from the 8
beads with the 5 to be added, pushing them behind and replace them with one
bead in the tens. Then the addition 8 + 5 is functioning mentally, and the abacus
is used as an aid to memory.
With subtraction one can likewise observe the abacus structuring the mental
act, though the learning process takes more time. Two ways of structuring can
be observed:

and
that is, from the 10 arisen by dissolving, the pupil immediately takes away 8,

or he splits the taking away process into two steps; first 3, then 5. On the
number line the first means taking away at the start, the second, at the end.

The didactical use of the abacus, as here demonstrated, is to suggest mental

procedures of this kind. When observing the childrens activities, the teacher
can reinforce the inclination to mentalise by useful interaction and can clear
away possible blocks.
There are a number of transitional stages between the abacus with the beads
and the column arithmetic on paper. A pictorial abacus columns with strokes
instead of rods with beads is one of these stages. In such a situation, where
no beads are being pushed and erasing is annoying, mental acts are stimulated.

120

CHAPTER 4

The pictorial abacus can be even more dissimilar to the genuine one: though
there are still drawn columns, digits rather than strokes are put in the columns,
such as

and similarly with subtraction.

The abacus is not the only tool used to suggest completing and reducing to
tens in addition and subtraction. This can be suggested in another way on a
number line where the tens are marked in a special way, or only the tens are
numbered and the numbers in between are indicated by strokes or dots.
18 + 5
is structured on the abacus as

but on the number line the closeness to 20 prevails,

a method also applied in mental arithmetic.

As long as algorithmics is dominated by the abacus or some intermediate
form, it does not matter in which order the various columns are processed

is one way, but there are many more.

can also be performed in another order. Proceeding from right to left recom-

mends itself in the final algorithm as the method that admits of the shortest
notation. The definitive algorithm is automatised to such a degree that while

working on a column, one does not pay attention to the next one at the left.
In mental arithmetic it is a habit to proceed from the left to the right. Sometimes instruction builds such a strong system separation between mental and
column arithmetic, determined by the horizontal or vertical position of the
summands, that pupils do not grasp that both operations mean the same. Of
course, numbers to be added or subtracted should not always be given in the

vertical position; pupils should have the opportunity to rearrange the given
numbers vertically and in order.

NATURAL NUMBERS

121

In the definitive algorithm haptic and visual auxiliary means are replaced by
borrowing and keeping in mind. The mental activity can be burdened with
multiple keeping in minds, the influence of more summands and more columns
in additions, repeated borrowing in subtraction. It is even more burdened in
mental arithmetic, when visual support is lacking and pupils may have forgotten
the result of the first partial operation as soon as they accomplish the next.
Poor achievement in column arithmetic may result from a lack not of understanding but of attention and concentration. Even though the pupils may be
fascinated in general, this attention can slacken, certainly at the age when
children are taught column arithmetic. Even good pupils can make a lot of
errors; less able pupils can be discouraged by their failures and finally stop
learning altogether.*
Learning column addition and subtraction has been viewed here as a I
do not say gradual but step-by-step development. By preference the steps
should be taken spontaneously by the learner for the course that was sketched
here, it has been shown to be feasible. As much as possible, the particular
steps should be observed by the teacher and made conscious to the learner, as
a means of reinforcement. Pupils who are not strictly led may develop methods
of their own. Once I observed a pupil who in subtractions almost systematically
rounded the subtrahend upwards:

Perhaps this method is even better than the usual one. Anyway a pupil who
contrives such things proves to act with so much insight that he can learn the
traditional way too.
On the other hand, pupils should not be pushed to take a step on the road
to algorithmisation unless they have really got there. These are general didactical
principles, though particularly relevant for learning column arithmetic. Replacing
insight with algorithms is a meaningful activity provided the algorithm has arisen
from insight rather than having been imposed and blindly accepted. If it does
not do any good, it does not do any harm, is not a convincing argument.
Algorithms should be learned by algorithmising, and this means most often by
progressive algorithmising, which is a special case of progressive schematising.
Even this is not sufficient. Once a performance has been learned, the way
in which it has been learned is readily forgotten. For algorithms this may mean
that their sources of insight are clogged. In the aftermath of algorithms, teaching
should aim at retention of insight.
* Another cause of failure, possibly even more important than lack of attention and concentration, which seems not to have received sufficient attention, is failure of short term
memory. This is not the place to advise remedial teachers. My experience has shown that
systematic training of short memory can be helpful.

122

CHAPTER 4

4.36. Spontaneous algorithmisations are less easily prepared for multiplication.

The operation is more difficult, and the natural motivation is not strong enough
to conquer the difficulty. Multiplying is defined as repeatedly adding, and if
4 times 8 or 4 times 8 florins can be obtained by simple adding, if the price of
20 Seasons Wishes at 0.50 florin a piece can be obtained by pushing a key
twenty times, extra stimuli are required to store the results of multiplication
in the memory. Once and for all I came near to saying but that is just not
what happens. We know that this is not impossible, imprinting something in
ones memory in one stroke a name, a phone number, a birthday if only
we concentrate on it. But it seems that learning the tables looks more like being
introduced to twenty people in a row there are not many who can tie their
attention to such a procedure. Learning the tables is a process of slow inculcation, of transfer from short term to long term memory. In column multiplication
the knowledge of tables is rewarded; so column multiplication might be a good
motivation to learn tables. It sounds paradoxical: The exercises by which
memorising tables is motivated do not start until the learner is supposed to be
familiar with them a gap that is not easily bridged.

I do not know serious research about how to guide the memorisation of tables
I mean research on actual learning processes. Pupils develop a great many
strategies in order to facilitate the ever fresh computations of the table products.
One of them starts with familiar products, such as n2 (
), which
are somehow attractive, in order to go up and down. But it can happen as well
that if, say,
is asked, the pupil leafs back to
which he recalls was
asked earlier. Some more knowledge about pupils' strategies might lead to more
effective techniques in teaching tables.

4.36a. In the original version, I turned at this point immediately to column

multiplication in general (Section 437). Remedial work with 5th and 6th
graders revealed to me a source of failure. It concerns certain apparently simple
automatisms of the positional system of N (and as important for decimal fractions) which are fundamental and indispensable if the algorithms are to function
at all.
I means the automatised rules

multiplying by 10:
multiplying by 100:

attaching a zero,
attaching two zeroes,

and so on, in general

multiplying by 10 . . . 0:

attachingas many zeroes as there are after

the one.

The converses, related to division:

division by 10:
division by 100:

dropping a possible zero at the end,

dropping two possible zeroes at the end,

NATURAL NUMBERS

123

and so on, in general

division by 10 . . . 0:

dropping as many possible zeroes at the end as

there are after the one.

Adults I observed from teachers in training to professors of sciences were

as familiar with these rules as they were unable to argue them, and even to feel
any need for arguing them a shortcoming which is perhaps more serious for
teachers in training, with a view to their future job, than it is with professors.
Some of them, perhaps even many, will have learned such rules by insight. It
looks like a typical case of clogging the sources of insight by rote exercise. I
should add that in the textbooks I consulted I met with the same lack of insight
that is to say, didactical insight in how to stimulate useful learning processes
towards these rules by didactic sequences.
Let us start with the phenomenon that the tables for 2 , 3 , . . . terminate with
the round numbers 20, 30, . . . . It is a phenomenon experienced by learners
with feelings varying from satisfaction to astonishment. Anyway, they are
keen to reach at the terminus the safe harbour of 10 times. Would it not be
wise to make good use of this emotional concern to have them to explain this
phenomenon?
If we do nothing, the road to the above rules is paved by the empirical
induction

thus ten times means a zero at the end. From here an easy too easy
generalisation leads to the multiples of 100, 1000, and so on. It is obvious,
however, that this learning process provides less insight than is possible and
desirable. What to do about it?
First of all, the phenomenon of the handsome terminus of the tables should
be explained, which may happen in two steps: first, commutativity known
from the rectangle model which transforms

10 twos into 2 tens,

10 threes into 3 tens,

and so on; second, coupling this to the promotion of

2 ones into 2 tens,
3 ones into 3 tens,
and so on by attaching a zero, which can be supported by the abacus or other
material. This, then, is the indispensable link in the learning process towards
the rule of multiplying by 10 while attaching a zero,

10 times 463 are 463 tens,

which requires the three units promoted to as many tens, the six tens promoted

124

CHAPTER 4

to as many hundreds, the four hundreds to as many thousands, which results

in attaching a zero. The promotion of

units to tens,
tens to hundreds,
and so on, by attaching a zero thus reflects multiplication by 10.
The next step would be understanding multiplications by 100, 1000, ... as
repeated multiplication by 10, in order to get the promotion of
tens to thousands,
and so on, by attaching two zeroes, reflecting multiplication by 100, and so on.
An (intelligent) 12-year old says that multiplying a number by 10 and once more by 10
boils down to multiplying by 20. She masters the rule of attaching one zero, but without

understanding it. She does not master those of multiplying by 100, 1000, and so on. She
does not know what is

She does not know how to divide by 10. She knows

almost nothing about decimal fractions.

Division by 10, 100, . . . , if possible, should be made explicit in this context.

This didactical sequence has its sequel in decimal fractions (cf. Section 5.24a).
A question, which will then be repeated, is whether in such a sequence the
power notations

and so on, should be used. The law of exponents is a powerful means, which,
however, can didactically lead the wrong way
. The reader
will have understood that the power notations in Section 437 are his business
rather than the pupils, but this does not exclude a final transfer to the pupil.
4.37. Column multiplication in the decimal system is based on the knowledge
of the tables, that is, the products
on the rule
and on applying distributivity. If two numbers m, n are given decimally,

the product
is built from the partial products

NATURAL NUMBERS

125

These partial products are computed and pulled together according to some
principle which assures that each gets its turn once and only once. The question

of how this can be done most efficiently should be answered by the algorithm
of multiplication. Let us forget the usual solutions and ask ourselves how we
would tackle such problem if they were new for us.

Well, the most natural image is a two-dimensional pattern, a table with two
entries: one factor is written in the ordinary way, horizontally, the units at the
right and progressing to the left according to powers of 10; the other vertically,
from below to above. At the crossing of the
row and the
column the
product
is placed, omitting the factor

It is these partial products

provided with the factors

that should be

pulled together.
Which ones among them bear the factor

? One gets

which, indeed, suggests a way of conbining: according to oblique lines under

an angle of 45 from below left to above right. I think everyone would do it
this way if he were allowed from this point of view to reinvent multiplication.

In fact, it is the way to multiply polynomials in x,

The boxes of this pattern contain two-digit numbers yielded by the tables. Let

us do it numerically

126

CHAPTER 4

and assemble

or vertically

which could also be done mentally, with a lot of keeping in mind.

It is a matter of technique how this assembly along the oblique lines is

performed. For instance, one could do it on the abacus. First the 45, then one
column to the right 40 and 54, one more column to the right 35, 48, 63, and
so on. Written down

That is to say, one attaches as many zeroes as correspond to the oblique lines,
but the zeroes can also readily be dropped. For safety one can place the corres-

ponding power of 10 at the row, columns, and oblique lines (Figure 22), thus

NATURAL NUMBERS

127

Adding according to the oblique lines can also be done mentally, which would
mean writing down the result in one line. That, then, is the so-called abridged
multiplication not new which I have presented here.
The method exposed here is mathematically the most natural, but it is not
the usual one. There are reasons why it is not. First of all are historical reasons.
Abacus traditions may have played as big a part as accident when the system
was chosen, but once the choice had been made, it became a tradition. It is
not easy to change such things. It is safer to have a teacher teach a method he

has mastered than one he must learn himself and that is not convincingly better
than the old one.
Didactical arguments can also be adduced in favour of the usual method, and
this is very likely to tip the scale. After learning the tables and products like
and so on, the first true column multiplications will be

The rectangular pattern shows no more than one column

while the usual method

is more perspicuous.
The next step in the learning process would be dropping the zeroes and shortening the procedure: Less on the paper and more mentally: 2 times 4 is 8,2 times
2 is 4, 8 times 4 is 32; write down 2, keep in mind 3,8 times 2 is 16, plus 3 kept
in mind is 19. The result is written in one line, without intermediate steps.
This then determines the sequel. The next step is

reduced to

which had been mastered earlier and is augmented by a zero. Similarly

128

CHAPTER 4

is dealt with.
Then

invites splitting into

brought into one pattern

where finally a zero may be dropped.

This is in principle the general method, but it does not mean hurrying to
apply it directly to

which initially is better dissolved into

Perhaps this even helps better ones understanding of

It is a quite natural course, this way. Viewed through the rectangular pattern
it is assembling the partial products according to columns, 0th column, 1st
column, and so on, and within the columns upwards. But this table of partial

products is not made explicit. I confront the two methods with each other
without drawing any conclusion. I myself do it according to the method I learned
at school; in spite of many efforts I never succeeded in chasing it away in favour
of the more efficient abridged method.

NATURAL NUMBERS

129

4.38. In a learning process directed towards reading the ordinary algorithm, the
rectangle model of multiplication need not be excluded. It is entitled to be taken
seriously because of its convincing geometric power as well as its significance for
the product as a means to calculate areas. Its use has many times been proposed,
and it has recently been elaborated in the Wiskobas curriculum.*
The product m n is visualised by a fabric with n warp and m woof threads
(Figure 23). The number of crossings can be counted and calculated more or less
adroitly. The threads are taken together in bundles of ten. Thus, for instance
24 8 becomes the Pattern of Figure 24, and by a similar, though more complicated, pattern 24 82 is illustrated. Counting the crossings adroitly is away
to structure the activity.

From the abacus the suggestion comes to combine ten thin threads into
the thick thread, which for 24 82 means (Figure 25) a way to structure the
counting activity even more sharply.

* Leerplanpublikatie 10, Wiskobas Bulletin 8, nr. 56, Nov. 1979.

130

CHAPTER 4

The method can be extended to factors with more than two digits, although
this can hardly be recommended. Instead, genuine algorithmisation should
have started in the meantime and even made fair progress. Finally the geometric
picture will fade away. In the long run, visualising may not interfere with
algorithmising. But if need be, the picture can be called in, in particular when
the area of the rectangle is dealt with. Then the fabric becomes, as it were, a
braided mat, the threads become strips, horizontal and vertical; say, one mm for
the thin ones, and one cm for the thick ones. This is not absurd fabric threads
too have a thickness and can be close enough to fill an area.
4.39. Among the four arithmetical operations, division is of course the most
difficult. Its algorithm is complicated and hard to memorise. To make it even
harder, it contains a particularly strange element that has no analogue in the
other algorithms: estimating partial quotients. Few pupils attain reasonable
proficiency and accuracy in long division, and after a lapse of time, with little
opportunity to practice, the algorithm is soon forgotten.
For learning the algorithm it does not matter whether dividing is understood

as repeated taking away or as distribution, though the first approach is more

appropriate to exploring self-reliantly and describing the applied procedure.
Dividing 56789 by 3
then means subtracting 3 again and again, though these subtractions can be
performed at a larger scale, say, 10000 threes at the first step in the present
example

together 18929 times, with the remainder 2. The pupil will initially fail to take
away the largest possible amount. Finally he will understand that the decision
how many times to subtract the divisor, boils down to an elementary division
problem. Large dividends are conducive to subtracting as much as possible in
one stroke.
The above pattern does not differ too much from the definitive one. One

NATURAL NUMBERS

131

writes the partial quotients above rather than at the side, while dropping the
zeroes to get the familiar pattern.

In this example the divisor was a one-digit number. The real difficulties
arise with longer divisors. This extension should not be attacked unless long
division with a one-digit divisor has been reasonably mastered algorithmically.
Technical difficulties can persist even though the principle has been understood.
In view of the mass of calculations required in long division, it may be taken
as a wonder if no mistakes slip in. One out of three long divisions correct may
be a discouraging experience, yet in fact this is the normal achievement. If it is
even worse, there is a big chance that the pupil stops altogether. Such long
divisions are of no practical use any more; if it is serious, one uses the calculator.
Apart from the mass of calculations there is another factor that makes long
division in general difficult. In the case of one-digit divisors the multiplication
tables suffice to find the partial quotients: when dividing by 7, all one has to do
is locate a number
between the rungs of the ladder for 7. This convenience
is lacking in general, say, with the divisor 47. The table of 47 is not something
to be memorised. If 331 is to be divided by 47 one has to scan the multiple of
47 that is just below 331. Rounding the divisor is one way to try it; in the
present case: dividing 331 by 50, or rather 33 by 5. It goes six times. 50 goes
6 times in 330 with a large remainder.
so the correct partial quotient
must be at least as large as 6. But 6 times
which is 49 less than 331.
Thus 6 was not enough. It should be 7, thus

A long reasoning with a lot of computations in order to get this result, which
with a longer dividend is only the first step. Trial and error, and finally the

experience of failing a new experience of uncertainty that is not to the benefit

of the work.
4.42. Divisibility too is a structure of N which is also algorithmically approached.
There are algorithmic rules for divisibility by 2, 3, 4, 5, 6, 8, 9, 10; that
for 11 is relatively simple, but that for 7 is not worth memorising. The threefolds of 37 are funny. The multiples of 142857 show nice regularities which
have profound roots.
The rules, particularly, for divisibility are well-known, but very few people

feel the need for asking why?. The rules are considered to be empirical facts.
Divisibility by 9 can be elucidated by the abacus: Transferring a bead from one
rod to another makes a difference that is a multiple of 9; for instance

Transferring all beads to the rod of the units yields the total of the digits, which
differs from the original number by a multiple of 9. Division by 9 leads to the
same remainder with both of them.

132

CHAPTER 4

In order to see whether a number is prime or to factor it, one tries the prime
divisors 2, 3,... systematically; at
one is allowed to stop.
The greatest common divisor of two numbers is delivered by the algorithm,
named after Euclid.

4.43. Without much ado we have based the algorithmics of N on the decimal
structure of N. I think it should be done this way. Innovators like to do a lot
with structures on other bases. They make one believe that it is mathematics if
one moves into another positional system. It is, however, only a slightly different
algorithm rather than an expression of mathematics. Some of them assert that
the principle of the positional system is better understood if it is embodied

more than once by a variety of systems and not only in base 10. There is,
however, not the slightest indication that they are right, though I do not exclude
other bases for remedial use. Unorthodox positional systems are rather a symptom
of innovation by new subject matter. If compared with mathematics resulting
from pondering more profoundly the subject matter and its relations to reality,
unorthodox positional systems are a mere joke. Jokes are a good thing in instruction. It is good didactics to motivate pupils by jokes, and an unorthodox

positional system may even be a good joke. Of course, other bases can have
their own significance, in particular base two, that is, for computers. A context
that justifies other bases mathematically will be touched on in Section 5.27.

CHAPTER 5

FRACTIONS

5.1-2. The Title

5.1. It is not a slip of the pen fractions rather than positive rational
numbers in the title of the chapter. It looks old-fashioned, this terminology.
To the present view rational numbers are the proper mathematical objects that
are meant here. This view is correct, as a consequence of how the mathematician
interprets his formulae. If a and b are numbers,

a + b is not the assignment add b to a,

rather it is again a number, to wit the sum of a and b. If this is understood,

3 + 2 is again a number,
which more briefly can be written 5, though if you like it you may write
as well, or

then should not be read

if I add 2 to 3 I get 5,
but

3 + 2 and 5 are the same thing

sometimes also formulated as

3 + 2 and 5 are different names of the same thing,

such as, for instance,

Amsterdam and capital of the Netherlands

are names of the same thing.
On the left and right of the equality sign, the same object occurs. Likewise
in

there is talk of again and again the same thing, only represented in various ways;
and this thing is a rational number. Well, one can agree to prefer the way
and
in general, for every rational number, the expression by means of a fraction
where numerator and denominator have the common divisor 1, the simplified
133

134

CHAPTER 5

fraction; as one prefers for the number 5 the expression 5 rather than 3 + 2,
10 5 and so on, though the others are equally well admissible. There is,
however, a difference: 5 is not only the preferred name of the number 5, it is
its first name, the name by which it has been introduced to me, and under
which I first made acquaintance with it, whereas 3+2 and 10 5 are

aliases by which I can also call it up.

, however, is only the simplest name
of a certain rational number, and I would not even be able to say about many
rational numbers under which name I first met them. This then is the reason
why the various fractional expressions of the same rational live so much more
their own lives, and why they are known under a special name: fraction.
But whatever one may feel about it, the mathematical object that matters
is the rational number rather than the fraction. Nevertheless, I put the word
fractions into the title, and I did it intentionally. Fractions are the phenomenological source of the rational number a source that never dries up.

Fraction or what corresponds to it in other languages is the word by

which the rational number enters, and in all languages I know it is related to
breaking: fracture. Rational number evokes much less violent associations;

rational is related to ratio, not in the sense of reason but of proportion,

of measure a learned context, and much more so than fraction.

5.2. In fact, fractions have much to do with ratio, and I hesitated about whether
I should not place the word Ratio under Chapter 5. Not as a substitute for
Fractions but as the subject that deserved priority priority for didactic
reasons but also on behalf of the exposition. I delayed Ratio to Chapter 6,
though repeatedly in the present chapter I shall anticipate it. From the start
I have struggled with problems of priority while I wrote this book, and I can
only hope that the damage caused by that struggle looks bearable. As a matter
of fact, I have turned the present chapter inside out several times. It is the
wealth of phenomena mastered by fractions and ratio that caused the trouble.

In order to write a phenomenology I have to pay attention to all these phenomena; organising them too systematically may mean simplifying so much that it

infringes on the phenomenological task.

It cannot be denied that the didactics of fractions is characterised by unifying
trends. As a rule, natural numbers are approached on a variety of tracks. If it is
the turn of fractions, pupils are supposed to be so advanced as to be satisfied
with one approach from reality. To my view, this wrong assumption is the
reason why fractions function much worse than natural numbers and why
many people never learn fractions.
It is my intention to present fractions in their full phenomenological wealth

I only hope that I do not drown myself in this ocean.

5.3. Fractions in Everyday Language

5.3.1

half as (by analogy with equally as, twice as,...)

FRACTIONS

135

followed by
..., much, many, long, heavy, old,...

compares quantities and values of magnitudes.

Less usual

a third as, two thirds as

..., much, many, long, heavy, old,...
5.3.2

two and a third times as

..., much, many, long, heavy, old,...

is as it were an extension of

twice as
..., much, many, long, heavy, old,....*

Yet

one third times as

..., much, many, long, heavy, old,...
can hardly be considered as belonging to everyday language.

5.3.3

half of, third of, fourth of,...

describes a quantity or a value of a magnitude by means of another. The indefinite or definite article adds nuances
..., a (one), the
half of, third of, fourth (quarter) of,...
a
the
cake, way, travel, hour, pound, money, million,....

So does
..., a (one), the
half of, third of, fourth (quarter) of,...
seven
cakes, hours, pounds, millions,....

Multiples can be formed

two thirds of, three fourths (quarters) of, . . .

a (one), the
cake, way, travel, hour, money, million,....
* In many other languages I would be able to add little, short and few to these lists.

136

5.3.4

CHAPTER 5

half a . . . , half the. . .

is used in the same sense.

5.3.5

From the noun or numeral to the measuring number expressed by

figures

m, kg, 1, sec, bottle, million.

5.3.6

A strange phenomenon I think in all languages is

time or times
after fractions. In Section 5.3.2 we already met with

. . . times a s . . . .

Times belongs to multiplication (cf. Section 531). With a natural number m

it occurs in
m times doing, undergoing, experiencing, waiting for,

something, for instance

m times seizing n marbles (m n marbles),
m times laying down a measuring stick (m times as long),
m times turning the key in the hole,
m times around the clock, the race track, the Earth,
m times rolling of a wheel,

m times swinging back and forth.

At a certain moment fractions are allowed for m. This linguistic extension is
more easily understood if m is a mixed number:
2 times as long

brings

in its wake.

looks unnatural, but the whole number in

FRACTIONS

137

suggest that this means

In cyclic or otherwise periodic processes

might be self-explaining without the whole numbers. In

the fraction suggests an action whose last phase has only partly been performed.
If this is applied to the movement of measuring - for instance the use of a

measuring tape

becomes clearer as a process of fitting a measuring tool periodically, where the

last phase is only partially performed.
5 3.7

The more natural terminology is

3 times...

and also

applied to a
number, quantity of objects, divisible object, value of a magnitude,
such as

7,

30 marbles,

a cake,

5 kg.

but arithmetic and mathematics are better served with one term only. In exceptional cases the times is replaced with of, as in

138

CHAPTER 5

3 packages of 5 kg each,

but this is another of than after a fraction. For the sake of uniformity and
following an old tradition
times takes over from of.
In textbooks this is most often simply prescribed:

In the preceding we sketched natural ways from

of

to

times:

One way is from

2 times
via

to

the other is the cyclic or periodic process:

2 times around the clock,

Later on I will deal with this question again when multiplication of fractions is
discussed.
5.3.8

In another way

of

or

out of,

or

in,

or

to

suggests a fraction in

3 (out) of (in, to) every 5

5 (out) of 100
35 miles to the gallon,
a scale of 1 to 1000.
one chance in a hundred,
3 out of 5 parts.

(people living in cities),

(5%)

FRACTIONS

139

In a mixture:

3 parts salt and 2 parts pepper.

5.3.9

A stronger looking terminology

every third lot wins,
every fifth man is Chinese.

It seems that this is the origin of the ordinal numbers as a means to indicate
denominators of fractions: Counting 1, 2, 3, . . . , 10 to count out the tenth;
all these tenth people or objects together form a (one, the) tenth of the
whole. Thus the tenth part is in fact the last of all of them. In an obsolete
terminology nine parts means
the remainder that is left if the tenth is
counted out. Decimate originally meant counting out the tenth (to be shot).
5.4. THE FRACTION AS FRACTURER

5.4.1. Causing Fractions

We have already explained how magnitudes are divided, with or without a
remainder. In order to divide substance, measured by magnitudes, many methods
are available: fracturing can be

irreversible, or reversible, or merely symbolic.

The equality of parts is judged

at sight or by feel,
or by more sophisticated methods. One of them is
folding in two

in order to halve,

folding in three
in order to divide in three equal parts;
repeatedly folding in two and three

leads to more fractions.

Heavy objects are halved by
weighing the parts in ones hands or on a balance,

while repeatedly correcting the lack of equilibrium. Similarly

comparing and correcting

140

CHAPTER 5

play a part if in general a substance measured by magnitudes is to be distributed;

for instance, a liquid over a number of congruent glasses, where the heights of
the liquid are then compared.
Planar and spatial figures or objects as well as large amounts are sometimes
distributed with regard to area or volume while using
congruences and symmetries;
for instance, the round cake into congruent sectors, which can be done
at sight or by feel.
In all these examples I disregarded proper measuring. I aimed at drawing
attention to more primitive methods. In the mental constitution of all kinds of
magnitudes, meting out fair shares seems to me an important link more important than what is investigated under the title of conservation by psychologists.
As far as I know, developmental psychologists have paid hardly any attention
to this aspect. I have observed many times that 7 to 8-year olds are able to
estimate one half or a third of an irregular area to be coloured; by this ability
they are mastering an important component of the mental object area, whereas
knowledge of the formula for the area of a rectangle as shown by 10 to 12-year
olds need not mean progress on the contrary it can equally well mean retrogression. Earlier I stressed the importance of break-make transformations for the
development of magnitudes as mental objects.
5.4.2. Whole and Part
In the most concrete way fractions present themselves if a whole has been or is
being
split, cut, sliced, broken, coloured
in equal parts, or if it is
experienced, imagined, thought
as such. In this complex of phenomena we will try a classification, illustrated
by examples.
The whole can be
discrete
or continuous,
definite
or indefinite,
structured or lacking structure,
which means extremes with a variety of transitions in between.
The attention can be directed to
one part, a number of parts, all parts.
The parts themselves can be

FRACTIONS

connected or disconnected.
The way of dividing can be
structured or unstructured.

5.4.3. Examples Definite Whole (Figures 2632)

141

142

CHAPTER 5

Out of a bag of marbles discrete definite whole I have taken a tenth; my

attention is fixed on this tenth, and perhaps also on the remaining nine tenths.
From the same marbles in front of me, lying or rolling, structured as a
sequence, I took arbitrarily a tenth an unstructured choice or the first
tenth part or every tenth of them a structured choice.
Out of 60 beads in a bowl
are red, white, blue a discrete definite
whole, structured according to colour, unstructured in space. The same as a
string structured as a whole, and if the beads follow each other regularly,
say 1 red, 2 white, 3 blue, also structured as to distribution.
A lottery discrete definite whole, structured by numbering the attention
is fixed on the parts that gain prizes.
A strip continuous, of definite length, with a linear structure with one
or more segments coloured, say, redwhiteblue connected parts, or disconnected here and there some red, white, blue spots divided with or
without structure.
The same with the circular disc continuous, definite, cyclically structured,
divided into sectors, which separately or taken together represent parts (roulette,

spinner, sector diagram).

The same with more or less structured geometric figures:
A square with or without an underlying squared paper structure, or a regular
or irregular polygon with or without an underlying lattice structure, regularly
or irregularly divided.

The edges or faces of a cube, combined into parallel quadruples and pairs,
respectively.
Curvilinearly bounded planar domains, or spatial domains bounded by
curved surfaces, regular or irregular, divided in various ways.

5.4.4. Examples Indefinite Whole (Figures 3334)

Mankind discrete indefinite whole divided according to blood groups, where
attention can be paid to one or more of them; the whole can further be taken as
unstructured, or as structured according to sex, race, geographical distribution,

and so on.

143

A string of beads of indefinite (possibly infinite) length discrete, indefinite,

linearly structured and as a string of finite length, divided according to rank
numbers or colours, which can be placed in a structured or unstructured way.
A strip of indefinite (possibly infinite) length continuous, indefinite,
linearly structured and as a strip of finite length with coloured segments
or spots, structuredly or unstructuredly distributed.
A wall or a tiled floor continuous, of indefinite extension, structured in
a pattern of bricks or tiles divided according to colour, gloss, pictures, material
disconnected pieces, structuredly or unstructuredly distributed with or
without structure.
The air continuous, indefinite, a structureless whole divided into gases,
oxygen, nitrogen, and so on, connected parts, structureless distribution.
The soil continuous, indefinite, a structureless whole divided according
to categories of use disconnected parts, structureless distribution.
Print in a certain language discrete, indefinite, a structureless whole
divided according to letter symbols, structureless distribution.
Time continuous, indefinite whole structured according to various
criteria.
Remark. All these examples are to be taken with a grain of salt. Where I
called something structured, it is possible to neglect the structure if it does

144

CHAPTER 5

not matter, and where I aimed at structureless, one can introduce structure.
There are transitions between discrete and continuous: particles can be so small

that the whole looks continuous. In a discrete whole, connection can be built
up from neighbourhood relations.
Remark. Most of the examples have been presented in a static way: something is, or is considered as divided. One can, however, read them also in the
sense of something being divided. In some cases this is even more natural, as
it is with the beads in a bowl, the lottery, and the colours of strips and planar
domains; in the latter examples (air, soil, print) it seems less plausible, but
one can imagine situations where even here the stress is on is being rather
than is divided.
5.4.5. Whole, Part, and Fraction
Fractions were explicitly mentioned only at the beginning of Section 5.4.2,
though it was the aim of the general exposition in Section 5.4.2 and the examples
in Section 5.4.34 to relate the parts and the whole to each other by fractions.
Parts and whole are numerically compared according to measures that can vary
greatly.
The question of how many times a part goes into a whole is meaningful only
if one has agreed on the condition under which parts are to be considered as

equivalent. The criterion can be

number
or
value of a certain magnitude.

This will be elaborated later.

In spite of the many sided classification and the wealth of possible examples,
the approach to fractions from the point of view of partwhole is much too
restricted not only phenomenologically but also mathematically this approach

yields proper fractions only. The traditional didactics of arithmetic restricts

itself to this approach, mostly even in the narrow sense of dividing a cake.
After these concrete cake divisions with proper fractions only the learner
is immediately introduced to dividing abstractly presented quantities and values
of magnitudes; with arbitrary decrees like
times means the same as of;
and with arithmetical rules a straight way is leveled to the rational number.
Some innovators inserted a stage of fraction operators as inverses of multiplication operators. This could have been a progress were it not that even they
are satisfied with too small a basis of orientation.
Pupils with a knack for digesting algorithms learn to operate on fractions anyhow, pupils who are less or not at all gifted in this specific way learn it by trial
and error or not at all. After one or two years of fractions, some pupils master
the algorithms though they have no idea what fractions mean and what you can

FRACTIONS

145

do with them; others do not even know the names of the particular fractions.

The phenomenological poverty of the approach seems to me largely responsible

for this didactic failure.
5.5. F R A C T I O N S AS C O M P A R E R S

5.5.1. Comparing Concrete Objects

The traditional didactics overlooks the fact that the concreteness of fractions
does not stop with breaking a whole into parts. As the linguistic analysis of
Section 5.3 showed, fractions also serve in comparing objects which are

separated from each other

or are

experienced, imagined, thought as such:

in this room there are half as many women as there are men,
the bench is half the height of the table,

the street is
times as wide as the footpath,
John earns half as much as Pete,
copper is half as heavy as gold.
Comparing is performed according to certain criteria,

directly and indirectly.

Directly: the objects which are to be compared are brought close together, or
are in some other way considered, as though the smaller were part of the bigger,
by which strategy the fraction as comparer is reduced to the fraction as fracturer
of one concrete object.
Indirectly: a third object, say a measuring stick, mediates between the two
objects to be compared by being, or regarded as being, transferred from the one

to the other.
The above examples admit of another formulation:

the number of women in this room is half the number of men,

the height of the bench is half the height of the table,
the width of the street is
times that of the footpath,
Johns income is half of Petes,
the (specific) weight of copper is half that of gold.

Rather than
objects with respect to number or magnitude value

we now compare

numbers or magnitude values themsleves.

146

CHAPTER 5

It looks like too much sophistication to make this distinction, and in the
unsatisfactory phenomenology of psychological research as well as in traditional
didactics it is disregarded. In our phenomenological analysis it is not superfluous.
One should fully realise that comparing with respect to number or magnitude
value precedes comparing numbers or magnitude values themselves, and that the
former remains, or should remain, imminently present in the latter as long as
fractions are to be more than a formalism.

5.5.2. Fraction and Magnitude

Earlier we explained how distributing into three equal parts can take place: with
small quantities at sight, with larger ones by alternately taking away equal parts,
or algorithmically by division, as the inverse of multiplication. If the division
terminates, no new problem turns up. If not, then in realistic problems the
question arises of what to do with the remainder. If its division is feasible, then
the mathematical distribution problem and its relation to the real one have
changed. It is no longer a finite set that is distributed; the finite set model does
not fit the real distribution problem any more. For instance rather than six

discrete loaves of bread that are distributed, it is bread, in a quantity that is

thought to be arbitrarily divisible and according to a rule that states when
quantities can replace each other in order to be considered equal.
The mathematical model that fits this task of distribution, is magnitude.
It was already discussed in Chapter 1, and it is a subject to be dealt with once
more. Meanwhile we repeat the essentials:
To constitute a magnitude in a system of quantities requires:
an equivalence relation, which describes the conditions for replacing
objects (for instance quantities of a certain substance) with each other
and which leads to equality within the magnitude,
a way of taking together objects (quantities), which leads to an addition

in the magnitude,
the unrestricted availability of objects with the same magnitude value (that
is, in the same equivalence class), which makes addition unrestrictedly
possible,
the possibility of dividing an object into an arbitrary number of partial
objects that replace each other, which leads to division by natural numbers.
Multiplication by natural numbers is a derived operation defined by repeated
addition; nth part of becomes the inverse of n times. By composing multiplications and divisions with each other one gets multiplications by rational
numbers.
If we restrict ourselves to mathematics only, then in order to define what
magnitude is we could be satisfied with postulates on addition and division. In a

phenomenological approach we must start with objects which by an equivalence

FRACTIONS

147

relation are required to form classes representing magnitude values. The unrestricted availability of such objects in each class is indeed indispensable. I
stress this point which via a defective phenomenology has produced a defective
didactics of fractions.
Our exposition shows an asymmetry between multiplication and division.
The operator nth part of can be applied to the object before it can to the
magnitude value. An nth part can be a concrete part of a given something.
On the other hand, n times cannot be realised by means of the given object;
one has to call in others, perhaps arbitrary ones, whereas nth part can be

realised within the object and only the choice of the part is arbitrary.
This asymmetry is so striking that no phenomenology can be allowed to disregard it and no didactics of fractions may pass over the results of this analysis.
It is, however, just the point where the traditional didactics of fractions shows
its defects, for which a defective phenomenology is likely to be made responsible.
The fraction as part of something is of such a convincing and fascinating con-

creteness that one is easily satisfied with this one phenomenological approach
and forgets about all others. In all examples, whether visualised or not, one
restricts oneself to fracturing. The nth part is exclusively seen or imagined
within the whole something that would not be feasible with n times .
Phenomenologically this approach leads to proper fractions
only. The
insufficiency would appear as soon as mixed fractions
are taught, but
when this point is reached, mathematising fractions and the operations on
fractions are already in full swing if not completed; the required extension to
mixed fractions is simply dragged along in the stream of mathematising, or
accomplished purely formally without any phenomenological bonds. Expressions
such as
are paper work, unrelated to reality, which is still visible in the
proper fraction.

The fraction as fracturer is not only too narrow a start, it is also one-sided.
It is strange that all attempts at innovation have disregarded this point. Modern
phenomenological analysis has carefully approached the concept of magnitude;

the part played by equivalence and fractions has been recognised, but this
phenomenological analysis has never taken a didactical turn. In particular, it
has not been realised that the didactics of magnitudes cannot be built on that
of fractions, which in turn require magnitudes to be approached didacticallyphenomenologically. The fraction as fracturer can be described by a quite
restricted equivalence concept; it does not require any more than dividing
something into n equal parts. But in the didactic reality an equivalence of
broader scope is needed, as well as the unrestricted availability of objects in
every equivalence class. So far this need has not been recognised in the didactics
of fractions and in the choice of didactical models.
5.6. Aspects of the Fraction
Let us summarise the contents of Sections 5.45 formally and replenish it.
As the mental stress is on

148

CHAPTER 5

acting or stating
the fraction appears

in an operator or in a relation
halving versus half as big.
Both the fraction operator and relation can work on and relate to each other,
respectively,

objects
with respect to certain characteristics (number, length, salary, weight . . . )
half of the stick, the bench is half the height of the table, and so on or

quantities and magnitude values

this length is half that, this weight is

If the objects to be compared are

times that.

part and whole

or are considered as such, the fraction appears in the

fracturing operator or relation.

If they are

separated,
it is better to speak of the

ratio relation.

If it is about quantities and magnitudes, the fraction occurs in the

ratio operator,
which transforms a number, length, weight into another one.
From the ratio relation as stated between objects one can pass to the ratio
operator, which acts on quantities and magnitudes, by an intermediate stage,
the fraction in the

transformer,
such as mapping a half scale, stretching :
performed

on the object itself,

though not by breaking, but by

mapping and deforming.

times. This operation is

149

FRACTIONS

If we leave the concrete sphere around fractions step by step we arrive at

the
fraction as measurer

preceding a unit

in

unit, as is the case with-

kg,

m,

cc,

bottle or without a

... that measure segments

on the number line;

the fraction operator as

inverse of the multiplication operator;

and the

fraction as rational number.

As we explained earlier, the traditional didactics knows the fraction only

in the fracturing operator, from which it passes straightforwardly to the end
of the sequence: the fraction as rational number.

5.7. The Fraction in an Operator

Among the aspects in Section 5.6 we meet an operator aspect on three occasions,
that is,
the fracturer

which claims to act on concrete objects by breaking them into equivalent

parts,
the ratio operator,

which puts magnitudes into a ratio with each other,

the formally defined fraction operator

in the number field.

The differences look sophisticated, but didactically they are not so the
medium in which the fraction operator acts is being stripped of its concreteness
in a stepwise manner. Initially it acts on the objects cited concretely, while their
magnitude aspects are the factors that check the fairness of the distributive
procedure. Next, the magnitudes themselves are objects, while the concrete
objects measured by them are disregarded or passed over in silence. There is
a remarkable intermediate stage, the transformer, which, as it were, preserves
the substance while changing the magnitude values proportionally. Finally, the
fraction operator acts in the pure number domain, where it satisfies the need
for inverses of multipliers.
In the ratio relation the ratio operator is, as it were, coagulated, from an
operation to a relation between the object operated on and the result. The

150

CHAPTER 5

fraction as measuring number, as spot on the number line, and finally as rational
number is the result of applying the fraction operator to a unit. In all aspects of
the fraction, the operator aspect is felt. In a didactics of fractions it should be
appreciated accordingly, and in modern approaches it is in fact done. Unfortunately, this is allied with misconceptions, expressed in such formulations as
the fraction as operator. Logically such interpretation is of course, feasible
number and vectors, too, can be interpreted as operators. Elsewhere* I have
shown the didactical rocks on which this logic must founder. The interpretation
of the fraction as an operator is untenable, as is the involved terminology. One
badly needs the fraction as a number, which for that matter may have arisen by
applying a fraction operator to a unit. This means that in the fraction operator
one must distinguish the operator from the fraction. The operator with a fraction in it cannot afford a second self in the form of the fraction as an operator.
It is a fact that the operator aspect is more important for fractions than it is
for natural numbers. In the constitution of the mental object natural number
the growing together of the cardinal and ordinal root is decisive, and only after
natural numbers have been constituted are they used in operators such as three

more than, three less than, three times (as much as, as many as).
Fractions, however, show the operator aspect from the start, which justifies
a didactics which calls itself by exaggeration the operator interpretation of
fractions.
An operation known as early as natural numbers is distributing. If a finite
set of equivalent objects is distributed into three equal parts, say, among three
persons, each part is a third, that is, a third of the whole a strange terminology
whose troubled source I have uncovered in Section 5.3.9 yet we are so accustomed to this strange use of the ordinal numeral that we are not aware any
more of its curiosity, let alone inclined to protest it or to ask ourselves why
year after year hosts of pupils do not grasp it.
5.8. Models of the Ratio Relation (Figures 3538)

* Mathematics as an Educational Task, pp. 260262.

FRACTIONS

151

The universal model of magnitude is positive number visualised on the number

line as length, although other models may be equally useful didactically, in
particular if fractions are concerned: area, volume, weight, time, to mention a
few. Lengths and areas have their own visualisations; with some precaution
volumes, too, can be graphically visualised, though proper spatial visualisations,
which at the same time can be rather palpabilisations, are highly recommended.

152

CHAPTER 5

Weights can be visualised linearly, on the scale of a spring balance or on the

beam of an old fashioned steelyard where a running weight is displaced; time is
visualised on the time axis, unreeling the clock dial, as it were. Each of these
models deserves our attention since it can be useful in the ratio relation.
I did not mention here the classical model for fractions, the pie distribution,
not to mention more recent ones, such the fractions boxes. In the didactical
practice they may certainly not be skipped. The pie distribution is the predecessor of general sector divisions of the circle that are applied as statistical sector
diagrams and in roulettes and spinners. As didactical models for fractions they are
especially effective if various sectors are to be taken together in order to make
assertions about m out of n parts or p parts of this against q parts of that.
On a spinning top, mixing p parts of one colour with q parts of another colour
to get a certain colour shade is an effective illustration of mix ratios. Likewise,
one can mix liquids in a given ratio and illustrate the mixture in a sector diagram.
Handsome illustrations are offered by strings of beads, walls, and other patterns
where beads, stones, and so on of various colours or shapes alternate regularly in
a certain fraction ratio three white and two black an indefinite whole where

no limits are suggested. If the subject is fractions, the particular shares will be
expressed by fractions. Likewise, the fraction box can be handsomely used to
display histograms, but I should say that I never saw it used in this way.
Whoever uses these traditional models should bear in mind that they do not
suffice. Their rude concreteness should not seduce him to trust this narrow
approach. The pie distribution takes place within the pie; the circle to be divided
is the universe that is divided into sectors. The clock dial can be handled more
smoothly: by the relation to time the restriction to one hour or half a day can be
removed; the dial can, as it were, be unreeled on the time axis. The fraction box
is the most restricted tool; it resists not only extending but also refinement. The
drawn rectangle has more opportunities, but as long as the rectangle is only
subdivided, it is not worth much more than the rigid fraction box.

Lengths and areas are the most natural means to visualise magnitudes with
respect to teaching fractions. Lengths arise from straight long objects by means
of congruence as an equivalence relation; if arbitrary long objects are admitted,

congruences have to be amplified by break-make transformations or flexions.

Areas arise from planar objects by the equivalence relation of area equality,
which will be dealt with in Chapter 13; congruences and break-make transformations contribute to the extent of this equivalence class. In the process of pie
division the circle sectors are compared by congruence, which should guarantee
the equality of area or volume.
Line segments are the most simple visual representatives of magnitude values.
Two magnitude values in a fraction relation are easily visualised by two line
segments in the same ratio (Figure 39); in order to make the ratio comparison
easier, relevant parts can be marked; the representing line segments are by
preference taken parallel. This, however, is not the only way. Two trees beside
each other (Figure 40) that are in fraction relation, which can be stated by

FRACTIONS

measuring or using intermediate scales, two books with thicknesses in fraction

ratio, ages on the time axis; weights on the scale of the spring balance are other
examples. Most of these representations show more than one linear extension,
which means that the other extensions can also be discussed. They are not as
thin as pure lengths, indeed.
Thinner lengths can be stylised by low rectangles, strips which are the same in
one extension and variable in the other (Figure 41). In order to systematise this
and to facilitate comparison, one may draw the rectangle on a squared paper
background (Figure 42), where comparing is reduced to counting. But again,

153

154

CHAPTER 5

this should not be the only way. One should admit figures which overlap or run
counter to the squared paper structure.
I stress once more that in all these cases the pairs of geometric objects line
segments, planar domains can be present in their own right to embody fractions or they can be representatives of other kinds of pairs of objects two
trees, two books, two heavy bodies, two time intervals which are to be understood in their fraction relation. Then quite concrete couplings can arise: weight
and price on the balances in shops, weight in the scale and length on the beam
or on the scale of the spring balance.
5.9. Models of the Ratio Operator
In the most natural way, of is realised by two figures, one of which is of
the other in length or area. Yet this procedure represents of unsatisfactorily
as an operator. It is as though one would illustrate a function not by a graph but
by one point of the graph. For a linear function this is, indeed, enough, but in
no way does it satisfy our expectations. To show the action of

of in its

whole domain, other devices are needed.

The most popular device today is to suggest a machine in the present case
it would be the
of machine. It is most often merely a verbal suggestion
illustrated by a conventional picture. The input of the machine is numerical
data, which however can also be represented geometrically. The machine itself
does not show any structure, geometrical or otherwise. It is a black box. As
far as my experience extends, textbook authors, teachers, and pupils use these
machines merely verbally, with no relation to any concretised fraction operation.
It is my impression that the machines owe their origin to attempts at introducing

the concept of function rather than functions as mental objects; the false concretisations which are then unavoidable have adopted here the form of a pseudo
concretisation: a verbal suggestion.
More concreteness is provided by the picture of flow distribution in order to
embody fractions (Figure 43). As a matter of fact, the magnitude flowing

in and out, exists only in imagination it is replaced, as it were, by an indefinite

time , but the branching image can portray the fractional part (and its complement) with a geometrical precision.

FRACTIONS

155

Whichever model is chosen, one is free to interpret it arbitrarily, for instance,

the flow image as length, weight, money, and so on.

5.10. Mapping Models of the Ratio Operator

A complete geometrical as well as global picture of the fraction operations is
obtained as soon as they are genuinely interpreted by geometrical operations.
If in order to do so, lines are mapped, there are a few possibilities, all of them
affine mappings (Figure 44).

central projection of parallel lines (lamp shadow),

parallel projection of, say, orthogonal lines (sun shadow),

composition of two parallel projections (such as used in the graphic representation of a linear function).

Performing the geometrical constructions in detail can be both advantageous and

disadvantageous: all the details become consciously clear, but the procedures
are protracted.
A more attractive way is to use
planes, rather than lines, that is, projection planes.

The detailed constructions are even more difficult to perform, but they can
readily be dispensed with if the pictures are differentiated to show clearly which
points correspond to each other in the original and the image (Figure 45). What
I mean is two pictures beside each other, one an enlargement or reduction of
the other, where the same ratio relation can be stated for each particular detail.
The same can be done in three dimensions by building models in different scales.
A danger one should anticipate if one uses such two- and three-dimensional
representations is the possible confusion of length, area, and volume ratios.
Nonetheless even if it is ratio of lengths that matters, planar figures are to be
preferred as means of representation because of their more global expressiveness;
in order then to stress length, one can relie on two artifices:
as planar figures one chooses narrow strips, which are transformed according to length only, while places are distinguished by means of ornaments,
or one takes plain two-dimensional parts, which are transformed according
to both extensions, and to which one attaches drawings that suggest one
extension, such as worms, snakes, whips, spectacle frames.

156

CHAPTER 5

157

FRACTIONS

5.11. Mathematical Theory of Rational Number from the Point of View of the
Ratio Operator
It is well-known how rational numbers are introduced, starting with natural
numbers (or integers): one considers pairs (fractions) of integers with a

non-vanishing second member and prescribes an equivalence relation

rational number are then the equivalence classes of these pairs. The arithmetical
operations are defined appropriately for the pairs, and accordingly for the
equivalence classes.
I now sketch how it is done if the multiplication operator is chosen to start
with, and an a priori genetic rather than a posteriori axiomatic way is followed.
Fractions then are not the result of a definition; instead they are discovered
and described.
We consider a magnitude S and within S multiplications by natural numbers
which form a set M, with composition as an operation in M. M then is
a
commutative semigroup
with identity and a
cancellation rule:
Such semigroups can in general be extended to groups, which is easily proved.

In the present case it is even easier because the elements of the semigroup
are given as multiplications within a magnitude S. I display the sequence of

steps (l.c. italics stand for natural numbers

(1)

):

k times is a one-to-one mapping of S into itself.

(2)

The inverse of k times is called kth part of .

(3)

All the k times form a set M; the kth part of a set

(4)
(5)

M is closed and commutative under composition.

(6)

Given a set T and one-to-one mappings

of T on itself, then from

one concludes:
If

(7)

and

commute, then
moreover

and

do so also, as well as

and

Applying (6) on S instead of T and two elements of M instead of

one gets

158

CHAPTER 5

with composition as its operation is commutative.

(8)

From the last part of (6) it follows that

(9)

One defines

which according to (7) can also be written

Here

is not yet meant as a symbol for a rational number. It is

rather an arbitrary symbol, expressed by means of m and n.

(10)

The multiplication rule

is derived from (9), (8), (7).

follows from (8) and (2).

(12)

The cancellation rule

follows from (10) and (11). This allows one to introduce rational
numbers as classes of fractions.

(13)

The

of) form a set N, which according to (11) is closed and

commutative.
(14)

of ) is a one-to-one mapping of S onto itself with {

of) as

its inverse.

(15)

N is a commutative group of one-to-one mappings of S onto itself.

It looks awfully complicated, though it reflects nothing more than the

occurrence of rational numbers in multiplication operators; addition is lacking,
and the rational numbers are not yet freed from their operator formulation.
However, the preceding sequence should not be understood in the way that

any of its steps would be made explicit, except if it is done paradigmatically.

If we take a closer look at what is didactically required in this line of thought,
then we get the following sequence:

FRACTIONS

159

the mental object object one-to-one mapping, albeit specialised to

stretchings and shrinkings of the number ray,
the mental activity of composing and inverting mappings,
the recognition of k times (for paradigmatic k) as a one-to-one mapping,
and the identification of certain mappings as k times.
the view of, and identification of, the inverse of k times as kth part
of or of (known as such from the division task),
the mental composition of k times and m times (for paradigmatic k

and m) and the recognition of the result as km times,

the mental composition nth part of and mth part of (for paradigmatic
n and m) and the recognition of the result as mnth part of ,
the definition and recognition as a mapping of
of as composed of
m times and nth part of, in arbitrary order,
the mental composition of
of and
multiplication rule,
the grasp of the cancellation rules,

inverting

of into

of and the grasp of the

of .

The only steps of the mathematical analysis that do not figure in this imaginary didactical sequence are those where commutativity is ascribed to certain
pairs of mappings. In most cases this property is so obvious that to make it
explicit would cause confusion. The only case where it is required to do this
is the commutativity of m times and nth part of.
It is perhaps surprising that in the mathematical analysis the inverse of m
times is not immediately called times but
of mathematically
viewed, nomenclature is not bound by any rules. We did so, because as has
been added in parentheses the inverse of m times must be first identified
with the familiar and visually rooted mth part of, and labelling this inverse
by of requires a motivation which must be prepared carefully.
The preceding approach can hardly be compared with that of introducing
rational numbers as equivalence classes of number pairs; the approach by
operators follows a didactical sequence, whereas the one using equivalence
classes accounts formally for an already acquired arithmetic ability.
To what degree can the sequence described above and justified mathematically
be realised? Well, this is a badly formulated question. As a matter of fact, this
sequence must be implicit in any didactics of fractions it is rather a check
list. The proper problem is that the sequence is fleshless. Restricting oneself
to this list would be a hobby, inspired by a mistaken hunt for purity of method.
The sequence is fleshless, its basis is too narrow. It is walking with blinkers
which, for that matter, do not sufficiently protect one against disturbances.

160

CHAPTER 5

Both the mathematical and the didactical sequence start with mappings
(multiplications) in a magnitude. This magnitude must be specified somehow,
and the most obvious specification is length, visualised as a number ray (or
number line), which we may suppose to be familiar to the pupils. There multiplications by natural numbers are readily recognisable mappings, as are their
products and inverses. This intuitive recognisability, however, is insufficient;

communication requires verbalisation, which initially might be ostensive, but

gradually should be refined by means of relative and functional linguistic
devices.* A system of arrows from (variable) x to mx to show m times remains locked in the ostensive sphere. More sophisticated linguistic devices are
required; for instance,

indicating a point by A,
its m times image by mA,
its
its

of image by
of image by

which boils down to plotting positive rational scales on the number ray and putting them into a mutual relation.
This could be a quite useful exercise in detail if it were not for the fact that
the number ray is already familiar to the pupils as the infinite ruler, where the
natural numbers are lodged, perhaps even intercalated by some fractions. This

presence cannot be obscured. In fact, it even gets systematised: by applying

the operation of on the natural numbers, which may be assumed to be
prefigured on the number ray, all the rational numbers on the number ray come
into being. This would not be inconvenient were it to occur rather later. As it is,
the rational numbers now fulfill a double task: numbers lodged on the number
line as well as linguistic parts of ratio operators. Of course, in the long run this
is unavoidable, and at a certain moment this consequence must be accepted and
made conscious; but then one must be able to choose this moment such that
the consequence can be made conscious in order that the rational number can
play its double role well and in an undisturbed fashion.
This, however, is not the major objection against the fleshless didactical
sequence. Intentionally I had the fractions marching on a broad phenomenological front. The phenomenological wealth should be put to good use. The
steps isolated in the mathematical sequence should be taken not in abstracto
but in a variegated context. Even if each of the steps could be taken paradigmatically, one should not expect that the didactical sequence in which I try to
realise the mathematical one didactically automatically contains the wanted
paradigms.
* Weeding and Sowing, Chapter IV, Section 15.

161

FRACTIONS

I wish to add that besides the didactic realisation of the mathematical sequence room must be created for

adding and subtracting fractions,

isolating the fraction as part of the ratio operator,
replacing

of b y

times.

Even then I would not yet have accounted for the algorithmisation or formalisation of fraction arithmetic.
I am now going to sketch a rich didactical sequence for the arithmetic of
fractions.

5.1213. A Rich Didactic Sequence for the Arithmetic of Fractions

5.12. Eight bottles of beer, three persons and each of them gets his fair share
a ten-year old girl reacted to this problem by setting up a long division and then
reproaching me that it did not terminate. To my answer yet they did share it,
she reacted as though she had awakened from a dream suddenly she noticed
more things between heaven and earth than are dreamt of in the arithmetic
lessons she had had so far.
She drew sketches of eight bottles beside each other, divided each of them in
three parts, gave each person eight thirds in fact she did not know this word,
but said little bottles and because it was suggested by the total problem
she gave the leftmost to A, the rightmost to C, and the part in between to B.
Possibly there would be pupils who assign all the lowest thirds to A, the
middle ones to B, and the highest to C. Can it be done otherwise?, may be
asked. The children track down a rich variety of solutions. (Permutations disregarded there are 280, but it is not the aim of the question to find all of them.)
The same problem can be posed with other numbers. It is practicularly
instructive to deal with the following ones next to each other:
24 bottles and 5 persons,
26 bottles and 5 persons.
In a visual context the children learn
with respect to an m-partition changing wholes into mths (for small m),
using additive splittings k into
with a view to getting additive
splittings

into

exercising, in particular, the splitting

off of wholes.

The initial notation is k mths; the notation

split off, the initial notations would be
The aim is to

is of a later date. If wholes are

to finish with

162

CHAPTER 5

transfer addition, subtraction, order relation from N isomorphically to

while slackening the visual bond, this can be supported by tables like

At this point simplifying fractions with denominators like 12, 24, 60 can be
practiced.

Then tables are again visualised on the number line where corresponding
points are joined (Figures 46 and 47).

163

FRACTIONS

Multiplications can be prepared as repeated additions; after posting

ask the question: How can you say this in other ways?
Cautious examples of division:
half of

of

of

and even more cautiously

a third of
5.13. The beer is distributed among couples and after the distribution among
both members of each couple a somewhat more compact sequence than the
former, which aims to

transfer addition, subtraction, order from

understand the isomorphism

to

and to

as the product of the isomorphisms

The same situation visualised by tree or flow models yields Figure 48.

5.14 A picture of a flock of sheep; the farmer sells one out of three (that is,

).

Strike them out. What is left? If it had been 120, how many were sold, how
many left?

164

CHAPTER 5

The field of hundred: Colour of the squares red. Can it be done differently?
red. Can it be done differently? Find beautiful patterns!
The same with a wall of bricks indefinite whole.
A lottery with 1000 lots. One out of five wins. How do you fix which ones?
One of the five gets at least its stake back, one third of these gets double its
stake. How many?
There are 10 first prizes that is one out of . . . ?
Take a strip; fold it in two, three. What part is the folded strip of the original
one? Fold it such that it is one sixth.
Strips below each other in a visual ratio m : n. If one is worth A, how much
is the other?
The aim of these problems is
Colour

Recognising and evaluating cases of the function

in numerical

wordings and visualisations.

5.15. So far - has occurred systematically as a number only in measures like

bottle,
strip. The following aims at
constituting, constructing, recognising the function
Here is the tree A.
Draw tree B half as tall as A.
Draw tree C three times as tall as B.
Draw tree D one third of tree C.
Draw tree E five times as tall as D.
Draw tree F one third of E.
I can also write

I possess a
lens

through which I see everything 3 times as big

lens

through which I see everything

and a

as big.

FRACTIONS

165

I look at the flower through both of them in a row (Figure 49).

What is the result in both cases?

A variety of examples should serve to
exercise the composition of the function
in an arbitrary
number, visually supported as well as numerically isolated, for instance,
evaluating (
of) (5 times) ( of) (2 times) applied to lengths and
numbers, and recognising it as

5.16. The following serves to

replace

of by -

times.

As has been mentioned earlier, most of the textbooks do not care to motivate
this equivalence. It is annoying that three times is a natural operation, as is
of , whereas the vernacular does not account for their similar character.
There is, however, one opportunity, as noticed earlier, where everyday language
as far as I know, every language admits of the passage from
of to
times, that is, in cyclic processes:
turn the key
times in the keyhole,
the big hand has gone
times around the clock,
the satellite has turned
times around the earth where is it now?
the merry-go-round has turned
times,
so has the big wheel where are you then?
Irregular circuits at the fair,

166

CHAPTER 5

the roller coaster,

the haunted house,
allow the same questions, as well as forth and back movements,
times swung forth and back,
times travelled from A to B.
This times can also be interpreted as a multiplier of numbers,
times around the clock how many minutes?
around the clock is 60 minutes,

around the clock is . . . ?

times around the earth lasts ... ?

one times lasts . . . .

times the merry-go-round how long does it last, how many horses?

times from A to B, how much does it cost?

And the winding stairs
times around, how many steps?
From the cyclic to the periodic processes,
hundred times ticking (typing, turning of the odometer, jumping)

and to rolling a wheel

how far after 1,2,3 times around,

times?

This leads in a natural way to

times,

times a given length,

which can also be given numerically.

So a natural language of
times,

times is established. As examples

so far I have taken mixed fractions, which I think is didactically important. In
times the whole part 4 suggests what operation should be performed, and
is dragged along. In the progression of the didactical sequence, however,
proper fractions should become more frequent.

5.17. Now

of and

times

are side by side; finally they shall

be identified with each other.

167

FRACTIONS

This can be done by applying them where both are meaningful (Figure 50):

It is critical that
the identification is made conscious
in order to recall it if mistakes occur. Likewise

consciously:

of

times

Identifying the fraction in the fraction operator and the fraction as a rational
number is, however, delayed. was introduced as

5 times

of or

of 5 times.

It shall now become

consciously: 5 times

times or

times 5 times.

In general, other examples shall be repeated in order to

consciously replace

of with

times.

5.18. Given a point A on the ray, the rational scale of the

constructed the expression is read as
times A.

In this picture (Figure 51)

A is systematically

168

CHAPTER 5

to a point

the m times image,

the times image,
the

times image

can be located, such that

to the scale of the rational multiples of A the operation
applied,

times is

in order to validate the formula paradigmatically

From this, by chopping A the

multiplication formula

is

paradigmatically made conscious.

Exercises in special cases such as

are included.
5.19. It seems natural to have the division of fractions join this sequence, namely

as the inverse of the multiplication

In the case

this problem was solved at the end of Section 5.18:

is solved by turning the fraction, representing a, upside down, and in order to

solve

FRACTIONS

169

this result has to be multiplied by b. This, however, does not answer the problem

of division of fractions didactically. Any hint that dividing is somehow related

to reality is lacking in this approach.
Interpreting b : a as a
distributive division,

that is, as a partition of b into a parts is equally meaningless unless a is an

integer. It is more to the point to understand b : a as a

ratio division,
answering the question

how many times does a fit into b?

for instance, if both of them are visualised as lengths. But then it is even more to
the point to ask this question honestly in the context of ratios, which we will
enter in the next chapter. Let us presuppose this context for a moment as a
didactical precondition, with the operational conclusion:
the divisions b : a and bc :

are equivalent,

that is, have the same solution. This is indeed an important principle, which does
not become meaningful until fractions are at stake it would not hold for
divisions with a remainder.

Of course this principle can also be motivated if division is understood as

the inverse of multiplication:

and
do have the same solution x. As well as in the context of ratio the principle
can also be motivated with simple approaches, such as

fits into

as often as 2 into 4

fits into 6 as often as 2 into 18,

fits into

as often as 4 into 7,

fits into

as often as 5 into 12.

The gist of this principle is that

reducing division of fractions to that of integers

via fractions with equal denominators
a procedure that is formally equivalent to

multiplying with the divisor turned upside down,

though it is didactically better motivated.

170

CHAPTER 5

5.20. Adding, subtracting, and comparing of fractions are supported by the

image of the number line, such as that prepared in Section 5.12:
understanding visually the mutual situation of

and

finding for paradigmatic p and q one r such that

and

are comprised in
Adding, subtracting, and comparing are, according to Section 5.12, performed within one

which is produced in each particular case.

5.21. Combinations of additions and multiplications are exercised in flow

models (Figure 52):

5.22. Finally the attainments are exercised in enlargements and reductions.

Pictures I, II, III of the same object,

transition from I to II by multiplication with factor a,
transition from II to III by multiplication with factor b,

from I to III?

(a, b paradigmatic fractions).

5.23. In the traditional didactics of fractions the multiplication is tied to the
rectangle pattern rather than to the fraction operator. In our particular didactic
sequence we chose the fraction operator; in this structure the rectangle model
is not easily accommodated. This does not mean that it should be neglected. It

can be linked to 5.20 if it does not come earlier, though certainly not in the
restricted traditional form of mere sub dividing one rectangle.
The following didactic sequence is based on a previous treatment of areas

of rectangles and similar figures (Figure 53). It is embroidered onto the pattern
of

the cartesian product of

and

171

FRACTIONS

The standard problem is:

calculate the area of a rectangle with sides

and

This can be joined by a sequence:

given a rectangle, find others with the same area.

These rectangles are constructed with a common SW corner and assembled in

tables (Figure 54)

Another version,

given a rectangle, find rectangles with a times its area.

This sequence will be reconsidered in the section on area of Chapter 13.

172

CHAPTER 5

5.24. Decimal Fractions

As early as in Stevins* proposals the decimal fractions have been closely connected to a decimal system of measure. They should again be dealt with in that
context and then problem should be tackled like: Why both common and
decimal fractions; the precision and the rounding of decimal numbers; percentage
and permille (per thousand); the standard notation. Here we restrict ourselves to
subjects closely related to the approach of Section 5.12.
The decimal fractions are consecutively introduced as elements of ever finer
nets
(

N (

31 written as 3, 1),

314 written as 3, 14),

3141 written as 3,141),

and so on. These transitions are parallelled by those from mm to cm, dm, m,

. . . , and from g to dg, hg, kg, and so on.

The connections are again made between the layers

that is
and so on,
thus
and so on.
Addition, subtraction, comparison are performed in each net

separately. The multiplication ties two nets to each other. Because of

one gets

* Simon Stevin of Bruges in his booklet La Disme (1585).

FRACTIONS

173

The algorithm of multiplication is augmented with a

rule about placing the decimal point.
In divisions one takes care, according to 5.19, that dividend and divisor
belong to the same net; that is, transforming the problem into the form

which is equivalent to

a:b.

5.24a. A Didactic Remark

In remedial teaching and observations at teacher training institutions (cf. Section
4.36a), it occurred to me that the usual didactics, which aims at teaching rules
for the place of the decimal point, can lead to a block of insight and of the need
for insight. Once these rules have been formulated and learned, it is almost
impossible to correct wrong applications by an appeal to insight. If they are
needed, such rules should be the terminus of a development, which cannot
be accelerated artificially. The rules should be understood on several levels.
The lowest is to start with the explanation that to the left of the decimal

point are the wholes and that the decimal point is followed on the right by the
tenths, hundredths, and so on, and preceded at the left by the units, tens,
hundreds, and so on. Multiplying by 10 and dividing by 10 change units into
tens and tens into units, respectively. This can be illustrated by an abacus with
a decimal point. Equally useful is a ladder of refinement

which can be related to the measures in the metric system. Multiplying and
dividing by 10, 100, 1000, ... are experienced as an action on this ladder. This
prepares mutual multiplying (positive and negative) powers of 10. It may be

174

CHAPTER 5

asked when the usual notation for powers of 10 should be introduced (cf.
Section 4.36a). However written, multiplying and dividing (positive and negative) powers of 10 should precede the formal introduction and the training of
multiplying and dividing decimal fractions in general. The reduction of multiplications and divisions in this domain to those in N by means of extracting
powers of 10 deserves to be preferred above memorising rules about placing the
decimal point.

5.2526. Decimal Development

5.25. The division of decimal fraction is by this means reduced to that of
integers, that is, to what is called the
development of b : a, or the fraction

in a decimal fraction,

which can be of infinite length.

So far decimal fractions have been dealt with as fractions with powers of 10
as denominators, which means that a division or a fraction is to be transformed
from
to

In order for this to be possible, the fraction in its simplified form must possess
a denominator that is a divisor of a power of 10; that is

the denominator may possess no prime factors other than 2 and 5.

Other fractions do not admit of such a finite development.
Transforming
into

is performed by means of a division

based on

that is to say, at the dividend and finally at the quotient one passes to
new units before performing the division.
In fact this happens successively:

as

after the first division of b by a the units of the remainder are changed
into tenths, with which the division is continued; the new quotient, being
a number of tenths, is put in the first position to the right of the decimal

point;

FRACTIONS

175

the remainder is changed into hundredths, with which the division is

continued; the new quotient, being a number of hundredths, is put in the
second position to the right of the decimal point; and so on.

If the denominator has no prime factors other than 2 and 5, the procedure
terminates with the result wanted. In other cases, an infinite decimal fraction
comes into being,

What is mathematically relevant here differs much from what has been dealt
with so far in this didactical phenomenology. It belongs to number theory and
infinite series, which, however, does not exclude a phenomenological approach
that fits into the present frame.
5.26. There is at this moment no need to place the infinite development of
fractions into the frame of infinite series or, for that matter, into that of infinite
decimal fractions in general. This can be resumed later. There is equally little
need to appeal to number theory in order to explain the periodicity of the
development. It is done in a more elementary way.
A division by n produces at every particular step a remainder which, consider as an integer, is a number
So among the first n partial remainders
there are at least two equal ones. There is a first time in the sequence of
remainders that a remainder equals a previous one. Let us assume it is the jth
that equals the ith. But then the whole procedure runs from the jth onwards
as it did from the ith, that is, the piece from
the ith to the

th quotient

repeats itself periodically. The decimal development of a rational number

eventually becomes periodic.
It can be

purely periodic
or the period is preceded by an initial segment. Examples of the first kind;

of the second kind;

How can we predict what happens?

176

CHAPTER 5

The examples suggest: The decimal development of the simplified fraction is

purely periodic or not

according whether the denominator n

does not

or

does

have a prime factor 2 or 5.

This appears to be correct: The period of the development of

comes into
being when for certain i and j the remainders after the ith and jth division are
equal, which means that
and

leave the same remainder,

when divided by n. In other words,

m is divisible by n.
If n does not have prime factors 2 and 5, this implies that

m is divisible by n,
thus
m and m leave the same remainder,
when divided by n. Thus the period starts immediately after the decimal point.
Conversely: Take a purely periodic development, with a period of, say,
length l. Let the period itself, considered as a natural number, be c. Thus

The expression e between the parentheses can be calculated as follows

thus

thus

Now

certainly does not have prime factors 2 or 5, nor has n. Thus

if the development of simplified is purely periodic, the denominator n does not have prime factors 2 or 5.

FRACTIONS

177

5.27. Other Bases

With respect to becoming acquainted with and working in positional systems
other than the decimal one, the arguments of Section 4.43 can be repeated,
though a certain difference is worth mentioning. In general, one may not expect
that a change of basis creates more insight, even with regard to terminating and
not terminating developments. If it has been understood which denominators in
the decimal system lead to infinite developments, why they are finally periodic,
and which cases are purely periodic, the transition to a new base g can open
new perspectives. The divisors of 10 are replaced with those of g, and this has
different consequences according to whether g is a prime number, the power of
a prime number, or otherwise composite. It depends on the total instructional
situation and in particular the special group of pupils concerned whether the
insights acquired in such a course are worth the trouble of introducing other
positional systems.

CHAPTER 6

RATIO AND PROPORTIONALITY

6.1. A Preface in Between

A first version of the present chapter has been the first specimen of didactical
phenomenology that I produced in 1973 in German. The immediate cause
was a theoretical exposition by an educationalist on instructional objectives,
where as a paradigmatical example the author dealt with ratio. He chose this
example because in the larger work from which it was borrowed, it was a subject
that could be covered by just one objective. I have repeatedly argued that
formulating instructional objectives should be preceded by observing such
learning processes as could reveal what is being, and thus what should be,
learned; and that for observing learning processes as well as for educational
development an indispensable precondition is a didactical phenomenology.
At that time instructional objectives, however, were distilled from prevailing
textbooks and test collections. In order to show how much is lost by this
approach, I seized upon ratio as an example to explain what didactical
phenomenology is or should be. Whoever reads what I wrote at that time
and it does not sound much different in the present version will be struck

by a tightness of style that was not my habit. With hindsight I should say that
this style was conditioned at least as much by the special subject as by my
intention to write a specimen of didactical phenomenology later on I will
give reasons for it.
In May 1975 I lectured in Berlin. It was the first time that I met Christine
Keitel, with whom I had already corresponded. I told her about a manuscript,
which was later published under the title Weeding and Sowing, and about a
didactical phenomenology of fundamental mathematical concepts, which would
be my next undertaking. I promised her a work written in a rigorous scientific
style, with no regard to legibility. Christine implored me dont with an inflection, as though she meant you are not obliged to. For a long time these words
preyed on my mind.
In the summer of 1975 the chapter on ratio existed, as well as a provisional
sketch of Fractions, but the first chapter had still to be started. Sets, of
course. I struggled with it but I did not succeed. The subject was refractory and
the tight style, which I had mastered successfully in Ratio deserted me. I
did not write a single line.
I made up my mind. No sets. Numbers no. Geometry no. Finally I chose
Length, and after a short while the chapter was conceived in detail. (It shows
gaps, it should be rethought.) But again I could not write it that is, not in the
tight style of Ratio an ideal that fettered my mind. Should I straightjacket
178

RATIO AND PROPORTIONALITY

179

a subject that was not created for it, only to have it look in a way that was not
its nature?
Moreover, I was not obliged to. I need not assume scholarly sounding
language to raise expectations of depth. I do not start a career where such a
language would mean a recommendation, and I do not feel happy providing
work for generations to come who would fathom depths in unreadable work.
My spiritual portrait is established, and my ideas on what is scientific need no
correction. Simplex veri sigillum I translate it as: what is true may be said in
plain language.
I knew what to do, but I still did not know how to do it. I read and reread
Ratio. It looked good and well-written. It was clear and the style was honest.
Why could I not write the same way on Length? I tried formulations, to no
avail. Why, I must free myself of this model. I knew Ratio, as it were, by
heart. I must close this drawer of my mind and open another.
So I decided to move from German to Dutch. I would write the phenomenology in Dutch, in order to translate it afterwards. Language is an infection. Dutch
is the only prose in which I never tried profundity.
After I had taken the decision on the language, fresh arguments emerged. The
phenomenology had been started and was intended to help the developers at
my side in their everyday work and in discussions about it; nobody else would
profit from it in the short run translated or not. It was meant for our colloquial talk and would be written in our colloquial language.
This was a preface in the wrong place. Another will be written when I look
back on this work.

6.2. The Logical Status of Ratio

Belatedly I understood that the logical status of ratio is far above those concepts
discussed so far. I also understood why I should separate ratio from fractions.
Ratio is a function of an ordered pair of numbers or magnitude values. So
are sum, difference, product, and quotient, but they are so in an algorithmic
sense: there is a recipe to figure out the function value assigned to a particular
pair, or at least to act as if you had indeed, what do you figure out if you
answer 3 : 4 by ?
Ratio can also be figured out: transformed into a quotient, that is reading

as 3 is to 4
as

3 divided by 4,
but this is the rape of ratio. Then ratio is deprived of what it makes valuable as
ratio.
Ratio is a function of an ordered pair of numbers or magnitude values. But
what about the values of this function? Again numbers, values of magnitudes?
One can interpret it this way, though it is the wrong way. Indeed, this would

180

CHAPTER 6

identify ratio with quotient. It is the meaning of ratio to speak about equality
(and inequality) of ratios without knowing how large the ratio is, to be able to
meaningfully say
a is to b as c is to d
without anticipating that

a is to b
can be reduced to a number or magnitude value

which then for

c is to d
is the same:

With the opportunity offered by numerosity and length I stressed that the
recognition of equality and inequality, of bigger and smaller, phenomenologically

precedes the operation of adding and measuring it is a pity that this simple
fact is impaired in its credibility by wrongly interpreted conservation principles.
If ratio should be taken as seriously as numerosity and length, then equality and
inequality, bigger and smaller, should play a similar role. Anyway the phenomenological exploration should uncover the same roots.
If these suppositions are confirmed they will then the logical status of
ratio in its phenomenological context would be paraphrased as follows:
ratio is an equivalence relation in the set of ordered pairs of numbers
(or magnitude values), formally indicated by
if the pair

is equivalent to the pair

We do not formulate the (axiomatic) postulates to be fulfilled by this particular

equivalence relation.

It is a fact that after choosing a unit e the equivalence class of the pair
can be expressed by one number (one magnitude value), namely that u for which

but this approach is an a posteriori insight, which in fact matters only if e does
not depend on any arbitrariness (for instance, if it is the numerical unit). A
priori, ratio depends on two data, and consequently each proposition on ratios
proportionality depends on four of them.

RATIO A N D PROPORTIONALITY

181

This complex behaviour was what I meant when in the first paragraph of the
present section I placed ratio, as regards its logical status, high above other
concepts dealt with before them. Quotients and fractions are a means to reduce
this complication, to lower the logical status, at the expense as it happens
of insight. One may doubt whether fractions can be insightfully taught if insight
into ratio is lacking it is this doubt that influenced the composition of the
chapter on fractions. The influence could have been stronger, but I did not
dare integrate the phenomenological analysis of fractions and ratio. Should I
not follow the chapters Fractions and Ratio and Proportion by a chapter
Fractions, Ratio and Proportion?
The logical status of ratio which I explained here implies that ratio and
proportion is more intensive mathematics, mathematics on a higher level than
what has been discussed so far. This fact, I think, influenced the tight style of
my first example of didactical phenomenology. By the choice of subject, the
most mathematical at an elementary level, I found my mathematical bread
buttered on both sides. Rather than by my desire to write a didactical phenomenology, the tight style was suggested by the choice of the subject. The attempt
to imitate it with other subjects was badly motivated and doomed to failure.
The reader will have to content himself with this chaotic alternation of styles.
It is rooted in subjects and views on subjects, rather than in states of mind.
6.3. Ratio as a Relation In and Between Magnitudes
In order not to overburden the exposition of the most relevant ideas, I start with
a few concepts, terms, and notations. I will use a rather loose language, with a
minimum of formalisation. For instance, I will speak of equal sizes if objects
of equal size are intended, of equal distances, weights, times where I should
properly say: paths of the same length, bodies of the same weight, intervals of

the same duration. It can even happen that I speak of the ratio of two objects,
where it should be the ratio of size, volume, or weight of the objects, or of the
ratio of two metals in an alloy rather than of that of their masses.
I start with a heavily mathematised example,
uniform motion:

(1)

in equal times equal distances are covered,

which is equivalent to
(2a)

distances are in proportion to times,

as soon as motion is assumed to be continuous, as it should be;

(2b)

distance is proportional to time

is only another wording of (2a), and

182

CHAPTER 6

(3)

the distance is a linear function of time

is again another formulation, as is

(4)

speed is constant,

though it looks different.

A brief comment:
From (1) it follows:

in twice the time twice the distance is covered,

in thrice the time thrice the distance is covered,

and more generally

in n times the time, n times the distance is covered.

Let
be the distance as a function f of the time t. We just noticed:

Replace t with

Then

which read

yields
in

of the time

of the distance is covered.

If in the last formula t is replaced with

one gets

Thus for each (positive) rational

in a times the time times the distance is covered.

Continuity guarantees the same for real instead of rational .

Take two times

Then

t. Put

RATIO A N D P R O P O R T I O N A L I T Y

183

which is the formulation (2a) or (2b). This can also be written

which is formulation 3, or

which is formulation 4.
There are two magnitudes concerned here: time and length; and a function f
that assigns a length to a time, namely the length of the path covered in the
time interval. The ratios considered here are those of pairs in one and the same
system (time or length); the ratios in one system are required to equal the
corresponding ones in the other this is the postulate of the uniformity of
motion.
We designate

ratios formed within a system as internal

to distinguish them from the external ones that are discussed below.
If
are times and
corresponding paths, the postulate of uniformity
says
If we are tempted to interchange the middle terms, we get

again the equality of two ratios, albeit ratios of path to time.

We designate

ratios between two systems as external.

The uniformity of the motion is now expressed by the postulate
the ratio path to time is constant.
Ratios can also be interpreted as
quotients.

In this interpretation
the internal ratio is a number,
the external ratio is a magnitude,
that is, in the present case of uniform motion,

the quotient of path and time: speed.

The whole reasoning, in particular interchanging the middle terms in a proportion, is quite familiar to us. I ask myself whether we sufficiently realise that

184

CHAPTER 6

it need not be as obvious to the learner. Former arithmetic instruction was quite
conscious of this jump. Rather than bridging the gulf, one invented two kinds
of division, ratio division and distributive division. Together with this twin
monster the former awareness of this problem seems to have vanished, and
since no-one today is conscious of the mental jump from internal to external
ratios, nobody raises the question as to whether it could not be too big for the
learner.
The geometrical tradition of Greek antiquity allowed formulations only with internal
ratios; algebraic operations or magnitudes were allowed only in a complicated geometrical
setting. It is a drawback of Greek geometry that, because of the lack of external ratio,
interchanging the middle terms in proportions in general was not allowed and had to be
circumvented by means of complicated procedures. The ancient tradition has maintained
itself in theoretical sciences for long times. Outstanding examples of this habit are Keplers
second and third laws:
in equal times the radius vector from the Sun to a planet sweeps equal areas;
the squares of the times of revolution are in the same ratio as the cubes of the long
axes of the orbits.
This tradition pervaded the theoretical sciences longer than it did commercial and technical
mathematics, where direct, non-geometrised algebraic operations and, in particular, external
ratios were admitted earlier; even today pure mathematicians often show little understanding
for calculations with magnitudes.

I used uniform motion as a paradigm. The generalisation may be left to the

reader. It will be clear enough what I meant by
internal ratio (within a magnitude)
and

external ratio (between two magnitudes).

It is equally obvious that in mapping magnitudes,

invariance of internal ratios,

and the equivalent
constancy of external ratios
means

linearity of the mapping;

in our example,

uniform motion is a linear mapping of time on path.

The linearity of a mapping f is correspondingly defined in two ways:

implicitly (postulatory),
to the sum corresponds the sum,

RATIO AND PROPORTIONALITY

185

explicitly (algorithmic),

for all x and a certain

Once more: all this is so obvious that as mathematicians we do not worry
about it any more, but let us not expect that it passes by mere diffusion from
our unconsciousness to that of our pupils.
Things are even more involved: A uniform motion has all time intervals of equal length,
wherever they might be, mapped upon equal path intervals. It is not explicitly mentioned
that the composite of two connected time intervals is mapped onto the composite of the
corresponding path intervals because it is implicit in the idea of motion. The same holds
for other pairs of magnitudes, such as volume and weight of some substance. If, however,
f is a function that maps magnitudes onto each other, I have already abstracted from the
particular time and path intervals (and similar ones); they have been superseded by lengths
and durations (and suchlike). So I am obliged to require explicitly that, just as in the sphere
of objects where composites correspond to composites, so in the sphere of magnitudes
sums correspond to sums. In a more formalised setting I could have formulated this more
sharply, but I would avoid too much formalism.

6.4. Expositions and Compositions

Ratio must be viewed in a broader context than that of relations within and
between magnitudes. I want to sketch it by such disparate examples as:

6.4.1

a set of animal species with their average weights (or other quantitative characteristics),

6.4.2

a set of flight connections with their prices (or distances),

6.4.3

a set of countries with their numbers of population (or their areas),

6.4.4

a set of articles with prices (or weights),

6.4.5

the set of components of an alloy with their masses,

6.4.6

the set of age classes of a population with their numbers,

6.4.7

the set of categories of soil use of a nation, with the corresponding

areas,

6.4.8

the set of diseases with the number of cases of each one,

6.4.9

the set of pairs of points of a plane with their mutual distances.

The common feature in these examples is

a set, in general indicated by

in the sequel,

and
a function, in general denoted by
values of a certain magnitude.

in the sequel, which accepts

186

CHAPTER 6

Between the first four (6.4.14) and the following four (6.4.58) there is
a profound difference:
In the first group the elements of are
objects in a primitive sense, while
is defined by common traits of its
elements (species of animals, flight connections, countries, articles),
in the second group the elements of

are

classes of a universe, formed according to certain criteria that are important

for that universe (ages in a population, and so on).
In the first group
the function

describes internal properties of the elements of

in the second group

the function
describes the size of the class (not necessarily a whole
number, cf. 6.4.5).
I will call, quite arbitrarily, the first and second kind, respectively,

expositions,
compositions,
The ninth example, a not unimportant one, is wholly different from the
preceding ones; in Section 6.5 we will return to it.
Expositions and compositions differ in how they are used. Usually they occur
in couples. Anticipating a general formulation, I will explain this by examples.

Couples of expositions:
a set of countries,
the function that assigns to each country the number of its inhabitants,
the function that assigns to each country its area;

the ratio to (population density) is variable: a country has in proportion

the same (a larger, a smaller) number of inhabitants.
a set of filled plastic bags in a supermarket, on which are indicated:
the price
the weight
the ratio to
(unit price) is variable, for bags containing the same article,
it will be the same; on these and are linearly dependent.
Couples of compositions:
We consider two alloys with the same components. The components of the
alloys form two sets

and

RATIO A N D P R O P O R T I O N A L I T Y

187

with the corresponding mass functions

for instance,

30 kg bronze consisting of 20 kg copper and 10 kg tin,

65 kg bronze consisting of 40 kg copper and 25 kg tin.
Both

and

are the set

{copper, tin}.

In the two alloys we have, correspondingly

In general the ratio to can vary; if and are linearly dependent, it is the
same alloy.
Two populations The Netherlands and The Philippines are partitioned
into age classes.
and

are formed by the age classes

{[0,1), [1,10), [10,20),...}.

and
are the number of people in the respective classes.
In one population there are in proportion fewer babies, more aged people,
and so on, than in the other.
The case of a couple of expositions consists of
one set
with two functions

on it;

whose mostly external ratio is considered.

The case of a couple of compositions consists of
class partitionings

and

of two universes, attained according to the

same principle and identified in a natural way with each other;

with two functions

on it,

whose mostly internal ratios are considered and perhaps compared.

6.5. Constructs
We pass to the example 6.4.9. It shows
a set based upon a strong preferably geometrical structure
a measure function.

with

188

CHAPTER 6

In our particular case was a planar figure, for instance, the whole plane,
the set of pairs of points, the distance.
Other possibilities would be:
the set of plane curves with

the set of rectangles with

as the arc length;

as the area.

I will designate such a system

as

a construct,
or more precisely
a

(if

is the structure on which it is based).

Constructs, too, are used in couples,
that
and

and

where it can happen

A couple of constructs:
is the set of pairs of points of a planar figure
is the set of pairs of points of a planar figure
are the corresponding distance functions.
Moreover, there is a mapping
f of

in

which extends itself in a natural way as a mapping

f of

in

A property of f that may be relevant, is

similarity.
As with uniformity of motion in Section 6.3, similarity can first be characterised
by the condition
f maps pairs, with the same mutual distance, on pairs with the same
mutual distance,
or
f conserves equality of distance,
or a richer, but equivalent, formulation

f maps pairwise congruent on pairwise congruent figures,

or

f preserves congruence.

RATIO A N D P R O P O R T I O N A L I T Y

189

This formulation does not yet involve ratio, but under the natural
condition of continuity this characterisation is equivalent to
f preserves ratios, that is,
This is

preservation of internal ratios,

taken in

and in

respectively. As in Section 6.3, in the case of magnitudes,

f being a

similarity
can be expressed by the

constancy of external ratios:

Another example:

a set of line segments, the length function,

the set of squares on these line segments, the area function.
The use of this couple is obvious.

6.6. The Occurrence of Ratios in Sections 6.35 Compared

In Section 6.3 ratios occur in and between magnitudes, in Section 6.4 in and
between expositions and compositions, in Section 6.5 in and between constructs.
The cases 6.3 and 6.5 resemble each other because of the underlying strong
mathematical structure, whereas in Section 6.4 these structures are weak.
Sections 6.3 and 6.5 also have in common that

proportionality and similarity

can be defined
without involving ratio
purely by
preservation of equality or congruence.

In the case of Section 6.4 this is not possible or it would require complicated
reasoning.
Compared with Section 6.3 the case 6.5 has the advantage that
congruence of figures can be visualised more strongly than equality of
magnitude values.

These distinctions will be seen to have important didactical consequences.

190

CHAPTER 6

6.7. Ratio in Similarities

In our phenomenology the stress is now shifted to didactics.
Ratio as a concept and even as a mental object requires a considerably high
developmental level. For all that, the feel and the look of ratios occur remarkably
early in development. According to Piaget, topological concepts should precede
euclidean ones. We anticipate that this holds at most for such spatial relations
as inclusion, exclusion, and overlapping, but these are relations which no mathematician would consider topological as psychologists do. The acceptance of
truly topological properties that is, stating equivalence by means of one-to-one
continuous mappings is certainly not an attitude that can be placed in early
childhood; it is much too sophisticated to be expected of little children. Piaget
and researchers who repeated his assertions or experiments were seriously confused. From the inability of little children to draw circles and squares so neatly
that they reasonably differed from each other, they drew the conclusion of
topological predominance. Yet at an early age children are able to distinguish
clearly circles and squares, which is the only thing that matters. It is true that

children judge drawings in books or made by adults by other criteria than their
own production a kind of system separation which is worth studying closely.
There can be no doubt, however, that children recognise early the different
sizes of objects, and their being larger or smaller. It is equally certain that they
can handle similarity as an operational equivalence. I would even go so far as to
assert that congruences and similarities are built-in features of that part of the
central nervous system that processes our optical perceptions. The immediate
reidentification of objects after a rotation (of the object or the perceiver) and
after a change of distance presupposes something in the brain like a computer
program for the elimination of this kind of mapping it is riddle to me what such
a program looks like; its existence, which I do not doubt, is like a miracle to me.
At a young age a child recognises drawings and models of animals, furniture,
cars, bicycles, ships as images of these objects it does not matter on which
scale, and whether they are pictured side by side on different scales. How big
is a whale really?, a child can ask, convinced that the picture, except for the
scale, is faithful. Well, sometimes whales are sketched by drawings in one line,
but even the difference between a photograph and a characteristic sketch is
grasped early.
Weighings with a spring balance, performed by Bastiaan (5; 6), were indicated by him on
a horizontally typewritten spring balance with a different scale. He noticed inessential
deviations in the figures (1 instead of 1) but neither the difference in orientation nor scale.
The typewritten image was structurally faithful.
After a sequence of sunny days Bastiaan (6; 1) sees clouds again, and says: It will rain.
I tell him: No, these are very high clouds, where no rain falls out; rain clouds are low and
dark. He: What height are these clouds? I (exaggerating): 10 thousand metres. He:
And rain clouds? I: Thousand metres. He (showing to the ground): So if we are here
and this (showing a height of about 30 cm) is rain clouds, then this (shows about 1 metre)
is no rain clouds.

RATIO AND PROPORTIONALITY

191

Without any hesitation children accept that objects at the blackboard are
drawn ten times as large as on the work sheet, that the number line at the
blackboard has a unit of 1 dm compared with that of 1 cm on the work sheet.
They accept number lines where the same interval means a unit, or ten, or
hundred, side by side. Children would, however, immediately protest structural
modifications that violate the similarity of the image:
what is mutually equal in the original,
should be mutually equal in the image,
which as we know, implies
the invariance of internal ratios,
characterising mappings as
similarities.
Children become familiar at a young age with these
ratio preserving mappings
as we shall call them, if they see planar or spatial figures pictured paintings,
copies of paintings, models of buildings. Systematic deviations from this mapping
principle are noticed; for instance,
the use of different scales in different directions,
the use of different scales for different figures,
the use of different scales for parts of the same figure.
This, however, is not done by making the scales explicit, but with formulations
like
the head is much too large that is, if compared with the trunk,
this is much too long that is, if compared to the width
objections regarding the lack of similarity though with no explicitation of ratios.
It requires more insight into geometrical relations to adduce other criteria,
such as:
what is a right angle in the original,
should be a right angle in the image.
With this feeling or eye for similarity, as I have termed it, the child is of course
still far away from similarity as a mental object, let alone as a concept. I indicate
a number of intermediate stages:
Recognising ratio preservation or non-preservation of mappings,
Constructing ratio preserving mappings,
Resolving conflicts in the construction of ratio preserving mappings,
Operationally handling,

192

CHAPTER 6

formulating,
relating to each other:
criteria for ratio preservation, such as
preservation of equality of lengths,
preservation of congruence,
preservation of internal ratios,
constancy of external ratio,
preservation of angles,
and deciding about the necessity and sufficiency of such criteria.

In the first steps of this sequence

ratios do not occur explicitly,
later on

equality of ratios (internal and external) becomes explicit,

and finally
ratios themselves are made explicit.
It is a sequence similar to those observed with length and other magnitudes.
The strong visualisation is an advantage of the geometrical context of ratio

compared with other contexts. What matters didactically is the

gradual verbalisation of visual reasoning.

Most often the contexts of ratio are not visual but are accessible to visualisation. Early familiarity with ratio-preserving mappings is a support to visualising
such contexts of ratio as are not a priori visual. This, however, requires that the
visualised ratio is somewhat loosened from the context of global similarities. In
order to build a bridge from non-visual to visual ratios, the strict visualisation by
similarity must be weakened.
Similarity, as mathematically understood, is a mapping that extends over
whole planes. In each visualisation one is satisfied with linear or plane figures
that by their size and structure suggest the whole plane: pieces of the real
world and their pictures, wallpaper patterns, and other structures that can be
continued inside and outside. For all activities and their various levels which are
enumerated,
a rich structure of original and image
and the fact, or at least the suggestion that the one is the image of the
other,

are if not required, then at least advantageous.

Too little structure may be an obstacle to the visual recognition of similarity
of figures this holds even for adults. In a wealthier structured material, the
more or less algorithmic criteria for the recognition of similarity can be isolated

RATIO AND P R O P O R T I O N A L I T Y

193

and exercised following the above course of activities, in order to become

operational in a less structured material.

It is a small step from the fact to the suggestion that something is somehow
the image of something else. The lack of any suggestion may be an impediment
to even think of similarities unless the dispensability and restoration of such a
suggestion has been prepared in a learning process.
If pictured rectangles are to be compared with regard to similarity, the
structuring addition of the diagonals alone can transform failure into success.
Path lengths on ground-plans and in reality are more easily compared than
circumferences of bare rectangles. The insight that all circles are similar can
more easily be acquired with structured circles, such as our various coins, which
on the obverse side are even similar in surface details. The bare circle is not a
good medium to reveal the internal ratio of circumference (or distance covered
by rolling it) to diameter. The approach via the external ratios and circumferences using different (and similarly structured) circles is more useful.
I resume this exposition with a list of activities:
Transferring what has been
exercised, recognised, made explicit
with respect to ratios and ratio preservation
in a richly structured context
into a less or poorly structured context.

On behalf of ratios and ratio preservation

enriching a poorly structured context,
introducing a geometrical structure in a non-geometrical context,
translating a non-geometrical context into a geometrical one,
understanding and using contexts that might be geometrical or not as
geometrical images of each other.

6.8. Relatively
Whereas one can go a long way with ratio-preserving mappings without verbalising all that can be seen, experienced, constructed as ratio, other contexts require
an early verbalisation (albeit not of ratio) of such ideas as
relatively (or comparatively).
As a mental object this may be supposed at the end of the kindergarten age.

This chocolate is sweeter because it contains relatively more sugar.

A flea can jump relatively higher than a man.
An air travel to South America is relatively more expensive than one
in Europe.
In the Netherlands there are relatively more bikes than in Germany.

194

CHAPTER 6

In the explicit formulation the word relatively can be lacking, since it is clear
what is meant. The terms
relatively more, as much, less

can be given various shades of meaning

from roughly qualitative to precisely quantitative.
In particular, to establish more or less, estimations may suffice, though
they can be refined by additives like
much, very much, a bit.

Relatively lacks a relation term, which can be

obvious or explicitly added.

For instance:
If compared with the number of inhabitants there are more bikes in the
Netherlands than there are in Germany.
A possible sequence of levels:
understanding that orders (larger and smaller, more and less) can be
relativised (relatively larger, smaller, more, less)
understanding relatively in the sense of in relation to ... , with the
criterion of comparison filled in at the dots,

using meaningfully relatively and in relation to,

completing relatively to in relation to ... in a context,
knowing operationally what relatively and in relation to mean in
general,
explaining what relatively and in relation to mean in general.
There is a number of stages
from roughly qualitative to precisely quantitative,
where finally according to the subject the criterion is

internal or external ratio.

Tasks where such activities can take place can have a visual character:
houses, people, trees on different scales
which ones belong together, and why?
a group of persons, and on another scale, clothes
what belongs together?

RATIO A N D P R O P O R T I O N A L I T Y

195

walls on different scales and of different thickness

which ones are thicker?

meadows with flowers, ponds with frogs, skies with clouds
where are there relatively more?
Other senses can play a part :
a large orchestra that produces relatively soft sounds.

6.911. Norming
6.9. A few examples will be given to introduce the complex of techniques,
wrong use of techniques, attitudes fostered (or rather, not fostered) by these
techniques a complex I designate by norming:
If we imagine the earth as a pins head (1 mm diameter), the sun appears
as a sphere with the diameter 10 cm at a distance of 10 m.

The scale reduction is meant to visualise drastic ratios; one chooses a familiar
unit to start with; it does not matter what the scale is.
If the development of life on earth is thought to have happened in a day,
man appeared one minute ago and human culture started a second ago.

It looks much like the first example: a time reduction which can be illustrated
by a linear drawing. For the largest component day has been chosen as a unit,
whereas in the first example it started with the pins head.
The examples

one out of five children born, is a Chinese,

one out of four cars is a Fiat,
show a preference for ratios normed by one out of ....
A recipe boeuf trois moutardes for four persons,
the meal unit is four, which in many cases will save conversions. (It is, however,
not recommended that one trust ratio in cooking and baking.)
The quality of drinking or swimming water is indicated by
this much salt in one litre
or

that many coli bacteria in one cc,

where the quantities that are actually drunk or bathed in are of quite another
size, and the quantities actually analysed are again of a different size.

196

CHAPTER 6

The production of refuse is measured by a vague unit such as

inhabitants equivalent,

which only serves to estimate and to tax the refuse production of families and
industries.
The power of nuclear explosions is measured in
kilotons of TNT,
a strange norming that serves to compare nuclear bombs with each other rather
than with conventional explosives.

If the cost of living is put at 100 in 1965, it is 147 in 1975 an example of

the much used index figures, where by preference a basis 100 is chosen. In other
cases an average is normed at 100, for instance, for the I.Q.:
the average score in a certain population (at a certain age) is put at 100

in order to measure individual scores with it.

This number 100 links up with the decimal system, while on the other hand,
decimal fractions are avoided as much as possible. In traditional instruction in
arithmetic, percentages and interest were closely connected. This, however,
is not an old tradition; interest was expressed rather by one to . . . (the
tithe means one tenth). By the decimalisation of money, percentage interest
arithmetic became effective. Today the most usual application of percentages is
in compositions:

the whole is put at 100 in order to express the parts numerically.

The aim is
to make different compositions comparable.

The comparison can be

supported by visualisation,

for instance, by sector diagrams. The need to make composition data comparable
is at present the strongest motivation for percentages; moreover, percentage is
a device that presents itself most naturally as soon as, on behalf of comparability,
totals must be normed uniformally. Thus not

if the Netherlands were as large as the FRG

nor
if the FRG were as large as the Netherlands
but

put the area (number of inhabitants) of both of them at 100 (or perhaps
1000), then ...

RATIO AND PROPORTIONALITY

197

Resuming the preceding analysis, one can state a few levels

with respect to making compositions and constructs more perspicuous by
norming one component,
with respect to making compositions comparable by norming the whole
(in general on 100), while the absolute data and the scale factor play such
a subordinate part that they are more or less disregarded:

understanding the norming,

understanding the rationale of the norming,
performing normings where they are required,
performing normings where they are useful,
understanding this activity operationally,
describing it,
and putting it into a larger frame.

6.10. A mistake related to norming is: forgetting about the unnormed data and
the scale factor:
absolute meaning is ascribed to data that depend on norming, in particular,
data derived by different normings are compared without renorming;
the number 100 plays an, as it were, magic part;
percentages derived from different norming procedures are added and
processed to unweighted averages;
it causes surprise and is not understood if, for instance, a party in an
election sees its percentages increase in all districts while the percentage
over the whole decreases;
double norming is applied as in the example taken from a newspaper:
in 1972 the national product per capita of Bresil increased 5%, but this
increase is in the greater part absorbed by the 4% increase of the population in the same time.
6.11. A more subtle and more dangerous feature is forgetting about the unnormed data, for instance, in statistics, if this includes forgetting about
the precision of the normed data:

one out of two, or 50% can have been obtained from a total of two, or by
a rough estimation from a total of a thousand or a million.
Problems of precision can be caused by measurements or by stochastics
as a matter of fact, the source of imprecision in measurements, whether exact
or estimated, is also stochastic. Precision will be dealt with in another chapter,
but meanwhile it makes sense to have touched upon this subject already in
connection with norming relative data. Even in the present chapter we will
touch on it once more.

198

CHAPTER 6

6.12. It can happen that normings take place or are asked for where they do
not matter or are even disturbing. Examples:
A string closing around the equator is lengthened one metre and again closed,
loosely, around the equator. Can a man creep through under it?
The problem is often answered with a question regarding the diameter of the
earth, which, in view of the linear relation between diameter and circumference
of a circle, does not matter.
John and Pete live and work at the same address. By bike it takes John 30
minutes and Pete 40 minutes to go from home to work. John leaves 5 minutes
after Pete. Where does he catch up?
The usual reaction is to ask for the distance between home and work, which
again for reasons of linearity does not matter.
An even more drastic example: a student who must switch from the metric
to the Anglo Saxon system of measures asks: how much is here?
The preceding can be summarised as follows:
Insight into the irrelevance of normings in the case of linear relations.

6. 13. Visualisations
Understanding ratios can be steered and deepened by visualisations. One can

illustrate
expositions by histograms and pictorial statistics,
compositions by sector diagrams and other planar divisions.
Example of visualised expositions: The EEC countries are represented, with
respect to their areas, by
rectangles with the same base and heights proportional to the areas
which are placed side by side as in a histogram; the numbers of population by
a group of human figures (for instance, each representing a million),

where both representations can be combined by

placing the human figures into the corresponding rectangles,
in order to visualise the different densities of population (ratio of number of
inhabitants to area).

Example of visualised compositions: A circle divided into sectors corresponding

to, and with respect to, area

proportional to the use categories of the soil of a country,

for various countries side by side, in order to illustrate
the differences with regard to the use of soil

(more or less agricultural, and so on).

RATIO AND PROPORTIONALITY

199

Such visualisations are a kind of ratio-preserving mappings, with ratios other

than those between distances of pairs of points considered in the last example,
on the one hand,
the ratios of areas, population numbers, use categories,

on the other hand,

areas of planar figures.
A sequence of levels could be:

understanding histograms, pictorial statistics, division of areas and similar

visual representations as ratio-preserving mappings of expositions and
compositions,
constructing such visual representations,
deciding conflicts in constructing them,
understanding the principles of such visual representations, and describing
them;
recognising preservation of ratio as the common principle in the visual
representation; and
describing it.
Furthermore, as regards comparing two or more expositions and compositions
represented in this way:
deciding questions on relatively more, as much, less by means of those
visual representations,
making such decisions possible by means of manipulating the material;
understanding the principles of such decisions; and
describing them.

6.14. Visualisations by Means of Constructs

Constructs can serve to visualise not only ratios and proportions, but also entire
linear connections. One can distinguish graphic and monographic methods:
the graph of the linear function (Figure 55),
the sun shadow (Figure 56),
the lamp shadow (Figure 57).

200

CHAPTER 6

Though used too little, these visualisations are particularly effective didactically.
They are models that fit quite well the ideas on the geometrisation of elementary
instruction. They have a good chance of being seriously exploited.
Internal and external ratios
and their mutual relations
can be efficiently
seen, understood, described
by these models.
When reading Section 6.15 one should remember this fact.

6.15. A Igorithmisations
The counterpart of visualising is processing numerically. Verifying preservation
of ratio of a mapping f is simplified by the remark that the validity of

need not be verified for all all pairs

Indeed the

validity for A, B and B, C

implies the
validity for A, C,
the transitivity of ratio preservation.

(In the case of constructs more simplification can be used which rests on
geometrical facts; in the plane it suffices to check ratio preservation for the
distances from two fixed points; the remainder is guaranteed by congruence
theorems.)
It is less trivial to grasp that preservation of ratio can be described by the
existence of a constant scale factor, that is, by an external ratio.
Another important insight is that the
composition of ratio preserving mappings again yields ratio-preserving
mappings
and to know
how scale factors (external ratios) behave under composition of ratiopreserving mappings.
In the case of magnitudes it is important to notice that the preservation of ratio
is essentially recognisable as
an isomorphism with respect to the addition of magnitudes.

R A T I O A N D PROPORTIONALITY

201

I am going to formulate a few levels:

Simplifying the verification of ratio preservation by means of
transitivity of ratio preservation,
geometrical congruence properties,
external ratio and scale factor,
isomorphism with respect to addition within magnitudes,
behaviour under composition of mappings;
simplifying the construction of ratio-preserving mappings by the same
principles;
deciding conflicts in applying these principles;
understanding these principles operationally, and describing them;
understanding relations between these principles operationally and describing them.
In the course of algorithmisation this is complemented by

understanding ratios operationally in the context of the arithmetic of

fractions; and
describing this relation;
understanding properties of ratio operationally as properties of fractions;
and
describing this relation;
understanding ratio preservation of mappings of magnitudes operationally
as linearity; and
describing it as such;
understanding their properties operationally as properties of linear mappings; and
describing them as such.

The converse, which properly belongs in the chapter on fractions, may

explicitly be added:
understanding fractions operationally in the context of ratio; and describing this relation;
understanding properties of fractions operationally as properties of ratio,
and
describing this relation.
understanding linear mappings in the number domain operationally as
ratio preserving mappings; and
describing them as such;
understanding their properties operationally as properties of ratio preserving mappings; and
describing them as such.

202

CHAPTER 6

Ratio-preserving mappings not only serve in visualisations, but also have their
own cognitive function as models, as shown by our first example, the uniform
motion as a ratio-preserving mapping of the magnitude time on the magnitude
length.
The ratio-preserving mappings themselves are illustrated
graphically (the straight line as an image of the linear function),
nomographically,
by means of the slide rule,
and algorithmised by
proportionality tables (proportionality matrices),
formulae for linear functions.
Levels to be mentioned might be
reading;
constructing;
understanding operationally the principles of the devices; and
describing them;
isolated and in their mutual connection.

6.16. Criteria for Ratio Preservation

The principles by which one
recognises and predicts
that a mapping preserves ratio are more profoundly rooted and less accessible.
They can hardly be cleared up without a prior didactic phenomenology of
particular magnitudes. The following discussion tries no more than to sketch
how this can take place.
I start with an exemplary list of adjectives, whose meaning will soon become
clear:
many, big, long, wide, high, thick, much, full, long-lasting, heavy, fast;
strong, old, sharp, blunt, soft, dense;
bright, warm, red, loud, wet, high;
sweet, beautiful, painful;
clever, interesting, sleepy, difficult;
valuable, expensive, rich.
Some of these words have several meanings (such as bright). The adjective
high appears twice in this list, in the first place it may mean a property of
mountains, in the second a property of sounds, but this does not matter here.
One can ask the questions:
Which properties admit comparatives?
Which properties admit doubling?

RATIO AND PROPORTIONALITY

203

(Doubling stands here as a paradigm; more general would be multiplying,

maybe also halving, dividing, finally also adding.)
How to check comparatives?
How to check doubling?
How to make comparatives?
How to make doubles?
These are questions on factualities, though with a considerable logical or linguistically analytical touch.
The central question is that of doubling. The process of doubling is that
of combining two equals. This is how to transform a tower into one of twice
the height, namely by putting an equal tower on its top. A weight of sugar
is doubled by adding an equal one. Temperature shows that it is not always

that easy ; the temperature of a liquid is not doubled by adding a liquid of the
same temperature; likewise the speed of a rolling ball is not doubled by uniting

it with one of the same speed.

Parameters that, when things are combined, behave additively are called
extensive
number, length, area, volume, weight, energy, brightness (of a light source),

electrical charge, all have this property; others like temperature, colour, sweetness
are called

intensive.
Yet even parameters like temperature, or rather temperature difference, can
be interpreted as extensive parameters, though of a process rather than of a
state. So what are combined are not the states but the processes. As to temperature, for instance, a difference of temperature which is obtained by means of
heating with a source of heat W during a time t, is doubled if the same process
is repeated (actually this holds only within certain limits). In the case of
vectorial velocities this combining with the aim of doubling looks different

again: if A with respect to B and B with respect to C have the same velocity,
A has double the velocity with respect to C.

The principle by which the ratio preservation of mappings can be recognised

and predicted can now be formulated as follows:
Two parameters which are extensive under the same operation ofcombining are in a ratio-preserving relation.
I do not claim that this digging has brought profound wisdom to the surface.

The result is, in a wealthy wording, the criterion to which each able teacher will
appeal, more or less consciously, if he wants to convince his pupils about where
they may use the rule of three and where not. He who works double the
time gets double the money he says for instance, and perhaps he puts twice the

amount of money under two equal intervals of the time axis. Or double the
distance, in double the time with a similar illustration.

204

CHAPTER 6

It is clear why one cannot draw any inference from the number of wives
of Henry VIII to that of Henry IV, since the rank number of kings of equal
name can never be explained as an extensive parameter by any combination.
The rule of three does not apply to the problem if a man covers a distance in
3 hours and his son does so in 2 hours, how long do they need if they walk
together? because going together, for instance by people who are equally
fast, does not change at all the time required. Yet also in the problem of the
working men who do certain work first individually and then together, the

central question is: does the required time double if two equals work together?
No it halves, so the reciprocal time emerges as an extensive parameter. And
so it emerges in the case of the man and his son, provided they do not walk
together but to meet.
I note down the following levels:
deciding on the ratio-preserving property of mappings in factual contexts
and problem situations;
recasting context and problems in such a way that ratio-preserving properties gain prominence;

deciding conflicts under these circumstances;

understanding principles of such decisions and constructions operationally;
and
describing them.
Auxiliary activities may be required on the following levels:
In order to become oriented to ratio preservation
considering pairs of parameters that are extensive under the same combination; and

looking for such parameters;

grasping the importance of such parameters for ratio preservation; and

explaining it.
In these auxiliary activities the following levels can be distinguished:
deciding with respect to parameters of states and processes whether they
are extensive according to a certain way of combining,
finding extensive parameters for given ways of combining,

finding ways of combining that make given parameters extensive;

finding parameters and ways of combining that fit with each other;

understanding what extensive parameters are operationally; and

describing them.

RATIO AND PROPORTIONALITY

205

6.17. Non-linearity
A great variety of phenomena suggests that proportionality, ratio preservation,
linearity are universal models; the faith in these models is reinforced by their
frequent use. Approximately at least the linear relation looks appropriate in
many cases as a phenomenal tool of description. We indicate cases where this
primitive phenomenology fails for theoretical reasons:

the non-linear behaviour of areas and volumes under linear multiplication;

the non-linear variability of precision in measurements and stochastic data
under multiplication of the sample size.
A historically remarkable example:

the bet on at least one six in 4 throws with a die was considered equivalent
to that on at least one double-six in 24 throws with two dice.
Another historical example:

the idea to solve the probleme des partis* by a linear procedure.

Faith, acquired by long practice, in linearity (or the rule of three), where
fresh principles were at stake!

6.18. The Use of Ratios and Proportions

The general use is to predict a fourth term when three in

are given.
The ratios on both sides can be meant as
internal,
and related to
equal or different magnitudes,
for instance
two strips a, b and two money values c, d,
or

two paths on a map a, b and two lengths c, d.

* Mathematics as an Educational Task, p. 584.

206

CHAPTER 6

Or they can be meant as

external,

for instance,
two strips a, c and two money values b, d,
or

two paths on a map a, c and two lengths b, d.

It can also happen that one of the ratio is
explicitly given as such,

which actually means two data only, for instance in the case of the external ratio

path to time,

weight to volume

explicitly as

velocity,

density.

Earlier in Section 4.19 we analysed a strategy of comparing approximately

cardinals of what we then designated as

k-homogeneous sets.
In our present terminology the relation between the cardinal # and the character
k of a set would be termed (approximately) ratio preserving. As announced
there, it can be used to estimate cardinals and ratios of cardinals.
A practical use of proportions in general includes

changing the middle terms in order to profit from the relation between
internal and external ratio,
processing the data, independently of the place of the known,
exploiting the definitory properties of internal and external ratio, using
visualising models,
composing and splitting up proportions,
estimating parameters by means of approximately linear dependent ones.

6.19. Ratio in Learning Processes

The present section has been suggested by experiences in teacher training institutions, though they are rooted in principles which will be illustrated later
on by more examples.
In Section 6.7 I stressed that in visual contexts children even at the kindergarten age can grasp the relative view and ratios (Figures 58 and 59).

RATIO AND PROPORTIONALITY

207

In the IOWO theme The giants greetings children estimate the giants
size (and many other related sizes) by the trace of the giants hand on the

blackboard. This direct approach is possible because no numerical data are

introduced by the text. The theme Camping ground, however, is much less
directly accessible because it introduces explicitly numerical data, and it addresses itself to children who have learned to identify measuring with the use of
the ruler.
A similar case:
Monica (5; 8) builds towers with congruent blocks. She is quite good at comparing towers

of different height, even if they are placed on different bases. She has put 11 blocks on top
of each other. I ask her to show me the height of a tower of 20. She shows a height 23
blocks higher. I let her continue building. At 13 I repeat the question; her answer is some-

what better. I ask her how many should be added. Her lips are moving. Obviously she is
counting from 14 to 20 and every time is disappointed again because she does not know
how many she should add. I teach her to raise the fingers while counting.

This story is told to unmask my incompetent didactical behaviour: the premature

and unnecessary translation of a ratio into a numerical problem. This is quite

characteristic of the traditional dominance of arithmetic over mathematics

instruction.
To teachers in training whom I observed, ratio is either a vague relation
which has not been made conscious, or an entirely algorithmised or automatised
phenomenon in the most favourable case expressed by proportion matrices.

208

CHAPTER 6

RATIO AND PROPORTIONALITY

209

The mental objects relatively and ratio have been blocked by numerical
associations. The student teachers have great difficulty in creating models by
which they can open to their pupils the entrance to the mental objects: they do
not even grasp the relevance of such models. Obviously this is a consequence
of their own process of learning ratio which has been directed straight on to
algorithms.
By this I do not mean that these students have never gone through a period of
insight with respect to relatively and ratio. There is no need to suppose
and it is even not probable that they have experienced these notions from
the outset in an algorithmic way (in order to automatise them later on). It is
more probable and this is typical of many learning processes, especially in
mathematics that the original sources of insight have been clogged, and the
way back to insight is blocked by the processes of algorithmising and automatising. Autonomy of algorithm and automatism is a strong inclination, which is
understandable: too much insight can be a hindrance under certain circumstances.
Anyway, we have to view critically the bad consequences of such blockages.
What can we do against them?
I will answer this question at several opportunities. It is most often necessary
but not sufficient that algorithms and automatisms are acquired by insight.
The learning process must be steered in such a way that sources of insight are
not clogged during the process of algorithmisation and automatisation. This
can be achieved, in my view, by returning again and again during the process of
algorithmisation and automatisation, and even afterwards where it fits, to the
sources of insight. This process aims at an ever greater consciousness of what
initially was subconscious, and an ever sharper verbalisation of what initially
was not verbalised at all. With regard to relatively and ratio, this means
that the visual models are repeatedly recalled and abstracted into thought
models. What is wrong in many methods is a satisfaction with the uniqueness
of certain decisive steps in the learning process and with repeated exercises
of the consequences of such steps, instead of repeating the steps themselves.
A corrective measure: repeating the step if something goes wrong with the
automatism. But more important is prevention: repeating the step from insight
to automatism before things go wrong, in order to guarantee the ability to
repeat the step.

CHAPTER 7

STRUCTURES: IN PARTICULAR,
GEOMETRICAL STRUCTURES

7.1. Chapter 7 was originally Geometry, which later was changed to Geometrical Contexts. I stepped straightly into a phenomenology, which was a bit
didactically tainted. But I soon had to exchange the phenomenological thread
for a methodological one. In order to make things clear I had to take so many
side steps that the frame of the chapter was in danger of bursting. I started
again. What follows now is simply mathematics or, as far as it might be valued
as phenomenology, it is one with its object at a very high level, the phenomenology of a quite advanced mathematics. I am afraid this will not be my last
struggle with the revision of the whole idea.
7.2. Without much ado I used the word structure many times. I will explain

it now more systematically.

Look at Figure 60. It represents a structure, a graph consisting of seven nodes

and seven connecting lines. The figure resembles the Big Dipper, which
was indeed the intention of the drawing. (The constellation, though more
extended, is properly named Ursa Maior, but this does not matter.) I could
have drawn the same graph differently (Figures 61 and 62), but then you
would not have seen any constellation in it. The three figures considered as
graphs are isomorphic or, in other words, combinatorically equivalent.

A graph is a set of nodes and edges with a relation each edge joins two
nodes. Visually the nodes and edges are rendered by points and preferably

straight connections. Isomorphism for graphs means that the one can be
210

211

STRUCTURES

mapped one-to-one on the other such that nodes, edges, and joinings correspond
to each other. A variant is the directed graph, where each edge is directed from
one of its nodes to the other. Then isomorphism includes preservation of direction under the mapping.
Graphs as combinatoric frames are a frequent phenomenon:

A city plan with the corners of the streets as nodes and pieces of streets
as edges.
The network of the Netherlands Railways with the stations as nodes and
the direct junctions as edges.
A box of blocks or a jigsaw puzzle with the particular blocks or pieces

as nodes and neighborhood as edge.

A cube with its corners or faces as nodes and its edges as edges.
7.3. Combinatoric structures are relatively poor. In general, physical or mathematical systems possess more structure. Take a trellis-work or wire-netting of
squares or hexagons (Figures 63 and 64).

They can be structured purely combinatorically, and then they are equivalent
to the graphs of Figures 65 and 66.

However, Figures 63 and 64 suggest more structure than Figures 65 and 66.
First of all, that of a rigid body: Figures 63 and 67 are congruent, as are Figures
64 and 68, that is, they can be mapped on each other such that all mutual

distances are preserved.

212

CHAPTER 7

Figure 69 is similar to Figure 63, as is Figure 70 to 64; that is, they can be
mapped on each other while all ratios of distances are preserved.

This kind of structure could have been observed with the network of the
Netherlands Railways; for instance, one could note at everyjunction the distance
in km or minutes or even provide each junction with a length proportional to
the distance.
So the structure can include

distance of pairs of nodes,

ratio of distance of pairs of nodes.

Correspondingly, isomorphism means
congruence,
similarity.
If we consider Figure 63 as a graph we can include in the structure the
relation of distance with respect to

all pairs of nodes,

joined pairs of nodes (edge lengths).
The second structure is weaker. In this sense the structures of Figures 63 and 71
are isomoprhic. They are equivalent under transformations we designated as
flexions.

STRUCTURES

213

Vertical fences like Figures 63 and 64 are mapped by the sun into shadow
images like Figures 72 and 73. The shadow mapping conserves rectilinearity
and parallelism. Such a mapping is called an affine mapping.

In general, structures are weakened if, rather than distance or ratio of distance,
they include only

rectilinearity and parallelism.

Isomorphism then means

affinity.
If the fences of Figures 60 and 61 are projected as slides on a screen, we can
get figures like Figures 69 and 70, at least if the slide and the screen are parallel.

If they are inclined to each other, we get figures like Figures 74 and 75. This
kind of mapping preserves rectilinearity.

214

CHAPTER 7

Structures like the preceding ones can be weakened so as to include only

rectilinearity
as a structuring relation. Isomorphism of structure then means

projectivity.
If the projection screen is not flat but bumpy, then the image of Figures
63 and 64 can look like Figures 76 and 77. What is preserved in this kind of

mapping? Only the connection of figures: a continuous curve passes into a

continuous curve; there are no cuts and no folds. What is preserved in such
mappings is

neighborhood the topological character.

Isomorphism here means
topological equivalence.

Topological and combinatoric equivalence look in some aspects similar. A

graph where two nodes are joined by at most one edge can also be defined as a
set of nodes with a relation of

being a neighbor.

It is a coarser kind of neighborhood than in the topological case. Figures 6062

are not only combinatorically but as polygons also topological equivalent.
But topologically they are also equivalent to Figures 78 and 79, a closed curve

with a tail.

STRUCTURES

215

More examples of topological structures and their isomorphisms:

The surfaces of a sphere and a potato are topologically the same, but they
differ from the surface of a ring, which as far as its topology is concerned can
also be represented by a rectangle (Figure 80) where the opposite sides are
imagined to be stuck together according to the arrow.

A cylinder is topologically equivalent to an annular domain bounded by two

circles (Figures 81 and 82), where boundaries correspond to boundaries; the
annular domain without its boundaries is topologically equivalent to the cylinder
without its boundaries, as well as to a cylinder of infinite length, and finally
also to a plane in which one point has been pricked out.

7.4. The examples of structures I gave were geometric or illustrated geometrically. I stressed this kind for didactical reasons but it would be a shortcoming
if I were to leave it at that. As a matter of fact I already dealt with various
structures in the number system: the order structure, the additive structure, the
multiplicative structure. I can also put these kinds of structures on other sets,
for instance, a multiplicative structure on a set of four elements e, a, b, c by the
multiplication table

which is to be read in the usual way. This then is the so-called group of four.
It is, however, unusual to define structures as explicitly as has been done
here (and in Section 7.3). Most often it is done implicitly; that is, one introduces
a set

216

CHAPTER 7

with

certain relations on it,

and requires these relations to observe

certain postulates.
For instance, a group is by definition a
set G
with
relations

and

as postulates,
associativity :
an identity element
for each element an inverse
This does not define just one group but rather the group concept, which can
be examplified by many (finite or infinite) models; and for each pair of groups
one can ask whether they are isomorphic, that is, show the same structure.
This implicit approach is more fruitful than its explicit counterpart. In order
to add one more geometric example, I take

metric space,
a

set R of points
with a
distance relation for pairs of points,
that is a function

such that

is a real number
subject to the requirements that

Metric spaces can again be compared with each other; isomorphic ones are also
called
isometric.

217

STRUCTURES

A weaker structure is
topological space,
a

set R of points
with a
relation of being close to each other,
subject to requirements which I do not specify. The most usual topological
spaces are in fact better handled if approached from metric spaces by weakening
their structure. Then being close to each other can be defined technically via the
metric:
V is called a neighborhood of p in R if there is an
points at a distance
from p are lying in V.

such that all

This transforms the quantitative distance into a qualitative closeness.

7.5. Mappings are important not only between different structures. The combinatorial structure of Figure 60 admits a mapping onto itself which interchanges
and while all other nodes remain in their places. This is an
automorphism,
of course, in the combinatorial sense; that is, if I am satisfied with the combinatoric structure of the graph. If I consider Figure 60 as a picture of the Big
Dipper with the correct distance ratio on the firmament, I am not allowed to
interchange and
The one-to-one mappings of a system onto itself that preserve structure,
are called automorphisms. They form a group.
The n-gon, combinatorically viewed, admits 2n automorphisms; that is, if I
number the vertices subsequently by

I get the n automorphisms

and the n automorphisms
Together they form the so-called dihedral group.
An n-gon with more structure will in general admit no automorphism except
the identity. Suppose the sides posess lengths which are considered part of the
structure. Only if all these lengths are equal are the combinatorial automorphisms
flexions also automorphisms in the sense of the richer structure. If the

218

CHAPTER 7

lengths of the sides and also the angles between the sides are both understood
to belong to the structure, the above automorphisms exists as such only if all
angles are equal.
The graph consisting of the vertices and edges of a tetrahedron admits all 4!
permutations of the vertices as automorphisms; if the lengths of the edges are
comprised in the structure, the group of automorphisms may shrink. The graph
of vertices and edges of a cube has a group of automorphisms that is twice as
large, under similar conditions as for the tetrahedron.
The euclidean plane and space are particular rich structures with the relations
of
collinearity of three points,
coplanarity of four points,
order on the line, in the plane, in space,
congruence of line segments,

congruence of angles, and so on.

Distance does not belong a priori to the structure of euclidean plane and space,

though it is a fact that assigning a length, say 1, to one single line segment
bestows unequivocally lengths on all line segments. That is what is called gauging.
Thus
gauging transforms the euclidean space (plane) into a metric euclidean
space.

A posteriori it appears that

the euclidean structure is determined by this metric structure.

For instance, the order relation

q between p and r

can be brought back to the metric relation

and collinearity of three points can in turn be brought back to betweenness.

The mappings of the (metric) euclidean plane or space on itself that preserve
distance the isometries or congruences also preserve collinearity, coplanarity,
order; they map lines on lines, planes on planes. So they are automorphisms of
these structures.
The automorphisms of the metric euclidean plane are the

translations, rotations, slide reflections;

those of metric euclidean space are the

translations, rotations, screwings, slide reflections.

STRUCTURES

219

If, however, the metric is dropped and only the euclidean structure is left,

the group of automorphisms increases; it becomes that of the similarities.

A weaker structure of plane or space than the euclidean is the affine structure,
whole defining relations are

collinearity and parallelism.

The automorphisms of this structure are the affinities a larger group than that
of the similarities.
If one starts, as we did, from euclidean space, then to get the affine structure,
one need not require preservation of parallelism separately. Indeed, mappings
of euclidean space that preserve collinearity map straight lines onto straight

lines, planes onto planes; parallel lines are by definition lines in one plane that
do not meet a property that is consequently preserved under such mappings.
In restricted parts of the plane or space, however, preservation of collinearity
does not imply that of parallelism; there non-affine mappings exist that preserve
collinearity, though they cannot be extended to the total plane or space unless
the plane or space itself is extended, that is, by adding so-called points at infinity
(lying on a line or plane at infinity). This produces the projective plane or space,
a structure with the relation of
rectilinearity,
and as isomorphisms the
projectivities.

One more step is maintaining in one of these structures euclidean plane or

space, projective plane or space only a

relation of closeness;

then a
topological structure

arises: euclidean plane or space, projective plane or space considered as

topological spaces,

with as automorphisms their

topological mappings onto themselves,

which form very large groups.

7.6. The group of automorphisms of any structure includes the identical mapping
of the structure onto itself. It can happen that this exhausts the group. The
examples of Section 7.5 showed large groups of automorphisms. For studying
a structure its group of automorphisms may mean a great deal. Congruence
theorems is an example: the fact that triangles are congruent can be the source

220

CHAPTER 7

of many aspects and properties they have in common. The midpoint of a line

segment is an affine, yet not a projective, concept, which implies that with
only the ruler one cannot halve a line segment, though it might become possible
as soon as one has at ones disposal an instrument that produces parallelism.
The importance of group theory for geometry was chosen by Felix Klein as
the theme of his Erlanger program. Felix Klein grasped and stressed that the
geometries dominating his work together with their mutual relations were to

a high degree determined by their groups of automorphisms and the mutual

relations of these groups. Starting, for instance, from projective space and its

group, one could pass to affine space and to the affine group as a subgroup of
the projective group by fixing one plane to be considered as the infinite plane.
Further, by fixing in this plane a non-degenerate imaginary conic, one could
pass from affine space to the heavier structured euclidean space, or by fixing
a real or imaginary quadric, to non-euclidean spaces, and to their automorphism
groups as subgroups of the projective group.
So the group of automorphisms of a geometry came to the fore a slogan:
geometry is the invariance theory of a group. It is a fruitful idea, which has

been fully elaborated with new perspectives in E. Cartans homogeneous

spaces. It is a sound principle, which applies to structures with such a degree
of homogeneity that they can be defined by the expression of this homogeneity,
that is by their groups of automorphisms.
But notwithstanding its importance, this is only one aspect of the study of

structures, in particular, geometric ones. In our exposition on structures this

aspect has been brought to the readers notice as late as Section 7.5. In Sections
7.24 we were busy with various kinds of structures; we asked whether systems
of the same kind of structure were isomorphic. Such knowledge too can be of
great importance: discovering that two structures obtained in different ways
are isomorphic, or managing by means of restructuring to make them isomorphic,
may create opportunities to transplant concepts, properties, extra structures
from the one to the other. In fact this use of

isomorphisms between structures

is much more comprehensive and in general more fruitful than that of

automorphisms within a structure,

which is tied to quite special, and particularly homogeneous, structures.

The automorphisms of a structure form a group. Finding out whether, and
stating or ensuring that, certain structures are isomorphic has nothing to do
with group theory. Nevertheless, these two things have often been confused.
It is quite a serious confusion; it concerns more than the terminology of what
to call a group and what not.
The germ of this confusion is already visible in Kleins Erlanger program.
(The name under which that inaugural lecture has become known and is cited
is also rather responsible for the confusion: it has been interpreted to mean

STRUCTURES

221

the program for geometry, whereas it was an annex to the program of Kleins
courses at Erlangen University.) Yet thoughtless copying is a danger that even
mathematics has to face, unless it is hard mathematics, supported by formulae
and logical reasoning. It is a sad story, but so far this misconception has not
been extirpated. Even in mathematics research it shows up as soon as the Erlanger
program is around. So we should not be surprised to meet it in the philosophical,

psychological, and mathematical-didactical literature. A large part of Piagets

work, not only that regarding geometry, is dominated by this misconception,
at least as far as it is concerned with mathematically tainted theory.
7.7. I will explain this more precisely.

The marvellous spaces with their beautiful automorphisms that we dealt with
in 7.5 are not aims in themselves, and to add immediately a psychological and
didactical argument neither are they starting situations. The spaces are to
lodge figures, which can be both starting point and aim in itself, but if anything
within these figures is worth being explored and is a challenge to exploration it
has little to do with the automorphisms of the surrounding space.
If two triangles (or tetrahedra) have correspondingly equally long sides
(or edges) they are congruent, which implies that this correspondence can
be extended to a length-preserving mapping of the space. For quadrilaterals
in the euclidean plane or space the analogue does not hold: the isomorphism
of flexion isomorphic quadrilaterals cannot in general be extended to the plane
or space in such a way that the flexion character of the data is reasonably
accounted for. Flexions of curves and surfaces within the euclidean space
are not susceptible to reasonable extensions into space; hence if more objects

are concerned, such flexions cannot be brought together in a group, and in

order to relate them to the group of automorphisms of euclidean space in a
more profound way, highbrow devices such as the theory of sheaves are required.
The Figures 78 and 79 are topologically equivalent, but there is no topological
mapping of the plane onto itself that realises this equivalence indeed such a
mapping cannot exchange the interior and exterior of the closed curves where

the two tails are situated. It could be realised by moving into space, but there
one can again find topologically equivalent figures that are not equivalent by
means of a topological mapping of the whole space onto itself. Moreover, the

topological structure of figures in space is much more important than the

unfathomable group of topological mappings of space on itself.
In particular, prejudice arises against the combinatoric structures if automorphisms are too much stressed. The combinatoric structure of Figures 60
and 61 is not accounted for by any classical geometry of the plane: there
is no reasonable mapping of the plane that continues the isomorphism of
Figures 60 and 61, and in general there is no way to embed the isomorphisms
of isomorphic graphs into a group.
A great deal of geometry is and occurs within space. It is true you can attach

to a planet a copy of space in order to interpret the planets motion as a motion

222

CHAPTER 7

of one euclidean space with respect to the other, that is, as a sequence of mappings of space upon itself, a sequence of automorphisms. Moreover this can be
extremely useful if you want to study the centrifugal forces exercised by, say,
the rotating earth on a free falling object or on air and water streams. It can
perhaps even be useful to attach a whole moving space to a driving car in order
to understand what happens if it goes into a curve. But a car also has wheels,
and there is no need to attach separate spaces to the wheels in order to drag

in automorphisms. Moreover, a car is flexible, its doors can be opened and

closed, and this mapping of the open on the closed car does not extend in any
reasonable way to space in order to be put into a group. Bodies of animals
in movement are flexible systems according to certain combinatoric structures;
the mappings expressed by their movements can be understood as isomorphisms,
but no part is then played by automorphisms, either in the sense of some classical
geometry or in any other more extended sense.
A drastic example of that confusion is found in almost all New Math text-

books: opening and closing a door as inverse elements of a group. What group,
and on what set does it operate? On the two states of the door? But what does
opening perform if applied to the open door and closing to the closed door?
This is the point where the prominence of the group of automorphisms,
as an aspect of the Erlanger program, fails: in all that happens within the space,

and which has not extension, or no relevant extension, to space as a whole.

CHAPTER 8

PUTTING INTO GEOMETRICAL CONTEXTS

8.1. La reprsentation de lespace chez lenfant

The title of this section is the same as that of a book by J. PiagetBrbel
Inhelder* which, in spite of serious objections about its global approach and
about many of its details, I rate highly. Even more than Piagets other work
that touches mathematics it should have deserved from mathematicians serious
criticism rather than mere shrugs of the shoulders.
Initially I did not mean to discuss Piaget too much in this chapter, but while
I was writing it, the need became more and more urgent. Nevertheless I will
restrict my criticism of Piaget as much as is feasible because in spite of the
variety of its aspects his work does not contain enough elements to take ones
bearings on it.
I put the title of PiagetInhelders book above the present section because
my first considerations will center around two words contained in the title,
espace and representation.

8.2. Space
Space is an expression that from the title to the last page occurs a thousand
times in PiagetInhelders book, often as an adjective. I did not put it into the
title of the present chapter, and there were reasons why 1 did not. Space, whether

as a mental object or as a concept, is rather the endpoint of a development,

though not in the sense of a bearing this would again be a wrong perspective.
Not until a highly advanced mathematical context is reached, does space

get a meaning. On long trajectories the word space can be dispensed with as
a mere term, and even as a mental object it is not required. In no way does
the constitution of the usual mental objects in geometry depend on that of the
mental object space, whatever this may be!
Greek geometry and philosophy do not possess an equivalent of our space.
The universe is finite and the fact that, according to one of Euclids postulates,
every straight line can indefinitely be extended does not imply anything about
the medium in which this should be possible. Mentally such a medium may

somehow exist, but no attention is paid to it, up to Cusanus and Newton, say.
The etymological root of space is spatium, which means distance. Space and its
* Paris, 1948. English Translation: The Childs Conception of Space, Routledge & Kegan
Paul, London, 1960. For a few quotations this English translation has been used, though
at some places it does not match the French original; at one place to be quoted, however,
the translation is an improvement on the incomprehensible French text.

223

224

CHAPTER 8

analogues in other languages originally mean a closed thing, and to indicate the
very big space modern terminology was enriched by outer space.

Before going further, I shall give examples of the often more technical use of
the word space, with the intention of showing the direction I wish to avoid.
The euclidean space so called because the geometry studied in and according to Euclids Elements implicitly presupposes at least something like this
includes points, connected by lines, contained in planes, forming circles as large
as you wish, all over the world right angles are equal, and what happens far away
with the geometrical objects can be predicted here, because in all triangles,
however big they might be, the sum of the angles is two right angles. From the
closest neighborhood one extrapolates to ever bigger distances, and this then is
the mental object, called Euclidean space, for which Euclid himself had no
name.
Geometry originally meant measuring the earth, as performed by surveyors,
but this practical use was never stressed in Greek geometry; it was rather held in
contempt. Eratosthenes managed to measure the whole Earth from a restricted
piece of land, and Aristarchos did the same with the distances and sizes of the
Sun and Moon. The proper domain of astronomers, however, was measuring
angles. How this was done, is not told in the Elements though angles of a certain
measure right angles occur in its theorems. Line segments of a definite
length are of course not met with in the Elements, and only with an illiterate
slave does Socrates speak about squares of so many feet.* What counts in
euclidean space is the equality and the ratio of line segments.
The euclidean space with all its objects is a rich structure, although it is poor
if compared with all I perceive around, its colours, polished and rough surfaces,
sounds, smells, movements. But thanks to the impoverishment it furnishes a
certain context, which for some reasons suits us extremely well this is a point
I still have to consider more closely. Anyway this context has been accepted as
geometry for centuries, this mental object of euclidean space as if it were an
objective datum, though efforts have been made to describe it more precisely
and more efficiently than Euclid ever did. More precisely this means axiomatics
like PaschHilberts; more efficiently this means the algebraic approach from
Descartes to the modern version of metric linear space of three dimensions.
Elsewhere** I have sketched this development.
This euclidean space has never been an aim in itself, but rather it has been
the mental and mathematically conceptual substratum for what is done in
it: for constructions with a pair of compasses and a ruler, or with only a pair
of compasses, or with a ruler only, for constructions by means of algebraic
equations or purely mechanical constructions, for deducing properties of such
figures, for proving or refuting hypotheses about them.
In the more recent development of geometry it was an important discovery
* So does Theaitetos when he quotes Old Theodores.

** Mathematics as an Educational Task, Chapter XVI.

GEOMETRICALCONTEXTS

225

that with the ruler alone one can go a long way, for instance in the theory of
perspective and for all properties relevant in perspective. A methodological
principle of mathematics, ever more systematically applied since antiquity, is
purity of method. In the example I just cited it meant that as soon as one
studies properties which depend on the ruler only, one must choose a substratum which is restricted to points, lines and planes in their mutual relation.
This then is an even poorer structure, the projective one, but the relative poverty
can be an advantage. Wealth, if dispensable, can be a hindrance. In the present
case, in one respect, the poverty was too pinching; to embellish the structure,
space had to be enriched by points, lines, and a plane at infinity, and this then
was projective space. A variant on that principle: if parallelism is included in
the fundamental concepts besides rectilinearity, one gets a structure affine
space which is poorer than euclidean and richer than projective space.
Another evolution was that towards non-euclidean space. One started doubting whether the neighborhood, as one thought to perceive it, determined the
remote depth of space as it had been assumed. If the sum of the angles in a
triangle would systematically differ from the supposed value of two right angles,
space would look different curved, whatever this might mean.
A third evolution away from euclidean and the other spaces was into more
dimensions, even to an infinite number of dimensions. Geometric language
became a suggestive and creative device to organise quite different domains

analysis in this case.

And then the fences came down: structures are created according to ones

needs, and if they are related to structures that had formerly been called space,
or if they involve visual elements to be uncovered or stressed, they are called
spaces: metric spaces, topological spaces, discrete spaces, and so on. There are
good reasons why mathematicians did this: insight can be deepened and terminology can be simplified if various structures are brought together under one
heading.

8.3. Reprsentation The Mental Object

In the English version of PiagetInhelder reprsentation has been translated by
conception which shifts the stress even more to concept attainment. I am not
sure whether this is correct.* If one looks only at the titles, theoretical introductions and conclusions of the chapters, and sections of PiagetInhelders work,
one can indeed get convinced that the authors have tried to investigate the
childs conceptual approach to space or rather to find out which features of
* Another work J. Piaget, Barbel Inhelder and Alina Szeminska, La gomtrie spontane
de lenfant, Paris, 1948 has become in the English version The Childs Conception of
Geometry. One may rightly doubt whether there is any spontaneous geometry in this work
but anyway conception of geometry is nonsense. In the footnote to Section 8.1 differences between the versions were signalled as a general fact; actually they are rather

frequent.

226

CHAPTER 8

some adult conceptual approach to space can be traced in the childrens mind.
This underlying adult conceptual approach is then the one the authors knew
from the literature, in particular the one about the Erlanger program.
However, I am not sure whether for PiagetInhelders work this conceptual
approach meant much more than an organisational pattern and some theoretical
frill. The requirements are rather designed to observe the childs representations,

Vorstellungen in Kants sense, intuitions as some others say, or mental objects,

as I prefer to call them. They repeatedly distinguish the perceptive, the representative and far away the intelligible space, but the aspect they focus on is
representation sometimes mentally and most often mentally-graphically
recorded.
8.4. The Mental Object in Geometry
In no part of mathematics do mental objects serve so long before, or even
without, concept formation as in geometry. Images and imageries are more
efficient if they represent figures and spatial constellations than if they represent
numbers. Small numerical quantities can be supported efficiently by images,
actual and imaginary ones, but this support does not reach far in the quantitative
world and so is soon renounced. The numbers 3 and 5 are unsatisfactory
paradigms of arbitrary natural numbers and their sum and product fail as
paradigms of operations on arbitrary pairs of numbers. On the other hand each
triangle that is drawn in a not too specific way is a good paradigm of the triangle,
each pair of line-segments is a paradigm of the pair of line-segments if the
aim is to show what the sum or the product is of two lengths. One can show
other people what a parallelogram is, a rhombus, a square, what are diagonals
and what it means to say that they halve each other, that they are perpendicular
to each other, that they are equal. Without bothering oneself or the other person
with concepts, one can introduce words to indicate them and restrict oneself
to examples to explain what the words mean. One can explore widely the
geometrical domain without forming concepts, so widely that finally over-ripe
concepts drop in ones lap. One can even disregard the formalism characterising
traditional geometry and for a long time be satisfied with demonstrative linguistic
means and wait for relative and more symbolic linguistic tools to announce
themselves. For these and many other reasons geometry is the field where one
can fruitfully look for symptoms of learning processes, were it not that each
investigator carries his own geometrical education as blinkers for himself.
By geometrical education I do not mean something that starts with the
first traditional geometry lesson. Many geometrical objects and concepts have
been formed early, most of them at the primary school age and some of them
even earlier, though they do not yet bear verbal labels, or at least not those
labels that we have learned to attach to them in our geometry lessons.
If one compares this with the content of Section 8.2, it becomes clear that
my aim is not the geometry, nor a system of geometries; before I can arrive at

GEOMETRICALCONTEXTS

227

space or spaces as mental objects, I must deal with mental objects which are
understood as geometrical objects, are lodged within the space. As geometrical
objects they will in a later stage be placed in a space, but as mental objects they
are first of all in a context, namely a geometrical context. I have indicated
earlier* the didactical significance of contexts, grasping a context as a necessary
condition for more than mere algorithmic action.

8.5. The Context of the Rigid, Congruently and Similarly Reproducible Bodies
as an Example
I have no doubt that geometric education starts very early, and that this early
start has much to do with the fact that the geometric context suits and pleases
us so well. Colour seems to be a more subtle case than geometric shape.
First of all, rectilinearity, in the natural environment of man exemplified
by the straight posture, the stretched limbs arms, legs, fingers the stalks
of plants and trunks of trees, and the straight way, which is the shortest, the
most direct. Among the first tools, made by man, is the arrow, paragon of
rectilinearity, and as civilisation progresses, so more frequently and forcefully
man is confronted with objects and processes and elicited to actions that suggest
or represent rectilinearity: sticks, pins, rims, edges, paths, folds, cuts, stretched
strings.
Flatness is perhaps even more frequently and forcefully suggested, by paving
stones, floors, walls, ceilings, tables, benches, roofs, sheds, boxes, lids.
He is confronted with parallelism as often as he is with rectilinearity, again
of borders of objects, roads, gates, in planar divisions, in wire-nettings, palisades,
rows of houses; right angles are suggested by perpendicularity but also by the
angles of more or less carefully made objects. Even in the natural environment
man has become acquainted with mirror symmetry, polygonal symmetry, and
axial symmetry; so did stone age man try to imitate them and by this means
educated others to see and appreciate them.
Objects that suggest circles are rare in the natural environment but they
exist: Cross-sections of trees, sun and full moon, the horizon. After the wheel
was invented, man was already in the cradle showered with round objects.
Balls and rolling playthings suggest spheres and cylinders; to tell a child what
is a cone, you say a clowns hat. However regularly or irregularly something
is formed, it is influenced by geometry and suggests geometrical shape. Natural
production, craftmanship, manufacture, and industry have taught us congruence
and similarity, in particular the similarity of playthings that imitate the world
of adults.
I will leave it at that. I shall return to these examples to discuss details. Here
they have served to make it clear how the geometrical context comes about
* Weeding and Sowing, p. 242 sq.

228

CHAPTER 8

I mean the context of euclidean geometry, or phenomenologically described,

that of the

geometry of rigid, congruently and similarly reproducible bodies.

They are acquired early, all these mental objects, so early that it is difficult
to distinguish that which perhaps is innate. Mental objects like rectilinearity,

flatness, parallelism, rectangularity, rectangle, square, sphere, cube, symmetry,

congruence, similarity are, as far as they are mental objects determined in a
simpler, clearer, and sharper way than non-geometrical ones like plant, tree,
animal, colour. If this is duly realised, we need not be astonished about the
certainty of our recognition of geometric objects.
Notwithstanding the early start of our informal geometric education, the
fact that the formal one starts so late is to be understood in the frame of the
history of education. Intellectual education, which included geometry, was
generally conceived of as concept formation. The mental objects, as an ideal
material to work with, were most often neglected and so they are even today,
in educational theory and psychology. Or rather, one disregards the distance
between mental object and concept, identifies them, confuses them and by
this means does not do justice to either.
There is a world of difference between our examples of geometrical context
and the laboratory context, in which Piagets subjects are placed, as well as
with the theoretical frame of the Erlanger program. In the world as we view it
the first and foremost things are bodies in space. Their congruence or similarity
is of course related to certain mappings, which may be made explicit according
to the needs that are felt. To be sure, these mappings can be extended to the
whole space; they are restrictions of those global mappings which together form
the group of similarities of space, but this is an idea, far away, at a remote
distance where geometry has already been sufficiently mathematised. (I admit
that even in PiagetInhelders work it is not operational group theory has
been dragged in as a mathematical frame and as an organising element.)
One may object that in any case the group of similarities does not clash with
that which we have in mind as the context of rigid bodies. This is correct. To
say it in a terminology I have used in another connection,* the Erlanger program
idea of the group of similarities is the apotheosis of this context. In the next
section, however, we introduce other geometrical contexts which do not fit
into the frame of Erlanger program.
8.6. The World of the Boxes

With much eloquence I have exerted myself to convince the reader of how the
mental objects of euclidean geometry are forced upon us, from bow and arrow
to baby spoon to television aerial. This demonstration, however, was a bit
* Mathematics as an Educational Task, p. 171.

GEOMETRICAL CONTEXTS

229

simplistic. One might be led to believe that there were nothing else but this.
If it has been understood this way, I have missed my aim. Yet I put into the
title of the section the warning as an example. There are other contexts,
however, and the following is one of them.
I chose the term boxes for the geometric objects I am going to consider
(right parallelepipeds), because this article is available in the richest variety
planks remind one of something with one aspect much larger than the other
two, and bricks are represented by too small a number of models. But if I speak
of boxes, one should think as well of chests, cupboards, rooms (of appropriate
shape), books (with flat backs), and as many other things as one would like to
invent. You may consider the box as a structure with 8 nodes and 12 edges, but
then of well determined lengths, you may if you like add the side and space
diagonals, you may number the vertices, orient the edges and concoct even more
variations. They can be quite diverging structures but the conclusions I wish to
reach will remain the same.
I said that mental objects like congruence and similarity are suggested to us
by the world we live in. It is true, but sometimes other kinds of equivalence are
even more strongly suggested as mental objects. In the world of the boxes we
are told that a box is a box. Yes, a box is a box, but not in the way that a cube
is a cube, or a sphere a sphere. Cubes are similar to each other, as are spheres.
Boxes are not. What then do we say about boxes?
In order to map one box onto another in a gentle way, we have to do some-

thing with its edges, shrinking or expanding, but of course the same with parallel
edges. What kind of mapping is it? Take an origin and a rectangular system of
axes, put a box with one corner into the origin and with its edges along the
positive axes. The mapping of one box on another is expressed in coordinates
on this system of axes by
a multiplication along the axial directions with factors
respectively.
It is a mapping extending to the whole space line-segments parallel with
the first, second, third axis are multiplied by
respectively. Let me
illustrate it by drawings in two dimensions, that is, a rectangle D instead of a
box, two axes, and
(Figure 83).

230

CHAPTER 8

Such a mapping is called a dilatation according to three orthogonal axes

let us briefly say: a dilatation. It is an affine mapping although not a very special
one, since each affine mapping can be split into a rotation and a dilatation.
Is this all we have to say about boxes and their mutual relations? No, I
can do more with boxes than change their edges. I can also displace them as
rigid bodies. By a translation such a box D gets a position
which does not
any more lean against the axes but with edges still parallel to the edges of
D (Figure 84).
too is a box and if I subject it to our dilatation f, I again get
a box; the two boxes fD and fD' differ only with respect to place; corresponding
edges of D and D' were multiplied by the same factor, so they are equal.

Rather than to translations, I can subject the box to rotations. By rotating

D I get it into a new position
(Figure 85), with edges which are no large
parallel to the aforementioned axes, a fact which has momentous consequences.

If now
is subjected to the dilatation f, the result is no larger what we have
called a box rectangularity is lost.

Dilatations apparently are things which belong to boxes. They are the mappings that characterise the world of boxes. But each box has as it were its own
stock of dilatations. A box is kept a box if it is diatated according to axes
parallel to its edges. Or, from another viewpoint: the dilatations do not form a
group, the product of two dilatations according to different triples of directions
need not be a dilatation any more. If you want to make a group of it, you get
the total affine group, which treats all parallelepipeds alike, and which does
not appreciate beautiful right boxes as such.

GEOMETRICAL CONTEXTS

231

Well, there we are: the euclidean geometry with its group of similarities does
not allow us to assert that a box is a box, and the affine geometry does not
know any boxes at all. What are we to do about it?

The mistake if one may call it a mistake is in extending to the whole

space the mapping that transforms one box into another. I stressed this as
early as Section 7.7. Here it is confirmed again. It is often meaningless or even
obnoxious to extend mappings of figures in space onto each other to space as
a whole think of the parked and driving car, the closed and open car! Why
then is it done in other cases and with much success? It is done in order to force
the whole thing into a group, to interpret the various mappings of figures, say
spheres (or cubes), onto each other as mappings of a larger structure, euclidean
space, onto itself extended mappings, which together form the group of
automorphisms of euclidean space. Yet if one tries the same with boxes rather
than spheres, one does not succeed as pleasantly: one gets embroiled with a
group that does not respect boxness.
On the other hand the suggestion that a box is a box is vigorous, as vigorous
as the suggestion of supporting the assertion by a good mapping which makes
from one box another one. These are strong visual stimuli in a stage where there
is not the slightest need for extending such a mapping to space as a whole.
Mappings have still to be viewed phenomenologically more broadly, but this
much can be said, that if mappings present themselves in any geometric context
whatsoever, they are first of all mappings of restricted parts of space, which
can be indicated or filled by bodies. Extending such a mapping to the whole of

space, even though it be possible and meaningful, is not something that goes
without saying how long have geometers tinkered with congruence theorems
where mappings would have made things easier? In history the step towards
mappings of the whole space has been consciously taken as late as the 19th
century this is witnessed by the projective space, which has been created in
order to be able to extend projective mappings to the whole space, and by the
device of Moebius geometry where space is augmented by one point at infinity
to account for inversions and make planes and spheres the same sort of things.
Historically it was an extremely important step to view mappings more than
locally, to embed them into a group, with momentous consequences, though also
with that of a dogmatically interpreted Erlanger program a bodice for the adult
mathematician, and an oversize suit, bought for growth, for the young learner.
I could have put this section also in Chapter 7 in or after Section 7.7 as
a counter example to the predominance of automorphism groups of structures.
After some hesitation I put it here, in order to stress its positive meaning, as
an example of a geometrical context.
I could give more examples, but I will wait; first I should say more about
geometrical context in general. At this moment only a few variants of the world
of boxes:
the world of rollers,

232

CHAPTER 8

(or cylinders, if you prefer), a continuous sequence, stretching from flat discs to
thin reeds,
the world of dunces caps,

that is cones,

the world of pointed roofs,

square pyramids and so on. Similar stories as those I told about the world
of boxes can be told about these worlds.

8.7. IntermezzoPiaget
Piagets way, at least regarding the structure of his work mentioned in Section
8.1, has been dictated by what he had experienced as the gist of the Erlanger
program, or to use our terminology of Section 8.6: he has bought a suit that
looks like a bodice but in fact is dangling around his laboratory experiments,
and as far as it is operational it functions as blinkers.
According to Piaget the development proceeds from the poor to the rich
structures, from large groups of automorphisms to small ones, according to the
sequence
topological, projective, affine, similar, congruent.

There is not any reason why this should be so, from poorer to richer. If there
were a definite line, one would bet that it is the other way round: the richer
structure presents itself with the greatest aplomb; impoverishing means abstracting, taking away. People and things are first singular objects, with proper names,
and they finish with sorts and labels from richer to poorer. Yet it is not that
simple. Initially all that moves, is a car; then VW, Duck, Peugeot become more
important concerns, but finally car wins. Initially each old man is grandpa;
then the term is restricted to two specimens at most, with perhaps the grandpa
of a friend added; finally almost every child appears to have a grandpa. Initially,
Utrecht, say, is the close neighborhood, and the other side of the Amsterdam
Rhine canal is The Netherlands; then Utrecht becomes a label for a vast array
of streets, squares, parks, which look like those of the neighborhood; later on
it becomes the proper name of a geographical unit enclosing all such spots,
and finally some political administrative unit.
It is neither a trend of ever more abstract, nor is it the contrary. It is
rather a case of developing new contexts, which may overlap as they may be
incomparable. In Section 8.5 I pleaded for the context of the rigid bodies with
arguments which I could have extended tenfold or a hundredfold, arguments for
its developmental primordiality, but if Piaget were right, this context would
be the highest rung of the developmental ladder.
As a matter of fact levels have to be distinguished for forming contexts, for
placing objects and operations into contexts Piaget speaks of the perceptive,

GEOMETRICAL CONTEXTS

233

the representative, and finally the conceptual space. He admits that perceptively
a child may be on a higher rung of the ladder than representatively. Representative is what I indicated as mental object, though Piaget uses it also in the
sense being able to draw the object. In Part One Topological space, for
instance, espace perspectif is followed by espace graphique as a second
chapter the quality, or rather the lack of quality, of drawing becomes the
criterion of the topological character. But even later mostly almost always
the drawn record is the only criterion of the mental representation: in order to
prove their existence, mental objects are to be drawn. I think this is wrong, but
I will delay discussing this point.
Piagets claimed priority of topological space is also a point to be discussed.
In one respect, I anticipate the discussion: what is presented as topology by
Piaget, is next to nothing, and partly it may even be asked whether it is topology.
I tackle this here because it is essential for the question of how far the development of topological space must have proceeded before that of projective space
can start a question which must be asked again for other pairs of rungs of the
Erlanger program ladder. If such questions are not answered, the claim that
this space precedes that does not mean much, if anything.
Is it topology if the child does not distinguish a ball and a potato, or if the
mental object topological sphere (closed surface of genus 0) has been formed?
It seems that Piaget is satisfied with the first. With regard to the straight line
(he says the projective line though it is the ordinary one), Piagets requirements are much higher. Here he requires, indeed, the formation of the mental
object straight line, if projective space is to become manifest, since for Piaget
projective space means the ability to look and draw according to perspective,
from ones own or an imagined viewpoint. This projective space is the precondition of affine space, which again involves so little that one wonders why so
much progress in perspective should be required as a precondition. Moreover
this too must be tackled later on the ability of looking and drawing according to perspective is a totally different thing from the constitution of mental
objects like straight line, plane, parallelism, congruence. Without laboratory
experiments everybody knows that perspective as an ability is much more
difficult and is acquired much later than the euclidean context surrounding the
rigid body. But let us skip over this trivial fact.
For the acquisition of the mental object straight line Piaget makes exorbitant demands. In Section 8.5 I casually said how straight lines appear as
arrows, trunks, sticks, pins, rims, edges, paths, folds, cuts, stretched
strings,
but none of these can meet with Piagets approval. Straight lines must be acquired
as vision lines, and even stronger, the global constitution of the straight line must
be preceded by the local one of the collinearity of three points.
Certainly, the vision line as a light ray is extremely important, and just
as certainly we will draw to it the attention it deserves. It would be of interest

234

CHAPTER 8

to know when this appears in the development.* But the straight line as vision
line is an advanced stage there are even many adults who do not know how
to use this property. Visual line is a property of the straight line but I refuse
to make it a constituting property.
For Piaget it is just this. Why? Because of the constraint of the system.
Projective precedes affine precedes similar precedes euclidean, projective means
looking and drawing perspectively, perspective vision takes place along vision
lines, thus it is required that the straight line is earlier as a vision line than as
the mental object that has to do with rigid bodies and their edges, with flexible
long objects in their preferential state, with movements directed to an aim.
To satisfy the system the straight line must be constituted as vision line and in
no weaker way.
Is this to be judged as a disastrous influence of pseudo-mathematics? Yes
and no. No, because it may be appreciated that somebody has investigated, or
tried to investigate, how the understanding of the straight line as a vision line
comes about. Yes, because this pseudo-theory may have prevented investigators
from looking for the true origin of the mental object straight line.

I will now leave the so-called projective space and turn to Piagets affine**
space. The experiments about it were made with the so-called Nuremberg

scissors or lazy tongs, in my terminology a context of flexibility rather than

affinity. Well, it has to do with parallel lines, and this may be termed affine.
One could even materialise a two-dimensional flexible lattice that as Nuremberg
scissors demonstrates a special affinity of the plane a very special one which
is still dominated by the idea of flexions since it is the diagonals rather than
the sides of the compartments that is variable.
Experiments with the Nuremberg scissors are undertaken in order to observe
conservation of parallelism by the subjects. It is the only place in Piagets
work where the transformations are made explicit with respect to which conservation is meant: it is affine transformations, albeit of a very special kind
and in a specialised materialisation. The subject must first predict what will
happen when the scissors are opened, or more widely opened, in particular that
rhombuses come into existence with conservation of parallelism of the sides.
As far as I can judge there was no account taken of whether any of the subjects
already knew this plaything. Anyway this conservation of the parallelism
structure under the movements of the Nuremberg scissors was the original
interpretation of conservation of parallelism, but while the experiments were
going on the interpretation shifted; the experimenters became inclined to
understand conservation of parallelism as the ability to copy parallel lines by
drawing in Sections 8.9 9 we will separate clearly this ability of reproduction from that of constitution of the mental object, which have been mixed
up here.
* When reading the reports of experiments from Piagets laboratory, I was astonished that
it is never mentioned whether subjects closed one eye while aiming.
** In the translation one reads affinitive instead of affine.

GEOMETRICAL CONTEXTS

235

Yet the experiment with conservation of parallelism is wrong as such;

it is produced by the same dogmatic interpretation of the Erlanger program
which we identified earlier: parallelism is an affine concept; so as a mental
object it is not well constituted until its invariance with respect to affine transformations is established. In the case of topology Piaget did not make such
heavy demands the topological mappings were not even touched. But the
demand itself is by no means justified, like that of the constitution of the
straight line as vision line. History at least is a proof to the contrary. Since
affinities are quite a recent discovery, may one conclude that mathematicians
before that time did not have a good conception of parallelism?
There is not the slightest doubt and in Section 8.5 I have tried to convince
the reader of it that parallelism as a mental object starts early; it cannot be
difficult to prove this by good conservation experiments. There is no reason
why this conservation should be related to the affine group. Parallelism is
perceived within the context of rigid bodies and the first and best way to
observe it is in just this context. A subject is shown the parallel edges of a ruler,
of a sheet, of a box and asked what happens to them if the object is moved. I
do not know whether anybody ever made this experiment, but anyone who has
studied childrens behaviour will not have the slightest doubt that this conservation of parallelism is constituted early. And this decides the question. It is
nothing but dogmatism to require more. The fact that parallelism is preserved
by a larger group than euclidean motions the affine group is no argument to
require invariance under this larger group as a criterion for the constitution
of parallelism. With the same right Piaget could have required the invariance
under the Moebius group for the mental constitution of the circle, which allows
one to consider straight lines as a kind of circles; or to require for the incidence
of point and line as mental objects that invariance is established with respect
to all contact transformations. A mental object need not wait to be pronounced
constituted until its invariance is established with respect to all mappings that
can be contrived in all possible contexts.
In the preceding I have tried to do justice to Piagets expositions on conservation of parallelism as far as they are intelligible and consistent. The experiments
have little to do with these expositions; the experimenters observed other things
than the conservation of parallelism. Moreover the theoretical text around the
experiments is often unintelligible, possibly because of internal contradictions,
which might be the consequence of double authorsnip. The experimenters
said they had difficulties in explaining to children what parallel meant, or
rather they were obliged to have these difficulties by the system (p. 316):
First of all, how is one to pose the question in order to make the idea of parallelism comprehensible? Presumably by asking whether the lines slant the same way, since any
mention of ideas as equidistance introduces far more complicated notions and measurements which themselves depend on assumptions about parallelism (i.e. the parallelism of
lines at right angles to those under consideration, thus yielding a circular definition of
parallelism in terms of equidistance).

236

CHAPTER 8

I stop here for a while to give the reader the opportunity to think about it.
The definition of parallelism by equidistance involves no circle at all, but the

reasoning between the parentheses contains a twist that deserves to be straightened out. I pass over the question whether one should define parallelism by
equidistance, whether it is didactically the best way or even a good way.
Anyway it can be done, and it has been done in the finest setting by Kuno
Fladt. Distance from a point to a line can be taken in the sense of shortest

distance or of orthogonal distance and there is not the slightest reason why
measuring along fresh parallel lines should be required a priori. The experience
that in euclidean geometry the lines along which the distances are measured, are
again parallel, is an a posteriori fact, which at a certain level can be formulated
axiomatically.

But let us continue reading:

Alternatively, to speak of two straight lines sloping the same way means introducing the
concept of both straight tine and space orientation [in the original: inclinaison ou direction].
Now we have already seen in Chapter VI how late the straight line comes to be visualized . . .

One would be inclined to say that if straight line comes so late, parallel lines
come even later, thus why is there any further argument (indicated here by the
dots, where I interrupted the quotation). But this should be understood as

follows: parallelism of straight lines is discussed with ordinary drawn lines,

whereas the lines that are said to come that late, are the vision lines of the socalled projective geometry. In order to comply with the system, where projective
space presupposes affine space, the straight lines are forbidden developmentally,
but somehow they must be allowed, because otherwise no experiments on
parallel lines could be performed.
. . . and as for the concept of orientation [in the original: inclinaison] this is a matter of
either measuring angles or else finding some other method of determining the inclination
[in the original: lidentit de direction]. But the idea of parallelism appears at the same

time as that of angles, and this is hardly surprising since a pair of straight lines intersect to
form an angle wherever they are not parallel . . .

Similarly one could say: the concept of straight line occurs at the same time
as that of radius of curvature because as soon as a line ceases to be straight,

a finite curvature can be calculated. Or: the concept of length occurs at the same
time as that of area, because as soon as a line ceases to be thin, it gets an area.
Let us admit that any idea calls up its negation at the same time. The negation
of being parallel may indeed be formulated as forming some (positive) angle

but this is far away from the idea of angle itself, which includes knowing at least
what equal angles are. As a matter of fact this whole story is flatly contradictory
to what has been said a few pages earlier on the relation between affine and

projective geometry:
... and in the next chapter we will see that these twin [original: complmentaires] concepts
are psychologically interdependent. If this is the case, it necessarily follows that the concept

GEOMETRICAL CONTEXTS

237

of angle cannot precede that of parallelism, nor can it serve as measure of parallelity of a
pair of oblique lines.

The French text is here incomprehensible probably by a clerical mistake (omitting a few words). The translator tried to make the best of it, though I would
have preferred serve to measure whether two lines are parallel. It continues:
This leaves us with the idea of identity in orientation [direction] but this is soon ruled out
when it is realized that the concept of spatial orientation [direction] is the foundation
[point de dpart] of the co-ordinate systems themselves. And as will be seen in Chapter
XIII, their development is an extremely complex and protracted affair . . .

A similar reasoning would be: Addition is difficult because multiplication rests

on it. Or: the area of a rectangle is a difficult concept because it is needed for
the definition of integral.
As to the whole story: anyone who has ever been busy with children and in
education knows that angles and angular measures are much more difficult
mental objects than parallelism. It is not a problem to explain, say, to a fiveyear old what parallel lines are: show him one example or two, and perhaps a
counter example. With somewhat older children one can even successfully
analyse the phenomenon parallelism. To do this one need not be able to
measure lengths, let alone angles.
Moreover the claimed dependence of parallelism on angles is inconsistent with
the claim that developmentally affine space precedes euclidean space. But let
us read further:
In short, it is no simpler to imagine the parallelism between two lines than between the sides
of a closed and well-organized figure like the rhombus. But it may be asked, surely it would
be simpler to perceive, even if not to imagine? Here the result of comparing perceptual

estimates [donnes] is very much to the point, for actual study of the perception of parallelism leads to the conclusion that the idea of parallelism precedes their accurate perception
rather than being a consequence of it as might have been thought.
Wursten (op. cit.) carried out the following experiment: twenty adults and twenty

children aged between 56 and 1213 were asked to compare the lengths of oblique
lines drawn on cards. Alternatively they were invited to draw vertical, horizontal, and

oblique lines, or else adjust pivoted metal rods in a parallel position. Wurstens findings
were as follows: first, parallels are never perceived entirely without errors, even by the
experienced adults. This is a further confirmation of the intellectual, logical character
[caractre rationnel] of geometrical concepts which govern and influence [informent
et corrigent] perception rather than being wholly dependent on it . . .

The last remark is correct, though not as a conclusion of the preceding. A certain
concept is independent of perception not because perception is liable to errors,
but because the concept enables us to establish the fact that the perception is
wrong.
. . . Second, and most important, comparison of variations in thresholds and constant
errors showed perception of tilt [inclinaison] and spatial orientation [direction] to be

extremely poor below the age of 78. The reason why young children are better than
adults at comparing the lengths of lines pointing in different directions is precisely because
they remain indifferent to their relative orientation.

238

CHAPTER 8

Here a particularly revealing foot-note is added:

Less difficulty is experienced with vertical and horizontal parallels. Hence in these two cases
it would seem that perception of parallelity precedes the idea [notion] of it.

Take a long draught of it. Vertical and horizontal here mean, as in practically
the whole book, even in interviews of children, directions not in space, but on

the table between experimenter and subject. Vertical is the direction of the
one to the other, horizontal that orthogonal to it, along ones chest. It is in
this world that the coordinate systems of Chapter 13 of the book arise. Walking
around the table, at least mentally, rotating the table, or the drawings on it, is
forbidden. It is a world locked up in a rectangle. In the whole book Reprsentation de lespace chez lenfant space practically does not occur the world is
flat and is most often a table, direction means orientation with respect to the
edges of the table. Only the little ones cannot be forced into this frame. They
perform better in the meaningful experiments, which do not depend on the
frame. By disregarding the meaningless frame they show more genuine mathematical insight than the experimenters allow the older and adult subjects to
show.
I leave it here, but I cannot but ask myself: Is this really Piaget, or did he
never see the proofs?

8.8. Reproduction Symbolic and Ikonic

Researchers piagetian and others often show that they did not grasp the fact
that lack of names for mental objects and actions or lack of knowledge of the
conventional names does not prejudice anything with respect to the possession
of the mental objects or actions themselves. But even grasping that fact does not
protect one against serious misconceptions. Even the so-called non-verbal tests
need not prove anything. One cannot measure whether or to what degree the
mental objects (or the concepts) circle, square, straight line, and so on, are
present by having the subject draw or in some other way reproduce that figure.
Anyone who has observed children, is familiar with the technical difficulties
they have in expressing their intentions with drawing instruments on paper or
otherwise intentions that can simply get lost in their failing attempts. I fail

to understand how some researchers Piaget included will dare to interpret

the failure to reproduce figures by congruence or similarity as a proof of priority
of topology on euclidicity. Meanwhile more critical investigators have shown
that young children are certainly able to distinguish a better copy of a circle

from a poorer one and to appreciate them as such. There is a big gap between
recognising two figures as congruent or similar and being able to copy them
as such, not only for little children but also for adults unless they are gifted with
extraordinary graphic talent. Yet there are still researchers who forget about
such clear differences.

GEOMETRICAL CONTEXTS

239

The misapprehension is, however, more deeply rooted, in the relation between
reality, image, and language, or rather in the way that this relation is experienced.
Everybody understands what a tree is, though there are border line cases where
one can doubt whether something is still just a tree or is not a tree any longer.
A picture of a tree drawn or sculptured will be recognised, at least in our
cultural environment an Indian from the Amazon or a Greenland eskimo
may view it differently. Here too there are border line cases: how far can the
artist go in the sketch or in the impressionist manner to get something accepted
as being the image of a tree? But besides trees and images of trees there is

the word tree, accepted all over the world by English speaking people to
designate a tree, though misapprehensions are possible between people who
pronounce it differently. And finally there is the graphic image t-r-e-e for the
word tree which in print or writing can look differently, known as such to
people who have learned reading and writing.
In each particular case we know very well which tree is meant. If the teacher
pronounces the word tree, it depends on the situation whether the pupil points
to a tree, or to the picture of a tree, whether he repeats the word tree or writes
it down. Of course miscomprehension is possible, though in general it is not
serious. It is more serious that most of the authors of set theory textbooks
for primary and secondary education got into troubles or dragged users of their
books into troubles with sets whose elements can be trees or pictures of trees
or names or graphic pictures of names of trees and finally all of them in the
same Venn diagram.
I tackle this here, not in order to identify this kind of misconception in
experimental tests (they are being made, especially with logic blocks), but
because I am afraid that misconceptions about reproduction might unfavourably
influence the communication between experimenter and subjects.
The child gets acquainted early with two fundamentally different ways to
reproduce objects and events that is, fundamentally different in our view:
the picture of a fire-engine in action on the one hand, and besides that the
printed text, which according to the reader contains the word fire-engine
and a story about extinguishing a big fire. The child himself can interpret the
pictures and he can check the authenticity of the story by having it read once
more, by the same or another reader. How does the child experience this patent
contrast patent to us between ikonic and symbolic means of reproduction?
I cannot answer this question. Is the contrast really felt as such or is the one
picture for the child just as much pictorial as the other? Is the adult more able
to look at pictures, in the same way that he can take longer steps, climb higher,
speak louder? Writing and drawing are often synonymously used by children,
as are reading and looking at pictures.
One can certainly observe with children, as regards their internal and external
means of expression, a development from the ikonic to the symbolic. Yet the
question that puzzles me is whether and when a child draws a border line between
ikonic and symbolic representation. If a 24-year old draws as it seems at

240

CHAPTER 8

random a few lines and asserts this is Uncle John or a doll-house or if he

answers the proposition draw a . . . with a seemingly disorganised system of
scratches, is it then true that he is ikonically busy, that he has ikonic intentions,
or does he act on the same legal grounds as the adult who claims to see a tree
where there is nothing that looks like a tree? Would it not be more likely that
a child who starts drawing or reproducing in some other way some reality,
is moving in a sphere where the ikonic and the symbolic are not yet separated?
I once observed such a separation becoming conscious:
Bastiaan (4; 8) who is encouraged to start early to make a list of suggested gifts for his
birthday: Writing by words or drawing by pictures?

But even if the separation starts early, how long does the process last? I am sure
there are people who never manage it completely, who remain convinced that
a tree is called tree because in some strange way the word tree is similar to
a tree, that the word graphically reproduces the tree it is kind of magic belief.
As a matter of fact set theory in textbooks is a proof of what difficulties even

adults can have in this field.

Even if a borderline is drawn between ikonic and symbolic reproduction,
it need not be the one we adults are used to. A drawn picture of a house, which
we think is ikonic, can perhaps be meant symbolically, or symbolism may have
dominated its production. Even school-children who are asked to draw their
own house may produce a stereotype which does not resemble their own house,
with stereotyped details such as corner curtains such as they can only have
seen in standard houses in picture books the symbol of a house.
How does a psychologist manage to have children producing or interpreting
pictures without being sure that the child understands the assignments as
they are meant, whether they conceive the ikonically meant ikonically, and
the symbolically meant symbolically, whether they draw the borderlines as

they are intended and whether they know such borderlines at all at least
operationally?
For the experimenter, a drawing of a circle with an inscribed equilateral
triangle has a structure determined by his geometrical experiences; for a child
who has not had much experience with geometrical figures, the figure can be
meaningless, or ornamental, or a picture ikonic or symbolic of something,
and the particular view that it has of the figure determines how it would react
to the assignment to copy it. The child may have seen symbols of Fiat, PTT,
VW, and recognises them by some structural resemblance, even though the
various instances are not at all congruent or similar. But the circle with the
inscribed triangle what does it symbolise and which details do matter if it is
to be copied? Must the circle be truly round, the triangle precisely inscribed and
equilateral, or which deviations are admissible?
Adults who have not the slightest difficulty to recognise the symbol of the
Netherlands Railways (Figure 86) do have the greatest difficulty to draw it from
memory, and even when copying it, they repeatedly look back at the model.

GEOMETRICAL CONTEXTS

241

Why? Because it is an arbitrary symbol with no clear context. The circle with
the inscribed triangle, however, can be placed into a geometrical context, built

up from geometrical objects.

Researchers have children copying models without checking whether they
know what it means to copy something. Perhaps up until that moment these
children have only drawn objects. In order to copy something, one must know

what matters. Is the model to be copied an ikon or a symbol? For instance, the
picture of a direction post where the angles of the various hands have to be
respected can instead be the symbol of a railway crossing, with the symmetric
disposition of the crossed arms? Learning to write letters and figures seems to be

a struggle between ikonic and symbolic reproduction. Though ikons of one

geometrical figure, the letters b, d, p, q, have different symbolic values.
Bastiaan (5; 6), who permits himself all liberties with the images of the figures 0, 1, . . . ,
9, protested when in a typewritten text the figure 1 was indicated by the letter 1.

Knowing what matters is a precondition of copying experimenters are not in

the habit of telling their subjects, perhaps because in their own world it goes
without saying, or perhaps because if they explained it, the problem would be
too easy. Whether Uncle Sam, if copied, gets the prescribed height of his hat,
is different from whether the stripes and stars are forgotten. It is a different

thing to draw some set of teeth, or a set of teeth as a symbol of Jimmy Carter.
There are many shades between the ikonic and the symbolic. Caricature can
show more resemblance than portrait, but then a resemblance with a person

that has become a symbol.

8.9. Reproduction From Geometrical Context
In Section 8.5 I dealt with the context of the rigid, congruently or similarly
reproducible bodies, in Section 8.6 the world of the boxes, also reproducible,

though not that rigid; we will learn of other contexts, even less rigid or so
individualised that no thought of reproducibility comes about. As a matter of
fact, the objects in our contexts are not necessarily bodies, even if the term
body does not include three-dimensionality. In any case for geometrical objects
the possibility of reproduction is an important feature reproduction by means
available or created for this special aim, sometimes with great difficulties. We
know wire models, plaster models, cardboard models of three-dimensional
figures, but the most usual reproduction is: on the blackboard or on paper,
in books and on sheets.

242

CHAPTER 8

The prescription which the adult is expected to observe is to represent things

visually in the way he sees them. Archaic and primitive art, however, shows
that it is not that easy to see things in the way one sees them. A geometrical
theory perspective has been created in order to deduce how things are
seen, and experimentally this theory can be confirmed by means of a camera,
at least as far as it is allowed to identify the lenses of camera and eye with
each other as well as the photo-sensitive layers of a film and a retina. Yet
developing a film and processing a retina image in the brain do not seem to be
isomorphic procedures. In any case the mental image of, say, a cube seems to
differ considerably from the visual one prescribed by the theory of perspective.
Seeing, interpreting, and producing perspective drawings is no nave ability,
but something that must be learned. By no means can I say what the mental
image of a cube looks like as a matter of fact it depends on a variety of
circumstances. Certainly it involves more and other features than that which
one sees or is expected to see. It involves as much as one needs to recognise,
to make, to produce, and to reproduce cubes. It includes six faces, though one
cannot see more than three at a time and may be unsure about the actual

number, four, or six, or eight. About a man one is facing, one knows he has
a back even though it is invisible; about a house, that it contains rooms and
stairs behind its walls.
One has more sense-organs than two eyes, and a drawing may be used to
communiate more than visual perception. The little child does not yet divide
his knowledge about the world into compartments according to the so-called
five senses. A 5-year old, ikonically precocious boy amplified his marvellous
drawings of airplanes with images of the noise produced by the jet engines.

A 6-year old reproduced the turning flashlight of an ambulance by a triple of

three lights, one in forward direction, one right, and one left. The desire to
depict things other than those which the eye perceives according to the theory,
is not restricted to childhood. The great problem each painter has struggled with
is to process his impressions in a way that they reproduce a more objective
reality than his impressions do. The symbolic is entangled with the ikonic;
viewpoints are chosen in order to have the ikonic doing justice to the symbolic
a rule even observed by the photographer.
56-year olds draw houses with a front and side facade. It looks like an
attempt at perspective though it can be an imitation of a badly understood
drawing method. It can also witness a reflection, although not yet governed by
perspective: from certain viewpoints one can indeed see two sides of a house,
but it does not yet matter how this happens in detail. It is difficult to ascertain
what is achieved by posing questions as the child readily produces ad hoc answers:
Bastiaan (4; 3) has drawn a house with many rooms and appartments. There is only one
bathroom. When asked about it, he answers: All bathrooms are the same, arent they?.

Traditional geometry instruction does not even face the problem of reproducing. The child is expected to have caught somehow and accepted the adult

GEOMETRICALCONTEXTS

243

methods of reproducing. A cube is drawn on the blackboard, allegedly in perspective, though to avoid too strange a look, with parallel edges. Afterwards
this can mathematically be justified by looking from an infinite distance, as
they say, though a cube so far away would look infinitely small. Even all the
contradictions are taken over from the adult methods of reproduction, even
those which can only be justified by their usefulness and intuitivity, such as the
usual representation (Figure 87) of the globe with equator and poles counter
to perspective but suggestive and convincing.

The perspective is reprsentation de lespace, not in the sense of a mental

object but of reproduction on a piece of paper, a method acquired by imitation,
which is systematically exercised by teaching the pupil to see what he (the
pupil) sees lines, planes, light, shadow and which is finally rationalised in
a fully developed theory. But to stress it once more, primarily perspective is
not a geometrical context but a kind of reproduction, side by side with others,
and this remains unchanged for a long time.

This does not hold only for perspective. Initially, and to a high degree,
reproduction is a matter of imitation, even before kindergarten. As soon as
the child comes into contact with older ones, his method of reproduction is

influenced. As a matter of fact adults too sometimes produce baby drawing

methods alongside the more familiar baby language in order to be imitated.
Daphne (5; 1) gave me the drawing of a house with two chimneys which as usual were
drawn orthogonal to the planes of the roof, rather than vertically. I took her to the window

and showed her a roof with a chimney. Immediately she made a correct drawing. This
spontaneous reaction is astonishing. Clearly the original drawing was nothing but imitation

of what she had seen from another child.

A somewhat complex drawing of a child is a composition of more or less

obligatory parts, combined in a somewhat functional way the reproduction
of a combinatorically flexible structure; what counts is the combination of the
parts: the eyes are in the head, the ears and limbs, perhaps also the belly are
systematically connected to it, the one at the right, the other at the left, the
one besides the other, the one below the other. Parts can be meant ikonically,

but their presence and location is symbolic. This is already a geometric context,
not topological but rather combinatoric, with parts determined by the flexibility
of the structure. The size relations are symbolically rather than ikonically
reproduced this is again not a matter of defective mental objects but of
principles of reproduction (as in the case of perspective) where the symbolic is

dominant or not yet separated from the ikonic.

244

CHAPTER 8

However one looks at a cube or turns it, one cannot see more than three
faces at a time. To make a cube, one needs six. The technique that accounts for
it, is networks. On the network of the cube you can see the six faces at a time,
and as far as it is not clear, one can indicate the edges that have to be identified,
mentally or by adhesive tape. This too is a reproduction of the cube, not a
perspective but a combinatoric one, from parts that respect the similarity of
reproduction. Blue-prints are the same: each storey of the building is indicated
separately to scale, where the combination of the parts is marked by special
signs as well as by stairs and lifts; flexibilities, such as of doors and windows
can also be indicated. Another method involves three projection planes; plan,
front elevation and side elevation.
Let us call this kind of reproduction that is combinatoric from ikonic
parts and accounting for flexibilities compository. It is more flexible than
perspective; and in geometry it is at least as important for reproducing objects.
The childs method is predominantly compository. If he wants to draw the
interior of a house with two stories, after the frontroom of the groundfloor
has been plotted, he has to solve the problem of the composition of the backrooms of the groundfloor and the upper front room. Somehow he solves it

you could say in a primitive way, were it not for the fact that most adults
do not know what to do either. There are techniques of reproduction required

to solve it, such as the network of the cube, the blue-prints, descriptive geometry,
or artistic sophistication.
But what I would stress here is that the compository method of reproduction,
for instance, of a cylinder by means of a rectangle and two circles, or of a cone
by a triangle and a circle, which somehow are attached to each other does not
at all bear witness to defective mental objects. On the contrary, this way of
reproducing can prove a better view of the mental objects than reproduction
by means of perspective acquired by imitation.
Another method of reproduction is the topographical one, as used on geographical maps, railway nets, motoring maps, most of them to scale but not
ikonic, with cities, towns and villages indicated by too big spots, while rivers,
roads, railways are shown by too thick lines; with airports symbolised by drawings of airplanes, ways out by circles, bridges and ferries by other symbols. Is
it then a mad idea of a 56-year old who draws a network of streets to lay the
stop and priority signs as it were on the ground? The symbolism in the adult
topography is more subtle but it is symbolic and most often conventionally
symbolic. If we do not draw from the adult topographic reproduction the
conclusion that certain mental objects, such as perpendicularity, are lacking,
we are not allowed to impute to the child such deficiencies.
8.10. Grasping Of, and Putting Into, a Geometrical Context

The examples of the context of rigid bodies in Section 8.5 showed how geometrical contexts come into being. Natural production, craftmanship, manufacture

GEOMETRICAL CONTEXTS

245

and industry have made us familiar with geometrical figures, their congruence
and similarity, with rectilinearity, orthogonality, symmetry, parallelism. Tables,
doors, sheets of paper, windows, beds are rectangularly produced and impose

the rectangle as a mental object on us; we are being prepared to accept the
name rectangle and to name each rectangle (even a square) a rectangle. Of
course things are not always as easy. Is a diamond a rhombus or a square?
Standing on a corner can be a more important property than having equal sides.
A long rectangle looks different from a tall one, a cylinder lying down different
from a standing one typical consequences of a geometrical instruction guided,
rather than by the objects, by their drawings. Rhombuses in jigsaw puzzles,
cubes and cylinders from the construction box, fit better into the context of
congruently and similarly reproducible rigid bodies.
Contexts should not be taken for granted, but once grasped, they can function reliably. This presupposes that the characteristics which matter in the
context are paradigmatically clear.
A child understands early what things are to be classified as chairs, but in
a certain context a chair can be appointed to be a locomotive or a ship. Words
like triangle, square, rectangle can be meaningfully used by little children,even
to recognise these geometrical structures where they are obscured by roughness,
imprecision and rounded corners. A bench (without a back) made from three
parallel planks with two interstices is seen as one rectangle or as three of them

depending on your preference. The gestalt forming procedures that are active
here, are not restricted to geometry. They do not differ at all from those by
which we interpret a constellation as a dipper, or a cloud as an elephant.
The context required for recognising and reproducing figures can be determined more or less sharply by the data, and whoever wants to interpret the
behaviour of others in such activities, should first analyse how the results are
determined by the suggested context.

Suppose a person is given material, say plates, that differ with respect to
external shape triangles, circles, squares, and so on,
finish rounded or sharp comers, rough or polished surface, with grooves
or prickles, and so on,
internal shape with various numbers of holes of various sizes and shapes
(triangular, round, squares) and in various arrangements,
thickness,
colour or colours,
matter wood, plastic, metal.
Conclusions shall be drawn from the way the subject classifies the material.
For instance one expects some subjects or age classes of subjects to classify
primarily and by preference according to certain criteria, and one undertakes to
test hypotheses on this behaviour. By varying the number of objects representing

246

CHAPTER 8

a certain class, by stressing more or less some differences between class characteristics, by controlling the distribution of the characteristics over the classes, by
coupling characteristics more or less closely with each other, one can influence
the results in a decisive way. It can happen that thickness is the most striking
characteristic, because only two thicknesses occur or because there are ten of
them that are regularly ranked. By admitting only two or three, strongly different
colours, the stress can be shifted to colour. If the material is appropriately
chosen the most striking features can be small and big, or sharp and rounded,
or with round and not-round holes.
Geometrical criteria of classification can be
congruence,

similarity,
affinity,
combinatorial equivalence,
flexion-equivalence,
topological equivalence,

but the geometrical context in which such criteria shall be applied, is not at all

self-evident. One can tell a 5-year old to disregard thickness, colour, finish, but
if such instructions are lacking, even a 13-year old might be unable to put
the material in a geometrical context, and this will certainly happen if strong
enough distractors are built in. This kind of experiment, if undertaken in order
to investigate developments towards geometrical contexts or within geometrical
contexts is a priori useless.
Ethologists have experimented with more or less vague pictures of owls
shown to singing birds to illicit fright behaviour; they can tell you how far
they can go with dropping certain characteristics, for a male stickleback the
red belly colour of a putative rival is the signal to defend its territory. Man
child, adolescent, adult recognises places, things, persons, and identifies
classes in order to classify, by means of a small number of criteria which
rarely become conscious. Yet with regard to geometrical objects, the mental
development can lead to making criteria of recognition and classification
conscious.
At least so it looks. Without expressing it verbally, one can make absolutely
conscious to oneself and others what is a triangle, a circle; what are intersecting
lines, what is the structure of a cube. But it is much less clear why we ascribe
to an ivory die with rounded edges and vertices the shape of a cube, or more
poignantly said, the same shape as to a wooden die with sharp edges and corners.
What are the criteria? How is the die placed into the geometrical context where
it is judged to represent a cube in fact it stands as well in the contexts of
gambling and of probability. How are we able to agree about how badly a
rectangle may be drawn to be accepted as such, where the tolerance terminates
and where sharper requirements are to be made?

GEOMETRICAL CONTEXTS

247

Bastiaan (5; 6) describes the shape of a piece of wood he says he needs, as like the front
of a car. It appears that he means a rectangle. Though adults would not say so, it is correct
that the front of many cars is roughly a rectangle.
Bastiaan (6;0) says about an empty beer can which is somewhat compressed from two
sides and shows roughly a square section: This is a quadrilateral.

Why are they rectangles and squares (this is what he meant). One could have
continued the conversation: A rectangle has sharp corners, but this can is all
curved. But I myself had admitted it was roughly a square, and he would have
answered the same. Under certain circumstances rectangle and square can
be excellent descriptions of things.
Bastiaan (6; 2) plays with a stick with a longitudinal groove and two bottle tops he found

in the forest, as though it were a machine gun and two bullets. What does a gun bullet
look like? he asks though actually he knows how to draw it. I add: A cylinder with a
cone on the top. I do not believe he knew the word cylinder. Anyway he only asked what
is a cone. I said: A clowns hat. Then I let him show cylinders: pieces of trunks of trees
and trash baskets. I show him flat discs. He agrees that they are cylinders though with the
reservation: We shall rather call them discs. I ask him what you see if you cut a cylinder
this way. His answers have nothing to do with the geometric shape; they are related to
the particular cases. I help him with the word circle, which he apparently did not know.
I show him examples like the section of a tree, the rim of a trash basket, a button, and
I mention sun and moon (or he himself did so), and finally I show him the circular hole

in the top of a beer can. He protested: This is a bit long. Indeed it was elliptical a
difference of less than 10%. The next day he used the word circle correctly.

The sections of trees looked of course much less like circles than the hole in
the metal. But then he did not protest; about the hole he did. Why? Clearly
you cannot require so much geometrical shape from the section of a tree as you
can from the sharply bounded symmetric hole formed by a smooth curve. The
section of the tree did not pretend to be a circle but the almost circular hole
did and consequently it had to be judged by sharper criteria something like
this must have been the background of his evaluation.
At the opportunity which I related in Section 1.28 I explained to Bastiaan
what is a half he did not know this word, at least not related to length or
distance by breaking a stick (not exactly) in half; he protested because one
half was a bit longer. Here again we observe the presence of the mental object
and the testing of the example the first example by the mental object.
Or do I abuse the term mental object and should I rather speak of visual
imaginations? Well, it is visual imaginations but then different from those we
have of animals, trees, stones. The context of geometry implies that they are
normative imaginations, something like Platos ideas, though I would not like
to argue about the origin of these norms it does not matter whether they
are objective, genetically determined or developmentally acquired.
Am I allowed to name imaginations with this degree of exactness mental
objects? It is a meaningless question. I would rather ask another one. What is
the next step in the development? The concept circle, square, half? A definition
like a circle is the locus of ..., or in a modern style a circle with centre M

248

CHAPTER 8

and radius r is . . .? No! The next step is a question the question is how
to make a circle, how to produce a square, how to halve something. One can
suggest the answer by handing our material, or letting the child choose from
the material that is offered. One can also aim for a mental construction by
sharpening the question to: How can you make it more precisely?.
Bastiaan (6;4) asks: Where is the centre of Netherland? (Possibly he had heard about
Utrecht as such.) I tell him it is not easy to determine, and then: What is your centre?
He shows on his top. I argue it should rather be in his belly. Then I ask him about the

centre of a tile of the pavement (Figures 88 and 89). First he denies its existence. Then he
shows what is approximately the centre. I ask him to do it more precisely. He produces
the groove between the next row of tiles and cuts it with an estimated mid-line between
the other sides. I explain to him that it is easier with oblique lines. He draws the diagonals.
I mention the word diagonal. At a bench I ask him to indicate the diagonal of its bottom.
He draws a line that forms an angle of 45 with the sides of the rectangle. I show indignation. He corrects himself immediately.

Is the context of geometry not grasped until the question of the precise
construction arises in order to be answered? Anyway the question is characteristic
of a certain context. Even then the answer can be different according to how

precision is mentally measured this too requires a context.

Summarising: Which symptoms indicate the ability to grasp a geometrical

context and to put objects into it?

By showing knowledge of what matters in the context, by way of
recognition,

classification,
material reproduction,
naming,
mental reproduction
of mental objects and processes and by
making conscious to oneself and
describing

these activities.

GEOMETRICALCONTEXTS

And how is what matters in the context determined? By

natural, craftsman, manufactured, industrial reproduction,

paradigmatic examples,
explicitation.

249

CHAPTER 9

TOPOLOGY AS A GEOMETRICAL CONTEXT

In the original version, I continued Chapter 8 by the question What is topology? The answer led me so deep into topology as a geometrical context that

the frame of the chapter was in danger to explode. Finally I felt compelled to
put topology as a geometrical context outside Chapter 8 as well as the planned
Topography as a geometrical context in order to resume the thread of Chapter
8 in Chapter 11.

9.1. What is Topology (Not)?

As a joke topology is sometimes defined as the art of drawing badly. I will be
glad to take up this joke, but before I do so, I must offer some serious mathe-

matics in order not to interrupt the discussion at an inappropriate moment.

What is a curve? A point moves in the plane or in space, and a (planar or
spatial) curve is the path described by the point. I mean of course a mathematical
point, rather than that of a pencil.
A point moves in the plane or space what does it mean? The mathematician is ready to answer such a question: the place of the point is a continuous
function of a parameter t (the time).
Unfortunately, Peano gave an example of a continuous function that maps
a line segment (a time interval) upon a square (or a cube). A point that moves
continuously according to this function can behave madly enough to describe
a whole square (or cube). No, it is no madness Peanos examples are quite
reasonable and intuitive.
But of course that is not what people mean if they speak about curves. Our
definition was invalid as a description of the mental object that is present in
peoples minds if they speak about curves. They mean something like a thread,
not a piece of surface, not a piece of space.
What is wrong? I mean: why is there a gulf between the mental object and
the concept, between intuition and logic?
The villain is continuity. Continuity too is both a mental object and a concept.
One has tried to define continuity in a way that justice is done to the mental
object, but apparently one did not succeed as one should have done.
Can the wrong be redressed? Can the concept of curve be defined in order
to exclude this kind of abnormality? Certainly, it can; for instance, by requiring
that the function defining the curve is differentiate, or that the curve defined
by the function has a well-determined tangent in each of its points the Peano
curves lack this property practically at each point. This prevents abnormalities
but at the same time it excludes a whole host of legitimate curves, for instance
250

TOPOLOGY

251

each broken line, which as it happens has no tangent at the corners. In order to
save them, more compliance is recommendable: for instance by admitting
piecewise differentiable functions, curves that lack a tangent in a finite number
of points at most. Yes, this is an escape, but properly said, it is a loop-hole.
It is not satisfactory. A concept like curve belongs to topology and should be
defined in a topologically invariant manner, which means that each topological
(one-to-one continuous) image should be of the same kind. Yet topological
mappings do not respect differentiability.
Curve should be some thread-like figure. The problem of how to define

such a thing reasonably, has been solved, but it would take us too far away to
explain and to justify it; it would require too much theory.
The curves we tried to define were continuous images of a line-segment.
This means that they may have multiple points, the moving point may cross
its own path. Let us turn to a more handsome kind, simple curves, as it were.
Let us define:
A simple arc or Jordan arc is the topological image (one-to-one continuous
image) of a line-segment.
A simple closed curve or Jordan curve is a topological image of the circumference of a circle.
Jordans name is attached to these objects because C. Jordan first proved
the famous

Jordans theorem: A Jordan arc in the plane does not divide the plane;
a Jordan curve in the plane divides the plane into precisely two parts.
The last assertion can even be strengthened: Let the plane Jordan curve K
be the image of the circumference C of a circle by means of the one-to-one
continuous mapping f. Then f can be extended to a topological mapping of
the total plane, which in fact maps the interior of C upon that of K, and the
exterior of C upon that of K.

One would not expect it otherwise. With the naked eye one can see that such
a Jordan curve divides the plane in two parts, both of which look like the
interior and exterior of a circle, topologically viewed.
Can you really see it with your naked eye? Figure 90 is such a Jordan curve,
but Figure 91 too, with more bends and trunks, and you need your finger or a
pencil to ascertain what is the interior and what the exterior.

252

CHAPTER 9

Topology as the art of drawing badly indeed in order to put a drawing of

Jordans theorem on the blackboard, one need not exert oneself to come out
with a beautiful product. On the contrary ones effort may even be directed
towards producing an ugly curve, neither convex nor smooth, with bends and
trunks, as Figure 91. Or rather, one is advised to spare no pains to make it
as wild as possible in order to show the learner how eccentrically a Jordan curve
may behave. In fact, this is not superfluous if one wants to convince him that
such an obvious looking theorem as Jordans needs a proof a proof that is
not at all obvious and even not easy.
Once he has understood how complicated a Jordan curve may be, he can be
satisfied with simpler models. Or can he? Is an extremely complicated drawing
really enough to convey the full idea of a Jordan curve? I recall an experience
from my own academic study, which gives me food for thought.

Take two similar logarithmical spirals (Figure 92), given in polar coordinates
by

with fixed
which turn an infinite number of times around the
origin, add the origin itself and join the other ends by a line-segment. It yields
a Jordan curve K going through the origin. Sure, it is a Jordan curve, as nice as
a circle, and the origin is for this curve a quite common point, though it does
not look that way.
Now my own experience with this curve: I was decently familiar with topology, knew that Jordans theorem requires a proof and knew proofs of it, knew
all that was known at that time on mappings of manifolds and mapping degrees,
and yet I was dumbfounded when I discovered that this was a Jordan curve
like others. Since nothing of this kind had ever be dreamt of by me as a portrait
of a Jordan curve, I got second thoughts about proofs of Jordans theorem as I
knew them. Possibly in such proofs appeals were made to too restricted
visual images of a Jordan curve, rather than to its formal definition, a serious
mistake or a source of mistakes. My suspicion was unfounded, all was correct,
and meanwhile I got accustomed to this kind of curve.

TOPOLOGY

253

The logarithmic spiral

is the topological image of the set of non-negative reals

and if
this
image approaches, for
legitimately, that is, the origin. The set of
non-negative numbers completed with a point at infinity is topologically the
same as a line-segment, even though in mapping the one on the other one has
to run through it at an increasing speed. The spiral together with the origin
is consequently the topological image of a line-segment, a Jordan arc. Two such
arcs that have only the endpoints in common (presently the origin and
together with a connecting line-segment make up a neat Jordan curve
nobody can doubt it.
Okay! But this mad point, the origin, which pretends to be a well-behaved
point of K, could it not behave badly with respect to the plane, that is, influence
badly how K lies in the plane? No, it cannot. The strengthened version of
Jordans theorem asserts that a Jordan curve divides the plane, locally and
globally, as a circle does, that is a given topological mapping of the curve
K on the circumference C can be extended topologically to the whole plane.
How then does the impression arise that the Jordan curve K composed by the
two spirals is lying differently in the plane than does a circle C? Well, ordinary
circles with the centre at the origin intersect K infinitely many times whereas

with C they would not do so more than twice. Yet this is not the way to look
at K if one wishes to study the neighborhood of the origin. One has to distort
the surrounding circles like spirals as one did with C in order to straighten
out the image.
All this then is topology:
the mental object of a closed curve with no self-crossings,
accounting conceptually for it by the definition of a Jordan curve (the
one-to-one topological image of a circle circumference),
the hesitation as to whether this includes the pathological K,
the confirmation that the definition includes K,

the question of what caused the hesitation,

the certitude that the hesitation was unjustified.
This is topology, and it is a non-trivial sequence of steps in topology. Nontrivial, because the course could have been different, which I will show by

another example.
Let us mount one dimension higher, where even bad drawings serve no
purpose, good or bad. Let us define:
A Jordan disc as the topological image (in space) of a circular disc.
A Jordan sphere as the topological image (in space) of an ordinary spherical
surface.

254

CHAPTER 9

JordanBrouwers theorem: A Jordan disc does not divide space; a Jordan

sphere does divide it into precisely two parts.
However, with the strengthening, as formulated for Jordan curves, it goes
wrong: the interior of a Jordan sphere can now behave very badly. It need not
look like the legitimate interior of an ordinary sphere, as appears from an

example of AntoineAlexander the horned sphere.

The mental object (or imagination) of a
sphere-like surface
is imperfectly accounted for by
the conceptual definition (topological image of an ordinary sphere).
So I can acquiesce in the fact that

my mental object is less handsome than I thought,

or try
a revision of my conceptual definition,

such as not to admit of the discovered pathology. This has indeed been tried by
distinguishing
tame Jordan spheres

which are lying as neatly in space as I intended with my mental object.

Properly said the difficulties started much earlier, with the concept of continuity (of functions and mappings), which I already indicated. The concept of
continuity evolved in history from a mental object that for centuries had been
clear enough in the minds of mathematicians that it needed no precise definition.
The reason why they could do without it was not the clear mind and vision of
these mathematicians, but the mathematics they cultivated. As long as discontinuity played no great part, there was no need for more clarity on continuity.
A continuous function f that is negative for a and positive for b, must vanish
somewhere in between (Figure 93) blind man can feel it, and also that the

same would not be true for discontinuous functions. The drawing to illustrate
it can be as bad as one likes it, or by preference even worse. It is a remarkable fact that Cauchy who gave the modern definition of continuity almost
simultaneously with Bolzano initially did not care to prove this theorem;
initially he appealed to the drawn image. On the other hand, for Bolzano the

need for a proof of this theorem was just the starting point for his analysis of

TOPOLOGY

255

the mental object continuity, which led him, too, to the modern definition.
Yet Bolzanos interest and strength in mathematics was the analysis of mental

objects rather than the creation of mathematical contents. Cauchy, however,

continued working with the mental objects even where he had performed
conceptual analyses in limits, continuity, derivative, integral, and so on.
Once the definition of continuity was accepted, it appeared to admit functions as continuous which trespassed the limits of primitive imagination. It
lasted for quite a while until the majority of mathematicians acquiesced in this
unavoidable consequence. That is to say, difficulties arose where the step from
the mental object to the concept had not been taken in an early stage. New
generations soon got accustomed to this. Taught by the accepted definition of
continuity, they revised the primitive mental object. But do not forget: the
primitive mental object was indispensable historically in the development
of mathematics and it remains indispensable, in the mathematical development
of the learner, as an endpoint, or as a stage on the way to the more sophisticated
mental object.
With objects that one habitually recognises as geometrical, it is different. As
I stressed several times, one can advance very far in geometry without transforming the mental objects into concepts. If finally one lends them more precision by
putting the mental objects as concepts into a logical system algebraic or
axiomatic one will not be taken by surprise. Made even more precise they
will show the expected properties.
Some restrictions should be made. The mental object rectangle might
originally not have included the square, but as one progresses one gets convinced
that it is better to include it, although in everyday language one would not call
a rectangle something that is clearly a square. One would prefer to count the
sign of a priority road among the rhombuses rather than the squares. It was
probably not anticipated in ones mental object that a parallelogram has no axes
of symmetry. But once drawn to ones attention, this is soon redressed.
A more serious thing is paradoxes, well-known in elementary geometry:
by the wrong imagination of lines intersecting inside rather than outside
a triangle, one can be led into contradictions. It is remarkable that the source
of this mistake is somehow topological. It is a shortcoming of the visual insight
into order properties order, indeed, is a weak form of topology.
Nevertheless it may be asserted that ordinary geometry admits of a deep
penetration before the mental objects must be sharpened into concepts, and
even this can be restricted to a local procedure. It does not require a system,
a global frame. Earlier on* I have shown how far this road can be followed even
in analysis, how long continuity as a mental object suffices, as does the integral,
as a geometrically viewed area.
In spite of all visuality in which topology can flourish, it is a different case.
Since the watch-word Modern Mathematics has been heard, topology has more
* Mathematics as an Educational Task, Chapter XVII.

256

CHAPTER 9

and more been propagated as part of mathematical instruction. The harvest

has not been impressive: tesselations, graphs, the Euler polyhedron formula,
the Moebius stip, perhaps even surfaces of higher genus, but in spite of all good
intentions, it remains short runs and dead ends. With regard to our preceding
exposition this is easily explained. Mental objects do not lead far in topology;
concept formation is required to pass beyond the limits, and then concept
formation which means more than local organisation. The concept former has
mental objects in his mind, and ever new mental objects are formed on ever
higher levels spaces and varieties of arbitrary dimension and structure and
their mappings, connected with algebraic structures in order to be conceptualised again.
And how much is not required for the conceptual sharpening, for limit,

boundary, continuity not to speak of dimension and connection. It is concept

formation on a high level, sophisticated alternations of quantifiers for each
there is a and then not separated from the mental object but starting with
it, keeping it in ones grasp, falling back on it, because unless much routine
is acquired, it is a hard thing to manipulate the concept divorced from the
mental object.
Or should we look for the conceptual sharpening of these mental objects
elsewhere, not in topology?
9.2. The Topological Context Is This Topology?
I started the preceding section with the pun of characterising topology as the art
of drawing badly. Perhaps I had looked too much at Piaget* who deduced from
the poor drawing techniques of the little ones that reprsentation de lespace
starts topologically. I explained meanwhile that the argument of the bad drawing
techniques rests on false delimitations and the paralleling of ikonic and symbolic
intentions of the experimenter who sets the task, and of the subject who performs it.
Meanwhile I myself have stuck to the argument of bad drawing although at an
earlier stage I had asked with regard to representation that we should distinguish
imagination and reproduction, and not identify the mental object with its
picture. Of course this meant that my topology is left stranded in the plane
because that is the domain where drawings are made. Properly said, I should
rewrite the preceding section, but you cannot keep changing. Moreover it is a
good opportunity to show the reader what was wrong. It is a habit I will never
unlearn, in spite of efforts to change my own life and that of others. Again I
started at the wrong end, at the topology which I know, to be sure at a low level
but then from its upper rather than its lower side. Topology soon suggests
topological mappings. It is readily understood as identifying objects by oneto-one continuous mappings and forgetting about more primitive ideas preceding
it. I took too seriously that topology to which Piaget had paid lipservice.
* J. Piaget and Brbel Inhelder. Chapter 8.1.

TOPOLOGY

257

Bad drawings you can never draw so badly that it becomes topology. One
can draw a rectilinear triangle so badly that it is neither right nor isosceles.
(Unfortunately not so that it is neither acute nor obtuse and this then creates
the chance to be misled by the mental object.) One cannot draw a continuous

function in a manner that it is only continuous, even not approximately. Well,

in the case of the Peano function one succeeds in suggesting by a few steps the
whole sequence, and all the so-called pathological cases are being approximated
systematically by non-pathological ones. But all this is far away: it presupposes
too much sophistication.
I shall start at the other end, not with the concept, in order to uncover the
mental objects where they are rooted, but in a more naive way. Yet how
naive can you behave if you have learned a lot of topology?
I note down a few words:
connection, arc, dimension, hole, border, tunnel, cave, path, circulation,
braids, knots.
Where should I start? How systematically? How unsystematically?

9.3. Connection
Connection looks the most primitive. Ones own body is connected, though

disconnected from others. A tree is a connected whole from the roots to the
top, separated from other trees, but you can cut it, into pieces. The network
of streets of a city a connected whole. A river from source to mouth with
all its tributaries connected. The pathways around a block of houses, but
mind! do not cross! Continent means connected; an island is a thing
detached. But how about a peninsula? Three sides water, the fourth land I
learned. How wide is the fourth allowed to be in order to leave it a peninsula?
And how if you pierce the isthmus? And then build a bridge over the canal?

9.4. Jordan Arc

A string is connected until it is cut, and then pieces can be tied together the
breakmake transformation. But the string can also be split, into connected
threads, side by side. Was the string disconnected before it was split? As a twine
of threads it was connected; if they are untwined, it is another object, which
is disconnected.

The string suggests a mental object

a curve, which may cross itself

if I throw it casually on the table, but if I avoid self-crossings it suggests a

Jordan arc,
the simplest connected figure.

258

CHAPTER 9

The simplest because it can be

split by a pointlike cut,

and because it can be seen as
part of any connected figure;
the
points of a connected figure are joinable arcwise.

Though the string is flexible, there is not much stretch in it. In this respect it is
not examplary for the mental object arc. A rubber string is better, but also
too special, because its stretching behaviour is a similarity.

All arcs are the same.

One can see it and understand what it means. One can materialise it
by laying them upon each other, with stretching and shrinking.
9.5. Continuity

One more step? I hesitate. The word same I used above can be put into a
broader context by defining for all that is connected, continuity of mapping:
continuous is that which nowhere breaks the connection.

(Etymologically continuous, indeed, means connected, this then was the

mental object that preceded the modern concept of continuity.)
Furthermore one defines for a mapping
one-to-one-ness: it neither folds nor glues; what is different, stays different.
A mapping can be

topological: one-to-one, continuous.

Two Jordan arcs are topologically the same.

9.6. Linear Order

If I cut a string, I get two strings, which I can again compose. Is the point of
the cut broken into two points, which are afterwards amalgamated? Well, the
concrete string is concretely cut; the cut is no mathematical point. Cutting the
arc can be described as you want it, among others, by having a point divided in

two points, which afterwards are identified. I may also demand that the point
of the cut is attributed to one part only. Then I get an
arc with an open end.

TOPOLOGY

259

Equally well can I imagine an

arc with both of its ends open,

such as
the infinite straight line.
If this one is cut without doubling the point of cut, I get two arcs

with one and two open ends,

respectively.
On an arc there is a natural concept
between;
b between a and c if by the cut at b the points a and c get into different pieces.
As a consequence of betweenness an arc possesses
two opposite orders.
Topological mappings

preserve betweenness,
an order passes into an order.

9.7. Mathematical Comments

How do the various objects depend on each other?
Connection looks like the most primitive concept though initially it is rather

vague.
Jordan arc, illustrated by rope or rubber string looks more sharply determined.
Afterwards connection is more sharply described: the arc is an example of
connection; moreover it is stated that two points of a connected figure are
arcwise joined as it were a criterion to test connection. It is, however, not the
official definition; there are examples of (not too mad) sets which one would
like to classify as connected though they are not so in the sense of arcwise
connection.
Continuity in the sense of preservation of connection is not the official
definition. One can make real functions that map intervals on intervals (accept
between two values any intermediate value) but which are not continuous in
any reasonable way. Continuity of real functions requires the originals (rather
than the images) of intervals to again consist of intervals. However, for one-toone mappings conservation of connection is a valid criterion of continuity in
the usual sense. Consequently, with the suggested definition of continuity the
definition of a mapping to be
topological as one-to-one continuous

260

CHAPTER 9

is valid. By this means

Jordan arc as a topological image of a line-segment becomes more explicit,

though the logical circle
arc
connected = arcwise joinable
continuous = connection preserving
topological = one-to-one continuous
arc = topological image of a line-segment
is not broken.

9.8. More About Connection

A closed rubber string is connected, a Jordan arc the ends of which are tied
together:

a Jordan curve.

One cut suffices to transform a Jordan curve into a Jordan arc. A Jordan curve
can be seen as the
topological image of a circle circumference.
The Jordan theorem says that a Jordan curve in the plane breaks the connection of the plane into an interior and an exterior a visual property which does
not ask directly for a proof.
Is a chain connected? Roughly viewed it is. But the fine structure of a loosely

held chain suggests a system of linked Jordan curves if tightened it looks

connected. If I am right to require that the points of a connected figure can be
joined by arcs, the loose chain according to its fine structure may not be dubbed
connected. But how to separate its parts? How to separate in general two linked
Jordan curves in space? Concretely, I open a link in order to close it after the
separation. A pair
of linked Jordan curves can be mapped topological
upon a pair
of unlinked ones:
upon
, and
upon
. This
mapping, however, is restricted to the curves themselves; it cannot be extended
to space as a whole. The
and
viewed as such are topologically
the same but they are lying differently in space a visually clear fact that
does not ask directly for a proof.
The example of the loose chain shows that connection as a mental object
is not as simple as it looks. The linked curves
though separated, cannot
be pulled from each other. The pair
is topologically equivalent with
the unlinked pair
. We consider the figure consisting of
and
as not connected, because we mean connection as an intrinsically topological
concept, independent of the situation of the figure in space.

TOPOLOGY

261

9.99.16. D I M E N S I O N

9.9. Three Approaches

Old geometry texts start with the somewhat classical definitions:

a point is that which has no parts,

a line is a length without width,
a surface is a length and width without thickness.
Another set of definitions is the following:
a line arises from a moving point,
a surface arises from a moving line,
a space arises from a moving surface.
Starting at the other end one gets the sequence of definitions:
surface is the boundary of a body,
line is the boundary of a surface,
points are the boundaries of lines.
All these definitions interpret visual experiences, but they do it in a way that
does not directly lead from mental objects to concepts, though for centuries

philosophers have believed in this possibility.

9.10. The First
As to the first sequence:
there are objects (spots) small enough to suggest something so small that
it can be divided no further,

there are objects (threads) the width of which pales in significance besides
its length, suggesting in this way something with a length without a width,
there are objects (sheets) whose thickness pales in significance besides
the other extensions, suggesting in this way something with length and
width and without thickness.
The suggested mental objects point, line, surface are conversely useful to describe

certain properties and aspects of real objects (spots, threads, sheets).

In order to sharpen the mental objects conceptually one has to start at the
visually clear height, width, thickness three aspects of use of the same
mental objects, which, however, are not interpreted as a measured magnitude.
How long, wide, thick does not matter; rather the extensions are meant qualitatively. The simplest object of this kind is a Jordan arc the simplest length
without width; the simplest surface would be the cartesian product of two
Jordan arcs. But one would certainly not be satisfied with these choices. A

262

CHAPTER 9

Jordan curve and much more complicated thread-like figures shall be considered
as lines, and spherical surfaces (which are not cartesian products of two factors)
shall be so as surfaces. This can somehow be redressed by interpreting the
definitions locally, but even then one is left with difficulties how should I
understand that an 8-like curve has a length but no width in the double point?

9.11. The Second

Let us pass to the second sequence of definitions. They are more precise than
those of the first sequence, but we already know that as early as the first step

the sequence goes wrong. The moving point the continuous image of a linesegment viewed as a time interval may cover a square and even a cube. This
can be redressed by adding differentiability requirements, so that the line
becomes a differentiable curve, the differentiably moving line a differentiable
surface, the differentiably moving surface a differentiable piece of space. With
the rise of Calculus lines, surfaces, spaces have, indeed, been interpreted in
this way. But adding a new motion parameter, lines, surfaces, spaces were

described, that is, by functions of one, two, three parameters, a variable

point with co-ordinates
was given by

respectively. The functions were supposed to be continuous, which in fact

included differentiability of any desired order. Moreover independence of
the functions
was assumed in order to exclude degenerations of spaces
into surfaces, surfaces into lines, lines into points.

With all these sophistications we are far away from the intended mental
objects line, surface, space.
9.12. The Third

Let us now look at the third sequence of definitions. The striking distinction
compared with the others is the start at the top, with the bodies, bounded
by surfaces, which are defined by this capacity, that is as boundaries. Well,
the definition yields closed surfaces only. This can be redressed by admitting
extended pieces of surfaces again as surfaces. Their boundaries in turn yield
lines primarily closed lines, but afterwards also pieces of them. Lines in
turn are bounded by points. It seems to work better than the first and second
approach. It starts at the top, with bodies, in three dimensions. The descent
to a lower dimension is systematic: the object is deprived of its fullness; the
peel is left; the thickness of the body is lost, it is reduced to its boundary, the
width of the surface is lost by leaving only its border, and the length of the line
by leaving the endpoints.

TOPOLOGY

263

The first two sequences are clearly inspired by the idea that as in arithmetic
one starts at 1, so in geometry one has to start at the point. But as a cognitive
development geometry certainly does not start with points. Earlier on I have
put the rigid bodies first and foremost in the development, and if there were
anything that I would allow to be detracted from bodies, it would be the solidity,
rather than the bodylikeness.

9.13. Surfaces
Surfaces occur primarily as the name says as faces of something, as walls,
tabletops, floors, waterlevels, peels, skins, clothes that wrap, bags that comprise,
barrels that contain something. Primarily, I said, because in the long run we can
detach the surfaces from the bodies of which they are boundaries, even while
using the word surface.*
Sails and flags in the wind, leaves of trees, sheets of paper, curtains are objects
that suggest surface without being surface of something. But unlike the surfaces
of something they have two sides, right and left, or upper and lower, and if
indeed they wrap something, inside and outside sides that can differ by their
look, but which primarily have to do with their situation in space.
Yet quite different physical objects can suggest surfaces: a fence, some
wire-netting, a railing. They delimit space though not in the strict sense of
inaccessibility; in spite of the holes they mark boundaries. The filled net is a

particularly striking example, a surface pervious to water, but not to fish. Even
one closed curve in space can suggest the surface it spans, and this holds to an
even higher degree for nets of closed curves as found in wire-netting.
One step further: Independently of any embedding in space a cut-out or an
atlas may suggest a surface, even surfaces which are not without deformations

or not at all realisable in space, such as the well-known rectangle model of a

torus (Figure 94) or the circle model of the projective plane (Figure 95).

9.14. Lines

After the efforts to show the phenomenological origin of surface as surface

of something one may expect, in order to get a well-shaped closed system, that
* The German language knows Flche and Oberflche. English and French are restricted
to one term, surface.

264

CHAPTER 9

a similar origin of line is postulated, that is as boundary or piece of a

boundary of a surface. Systematics is excellent in an a posteriori synthesis,
but not as an analytical principle. Systems are artful and artificial and for these
reasons sometimes useful, but let us not yield to this seduction.
There are, indeed, numerous examples of lines suggested by borders or pieces
of borders; in particular circles first appear as rims of cups, dishes, bowls, buckets,
wheels, of the sun and the full moon, and as the horizon. If we use the term circle,
but also triangle, square, rectangle, we often do not know whether we mean the
surface or its boundary. The child is early taught to reproduce an object by
drawing what appears as its circumference perhaps against a natural inclination
to draw a whole surface, which attempt is interpreted by the adult as scribbling.
On the other hand we know that for other lines, in particular straight ones, a
quite different phenomenological origin can be indicated. In Sections 8.5 and
8.7 I have argued this forcefully for the straight lines. As objects and processes
that suggest straight lines, or at least rectilinearity, I mentioned

arrows, trunks, sticks, pins, rims, edges, paths, folds, cuts, stretched
strings.

Some of them can, depending on the actual situation, also suggest curvilinearity,
for instance strings if not stretched. Rims, edges and folds derive from circumferences; cuts too, and certainly so if something is cut out. Nevertheless there
are enough examples left of another kind of origin of the mental object line
than as a border. In order to systematise this wealth of examples, I put on

record four roots of the mental object line:

arrow,
string,
path, cut,
border,

and their mathematisations

arrow: line segment;

string: continuous image of a line-segment with a surveyable number of

self-crossings;*
path as well as cut (viewed as covered in time): continuous image of a time
interval with a surveyable number of self-crossings;*
border: boundary of a piece of surface.

Whereas in the case of surface the phenomenological analysis led to a primary

surface of something and a secondary abstraction from the spacelike substratum, we recognise in the case of line a fourfold root the components of
which are phenomenologically equivalent or at least almost equivalent.
* To avoid in an informal way pathologies like the Peano curves, which fill whole squares
and cubes.

TOPOLOGY

265

9.15. Points
Here we can put it briefly. Two possible and equivalent aspects can be distinguished: the point and the spot. The point as the end of a line-like object (the
point of a pin) or as the end of a surface- or space-like object (the peak of a
clowns hat or a pinnacle). The spot as the smallest piece of surface or space,
perhaps produced by the point of a pen, a pencil, or a pin.

9.16. Mathematical Comments

Mathematically two aspects of dimension must be distinguished, the differentialtopological and the purely topological aspect.
In differential topology the objects to be studied are created by mapping
line-segments or products of line-segments, where more stringent requirements
are imposed on the mappings than continuity only: differentiability of some
order and independence of mapping functions. The dimension of the object
is then defined by the number of independent parameters used to describe the
object.
It is unsatisfactory that a seemingly primitive mental object has to be sharpened into a concept so far away, in analysis. This has first been made conscious
by H. Poincar, who at the same time proposed a new approach, afterwards
realised by L. E. J. Brouwer, P. Urysohn and K. Menger, in so-called dimension
theory. The approach is closest to the third sequence of definitions, in Section
9.9. In order to stress what is visually essential, I avoid exaggerate precision in
my exposition:

A pointset is by definition n-dimensional if it admits everywhere cutting

out arbitrarily small pieces by cuts that are at most (n l)-dimensional

(but not already at most (n 2)-dimensional).

This is a so-called inductive definition. One has to start somewhere:
The empty set (and no other) is ( l)-dimensional.
Thus a set is 0-dimensional if everywhere arbitrary small pieces can fall off
without cutting. So is for instance the set of rational numbers, and Cantors
discontinuum. A line-segment, a Jordan curve even the union of a countable

set of Jordan curves is one-dimensional, and so it gets on to higher dimensions.

It is, however, a non-trivial fact and hard to prove that the cartesian product
of n line-segments (the n-dimensional cube) is n-dimensional in this inductive sense. This again creates an enormous distance between the concept of
n-dimensionality and the mental object of which it should be the conceptual
sharpening, though this distance is not yet felt in the definition itself but in
essential applications.
Cutting out and arbitrarily small as meant above require more precision

266

CHAPTER 9

the first by means of the concept of connection, the second by means of that
of topological space.
9.17. Dimension and Measure
It looks out of proportion paying so much attention to dimension and it would
indeed be exaggerate were it not that dimension is an indispensable tool if
magnitudes and their mutual relations are at stake.
The first sequence of definitions, of Section 9.9, is the traditional approach
to formulae like
area equals length times width,
volume equals length times width times height,

definitions which are of course restricted to rectangles and planks, and which
then are supplemented by formulae for
circumference (of the rectangle)

surface (of the plank),

and perhaps more formulae for elementary figures.
Didactically this algorithmic treatment can of course only be the end of a
development which runs through various stages.
From magnitudes one can form new magnitudes by mutual multiplication
(and division). By putting twice or thrice length as a factor one gets the
magnitudes

area
and
volume.
If measuring these magnitudes one chooses the units again as products of the
length-unit, for instance,

as a unit of area and volume, respectively.

If magnitudes are to be measured, the first thing that matters is to know what
kind of magnitude it is this is a general requirement length, area, volume,
time, velocity, weight, work, and so on. At this point dimension comes in, at
least as regards the first three: What dimension does the object to be measured
have? Or rather: Which dimensions of the object are relevant? For instance a
road can mean for me

a length if it is leading me from here to there,

a surface if I must pay asphalting,
a space if a groove must be made for it,

TOPOLOGY

267

and correspondingly what is measured, are

lengths, for instance in km,

surfaces, for instance in m 2 ,

volumes, for instance in m 3 .

There are more primitive aspects concerned, which should be stressed for
didactical reasons,
the behaviour of geometric measures under geometrical multiplication,
depending on the dimension.

Linearity is such a suggestive property of relations that one readily yields to the
seduction to deal with each numerical relation as though it were linear. Understanding that multiplication of
length by d,
areas by d2,

volumes by d3,
go together with the geometrical multiplication by d, is mathematically so
fundamental, that, phenomenologically and didactically it should be put first
and foremost. This fact rather than formulae for circumferences, areas, volumes,
should be primary. The behaviour of various numerical characteristics of the

same object, depending on the dimensions they are related to, plays a part not
only in physical enlargements and reductions but also in reproducing and reading
reproductions of such objects. We may conclude this exposition on dimension

with the somewhat paradoxical remark that dimension, though conceptually

located by us within topology, has its significance as a mental object by its
relation to measures lengths, areas, volumes rather than topologically.

9.18. Border and Boundary

The mental object border (or boundary) has been mathematised in two
divergent ways: set theory topologically and algebraically topologically. The
first version is about the boundary of a set S in a topological space R: a point
p of R is called

interior to S if there is a neighborhood of p entirely inside S,

exterior to S if there is a neighborhood of p entirely outside S,
and the remainder is the boundary of S with respect to R, thus p is called a

boundary point of S if any neighborhood of p possesses points in and not

in S.
Boundary points of S with respect to R can or cannot belong to S itself.
An essential feature of this concept of boundary is its relativity with respect

to the embedding space. A circular disc, considered as part of its plane has the

268

CHAPTER 9

ordinary circumference as its boundary, the same disc as part of the (ordinary)
space is entirely boundary. A hemisphere as part of a spherical surface has a
circle as its boundary, but as a part of a full sphere it is all boundary.
In contradistinction to this relativity the natural mental object border has
an absolute meaning: the border of a sheet, of a footpath, of a bucket, of a
city depend only on the object of which they are considered as a boundary,
independently of any embedding in a larger something. At least, so it looks.
However, one can view it also in another way:

if we speak of the border of an object we relate it to a natural extension

of this object, we view it as naturally embedded into a larger whole.
Such an embedding is most natural for flat objects in space: the border (or
boundary) is determined with respect to the plane in which it is included. The
endpoints of a line-segment form its boundary with respect to the straight
line that extends the line-segment.
This then is the viewpoint of the algebraic version of the border concept. One considers
points, line-segments, triangles, tetrahedra and higher dimensional analogues and defines

their border in a natural way, independently of any embedding, albert with a certain
algebraic sophistication: the line-segments, triangles, tetrahedra, and so on, are assumed
oriented; they are linearly combined, and borders are defined as linear combinations
(Figure 96):

In the case of flat or rectilinear objects a natural extension is uniquely determined to wit as a plane or a straight line. With a certain latitude the same can
be said of many objects which can be given, even if not precisely, yet with
a certain clarity, and delimited in a larger whole: parcels in a landscape,
countries on the globe, rooms in a house, fences, landmarks, and walls can
represent or suggest borders either isolated or in relation with each other.
Territories, as we shall call them, can be delimited against the world around or

TOPOLOGY

269

mutually by concrete, marked, or symbolic frontiers, restricted accessibility and

defense conditions. Romulus kills his twin brother Remus who steps over the
not yet concretely operational wall of Rome, in order to stress its symbolic
function.
Bastiaan (2; 5) in the park; I drew a circle around him with a stick. You are locked up,

you cannot go out. He accepted it. Only after I wiped out a piece as a door, he stepped

out.
A barrier across a road determines a frontier, though concretely I can pass under,
above, and around it. By his mere presence a policeman closes a street, with his

stretched arms a child marks a blockade of the footway. The law requires a white
line on the street before a traffic-light; a no-entry sign suffices to forbid entrance.

According to the circumstances the two halves of a street are one territory or
two with a continuous or interrupted line as a border. To decide whether a
person is indoors it does not matter whether some window is open or closed,

unless he leans far outside, but even with closed windows the frontier can be
violated if due to too much attention from outside, or if with a draught one
feels as if one is sitting on the street.
The kerb of a side-walk has a certain width; the frontier between two countries is actually a strip of no mans land. On the other hand land territories

extend to an undetermined height and depth into the soil and the air. It depends
on the functionality of the border how strongly the bounded territory is of a

higher dimensionality than the boundary.

The words border and boundary from the title of this section have been used
synonymously, sometimes the one preferred to the other a third word I used
was frontier. Border accentuates the separation of territories or of inside
and outside; boundary is rather the end of an object. The extension of a
territory beyond its border is concretely given again a territory. Extending
a cup, or a vase, a dish, a cap beyond its border or boundary is a matter of
imagination. In order to extend such objects so unambiguously that a boundary

can be claimed with respect to the extended object, another latitude is required
than in the former examples. It now matters how the cup, or vase, or dish, or
cap would continue; rather than how far or how thick, it matters how curved,
how vaulted. Or rather, even this does not matter because finally what is the
border does not depend on the way of extending. One can see and feel where
the objects terminate. Two-dimensionally viewed these are surfaces with a
boundary; at ordinary points one can move in all directions, there are disc-like
surroundings; at the boundary the freedom of movement is restricted, in some

directions one would drop off. But also three-dimensionally viewed the border
of a cup, vase, dish, cap behaves differently at the boundary: the surface of this
body is sharper curved in the points one would call boundary than in their
neighborhood.

Let us summarise: The set theory topologjcal concept of boundary rests on

a relation with the embedding topological space. The mental object border

270

CHAPTER 9

can be sharpened to this concept, if as embedding topological space a natural

extension is presumed which is either concretely present or derived from the
structure of the object by imagination.
In the following, border and boundary will be viewed from the aspect of their
coming into being.
9.19. Prick, Cut, Slit, Passage, Hole, Tunnel, Cavity

The words in the title of this section aim at phenomena which can be concretely
present in some object or concretely or in the imagination caused by acts
which are described by (similar) verbs
pricking, cutting, slitting, drilling, digging.

Inverse acts are

fusing, stitching, gluing, shutting, filling.

The prick in a one-dimensional object makes it fall apart at least locally;

a new one-dimensional object comes into being, or two or even more. A twodimensional object, say a sheet or a disk is pricked in order to attach it somewhere or to have it turning around the prick axis. Sewing is stitching together
by joining prick holes by threads how is this done by the sewing machine?
A three-dimensional object can be pricked in or through a candle pricked in
the pin of a candlestick, chestnuts pricked through by means of a darning needle
but curved pricks, too, are admitted. The prick suggests a damage of the
lowest possible dimension 0-dimensional in 1- and 2-dimensional objects,
1-dimensional in 3-dimensional ones. The damage caused by a cut in a sheet
or in a body by a slit is one dimension higher. Cutting can be cutting in as
well as cutting through, slitting can be sawing in or through. Cutting in a sheet
at the border does not change the structure too much; if the sheet is elastic one
can stretch the cut-in sheet to look like the original one. However cutting in
the interior of the sheet creates a slit within the sheet, a stretching of the prick.
Cutting out creates a hole, which can be viewed as originated from stretching
the prick or the slit. The more holes I make in a 2-dimensional object, the more
complicated its structure becomes; there are 2-dimensional objects with systems
of holes: sieves, nets, curtains.
This is the starting point for a full-scale mathematisation: the function theory
plane (or spherical surface), from which a closed set is taken away a system
of punch-holes, point-like ones, line-like ones and true holes it does not
matter but such that the surface does not fall apart. If the number of punches
is finite, say n, the surface is n-fold connected, which means that n cuts from
border to border cause the surface to fall apart: one cut for the once punched
surface it is simply connected; two cuts for the twice punched surface it is

TOPOLOGY

271

doubly connected; for the thrice punched surface (the swimming trunks) three
cuts it is triply connected. These are visually obvious facts, but a proof that p
cuts from border to border certainly split the spherical surface with p holes,
is not easy.
Likewise 3-dimensional objects can be split superficially or thoroughly.
Superficial splitting does not change them essentially; if cut through, they fall
apart. Splitting is a 2-dimensional damage. A 3-dimensional damage is suggested
by drilling, which again can be superficial or piercing. A pit does not change
the terrestrial surface essentially nor does a blind alley; a tunnel, however, which
emerges elsewhere, does. Tunnels can appear in various ways: a tunnel is produced
by the handle of a cup or bin; with the legs spread or the hands clasped a tunnel
is formed; children when playing form an archway passed by others; front
door in, back door out is as it were a tunnel through the house. The power
line crossing the road between two poles looks one-dimensional, but the radio
black-out caused by it makes it a true tunnel. Underpasses for pedestrians
below city squares form a branched system of tunnels; the table with its four
legs on the floor creates a similar tunnel system, with well distinguished passages.
Or, even more involved: the tunnel system produced by the edges of a cube
the passages through it are determined by the sequences of faces of the cube
which are being pierced.
The surface of the once or multiply pierced ball deserves to be considered: the
surface of the once pierced ball is essentially a ring surface, or in other words,
a torus. The twice pierced ball has a surface which is known as that of a pretzel;
it may also be represented by a cup with two handles. A ladder with
rungs
is equivalent to the p times pierced ball; its surface is of genus p.
This kind of surface can also be subjected to a complete mathematisation.
They are, apart from the punched sphere, closed surfaces, though of another
kind than the spherical surface. By a return cut the torus can be changed into
a cylinder, which in turn gets simply connected by a cut from border to border,
and by one more cut falls apart. In other words by two return cuts the torus can
be made simply connected. For the pretzel it can be reached by four return
cuts. In general: in order to make a surface of genus p simply connected I need
2p return cuts; it is
connected.

Two return cuts make the torus simply connected (Figure 97).

Four return cuts make the pretzel surface simply connected (Figure 98).

272

CHAPTER 9

Moreover if such a surface is q times punched, it becomes a bordered surface,

which is
connected.
Again the mathematisation requires profound and difficult reasoning.
Among the words in the title of this section there is one left that requires
some comment. It is the word cavity, by which I mean a place you cannot get
in if you are outside, nor out if you are inside: the cavity within a soap bubble,
in a closed tent, in a closed room. It need not be spherical as is the interior of
a spherical shell; the cavity may be a tunnel within a globe, or within a tunnel.
It may be a multiplicity of cavities as those in foam.

9.20. Multiple Connection

Connection was the starting point of our sequence of topological pictures.
In the course of our exposition the concept of connection underwent a sharp
differentiation.
A sheet of paper and a spherical surface are called simply connected because
they fall apart by any cut from border to border or by any return cut, respectively. A sheet with n holes requires n cuts to make it simply connected, a
surface of genus p requires 2p return cuts in order to be spread as a simply
connected domain on the plane. Conversely, these surfaces closed or bordered
can be built from plane networks by sticking the cuts together (Figures
97 and 98).

This homological approach to connectedness can be confronted with the

homotopical one: On the surface one considers closed paths starting and
finishing in a fixed point, which are allowed to be deformed while the start and
finish remain fixed. In the plane and on the spherical surface each closed path
can be contracted into a point, the constant path. On a torus (Figure 97) this
is obviously no longer true: the paths indicated by 1231 and 1451, briefly called
u and v, cannot be contracted, nor can they be deformed into each other, or
into their inverses
that is, the paths u and v completed in the inverse
sense. The paths u and v can be combined to form new paths, like

which, read from the right to the left, means completing first u, then twice v,

TOPOLOGY

273

then again u. The paths obtained in this way, however, are not all different in
the sense of deformational equivalence. For instance uv and vu can be deformed
into each other. So u and v can be considered as commuting; thus by deformation all paths can be reduced to
umvn with integers m, n.
Similarly Figure 98 of the pretzel surface shows paths u, v, x, y, from which
all other paths can be combined while

is a contractible path. From x, y, u, v one can form all words not in a

commutative way but as though (*) represents the unit.

The same kind of concepts apply to bordered surfaces. Take for instance
the swimming trunk, that is the twice punched plane (Figure 99). The paths

considered start and finish at 0; the path u circulates in a certain sense around
the one hole, the path v around the other one,
and
are their respective

inverses. They can be combined to form new ones, while paths that can be
deformed into each other are considered as the same. All closed paths starting
and finishing at 0 can be combined from u, v and their inverses, while uv and vu
are now to be taken as different.

The paths considered on such closed or bordered surfaces constitute

what is called the fundamental or first homotopy group of the surface. I have
briefly tackled this subject in order to show another variant of the mathematisa-

tion of multiple connection: besides the homological one, by cuts, we also have
the homotopical one, by paths.
9.21. Circulation
If I travel along a closed path in the plane, there are points that are orbited,
and others that are not; I can even distinguish whether the path turns left or
right around some point. The paths considered are allowed to cross themselves
see the examples Figures 100105.

274

CHAPTER 9

In order to provide more substance for the present section, I am going to

tackle a mathematical problem, which I will solve in all rigour.
There is a celebrated topological problem, the so-called four colour problem:
how many colours are required to colour a map that is, a geographical map
picturing countries such that each country gets a colour and countries with
a common border are differently painted? For instance, the map of Figure 106
requires four colours. It is an old problem as to whether four colours suffice
in all cases: the four-colour problem. After numerous fruitless attempts the
problem has recently been solved, as it seems: the answer is positive.

The problem I am going to tackle is much easier. There are maps that can be
coloured with less than four colours. I will indicate a kind where two colours

suffice, say black and white (if these are rightly called colours).
A closed path divides the plane in a number of domains. If the path shows
a finite number of self-crossings, two colours suffice. This is clearly seen in
Figures 100104, whereas Figure 105 requires three colours.
In Figures 100103 you may notice certain numbers the meaning of which
I am going to explain.

TOPOLOGY

275

If C is a closed path and p a point not lying on C, I can ask the question how
many times does C revolve around p? I draw a ray from p to the point x that
travels along C and I count how many full turns (of 360 each) the ray px
has completed if x is back to its starting position. As usual in mathematics a
clockwise turn is counted negative, an anti-clockwise turn positive. It can happen
that the ray px runs back and forth, but what matters is the final result.

In Figure 107 the path C circulates 4 times around p; if C is replaced by its

inverse, it becomes 4 times. Around the point q, however, the path C circulates
2 times: starting with x at a the first revolution is completed if x arrives at b,
the ray qx starts running back at c, from d to e it resumes its forward motion,

from e to f it runs again backwards, and from f to a forwards to complete its

second revolution. (Look now at Figures 100103 and verify the numbers in
these figures.)
An easier way to determine the circulation number is the following: Draw a
fixed ray S from p. The path C pierces the ray S several times, sometimes in the
positive sense, sometimes in the negative one. Add these numbers while taking
the sign into account. The result is the circulation number. For instance, if
in Figure 107 the ray S is drawn from p horizontally to the right, it is just 4
times. If S is drawn from q horizontally to the left, it is three times positive and
once negative, which makes it 2.
The number I get by this prescription does not depend on which ray I have
drawn from p. If I change the ray S, I can lose or gain piercings by C but this
happens always pairwise, a positive and a negative together look at the six
positions of S in Figure 108! (Look also at Figures 100103!)

276

CHAPTER 9

But what happens if I move point p? I take two points

how many times C circulates around each of them. I join
segment L. Suppose L is not touched by C (Figure 109).

and
and

and look
by a line-

I produce the line-segment L to get a ray S, which as a matter of fact is

intersected by C the same number of times whether it is seen from
or from
. This means that C circulates the same number of times around both of them.
The same holds with respect to
and , which can be joined by a line-segment
that is not met by C. It again holds for
and , for . and . All of them
are circulated around by C the same number of times. This shows:
Points of the same part of the plane, determined by C, are circulated around
by C the same number of times.
What about neighboring parts? Let p and q be two points of this kind (Figure

110). Draw the ray from p that contains q. The numbers of piercings of this ray,
considered from p and from q just differ by a positive or negative unit. Thus:

The circulation numbers for points of neighboring parts of the plane differ
. (Consider also Figures 100103.)
We undertook to prove that two colours suffice to paint maps created by a
curve C with a finite number of self-crossings. Paint those parts of the plane
white where points show an even circulation number, and black where it is odd.

by

Then each part has one colour. For parts with a common boundary the difference
is
. If in one part the circulation number is even, it is odd in the other one,
and conversely. Parts with a common boundary are differently coloured.
This, indeed, proves that maps produced by a closed path with a finite number
of self-crossings can be coloured with two colours.

TOPOLOGY

277

9.22. Weaving, Braiding, Knotting, Linking

Warp and weft the warp is a system of parallel threads, which are divided into
two layers between which the weft is shot. The result is a fabric which can

combinatorically be described as follows: if the warp threads are numbered 1 to

n, the way of each weft thread is described by indicating whether the warp
thread is passed above or below, thus by a sequence of minus and plus (for
below and above). This sequence of signs changes from one weft thread to the
next, according to a periodic pattern if the fabric is regular. The simplest pattern
is the chessboard distribution of plus and minus, which is also used to darn
socks. Plaiting mats a kindergarten activity of former times, perhaps not yet
entirely abolished is a primitive example of weaving: the warp is a coloured
sheet with parallel slits, the weft consists of paper strips of various colours which
are interlaced by means of a darning needle.
From weaving mats and baskets one is led to braiding pigtails. The most
primitive braid consists of two strands (Figure 111); strand 2 passes alternatively
above and below strand 1. A more solid braid is made up of three strands (Figure
112): strand 3 passes first above strand 2 and then below strand 1, which is

continued with a cyclic permutation of the strands: strand 2 passes first above
strand 1 and then below strand 3; strand 1 passes above strand 3 and below
strand 2; and so on. The number of strands and the pattern can be variated ad
lib. This combinatoric description can be built into a whole mathematical theory
of braids.
Knots can be described in a similar way. Mathematically a knot means a
closed curve in space. To conform with this terminology the ends of the knotted
string are joined to meet. The simplest non-trivial knot is the so-called clover-leaf
knot, which exists in a righthand and a lefthand version, depicted from the right

to the left in Figure 113. It seems to be a matter of taste which one is called
right and which left, but it is not such questions of orientation will be dealt
with later on. It is remarkable that right-handed people prefer to tie righthand
knots, that is, knots where the end of the string in the right hand is led above
that in the lefthand.

278

CHAPTER 9

Knots can be contrived arbitrarily. If the lower and upper ends of the braid

of Figure 111 are joined, one gets the knot of Figure 114. Knots can be described
combinatorically by the pattern of crossings. Whether two knots are the same

in the sense of deforming the one into the other, is a question that in general
cannot be decided at sight. For instance a profound proof is required to show
that the right and left clover leaf knot are different. The theory of knots has
been developed to a high degree of sophistication. There is, however, no general
method to decide whether two knots are equivalent.

The last term in the title of this section is linking. As one closed curve
may be knotted, a pair of closed curves may be linked. So are, for instance,

the links of a chain, a key and a key-ring, locked arms. But there are more
sophisticated linkings, such as exemplified by Figure 115. Again it requires a lot
of mathematical theory to prove that closed curves that are obviously linked
cannot be deformed into an unlinked state without crossing each other.

TOPOLOGY

279

9.23. Conclusion
I have been reviewing quite a lot of topology mental objects, processes, relations of a forceful visuality, which however do not demand the birth of
topological concepts. Though mathematics had for centuries been pregnant
with topological objects, neither philosophical speculations on dimension,
point, line, surface, nor more mathematically tainted ones on knots and links
gave birth to concepts; labour, if there was any, ended with miscarriage. The
topological concepts require difficult procedures of constitution, whereas on
the other hand there is no urgent need for such concepts unless we move far
away, into a theoretical sphere. The mental objects, processes, and relations
which have been discussed here, are certainly important. We are accustomed to
them, but we are not accustomed to realise this fact for ourselves and even less
in a didactical context.
Notwithstanding the wealth of phenomenology, displayed in this chapter,
there was hardly any didactical phenomenology involved, or as far as it was
didactical phenomenology, it were so in a negative sense. What I have sketched,
has scarcely been tested on an elementary level. Everybody knows what inside
and outside means, what is a line, a surface, connection, border, knot, link.
It turns up spontaneously in ones mind, and since there is little need for conceptual precision, it is not easily seen from the point of view of teaching matter.
Topological subject matter such as that offered in more advanced expositions
as teaching matter, lacks the character of necessity while the mathematics
developed from it with the sole aim of doing some topology in mathematical
instruction, lacks sufficient motivation and is leading to a dead end. A popular
subject combinatorics rather than topology is travelling through a graph
such that each edge is used once and only once a nicety known as the
Knigsberg bridges. It does not take much trouble to find a necessary condition
for graphs to be travelled in this manner, but proving that this condition is
sufficient is much more than can be expected from a non-professional mathematician. Moreover, it is an unmotivated and isolated subject.
If I try to survey the examples of this chapter, the only one that promises
more in a mathematical didactical respect, is the circulation number, which is
motivated by a true problem and can be placed into a larger mathematical
context. I do not discuss at which age or on which level it could be realised.
This does not mean that the other examples of topology would be didactically
irrelevant. They should be appreciated as what they really are mental objects,
processes, relations and not wantonly subjected to badly understood mathematisations. The context in which they may play a part among others is
the theme of our next chapter.
The time has come to discuss once more the developmental priority of
topology, compared with classical aspects of geometry, such as is suggested by
Piaget and stressed by his followers. Earlier on I explained that Piagets claim
rests on confronting quite diverging levels with each other: in topology he is

280

CHAPTER 9

satisfied with the constitution of quite primitive mental objects, yet with respect
to what he calls projective, affine and metric geometry, he makes high demands
of concept formation. It has been shown in the present chapter that even with
more sophisticated mental objects in topology the need for, and accessibility
of, concept formation is far below that which we know in the didactics of
classical forms of geometry.

CHAPTER 10

THE TOPOGRAPHICAL CONTEXT

10.1. Order of Coexistence Order of Succession

The pair of terms in the title of this section stems from Leibniz correspondence
with a newtonian. Leibniz attacks the absolute space and absolute time of

Newtons Principia mathematica philosophiae naturalis. For Leibniz space is an

order of coexistence, time one of succession, motion is displacement of an
object with respect to other objects that may be supposed fixed. I am not
going to tackle things philosophically, I am rather dealing with Leibniz exposi-

tion as an expression of common sense and as starting point to describe what

I mean by topographical context.
Time, one-dimensional, looks simpler than space: the whole order of succession is described by before and after, earlier and later. Roughly described,
that is: refining the relation is not that easy. Gradually in the development of

the individual the past and the future become differentiated. Philosophically
time appears even more difficult a famous pronouncement of Augustine
witnesses this uneasiness. That which was is gone, that which shall be is not
this expression of helplessness can be mitigated by saying: that which was is
already gone, that which shall be, is not yet.

But let us stay away from philosophy. The phenomenon time can be
spatially caught by clock, calendar, and time axis for practical and didactical
aims, and the only thing to be careful of is not to disturb the spatial catching
of time by the spatiality of space.

Space, with its three dimensions, is much more complicated, a complicated

order of coexistence.
Bastiaan (6; 4) after a talk at a three-forked road, says, about three- and four-forked roads
in general: There is much more than right and left, front and back. How much?
Certainly twenty. (I do not remember whether he also mentioned above and below.)

Some of the coexistences can be stated at a glance, but what I mean by the
topographical context includes

the coexistences that mentally must be constituted,

the mental system of coexistences,
coexistence as a mental possibility,
the mental system of coexistence as a mental object,
and perhaps some more variants of these subtle distinctions.
(To avoid misunderstandings I stress that if I say space I mean space.
The topographical context is not restricted to two dimensions nor to two flat
281

282

CHAPTER 10

dimensions decorated by light bulges in the height and depth. The context I

mean may include the layer structure of a building, the aerials on the roof, the
airplanes, sun, moon, and stars if need be, even the world as seen from a frogs
perspective, or from a birds nest, or from the moon, it can extend to the
interior of Earth, Sun and planets.)
In a few words I shall continue with Leibniz order of coexistence. Leibniz
shows how place is defined: Two things A and B are at the same place provided
they possess the same references of coexistence with respect to things C, E, F,
G, and so on, which have not had any reason to change their references of
coexistence. Then he continues: One must here restrict oneself to the references
of coexistence rather than admitting all references since otherwise A and B
would be the same according to the principle of indistinguishability.
Leibniz knows that his definition explains equality of place rather than
place itself, like Euclids (properly said, Eudoxos) definition of ratio, which
aims at equality of ratio rather than ratio itself a way of concept formation,
which mathematical methodologists have become used to in the meantime.
With the definition of place the coexistence is detached from the succession.
The order of coexistence must also be established, at least in principle, for
objects that are not simultaneous or not simultaneously perceived or perceivable

or even imaginable, and this happens, as is usual in the physical world, and is
systematised in physics, in an indirect way, by intervening objects and sequences
of objects, which can be placed pairwise coexistently in the order of coexistence.
In order to illustrate it by a quite concrete example, I mention the procedure
of the surveyor, who establishes the mutual position of two far distant points
on the terrestrial surface by linking them through a chain of small triangles,
with vertices in each others neighborhood. Well, at present such things are
done by photogrammetry from airplanes, but finally the pictures from such
a film must be knitted together to make up a total picture an order of coexistence constructed indirectly, built from direct orders of coexistence. Direct
ones? No, as soon as details are considered, a pair of points of any particular
aerial photograph is not coexistent in the sense of visual perception of perceivability, but they are so only in the mental relation.
This, then, is the topographical context, briefly summarised:
the catching of space (or of the space) as a mental coexistence of places,
that is, of
places of objects,
places of objects and perceivers,
places of perceivers
in their mutual physical and mental relations.

Topography literally means place description. Is this translation of topography not all we need? Or is this section a somewhat highbrow philosophical,
or philosophically looking, introduction to a chapter with a potentially rich

THE TOPOGRAPHICAL CONTEXT

283

content that will be imperfectly realised, by means of examples rather than

by a long list which anyway would be unsystematic because of the lack of any
classification, and boring to the reader? As mentioned before, this chapter has
come as a separate birth out of the womb of geometry. Its content is geometry,
though of a kind that does not lead to concept formation in the sense of
geometry the geometry of rigid and flexible bodies, of motions and other
transformations, of topology and combinatorics.

10.2. Means of Expression

Little if anything seems to be known about the development of the topographical

context in general. Obviously its development will heavily depend on the world
of the developing individual, whether it is restricted to one room, to an apartment, to a house with many stories, with or without a garden, whether this
world is a plane or a mountain, whether the way from home to kindergarten
is travelled on foot, or an mothers bike or by car, whether the environment
of the home is explored at a young age, whether the outside world is imitated
by picture books and television and by construction kits. However it develops,

the mental context which I have called topographic will not show in the end
much dispersion among members of the same cultural community. The verbal
and other means of expression, however, by which this context is understood
and described, as well as the variety of related concepts will sooner or later

diverge greatly, depending on the general development of the individual.

A few simple examples may show the tension between the possession of a

topographical context and the means to express it:

Monica (4; 4) wants to indicate some place by means of the way leading to it. The index

finger lifted but motionless, she says: ... and then you go so, and then you go so, and then
you go so .... Obviously she sees the way clearly with her minds eye but she lacks verbal
and even mimic means to describe it.
I want to go with Bastiaan (6; 4) to a certain place. He agrees that the shortest way is

over the locks. And then you go so, that is also a shortcut. I understand that he means
after the locks, and if I had asked it he would have been able to elaborate it mimically.

Even adults can experience difficulties if they have to show another, say a
motorist, a way, but then the reason can be gaps in the mental topography
how many corners, how many traffic lights, a one-way street, striking characteristics of spots where the direction is to be changed; for instance, traffic
signs leading to a goal beyond the desired one.

Modern manuals, from kindergarten onwards, show an increasing appreciation

of the topographic context. The stress is on teaching passive and active mastery
of verbal means of expression. This is welcome but not enough. The context
itself deserves more attention. Moreover more means of expression deserve to
be paid attention to in dealing with this context. On a lower level than, and as
a preparation for, the verbal means of expression, there are
ostensive means the way, the size of an object is shown

284

CHAPTER 10

accompanied or not by
a verbal component there, that large
and
a mimic component the gesture this way around the corner, that far
away,
possibly

accompanied by these words.

More sophisticated visual means of expression in the topographic context
show a great variety:

diagrammatic sketches,
views according to various principles,
ground-plans,
blue-prints,
maps,
atlases,
globes,
models,
supported by symbolic means such as
coordinates,
and verbal means, such as

legends.
Finally there are

exclusively or preponderantly verbal means of expression

in colloquial language,
in a language created for topographical aims,
in mathematical language.
It is impracticable to illustrate this by examples detached from the applications,
which will be more closely viewed.

10.3. Polarities
The naive space shows the polarity
above below,

which I can transfer from one place to another, thanks to my balancing-organ

and the perpendicular fall of heavy bodies. Geography and cosmography have

THE TOPOGRAPHICAL CONTEXT

285

taught us otherwise, but the gulf between naive and cosmic space can be bridged
only by theoretically educated intuitions, supported by topographic means.
Things display polarities, which can be determined by their normal situation
in space as well as be independent of space. Man can stand upright in space,
from head to toe, but he can also lie down or stand upside down. Houses, trees,
mountains stick fast from cellar to attic, from root to crown, from foot to top;
cars move such that normally their abovebelow coincides with that of space.
Tables, chairs, cups, and bottles possess an abovebelow, which keeps its value
independent of their situation in space, even though among these situations
there is a preference for those that match the space polarity.

The polarity abovebelow determines a gradient and an order. A thing

is above or below another directly above or below, or higher or lower. What is
the common basis from which this higher and lower is measured? What
part is played in this by the oblique lines running from high to low, whether
they are concretely realised, realisable, only mentally present, or abstract mental
or theoretical conceptual constructions?
How are the outside visible vertical lines of a building related to the layer
structure of the stores, to the staircases within the building, to the leaping
numbers in the elevator?
Notwithstanding the necessary stress on space, I must not neglect a usual
restriction of the polarity abovebelow to the plane I mean not the plane
as such but the drawing and writing plane, the plane on which communications

are written and objects pictured. How has this plane acquired this abovebelow
polarity? Obviously it has been transferred from walls and other standing planes
as communication and picture planes with their natural abovebelow to
more or less horizontal planes, and with some arbitrariness the side near the
viewer has got to play the part of the below. As a matter of fact, in perspective
representations this can produce a conflict between the polarities frontback
(foreground background) and abovebelow. Young children experience
difficulties with this overlap, even the convention of identifying near the
viewer and below is only slowly accepted in their own production.
Another polarity is that of
head tail

in living creatures, transferred to other objects such as arrows, pencils, roads.

A sausage or a rod has two ends which can arbitrarily be assigned the part of
head and tail. Roads can be neutral, or one direction can be stressed: to town,
out of town, up hill, down hill, one-way traffic.
Abovebelow and headtail are
polarities of direction,
mental objects, which can be sharpened to the mathematical vector concept.
Another kind is the

polarities ofside and opposite side,

286

CHAPTER 10

of a street, a river, a wall, a coin, a leaf. With respect to ones body this polarity
is sharpened into
front back,

and in a more neutral form to

in front of behind.
With a curved or closed surface the polarity of side becomes

inside outside,
of a room, a house, a cup, a bend in a cycling-track, a river, a garment, a bag,
a fence. This static polarity determines a movement

from one side to the other,

from inside to outside
crossing the street, uptown. These look again like polarities of direction, but
it is a more vague direction: one can cross the street in various ways, there are
many directions across the street, to the reverse of the medal, in and out of the
forest. The distinction, however, is not sharp: there are also many ways from the
head to the tail of an animal. It depends on the stress whether some polarity

is understood as one of direction or of sides.

Things that show a polarity of direction can be symmetric around this direction (vases, bottles, pots, arrows) or indifferent (mountains, bushes, or they
can possess a second polarity. Two polarities of direction mathematically two
vectors determine a plane, and consequently a polarity of side and opposite

side. It is a well-known feature of our own body. Besides the polarity of headto-toe it possesses that of front and back, which is transferred as such to our
environment. They become conscious and are verbalised at an early age. Because
of the near-symmetry of our own body the polarity rightleft takes longer, and

quite a few people experience it as a difficulty, though mathematically the

polarity rightleft is determined by the two former ones, abovebelow and
frontback. Mathematically, indeed, though what is ones right and ones
left side, has to be learned empirically. As regards our own body two phases
can be distinguished,
the operational and
the verbal

distinction of right and left, similar to the distinction of colours: distinguishing

colours and knowing their names. Even if the child is used to writing with the
right hand and shaking the right hand, the question which is your right hand?
or which hand is this one? can cause hesitations, and a correct answer to this
question does not imply that the child can indicate his right foot, eye, ear.
New difficulties can arise as soon as the relation to ones own body is severed,
or the distinction between right and left must be coordinated with motor

acts or experiences of ones own body. I used to remember the fact that as late

THE TOPOGRAPHICAL CONTEXT

287

as the age of 910 I had trouble to distinguish right and left (although I am
righthanded). If now I analyse this recollection, I think the case was different.

My recollection is tied to gymnastics lessons: I had to imagine myself standing

in the drill-hall in a certain direction in order to decide what is right and left.
Yet this relation to gymnastics indicates something else: understanding the
commands right turn, left turn. Neither operationally nor conceptually, indeed,
does the distinction rightleft imply that of right turn left turn. It is a mere
definition that right turn means that the front vector, that is, the sight or the
stretched arm, moves on the shortest way to a position at the right a conceptually acceptable definition that need not be operational. Certain tools
(keys, taps, corkscrews) have to be turned right to carry out certain operations,
and likewise the clock hands turn right, but that is only metaphorically related
to the right turn of ones body a point we shall deal with later on.
The distinction righthandlefthand can be transferred from ones own body
to other bodies and objects by means of displacements or other continuously
acting transformations. As soon as the polarities abovebelow and frontback
are given for any object, the polarity rightleft is also established. Somebody
who faces me eye-to-eye has a frontback vector that is opposite to mine,
while his abovebelow coincides with mine, and this means that his right
left is the opposite of mine, his right hand touches my left hand. If I am lying
on my left side and somebody else at my side but with his head at my feet and
looking the same direction as I do, then his abovebelow is opposite to mine,
his front-back the same as mine, thus his rightleft opposite to mine, which
means that he is lying on his right side.
With non-living objects we behave less consequentially. Of a cupboard in
front of me, the side that faces me is considered as its front. But I would not
call that side of the cupboard right or left which I would do if the cupboard
were a man. On the contrary the side of the cupboard that is on my right
is called right, and that on my left is called left. With a portrait group in a
newspaper one may doubt what is meant by to his/her right if it is not somehow explained. Books on a shelf are placed with their backs in front of us,
with the consequence that the pages succeed each other from the right to
the left and in the usual arrangement of volumes from the left to the right
the first page of Volume 1 is close to the last page of Volume 2.
Inconsistencies like this one are also found in the use of in front of behind.
Somebody sitting in a room, seen from the street is he sitting in front of or
behind the window? Which one of two objects, seen in a straight line, is behind
the other? (The English language knows the subtlety of in front of, lacking
in other languages.) If in some respect the objects form a procession or a queue,
the frontback is uniquely determined. A house has a front and a rear. What
does in front of the house mean for somebody standing at its rear? Does
in front of the station mean the main entrance? Does somebody who lives
across the river say at home that he lives across the river? Does a citizen of
Transvaal live on the other side of the Vaal?

288

CHAPTER 10

The background of these difficulties is a polarity, which I confess I had

almost forgotten; clearly because I was too strongly preoccupied with the
three-dimensional structure of space and bodies in space. I mean the polarity

here there

if related to objects: this that. We know nothing about the sequence in

which the various polarities which I enumerated, are experienced by our children,

mentally seized upon, and conceptually structured, in which order they are
refined, entangled, isolated. Piaget claims that the topological polarity of inside
outside precedes the others, and in certain respects he may be right. If I should
assign priority to one of them, albeit on shaky grounds, it would be that of the
herethere. One of my arguments would be that in my first approach I had
forgotten about making it explicit, probably because it is more deeply rooted
than the others.
The polarity herethere can be weakened to herenot here or refined
in the way that the there means a certain direction or a more or less precise
region, or quantitatively be refined to nearby and far away. The here
in herethere can be meant mentally rather than actually, which results in
confronting two theres with each other, one of which is more here than the
other. A variant of it is: thisthat direction, place, region.
The herethere is the means to relativise the other polarities. Misinterpretations of some polarities can sometimes be explained by implicit assumptions
involving herethere decisions. In the antipodes the abovebelow differs
from what it is here. The in front of and behind the table depends on which
here is stressed. For the child constructing or viewing a departing procession

the behind might be ambiguous. The rightleft side of the cupboard is determined by the here of the user. The man who says he lives across the river,
locates the here in the centre of the city.
I am going to reconsider the polarity rightleft, in order to answer a wellknown paradoxical looking question: why does the mirror interchange right and
left, though not above and below? I recall that the polarity rightleft derives
from our body. Other living and dead objects can be assigned a rightleft

by comparison, as it were by mapping my own body on them; that is by assigning to my own abovebelow and frontback polarities corresponding polarities
in those objects. In some cases such an assignment can be defined in a natural
way: if the other object is a man I have my head-to-toe and my frontback
to correspond to his, and then his rightleft is uniquely established. In other
cases the assignment may be more arbitrary, but once I have settled, say, about
a box, what is the abovebelow and the in front ofbehind, then its rightleft,
as seen from the box, is unambiguously determined.
What then does it mean that the mirror interchanges right and left? Well,
I suppose that in the image that the mirror reflects of my body, right and left
are interchanged, that is, not determined according the rule I am accustomed to.
However, at an object like my mirror image I can distinguish right and left

THE T O P O G R A P H I C A L CONTEXT

289

only on the basis of some presupposed abovebelow and frontback. On the

other hand, it is quite natural to understand the abovebelow of the mirror

image as the direction mirror top to mirror bottom, and the frontback of the
mirror image as the direction mirror man front to mirror man back, and then
for the mirror man what he has to call his righthand and lefthand is established,
provided I apply my own rules. Then since the abovebelow of the mirror
man equals mine and his frontback is the opposite of mine, it follows that the
mirror man, if he lived in my world would call his right hand the one that
mirrors my left hand. In this sense the mirror interchanges right and left: The
mirror image of my left hand is that hand of the mirror man he would call
right if it is supposed that he lives in my world. It is not an isolated rightleft
that is interchanged, but the rightleft of a body, coupled to an abovebelow
and a frontback which remain unchanged as it were by definition.

With an object where the abovebelow and in front ofbehind are not as
naturally determined one will reach different conclusions. If a column is placed
between a mirror and myself, I would probably call the sides of the column
and the mirror column in front of me their front sides then the mirror would
interchange front and backside and leave rightleft unchanged. One could also
suspend a box above a horizontal mirror between ones eye and the mirror and
now call the visible sides of the box and the mirror box upside then the mirror
would interchange above and below and preserve frontback and rightleft.
This then is the answer to a paradoxical question. We will resume it as soon
as we have dealt with the screw sense in space.
10.4. Connections
Connections are as it were syntheses of polarities but they are more than this.
They can remove polarities but they can also create new ones, by their existence

or lack of existence. The country is structured by land and water ways, railways
and highways, the city by streets, lanes, boulevards, alleys, which are meant as
connections. Less rigid are the unbeaten paths across meadows and through
woods, the furrows and trenches, the airways sometimes marked by white
condensation trails. Crossing a parking lot, between two cars, then diagonally
and again between two cars using gaps and spacious fields is another mode.
Low walls and curbs, meant as borders, are interpreted by little children as
connections to walk on. Ladders, stairs, elevators are connections between
below and above; bridges over rivers and viaducts over roads, zebra crossings
connect this side with that, but holes can do likewise, the hole in the fence
where one must bend down, the longer one is the deeper, to creep through,
the service-hatch, the ticket-holes, the window, the key-hole for spying, tunnels
and subways.
Connections may be composed of pieces which are mentally, or by cartographic means, stuck together, a road visible at a distance, marked by driving
cars, which appear and disappear behind bushes, houses, hills; the Champs

290

CHAPTER 10

Elyses at lEtoile, at Rond Point, at Place de la Concorde, the highway E8

at the Hague and at Hengelo, the Rhine at Basel and Cologne which, when
travelling by train, one looses sight of and meets again and again. Underground
sewer canals are witnessed by arrays of manhole covers, mole-tracks by hills
in the meadow. Game tracks and traces of a fugitive can stop and continue
again.
Connections can form networks like that of the railways, of a river with
its tributaries, the underground systems of sewerage pipes, waterpipes, gas,
electricity, telephone, which can cross each other in a chaotic arrangement,
or above the ground the roads at a large roundabout, but also the indoor
system of passages inside and outside doors, corridors, staircases, possibly
even the windows (to climb on the roof), chimneys and ventilation channels
can be a display of connections.
The detour structure, as I called it, is the childs first experience in this
field: two paths between two spots, the normal one mother goes, and the other
one, diverging, a childs adventure, safe-guarded by the knowledge of coexistence,
visible coexistence if there are no obstructions between the two paths, partly
visible and invisible coexistence if the paths are separated by bushes or walls. A
similar experience the closed path, pieces of which are simultaneously visible
and invisible but mentally coexistent; in a suite of rooms, one room in and the
other out; around the block where one is living. Decisions which one of two
ways is longer can be difficult for children of the age of 67 even in spite of a
lot of experiences; this holds for adults, too, in involved connection structures
if no cartographic support is available.
10.5. Standpoints

I do not mean opinions points for a disputant to stand on. I really mean
spots where one can stand, from the concrete here from which I look at the
world, up to the mental one at which Archimedes said he could move the
Earth by a lever. Well, here, on this spot, I can turn objects to view them from
all sides, I can touch one side while I view the other. Gradually I learned to
coordinate sense-organ and tool, eye and hand, and if need be, to isolate them
from each other, to check with the eye what I did by touch, and to have my
perceptions elicit actions, the wanted and unwanted consequences of which
can be approved or disproved by my eye. I have learned to point my ears to
the sound and my nose to the smell, to constitute objects as far as they can be
constituted from the here.
My subject is not the constitution of spatial objects in general, and the geometrical objects are only a special case which will be dealt with later on. The constitution of most objects requires a variety of standpoints. For the present I am not
interested in the objects but in the standpoints in their mutual relatedness the
space as order system of standpoints experienced or posited as coexistent.
Standpoints are connected with each other, primarily by the path which

THE T O P O G R A P H I C A L CONTEXT

291

leads from the one to the other, physically or mentally. The continuity of the
path is certainly an important factor by which images not only visual ones
perceived at different standpoints are mentally related as derived from the
same object: the constitution of an object in space and on its place in space.
Yet this continuity is not a strict requirement: as early as the age of four if
not earlier a child is able to identify a familiar spot at, say, one side of a canal
from the other side and to experience this identification as a meaningful activity.
Experience and knowledge regarding the interrelatedness of standpoints is
required to anticipate from a standpoint here a phenomenon such as it would
appear from another standpoint there. In order to better see, hear, feel, smell
something one changes ones standpoint.
Bastiaan (5; 4) and Monica with me on the way to the playing ground. I wonder whether
it will be open I say at a distance of 50100 m. We were somewhat obliquely crossing a
meadow in the direction of the door of the playing ground. The fence of the playing ground
is hidden by hedges and bushes, which also hide the entrance unless one approaches it
perpendicularly. After my words Bastiaan ran 1020 m parallel to the fence until he could
look straight into the entrance and state that the door was open (Figure 116).

One runs to the window, to the front or rear, upstairs, in order to locate the
origin of a sound or a light effect, down the street, around the corner, on the
roof, in order to remove obstructions to perception, which may be effective
at one place though not at an other. The child asks to be lifted in order to
perceive something beyond a fence, or wall or the ridge of a roof. Obstructions
and the way the foreground is delineated on the background, are also used to
locate some object for anothers eye: Stand here, then you can see the star
exactly above the aerial.
Misinterpretations of these standpoint procedures are worth mentioning
as well:
Monica (almost 4; 8) asserts, while it is still quite clear twilight, she sees Venus near the
crescent; she is probably right, whereas my visual acuity fails. She asks me to sit on my
heels in order to see the planet from her standpoint.

Something is approached to be better perceived, or viewed from a larger

distance for a better global sight. A six-year old knows that at a larger distance
objects look smaller, and in a way he knows how to draw conclusions from this
fact, in order to explain phenomena. However, that a line-segment viewed

292

CHAPTER 10

obliquely looks shorter than viewed straight this correction of the mental
by the visual reality, which is manifest in perspective drawing, seems not to be

made by six-year olds. They know too well that the rim of a cup is a circle to
be able to see it from various standpoints as an ellipse. The primitive realism
where object and perception are identified, is still waiting for a process of
sophistication: disentangling object and perception. Mirror image and shadow
familiar phenomena at an early age do not yet disturb the harmony of
object and perception, since they belong to the objective world, but with the
knowledge that objects look smaller at a larger distance, with the phenomenon
of the apparent size, a step the first? is being made on the road of a still
primitive realistic distinction of essence and appearance. At the same time
the child learns to distinguish between true stories and fiction. It would be
worthwhile investigating the development of the distinction between essence and
appearance, truth and fiction I do not remember having ever read anything
about it that drew my attention.
Whereas it can cost time and trouble to have a child grasp the fact that from
another standpoint things may look different, it can cost as much time and

trouble to have him accept and understand that elsewhere in the holiday

resort the sky is the same as at home the sky, that is, not the clouds but
the starred firmament. The 67-year-old knows and grasps that with each
movement the world around him changes, that is, the near world. The cathedral
tower at a few kilometers distance, however, does not change if I move a bit.
And qualitatively proportional: Sun, moon, and stars are so far away that a
long journey like that to the holiday resort counts as much or as little as winking
ones eyes does with respect to the cathedral tower again an application of the
polarity nearbyfar away.
The most surprising event I came across when observing the mental change of
standpoint with young children, again concerns Bastiaan (7; 4).
Back from a long walk we cross a slightly ascending bridge Monica and myself on the side
walk, and Bastiaan on a small wall along the side walk, ascending discontinuously by steps
with in between horizontal pieces of about 5 meter (Figure 117). At a certain moment
Bastiaan says: Now I am higher, but then you win be higher, it amounts to the same. In
fact we were always lower. He meant the difference between the continuous and the discontinuous ascent. His verbal means of expression were still imperfect, but it is clear what he
meant: we are ascending gradually, whereas he lags in order to catch up at the next step.

10.6. Reciprocity of Change With Respect to Place of Object and Standpoint

One can hang a picture upside down but the same effect is attained by oneself s
standing upside down. There are situations where one inclines ones head to the

THE TOPOGRAPHICAL CONTEXT

293

right or to the left in order to get a straight impression of an object that is

somehow inclined. One can turn a sculpture to consider its rear but one can
as well change ones standpoint and walk around the object. More generally:
In order to attain an intended relation between an object and its perceiver
or manipulator, there is a choice between
change of place of object, and opposite change of standpoint,
and such a relation can be maintained by
change of place of one of them and a corresponding change of the
other.
This is something that in concrete cases can be conscious to little children. A
more difficult case is the apparent motion: sitting in the train one gets the
impression that things in the foreground move across a fixed background;
ditches perpendicular to the railtrack seem to turn.
A general principle:
In order to modify the relation between an object in the foreground and
its background one has the choice between change of place of one object
and the opposite change of standpoint.
This is also conscious to young children. A different thing is:
In order to explain a change of relation between an object at the foreground and its background, one has the choice or feels the dilemma
of interpreting it as a change of place of the object or as a change of
standpoint.
With six-year olds I could not observe any symptom of awareness of this choice
or dilemma. They are able to repeat that rather than the Sun that rises and
sets, it is the Earth that turns, but this is done with no insight, although they
are able to understand the monthly course of the Moon along the planets as its
rotation around the Earth.
10.7. Obstructions
By obstructions I mean obstacles for ear and eye, for walking and acting. In
Section 9.26, Border and boundary, I already gave examples. Obstructions are
not necessarily absolute limits, they are rather obstacles that thwart the penetration into another territory, the transition from here to there.
A wall, fence, street can be an obstruction to sensory perception and motor
activity, but there is a chance left that one can go around it, across it, through
it. A mountain is an obstruction in the landscape, passable by crooked paths,
pierced by tunnels. A river is an obstruction to direct passage from one bank
to the other; the traffic across the river is restricted to bridges and ferries, but
by swimming and flying, people and animals can reach the other bank. Swamps

294

CHAPTER 10

and deserts are moderate obstructions to any traffic, a highway is such for
pedestrians who wish to cross. No entry, a series of poles on short distances,
a chain obstructs motor vehicles and yields to pedestrians. A red light is an
obstruction to traffic that respects it. Some roads obstruct only heavy traffic.
A barrier blocks traffic, but the blockade can be lifted.
A water or sewer pipe can be obstructed. A crooked pipe obstructs putting in
straight bars, but a flexible long object can conquer the obstruction; the more
crooked the pipe is, the more flexible transversally and stiffer longitudinally
the object must be to go through.
A window and a door is passable for small objects rather than for big ones.
Things that cannot pass transversely may succeed upright. An object may not
exceed a certain length to be carried around a corner in a corridor; bends and
corners can form obstructions.
Chains rest on the principle that each link obstructs the removal of the next.
Threads twisted in a rope obstruct each others stretch. By means of ropes and
chains persons and objects are obstructed to move. Springs and spirals grant
a greater margin, a lock between the spoke of a wheel obstructs turning. A plug
in the floor obstructs the swing of a door, a bolt or hook its opening. A door,
once opened, can obstruct opening another, or the passage through a corridor.
Aiming is a procedure by which one attempts to just or just not obstruct the
view on an object. Obstruction and non-obstruction may be mutual or not.
If A can see B, so B can A. Or can he? The peeper at the keyhole or behind the
curtain sees but it is not seen himself. Viewing eye to eye, the line from the
other is not obstructed, but the remainder can be partly so. One-sided transparent
windows and mirrors, as well as periscopes interrupt or obscure the reciprocity
of viewpoint and object, eavesdropping instruments that of the path of sound.
The way from above to below can more easily be passable than the converse.
10.8. Combinatorics
Line, plane, and space are gifted with combinatoric structure: that of the immediate coexistence, the vicinity. Vicinity should be understood with a pinch
of salt: beads at a string that touch each other, tiles in the paving or countries
with a common frontier, bricks in a wall separated only by a layer of mortar,
faces of a polyhedron with a common edge and their networks, bones of a
skeleton in a common joint, vertebra joined by a spine, but also neighbors in
a street, possibly across the street, towns along a road, stations and airports
communicating by direct traffic.
In Sections 7.23 I paid attention to a mathematical tool to describe such
structures: graphs. Mathematically viewed there is not much to be added, since
no profound properties of graphs matter in this context. It should, however,
be mentioned that the geometrical image of the graph also serves to visualise
non-spatial relations of vicinity. Genealogy is visualised by pedigrees, processes
by flow diagrams, electrical circuits by networks, which in no way reproduce

THE TOPOGRAPHICAL CONTEXT

295

the spatial situations of parts but electrical vicinity, a social system with its
components as nodes and its influences as paths.
10.9. Conclusion
In which way does this chapter fit into a didactical phenomenology? In the
chapter Topology the didactical component was almost lacking. Its aim was to
signal the tension between mental objects and concepts, which is not probably
cannot be bridged by didactical means at school level. Those mental objects
have again emerged together with others in the present chapter, purely
phenomenologically, unrelated to mathematical concept formation, but also
without tying more than a few didactical connections. This does not mean
that they do not exist. On the contrary, they do exist, more implicitly, more
globally than if numbers in the arithmetical and figures in the geometrical sense
are being discussed. Arithmetic and geometry can boast explicit didactics,
elaborated in detail. I can deal in detail with didactical aspects of counting,
of arithmetical operations, of perspective. I have illustrated what I called the
topographical context by examples but I would not be able to divide it according
to details of didactic action. To say it more concretely: what is above and below,
front and back, how to come from here to there, how to look around the corner,
what I can see through a key-hole, what is a bridge, a viaduct and so on and
so on is not being learned, let alone taught, point by point but in a rich
context, which presents itself, which if presented is modified, which is recreated
from the bottom upwards, and in which consciously and unconsciously
the wanted elements are processed, in an integrated way. Such a context can
serve to form the mental objects, relations, and operations, to become aware of
them, to verbalise them, to learn using and understanding them as tools of
understanding and acting and finally to remodelling them conceptually.
This chapter has been didactical phenomenology in the sense that it aims at
making conscious the topographical context as a global didactic medium.

CHAPTER 11

FIGURES AND CONFIGURATIONS

11.1. Abstracting
How many properties must be abstracted to make a thing a figure? Colour,
material, weight, taste, smell, irregularities, roughness, the place of the thing,
the place of the perceiver, which determine its appearance, use and emotional
relations. What is finally left? Shape and size. Or should the size also be wiped
out by abstraction in order to get the bare figure? Yes and no. The child copies
the same figure as drawn on the blackboard on a smaller scale. But stepping
through Madurodam* and a true city are different things.
Classification tests bear witness to the fact that six-year olds are able to
perform this kind of abstraction. They classify objects primarily according to
shape and size characteristics (length, width, thickness, height); they understand
explicit instructions to classify according to shape and size (but also according
to other criteria) and they understand the words shape and size. Only if
they are asked for new classifications, can it happen that the place of the objects
is introduced as a new aspect of classification, but even this is easily averted.
The objects from the topographical context are tied to their places: the
ground-plan of a room, the faade of a house, the net of streets, the solar system,
are mapped in the plane or in space according to their more or less pronounced
individuality. A line, a circle, a cube is not tied to any place; models of these
figures, even if specified by one specimen, are also models of mental objects.
Of course, there are intermediate cases: the pattern of a crossroad or a roundabout with right-of-way indications and traffic lights, is not as closely attached
to one place as are many other topographical representations.
I already explained how geometrical contexts arise, in particular that of the
rigid congruently and similarly reproducible bodies, how accidental features
and, if need be, thickness and width are wiped out by abstraction aspects
which with the greatest ease can again be included, as can colour, material,
weight, taste, smell. Place is more difficult; its influence has been eliminated so
early and in a so strongly implicit way, that it costs a lot of trouble to make
it conscious again, for instance in the use of perspective.

11.2. Production, Reproduction, Occupation, Manipulation, Transport

Several times I stressed that at a young age and in many ways children are
confronted with things that suggest geometrical mental objects. As products
* The midget city near The Hague.

296

FIGURES A N D C O N F I G U R A T I O N S

297

they are the result of production and invite reproduction, occupation, manipulation, transport concrete actions which can mentally be repeated with
corresponding mental objects, and this mental element is the one that counts
in the geometrical context. It is a rich variety of actions which I shall display
in the sequel probably in a quite unsatisfactory way. We are satisfied too soon
if we have introduced a mental object concretely in one way. On the other hand
we ask the child to be receptive for a multiplicity of approaches whereas in the
didactical situation we prefer to narrow our own view.
The didactical feature of the present phenomenology is the multiplicity of
concrete approaches. Among the numerous terms that describe the concrete actions none is redundant; on the contrary I am sure there are not enough of them.
11.3. Planes
I start with planes, or rather pieces of what one imagines to be infinitely extended
planes. There are reasons why I do not bestow priority on lines planes come
earlier. First of all, in the topographical context, horizontal and vertical planes,
floors, ceilings, walls, bottoms, covers. Among the oblique planes the most
striking are roofs, covers of chests and slides. Objects with faces can be bounded
by oblique planes, depending on their position. Water in a vessel does not
behave as a rigid body; its surface remains horizontal even if the vessels are
inclined. (A glass with powder, beads, or peas behaves as though it were halfway
between liquid and solid matter. Contrary to what Piaget claims it has nothing
to do with logic but all to do with physics whether such a surface is horizontal
or inclined and how much it is inclined.)
A remarkable feature of planar pieces is their unique continuation as it is
of straight lines. In fact, it is the criterion of flatness of a planar piece whether it
continues its own sub-pieces in an virtually unambiguous way, and the same
property plays a part in the mental constitution of planes. According to my
experience 67-year olds are able mentally to extend planar pieces. Some
evidence will be displayed later on.
Thanks to unambiguous continuation, planes can be laid down and slide on
other planes.
If a surface is everywhere horizontal, it is certainly flat level as the
carpenter and bricklayer call it after the instrument by which they check it.
Whether something is vertical is checked with the plumb: a vertical wall is as
it were a combination of plumb-lines along a level-line.
Knives and saws make planar cuts because they are flat themselves and
continue unambiguously flat indentations. The shaving tool, called a plane,
transfers its planar bottom to the shaved material. Scrubbing is removing roughness in order to glide, a thing must be flat. The imprint of a plane is again
a plane this is in general the simplest method of checking and reproducing.
Smoothing upon a plane creates again a plane; a table cloth on a table is a plane
as is the table.

298

CHAPTER 11

Among the surfaces the planes are characterised by the property to comprise
with a pair of points the connecting linesegment planes are closed with
respect to rectilinear connecting. Through two intersecting straight lines a plane
can be laid by pointwise connecting the lines. Three points, not lying on one
line, determine a plane by means of their connecting lines. The straight lines
drawn from a point to a line-segment or line, determine a planar piece or plane.
The rectilinear connectivity of the plane, is the reason why rolling a soft
matter (dough) on a plane produces a plane: A roller is a cylinder, described
by straight lines, which are imprinted in the soft matter; rolling repeatedly in
different directions assures rectilinear connectivity. Cylinders and cones can
be rolled upon planes such that at every instant a generating line is lying in
the plane. (Surfaces with this property are called torsi; apart from cylinders
and cones these are the surfaces composed by the tangents of a curve in space.)
Tests of rectilinearity which will be discussed later can also serve to test
the flatness of a plane: applying the ruler, stretching a rope, aiming. Aiming
along a plane gives a straight line.
Planes can be suggested by apertures: a gate, an arch, a football goal, a
sequence of lampposts. Planes can be piecewise given, by housefronts, interrupted
by sidestreets, or by networks of lines such as wire-netting. A linear object
together with its shadow determines a plane.
A plane can be suggested by mirror symmetry, that is, as the plane producing
the symmetry. Such a plane can be explicitly marked within the symmetric
object or the object can invite the marking of it mentally, as happens if an
object is divided into congruent parts.
Planes can be flexible; they may be rolled into cylinders, cones, and more
generally, torsi; by this process soft planes acquire a certain stiffness, which
they did not have before.
Planar pieces divide the space locally as do the total infinitely extended
planes globally. If in a topographical context they are horizontal, the parts
may be called above and below even though the planar pieces are somewhat
inclined. For vertical or almost vertical planar pieces the predicates rightleft
or frontback may fit, according to whether the observer imagines himself
placed within the plane or in one of the spatial parts.
Planar pieces can be lying as though one produces the other, that is, as parts
of the same extended plane. They can be parallel parallelism will be dealt
with later on or they can meet. Depending on how far they extend, this can
happen in various ways: in a single point, in a number of points, in a number of
line-segments. In the truly mathematical context one means the infinitely
extended planes, which if not parallel meet in an entire straight line. It is a
well-known fact that this meeting along whole lines is obscured by drawings
where planes are symbolised by parallelograms. I experienced, however, with
67-year olds that if a pair of planar pieces are concretely given in space, they
can indicate where the extended pieces would meet each other and devise
methods use of the ruler to carry out the construction more precisely.

FIGURES AND CONFIGURATIONS

299

(This led me to assume that children can perform the construction of planes
from planar pieces mentally.)
11.4. Polyhedra
Polyhedra with which the child gets acquainted at an early age, are blocks,
prisms, pyramids. The context is often topographic: blocks of houses, roofs,
pointed roofs. Toy building bricks provide an opportunity to detach such
figures from their topographical context and to place them in the geometrical
one: to relativise the phenomenon of base, to have blocks balancing on edges
and corners, to view roofs as lying prisms, to have pointed roofs resting on
lateral faces. It is well-known that older pupils, after many years of planar
geometry, can have difficulties with these changes of perspective as soon as they
are supposed to learn solid geometry.
It does not take much trouble, even at a young age, to recognise vertices,
edges, and faces and to name them as such; there is little reason to stick to other
terms that are believed to be more adapted to childrens language. Likewise
the relations between vertices, edges, faces (lying on, passing through) are
accessible at an early age. The same holds for networks, building polyhedra
from, and splitting into, networks, understanding a network as a combinatoric
pattern.
In spite of the didactical value that the network derives from its manageability, there are reasons to warn against the flight into the network. Polyhedra
are primarily surfaces of solids and must be mentally constituted and conceptually approached as such. This is the way to start; if this has successfully
been undertaken, the analysis of the surface into a network may follow. (This
does not exclude confronting the learner incidentally with a network and the
indications to build from it a perhaps surprising polyhedron, but it should
not be recommended as the normal approach.)
Let us take the case of a not necessarily triangular prism. Twelve-year
olds, and even older ones, do not lack the vocabulary but rather the mathematical ability to describe this class of surfaces. Supplying the child with a
description bears witness to a lack of didactical understanding, but the flight
into the network is no more justifiable. One should rather exploit the fact that
they are surface structures of solids, and this is done most efficiently by constructing the solids themselves, from clay or potatoes. This then is the way
towards a conceptual analysis of the prism as a class of surfaces. It starts with
modelling from clay, or cutting out of a potato, a disc, which can be irregularly
bounded at its sides. In order to arrive at a prism, one remodels the sides: cutting
away pieces, perpendicularly through the disc, in order to get a right prism.
This construction implies a conceptual description: congruent base and top
polygons connected by rectangular walls. (The parallelism of the edges in
the usual approach the primary element is now a consequence.) Piling up
prisms of the same kind or sawing parallel to the base and top side produces

300

CHAPTER 11

new prisms. These parallel cross-sections are congruent polygons related to each
other by right or skew translations a relation that leads to a new definition
of prism: a polygon moved in space sweeps out a prism a right one if it is

moved perpendicularly to its own plane. At a later stage this conceptual analysis
leads to the definition of the solid prism as the cartesian product of a planar
polygon and a line-segment, or even an infinite line.
Similarly one can analyse and define pyramids and truncated pyramids. The
cross sections parallel to the base are homothetically situated. So a pyramid
arises by homothetic displacement of a planar polygon towards one point, the
top. This leads to a unified description of prisms and pyramids, finite and
infinite ones.
These solids are an appropriate start for simple, and thus particularly effective,
logical-geometrical analysis. Other polyhedra are by their characteristics didactically important to learn structuring: in order to count the vertices, edges,
faces, for instance of a cube, the sets are structured:
four vertices below, four above,
or four vertices at the front-side, four at the backside,
or four vertices at the right, four at the left,
four edges below, four above, four upright,
or four edges in the width, four in the length, four in the height,
a ground face, a top face, four around,
or two faces, front and back, two right and left, two above and below.

The set of the twelve face diagonals is structured

two in each face,
two sets of six forming two tetrahedra,
that of the spatial diagonals,
one by each pair of opposite vertices.
A more difficult problem is the structure paradox:
Six faces with four edges in each of them, is 6 4 edges, rather than 12.
Young children do not experience it as a problem but from the age of ten
onwards they may be susceptible to it.
By their great diversity polyhedra offer a rich context. The cube is a good
start but other polyhedra are as well worth considering. That is, not only closed
ones (with no boundary) but also polyhedra with a boundary and non-orientable
ones (the Moebius strip), and polyhedra given combinatorically by networks
which are not embeddable in space (the projective plane).
Polyhedra can be concretely represented and constructed: as
closed surfaces of solids,
surface structures (cardboard models),
edge structures (wire models).

FIGURES AND CONFIGURATIONS

301

Structures within a polyhedron (space diagonals in the cube, face diagonal

tetrahedra in the cube) are patently visible in wire models.

Polyhedra, in particular cubes, can be piled upon each other to get buildings
(Figures 118120). Structuring such buildings is a significant activity. Describing
them can be an inducment to create linguistic means. Strangely enough, does

the toy industry know better than the educators do how to profit from the
childrens building activity?
Tilting polyhedra over edges on a plane is also a fruitful subject. In fact it is
a predecessor of unrolling certain curved surfaces on the plane.
Particularities of convex, regular, and semi-regular solids will be dealt with
later on.

11.5. Direction
In my original design plane was followed by straight line, but while I
wrote that section, I became once more critical of my preconceived judgment:

caught within logical geometrical structures, which phenomenologically are

302

CHAPTER 11

prejudices. Constancy of direction is one of the roots of the mental object

straight line but compared with the other roots direction sprouts much
more than mere rectilinearity, to wit the directed straight line and parallelism.
Axiomaticians may wrinkle their brows if one dares to define the straight line
by constancy of direction. In fact, in their world direction comes much later,
after congruence, as a consequence of angles. I would ask those virtual critics
to stand on their toes and to look over the wall of geometrical axiomatics.
In differential geometry, for instance, direction and transport of direction are
presupposed and straight lines are derived afterwards let us call them straight
ahead lines under the term geodesics. I feel that this approach of space
from the infinitesimal nucleus matches more closely the natural local approach
required by phenomenology.
Straight on running is a natural activity of people and animals and thanks to
inertia even a habit in the non-living nature as long as there are no forces to
disturb this motion.
Straight on towards the goal means
globally: that at every moment the fixed goal is invariably kept in view,
locally: that at every moment the once chosen direction is continued.
This unique continuation recalls that which characterised the plane, but while
the planar continuation was all-sided, this one is one-sided. One goes on, or

FIGURES AND CONFIGURATIONS

303

something goes on or is moved forwards, yielding neither to the right nor to the
left, neither up nor down; something rises straight or descends straight, and the
continuation is determined in each point not by an artificial track but by the
geometry of space.
Well, one can retrace ones steps, backwards in ones own footsteps, or after
a turn interchanging start and finish, at each step take the opposite direction, the
opposites of the same direction are again mutually the same; the opposite
of a directed straight line is again a directed straight line for people, animals
and in the non-living nature the straight ahead line is also a straight back line.
The things moving or moved in that way are points or something that is
considered as points, though it extends to rigid bodies. The body moves straight
ahead, without yielding or wobbling: all its points move in the same direction.
Equality of direction generates not only directed straight lines, but also
parallelism, the phenomenon of being parallel. Earlier I cited parallelism among
the mental relations which from early childhood onwards are suggested by
natural objects, and even more frequently by the products of human technique.
Direction also belongs to this list, and even vector the arrow that possesses
length as well as direction and parallelogram the figure that claims that
vector displaced along vector yields the same result as vector displaced
along vector
Let us review these mental objects later on, since at present the stress is on
how much direction contributes to the mental object straight line.
The order plane direction methinks is phenomenologically the right one.
Plane the static element, the resisting wall; direction the straight ahead
element, against the wall, through the wall. But thanks to its comparability
everywhere the direction is not bound to its carrier line. The same direction is
everywhere recognisable, and its carrier line that pierces the wall or screen, can
be restored behind them.
The comparability everywhere of direction is suggested by the parallelograms
in our technical environment. Actually we are living on a sphere (and perhaps in
a curved space), which does not allow for a global parallelism, but this is a
secret, not betrayed by the small parallelograms which we manipulate again
a point to be elaborated on.
11.6. Straight Lines
Earlier on I made a choice among the great variety of objects and situations suggesting straight lines. I am going to enumerate the ways straight lines originate.
Origin of the straight line:
by copying (drawing by ruler),
as intersection of planes,
as cut line,
as fold line,

304

CHAPTER 11

as straight-ahead line,
as shortest line,
as stretched string,
as vision line,
as symmetry axis (in the plane),
as rotation axis (in space).
These modes of origin are not independent of each other. The sharp edge of
the ruler is something like the intersection of two planes. The cut line is as it
were a cutting copy of the sharp edge of the scissors. The fold line arises when
on a piece of paper is imposed the shape of two intersecting planar pieces, and
the same explanation seems to hold for the rectilinearity of the symmetry
axis.
Like the straight ahead line the shortest line (shortest between each pair of
its points) is a familiar mode of the straight line, in particular in inorganic
nature. The rectilinearity of the stretched string can be explained by physics
as well as a mode of straight ahead line, and of shortest line.
The most subtle in this list is the straight line as vision line or light ray, that
is the rectilinearity of the propagation of light it would require much more
comment as would the symmetry and the rotation axis. As in the case of the
plane as the symmetry plane in space, one can imagine a great many situations
where symmetric planar figures suggest or explicitly display straight axes.
Likewise rotation axes are a frequent phenomenon though few people will
be aware of the fact that any rotation in space takes place around a fixed axis.
As a matter of fact it takes a great deal of didactic trouble to make it conscious
again a point worth being discussed.
The first four entries of our list are as it were mechanical, whereas the other
six are more theoretically related to the physics of our common sense space,
though they certainly do not characterise it. In more general metric spaces
in spaces with a Riemann metric it still holds that straight ahead lines and
shortest lines are the same; the stretched string on the globe (and on other
curved surfaces and in curved spaces) is also the skippers straight forward course
and the path of shortest distance. The rectilinear propagation of light can be
explained by similar principles as the straight ahead line or the shortest line.
Straight line as symmetry and rotation axes exist also in non-euclidean spaces.
Up to now I did not distinguish between lines and line-segments. The unambiguous continuation of the line-segment involved in the idea of straight
ahead line is a fact experienced early. It is a mere technical shortcoming if
even in the higher grades of primary education pupils who measure distances do
not pay attention to the rectilinear continuation as soon as the measuring
instrument is repeatedly laid down. It is a striking feature that they proceed
more carefully if the measuring instrument is to be laid down parallel to the
preceding situation than if it takes place in its extension.
Between each pair of points there is a rectilinear connection, which is unique.

FIGURES AND CONFIGURATIONS

305

This uniqueness is suggested by the straight ahead towards the goal as well
as by the shortest path if compared with detours. I would guess that this
uniqueness exists mentally at an early age.
Logically equivalent with the uniqueness is the statement that two straight
lines have at most one point in common.
With respect to lines that do not meet, it is usual in solid geometry to distinguish between parallel and skew lines according to whether they are comprised
in a common plane or not. This is a typical case of didactical inversion: a
sophisticated logical approach is preferred to a phenomenological one.
Bastiaan (7; 4) spontaneously though perhaps taught at school: Equidistant lines* never
meet. I ask him whether lines that never meet, are always equidistant. After some fumbling
with two forks he exclaims: Two highways above each other, they never meet, but go far
away from each other. The same question discussed for planes.

An important aspect of the straight line is the direction, as explained in

Section 11.5. Constancy of direction was mentioned as one of the origins of the
straight line. Equality of direction of directed straight lines is an intuitively
primary phenomenon. The discrepancy of direction of intersecting as well as
skew line pairs is a striking phenomenon. The fact that there is a plane through
parallel, though not through skew, lines is comparatively secondary and not
immediately obvious. There may be reasons to choose this property to define
parallelism of lines in a logical system of geometry, but they are not at all
compelling. One can equally well imagine a system in which direction or equality
of direction is one of the fundamental concepts.
As I pointed out at another opportunity, Piaget designed his experiments as
though the traditional logical structure of geometry reflected the development
of the geometrical mental objects. Whoever is familiar with the mathematical
method and methodology, knows what part is played by the inversion: the final
result of the developmental process is chosen as the starting point for the logical
structure in order to finish deductively at the start of the development. This
genetic-logical inversion expresses itself as a didactical or rather antididactical
inversion.
The traditional definition of parallel lines is one among a lot of examples of
this inversion. Phenomenological analysis is a means to uncover them. No doubt
direction plays an important part in the constitution of the mental object
straight line the straight ahead line; straight towards a goal, walking along
a most often straight line are early activities in the development, which are
verbalised just as early and instrumentally imitated, though perhaps insufficiently.
Compared with these symptoms the vision line is so late in the development
that it is a preposterous attempt to have the constitution of the straight line
conditioned by that of the vision line. Earlier on I criticised the fake arguments
by which Piaget lets the vision line precede direction as a mental object
* This is the literal translation of the Dutch term for parallel.

306

CHAPTER 11

arguments taken from a traditional structure of geometry which is considered

as though it were compulsory. There is, however, sufficient reason to pay
attention to the vision line as well as to the symmetry axis and the rotation
axis in a special section.
11.7. Vision Lines
Light propagates in straight lines a fundamental statement in physics, which
immediately requires amendments. First of all in geometric optics: light does
not propagate straight ahead but, according to Fermats principle, along the

fastest way, which in media with varying speeds of light and in the presence of
reflecting surfaces, means deflection from the straight line by refraction and
reflection. In wave optics interference phenomena cause the light to bend
around corners. For many observations, however, and certainly in the naive
view, the rectilinear propagation of light is the zeroeth order effect compared
with which the others are negligible, or accounted for as first order perturbations.
Between the experience that an object can be made invisible by an intermediate screen and again visible by an appropriate hole in the screen (the hand
befor the eye with slits between the fingers) and the statement of the rectilinear
propagation of light and its applications, is a long journey, aided by a long
sequence of discoveries. First of all I will formulate the

vision line principle:

If the eye O sees the object B covered by the object A, then O,A,B
are rectilinearly connected.

If the eye sees the objects A, B, C, . . . covering each other, then

A, B, C, . . . are rectilinearly connected.
If a long object is seen as a point, it is rectilinear.
These principles, resting on the rectilinearity of light propagation, can serve

to establish whether objects

are lying on a straight line with the eye,
are mutually lying on a straight line,
are rectilinear,
to manage, by displacing one object or some among them, or all of them,
or the eye, or by deformation, that objects
are lying on a straight line with the eye,
are mutually lying on a straight line,
are rectilinear.

Here it makes a difference whether

the eye and the objects are

FIGURES AND CONFIGURATIONS

307

concretely realised,
or experimentally simulated,
or partly or totally imagined,
or symbolised in drawings or otherwise,
to occupy certain places.
I would not be able to tell how and under what circumstances and in which
order these principles are being
discovered,
operationally applied,
refined,
transferred,
made conscious,
verbalised.

In vain I have looked for any research on such questions. We observed, however,
that learning processes for such activities,
are possible with seven-year olds,
can be necessary for eighteen-year olds.
At all levels of the primary school (1st6th grade) and in the two lower grades
of the secondary school the IOWO people paid attention in experimental and in

elaborated instruction situations to the geometry of rectilinearity under the

aspect of the vision line.
The picture of an island with all that is on it, is hanging in the class room:

What do I see if I stand here?

At which point am I standing if I see this or that?
How does one object move with respect to the other if I move this way?
How shall I move in order to have one thing displace itself with respect
to the other?
Pictures of the school with a tall building behind it, of apartment buildings,
supermarkets, public gardens, bridges, churches, sculptures, monuments, and so

on, in the city quarter around the school pass in review:

Where was the photographer standing?
Was he closer when taking this picture or that?
Was he more to the right or to the left?

Was he on the street or on this or that floor?

Where should he have stood to get a certain picture?
Photographs of block buildings (piled cubes) are analysed:
From which side was it taken?
Can there be more cubes behind this one?

How should it look from the backside, from the right, from the left?

308

CHAPTER 11

Photographs of the coast and the harbour taken from a ship travelling along,
which must be arranged in order:
From which place can you see the lighthouse, the mill and the church in
this mutual situation?

Aerial photographs of a landscape:

Where was the plane?

In which direction did it fly?
How should it fly in order to see this or that?
Photographs with a camera turning around a horizontal or vertical axis:
How to arrange these photos?
How to interpolate?
How to extrapolate?

The sky is lightly cloudy:

Why does it look heavily cloudy at the horizon?
An analogue-model, an ape behind bars (Figure 121):

What from the exterior world does the ape see?

A hole, a window, a mirror:

What can you see if standing here?

Where should I stand to get a certain picture?
Is there a place or not where you can see a certain object?
A tree, screen, wall:

Where can you hide?

Where shall I stand to see somebody hidden?
The preceding examples are activities embodying the principles proposed
earlier.
concretely,
imagined,
symbolised in drawings or otherwise.

FIGURES AND CONFIGURATIONS

309

There is, however, more involved in them than the vision line principles such
as formulated, to wit:
the mutual relations of vision lines,
right and left,

above and below,

obliquely right above, and so on,
of each other. What I formulated in Section 10.5 under standpoints and in
Section 10.6 as reciprocity under change of place of object and standpoint,
is under the vigour of the vision line principles sharpened to
viewpoint
and
reciprocity under change of place of object and viewpoint.

There is even more to it, resulting from the principle of the straight line,

X, Y, Z are on a straight line

can also be pronounced as

X is lying on a straight line with Y and Z,

or
Y is lying on a straight line with X and Z,
or

Z is lying on a straight line with X and Y,

statements that are equivalent, which for the vision line means:

if the eye O sees the object B covered by the object A,

then the eye at the place of B sees an object at the place of O covered by
A,
or with the object B interpreted as anothers eye:
if the eye O sees the eye O' covered by the object A,

then the eye O' sees the eye O covered by the object A,
or more symmetrically:
A is between O and O'.
Of course, it is not as simple as that, and this is early understood. O can spy
on O' around the corner A while O', looking in another direction, does not
perceive O. Or O can see around the corner A parts of the body belonging to

O' though not O' and by this way shield itself against O's look.

310

CHAPTER 11

11.8. Lightrays
I could not devise a better title shadow line would be misleading though
actually I mean precisely the fact that light source, object, and its shadow
are in a straight line. In the case of what I called the vision line the object seen
or to be seen is a secondary light source lying on a straight line with the
observing eye and the screening object. Now I mean the line between a primary
light source, such as the sun or a lamp, the object, and its shadow, which is
cast because the object screens the light source. In both cases we are concerned
with light rays and their rectilinearity, and in both of them I could have spoken
of the light ray as a straight line. In the first case the term vision line was
available, whereas I could not find as good a term for the second case.
I have good reasons to deal with light ray separately and after vision
line. Whereas the vision line is operational early and becomes more or less
conscious at the age of about 7 years, understanding shadows requires more
time. Shadows, as well as mirror images are perceived early and their origin is
qualitatively understood early, but I could not observe any understanding of the
geometry of shadow production at the age of 78-years. Even less than in the
case of the vision line, would I be able to indicate when this phenomenon is
discovered,
operationally applied,
made conscious,
verbalised.
What matters here is the
shadow principle
with its consequences:
light source, object, shadow are lying on one straight line,
from two of them conclusions can be drawn regarding the third,
displacement of light source or object correspondingly induces displacement of the shadow, and change of shape and size,
deformation of the light and shadow-receiving-surface correspondingly
induces deformation of shadow images.
The adverb correspondingly is a bit vague, but I used it in order not to be
drowned by details, and nobody will doubt how the correspondences are
meant
shadows get longer if the sun is setting or objects farther away from the
street lamp what is the common element in these phenomena?,
the elliptic shadow of a circle, the perspective image of a circle, the truncation of the cone what is the common element?,

F I G U R E S AND CONFIGURATIONS

311

the solar image of a hole at the opposite wall, the track of dust in a sun
beam, the sunlight circles under foliage how are they connected?,
the increasing vagueness of the shadow on the ground with increasing
height of the object,
the clouds in the sky and the changing illumination of the landscape
how are they related?,
the phases of the moon where is the sun?,
at which side does a lunar eclipse start?,
silhouettes and shadow images how are they connected?
a large number of queries, as meaningful for 18-year olds as for 8-year olds
meaningful pieces of geometry, even for adults.
11.9. Straight Lines in Views
The three-dimensional world projected on the two-dimensional retina is in the
central nervous system reinterpreted as a three-dimensional pattern, thanks
to sources of experience other than visual. We even succeed in suggesting this
three-dimensional world by means of two-dimensional pictures of this world,
thanks to an involved system of experiences and conventions, which is in no
way watertight, as appears from experiments on optical illusions.
How do we find out whether two lines in the optical field are skew to each
other? An eye movement can inform us about the intersection, whether it is
genuine or apparent. Homogeneities in these lines can inform us about which
passes before which, but there may be quite other experiences that contribute
to this mode of structuring. Conversely, such experiences can also be exploited
to provoke wrong judgments.
I do not intend to analyse here the theory of perspective and the methods
of descriptive geometry. Both subjects have in common the fact that they try
to do justice in two dimensions to the three-dimensionality of space and that
these attempts are astonishingly successful. The parallel-perspective image of a
cube, the central projection of a street is accepted as the representation of a
piece of geometric reality and reproduced in submission to the same principles
principles that may, or may not, have been made explicit.
A third means to get a grip on space in the plane, is folding down planes,
which when performed successively leads to networks of polyhedra, which can
be rebuilt from them. Shortest lines on polyhedra are recognisable in the
network as shortest, that is, straight lines, and as such they can be reconstructed
on the polyhedra themselves.
As an exam of the freshmen year of a teacher training college students
with a quite rudimentary mathematical education were asked the following question:

312

CHAPTER 11

This is the frame for a tent (Figure 122). The canvas is stretched
straight on from the oblique roof to the ground with no kinks or
folds. Construct the line in the drawing where it reaches the ground.

Terms like plane have been avoided because they suggest too much technical mathematics. The problem was thought to be difficult, but in my
view it was even too difficult. I observed a student perform the required
constructions without any hesitation. Though it was clear that he fully
understood why the construction was correct, he was not able to justify

it. He did understand, however, an a posteriori explanation. Likewise

other students who succeeded in solving the problem, acted by mere

insight, though because of lacking a geometric terminology they were not

able to prove the construction formally.
The two lines asked for are found by connecting points which themselves are intersections of two lines: a line g on the ground which connects
the feet of two poles behind each other and another line t that runs
obliquely from the roof. The two lines g and t intersect on the paper, but
is there anything in reality that corresponds to the intersection point on
the paper? The students saw that it was the case but were not able to
motivate it. With some elementary knowledge of solid geometry it is a
simple trick: imagine a plane through the two poles. Indeed, if this is
possible, g and t cannot be skew. But why is there a plane through the
two poles? Well, the plane is clearly visible, it is determined by the poles
and it extends to the triangle in the roof, containing g as well as t.
When solid geometry was still taught in the upper grades of our gymnasia,
paper and pencil constructions of plane sections of all kind of bodies was a
popular subject popular because it was an opportunity even for poor pupils

to achieve something by viewing and doing. Descriptive geometry was a similar

case at Modern secondary schools. These activities rest on recognising, verbalising
and processing such simple principles as:

FIGURES A N D C O N F I G U R A T I O N S

313

in order to find the intersection of two planes, look for lines in the planes
that intersect each other,
in order to find lines that intersect, look for a plane comprising both
of them,
in order to find planes, look for lines that intersect or are parallel and
imagine a plane comprising them.
The didactical folly of this subject was that it was offered to 17-year olds
rather than to, say, 1012-year olds, who in turn were tormented with a deductive system of planar geometry. Why? Possibly because in Euclid plane geometry
precedes space.

11.10. Polygons
We already met polygons as a combination of line-segments under the term of
graphs. One need not restrict oneself to the plane if discussing polygons. The
edges of a polyhedron form a polygon, which can be studied with regard to
combinatorics as well as flexibility. Even a closed skew quadrilateral can be a
rich source of experience. With its diagonals it forms the edge polygon of a
tetrahedron.
A line track is a directed graph, with vertices

and segments

it is closed if by chance
Self-crossings need not be forbidden,
though with what is usually called polygon, they are not admitted; figures like
the pentagram (Figure 123) are called star polygons.
Plane closed polygons with no self-crossings divide the plane into an interior
and an exterior domain, as do Jordan curves; the name polygon is also bestowed
on the interior; then the line track is called its circumference.
Plane closed polygons may possess what is called re-entering angles (Figure
124), but most often if one speaks about plane polygons, one means convex
figures. Convexity will be dealt with in the next section.

11.11. Convexity
Convex and concave first prompt associations with lenses. Both terms also
apply to arcs, and then it matters from which side they are viewed. If it is the

graph of a function it is understood that this happens from below to above.

314

CHAPTER 11

In Figure 125 arc 1 is convex and 2 is concave. A closed plane curve such as
Figure 126 behaves convexly at 1 and 3 and concavely at 2 and 4. Similarly with

a dented ball one can distinguish the dents as concavities from the convexity.

Convexity such as is defined in general for plane or spatial figures is a characteristic example of an unusually successful mathematisation, close to visual
reality, of a mental object. One defines:
a set S is convex if with any pair of points it contains the connecting linesegment.
(So the line of Figure 126 does not enclose a convex set since the line-segment

pq is not contained in it.)

A fundamental fact:
the intersection of convex sets is convex.

A plane in space determines two parts,

half spaces both of which are convex.
A line in the plane determines two parts,

half planes both of which are convex.

A point on a line determines two parts,
half lines both of which are convex.
By forming intersections of suchlike sets one gets new convex sets:

The intersection of two half spaces an infinite disc bounded by parallel

planes or a spatial angle bounded by intersecting planes; the intersection of

two half planes in the same plane a strip bounded by parallel lines, or a
planar angle bounded by intersecting lines; the intersection of two half lines
on the same line a line-segment. Three half spaces, if not too particularly

situated, have a solid angle in common; a tetrahedron as intersection of four

half spaces is convex, as is the triangle surface as intersection of three half
planes in the same plane.
At a glance as it were one can see whether a planar or spatial figure is convex.

If, however, such a figure is given mathematically, the problem arises as how
to prove convexity. This is not at all easy if the figure is bounded by a curve or
curved surface say, a circle, ellipse, parabola, sphere, and so on. For instance
in the case of the circle one has to prove:

FIGURES AND CONFIGURATIONS

if p and q have a distance

pq have the same property.

315

from m, all points of the line-segment

This in turn follows from:

for x on the line-segment L the distance (m, x) assumes its maximum at an
endpoint of L.

In other words:
for x on the straight line L the distance (m, x) assumes nowhere a (relative) maximum.

11.12. Circles

Acquaintance with and mental constitution of the mental object circle is an

early developmental phenomenon. The term circle for this mental object
appears relatively late, because parents and educators as it seems in all languages believe that round thing or suchlike words are easier propagating
baby language is an educational phenomenon worth being investigated.
Even the construction of circles can be a relatively late phenomenon, drawing
round a circular object can precede the construction by means of a pair of
compasses. Incidental observations showed me that 8-year olds had much
trouble in using a pair of compasses but in a third grade where the children

were familiar with the use of the instrument, nobody had any trouble. The
precise construction of circles is of course preceded by free-hand drawing
(or foot drawing) of circles on paper (or in sand).
Attempts at provoking definitions are useless even with children that can
handle the compasses, or do not yield the result that one would expect. One
gets statements like

circles are round,

circles are everywhere as wide (or thick),
circles are everywhere as round.

If one asks what as wide means, parallel tangents (or support lines) are drawn
and their distances are indicated as equal. Of course a figure with this property
need not be a circle: there are a great many other figures of constant width
(Figure 127).

316

CHAPTER 11

Everywhere as round is a good local characterisation of the circle a curve

of constant curvature, where curvature is an intuitively visual datum. S. Papert
has stressed this local definition; in his approach it is the computer that draws
a circle on the screen:
the turtle gets the instruction (Figure 128)

a little step of fixed length forward,

a turn over a little fixed angle.
The result is a polygon with equal sides and angles indeed, a circle in the limit.
Curvature of a curve is defined by differential geometry as the limit ratio of
tangent turn and arc length; Paperts program expresses the constancy of ratio

turn of direction to path covered which indeed in the limit gives the constancy
of curvature. As a matter of fact among all plane curves the circles are the only

ones with a constant curvature. Everywhere equally round is a good local

definition of the circle.
A circle cut out fits in many ways into the hole it left; I can turn the circular
disc transitively in the hole, that is, such that each border point passes into
each other point. I can attain the same by turning the cut out circle around a
diameter (a reflection at the diameter).
This is a more global definition of the circle, which needs profound mathematical reasoning to be derived from the former local definition. Visually the
connection can be made via Paperts approximating polygons.
The usual global definition of the circle as the set of points x that from a
fixed point p (the centre) have a fixed distance (the radius length) is not at
all obvious. The vast majority of circles in the environment do not betray their
centre.
How to discover the centre? From people who have not been trained in
geometry, one may not expect the construction that is taught in the geometry
lesson. The usual attempts show: drawing parallel tangents, connecting the
contact points and halving this line-segment. If, however, the circle is given or
suggested to be cut out, 8-year olds get a diameter by symmetric folding. This
is the straight line as a symmetry axis a phenomenon which will occupy us
later on.
The query how to find the centre becomes even more tricky if the circle is
given by an arc that must be completed. It is natural to seek the centre on the
axis of symmetry, that is on the orthogonal bisector of the largest chord of the

FIGURES AND C O N F I G U R A T I O N S

317

arc. But where on the bisector? A change of perspective to a smaller partial arc
is of help: the orthogonal bisector of a smaller arc as a second locus for the
centre. Finding the intersection of the two bisectors boils down to the classical
construction of the centre of the given circle a solution not parachuted
but suggested by the seemingly more difficult problem of completing a circular
arc to a whole circle.
From here it is a little step to the circumcircle of a triangle and the insight

that a quadrilateral in general does not possess a circumcircle.

With this exposition we have run far ahead on a didactically warranted approach to circles. The approach by means of a circle drawn on the backboard is
much too narrow. One can have children experimenting with concrete circles

(round discs) of various sizes, paving the plane with a large number of congruent
circles, discovering that each circle is touched by its neighbors in a regular
hexagon, that opposite contact points are connected by a diameter, and that
these diameters intersect in the centre. One can also have the children stand in
a circle and order one of them to stand in the centre where is it and how to
find it precisely? One can have circles drawn with a string, with a pinned up
strip, or by enlarging and reducing a given circle, before handing them out the
pair of compasses.
With the pair of compasses one can stake out and transfer distances. The
shortest distance of a point p to a set S has something to do with circles: the
compasses pinned in p are being opened until meeting S. The set S can be a
straight line, of course the shortest distance is: straight ahead on the line. The
open circle just touches the line; the line S is the tangent of the circle, the
shortest distance is the radius, tangent and radius are orthogonal a principle
that extends to functional analysis: the orthogonality of the line of shortest
distance. Let S itself be a circle; finding the shortest distance from p to S
produces a circle contacting S. Both contact each other, both radii drawn
to the point of contact are orthogonal to the common tangent, so the radii
are extensions of each other. And so it goes on there are a great many approaches to the geometry of the circle besides the one that is prescribed by
school geometry.
Measuring of arc length was discussed in Chapter 1. One can measure the
circumference of the circle by unreeling a string or rolling it along a straight
line. Rolling circles on straight lines or on circles is itself a source of many
phenomena which are worth being considered.
A well-known puzzle asks: If I roll a moveable florin around a fixed florin, how many
turns will it have performed when it has returned to its original place? The surprising
answer is: two. How to explain it? Though the answer does not fit in the present context,

let me anticipate it. One can observe the same phenomenon with, for instance, congruent
regular 3-, 4-, 6-gons, which makes it more perspicuous. Localising is even more efficient.
I roll off the angle C on the angle C (Figure 129) how much is the turn? It is the sum
of their exterior angles. A simple closed polygon has the sum 360 of its exterior angles.
If they are such that the one can be one-to-one rolled around the other, the total of turns
is 360 + 360.

318

CHAPTER 11

I said one-to-one. I did it intentionally. Take a triangle and a hexagon, both regular
and with same side length. I roll the triangle around the hexagon. How many turns does the
triangle perform, and why?

Among all plane figures the circle shows the highest degree of symmetry. It is
easy to disturb the symmetry and it is an art to profit, as least partly, from this
fact by drawing regular polygons and stars in the circle and to surround one
circle by a wreath of circles which touch and intersect each other phenomena
to be considered elsewhere.
11.13. Venn-diagrams
This is no section that got lost and properly belongs to Chapter 3. I once
introduced the term natural Venn diagrams* in order to distinguish Venn
diagrams as used as a tool in serious mathematics from the artificial ones, which
in school mathematics is an aim in itself. A natural Venn diagram appears if one
draws a simple closed curve and pays attention to its interior as a point-set. This
means that the set A is the point-set within a curve which is labelled A rather
than a set of objects or letters placed or indicated within the curve. Only these
natural Venn diagrams occur in normal mathematics although the plane sets
within the curves can be models for other kinds of sets whose unions, intersections, and complements are to be illustrated.
I put Venn-diagrams between quotation marks because I mean that I have
the figures in mind, primarily as geometrical figures, while the opportunity to
exercise set theory operations on them is a bonus. How shall I teach sets in
the first form of secondary education?, a teacher sighed after my well-known
criticism of what usually happens with sets. There is a vast choice for everybody
who does not suffer from what I called the constraint of the system. The system,
that is, straight lines, half planes, circles, circular discs, and so on, at the far
end of a long development which starts with sets at its first step. I need not
argue that system constraint is contrary to didactics based on phenomenology.
The intersections, unions, complements of geometrical figures which I am

going to exemplify are again meant as geometrical figures. Terms like

* Mathematics as an Educational Task, p. 342.

FIGURES AND CONFIGURATIONS

319

set of the . . . with the property that ...,

inclusion,
intersection,
union,
complement

will serve as linguistic means as is the case almost everywhere in mathematics

to

describe given figures by means of others,

formulate processes of construction,
explain connections between figures.
Training these linguistic means can be useful by transfer to other subjects where
they apply, the figures and the operations performed on them can be models
for vaster abstractions.

Our elementary sets are

line-segments and straight lines,

half planes (it does not matter much whether the boundary is included),
circles,
circular discs,
where the last are explicitly defined as

set of the points at a distance of p,

set of the points at a distance
of p.
Starting there

angular parts of the plane, polygons, polygonal pieces,

circle segments, circle sectors,
circle annuli,
types of lunules and their complements,
circular triangles and polygons and their complements
are

constructed by means of set theory operations,

analysed with respect to their mutual set theory relations,
described by set theory means (see Figure 130).

11.14. Spheres
In the development of geometrical mental objects the sphere as a ball
precedes perhaps even the circle. In traditional geometrical instruction the
sphere was a latecomer, dealt with if at all only through formulae for surface and
volume, which were learned, confounded and forgotten.
A local definition of the sphere, as of the circle, would be that it is everywhere

320

CHAPTER 11

as round. There are, however, with respect to the sphere experiences which
are more original than constant roundness. A ball can roll and differently
from cylinder and cone as easily in all directions. Or does this hold of all
round convex bodies ellipsoids, eggs, acorns? No, the sphere at least the
substantially homogeneous ball does not know any prefered position; at
which ever point it touches the ground plane, it is in indifferent equilibrium,
whereas other convex round bodies are distinguished by certain positions of
stable equilibrium. Mechanically this can be expressed as follows: the sphere
is the only round convex homogeneous body, the centre of gravity of which
is as far from all boundary points and this point then is also the centre of the
sphere itself.
But let us review the characteristic of being everywhere as round. In the case
of the circle the degree of roundness was mathematised by the concept of
curvature the curvature at a certain point was the limit ratio of the turn of
the tangent and the arc length, where instead of the tangent vector one could
as well use the normal vector. In differential geometry the curvature of surfaces
is defined in a similar way as that of curves: one imagines in each point of the
surface a normal vector of unit length; by assigning to each point of the surface
its normal vector, the surface is mapped on the unit sphere; the inverse limit
ratio at a point p of the areas of a piece of surface around p and its image on
the unit sphere is by definition the Gauss curvature of the surface of p. For
a sphere with the radius r it is everywhere

FIGURES AND CONFIGURATIONS

321

I just mentioned this kind of curvature though it does not reflect what we
mean by everywhere as round. The Gauss curvature accounts only for the
intrinsic properties of a surface, not for its appearance in space. A cylinder and
a cone have everywhere vanishing Gauss curvature as has the plane, and they
can be unreeled upon the plane without stretching or shrinking at least locally.

The sphere has a constant Gauss curvature but there are more surfaces with
the same constant curvature, which can be unreeled upon each other.
The everywhere as round means the appearance in space. The sphere

shows locally everywhere the same image. As a matter of fact, this holds also
for the cylinder, but for the sphere it does so at every point in each direction.
A mechanical production of spheres rests on this being everywhere as round.
If two stones or pieces of metal are ground upon each other, they will turn out
to display finally two congruent spherical surfaces by chance they might
be planes.
Once the circle has been identified by means of its centre and radius length
as the set of radius distance from the centre, the same characterisation of the
sphere is natural. Of course the centre of the sphere is less accessible than that
of the circle, but a pair of antipodes can be found on a concrete sphere, and so
can a diameter.
Cuts of spheres or the water level in spherical bottles show that the plane
intersections of spheres are circles, but this fact is also mentally understood as
soon as it is clear how to find the centre of this circle: the foot of the perpendicular from the centre of the sphere on the plane. Or, more perspicuously: viewing
the sphere as a solid of revolution with as its axis the diameter perpendicular to
the plane. Likewise the intersection of two spherical surfaces is recognised as a
circle; then the axis of revolution is the line connecting both centres.
Shortest paths on any convex surface can be obtained experimentally, by
held strings. It looks natural to consider such a path also as a straight ahead line.
There is, however, also an independent, intuitively acceptable, definition of
straight ahead line on a curved surface. Straight ahead line is a local concept;
if two surfaces are being rolled upon each other a straight ahead line is copied
as a straight ahead line; if a curved surface is being rolled upon a plane such that
the contact points form a straight line, the corresponding line on the surface is
a straight ahead line a geodesic.
Applied to spherical surfaces, this shows that the great circles are the straight
ahead lines. It requires more profound arguments to show that arcs of great
circles are also shortest lines.
It follows from the foregoing that spheres cannot be unwound to become
planes, even locally. A spherical triangle unwound upon the plane would
represent itself as a straight triangle with the same lengths of sides and angular
measures; this would result in a sum of angles of 180 whereas in a spherical
triangle the sum of angles is certainly larger. The fact that the sphere cannot be
unwound upon the plane is responsible for the variety of projection methods
by which cartographers try to map the globe.

322

CHAPTER 11

Nice partitions of the sphere most often use arcs of great circles not because
they are the straight ahead, and the shortest, lines but because the planes of the

great circles are symmetry planes of the sphere and the circles themselves symmetry lines on the surface.
Well-known partitionings of the sphere are the so-called dihedra great
circles through the poles together with the equator (Figure 131). The regular

bodies inscribed into the spheres project themselves from the centre onto the
surface as regular partitionings of the spherical surface. Similar effects are

obtained with semi-regular solids; the best-known is the football, a partitioning

by 20 regular pentagons and 12 regular hexagons. There are other opportunities
to deal with this subject.

11.15. Revolution Surfaces

Revolution surfaces arise if a plane curve the profile is rotated around an
axis lying in the same plane. The spherical surface is an example, with a diameter
as axis and a great circle as profile.
If, however, a circular arc turns around a chord, which is not a diameter of
the circle, one gets a lens or a vault. A decent surface does not arise unless the
axis is lying outside the circle; it is a ring, also called torus, next to the sphere
one of the most familiar surfaces. Revolution surfaces are produced in a rich
variety on the potters wheel; the rotating clay is being profiled into the wanted
shape. There is a striking difference between points where the profile curve is
convex or concave with respect to the axis. In differential geometry these points
are distinguished by the sign of the Gauss curvature: positive and negative,
respectively.
If the profile curve is a straight line-segment, the surface arising is a
truncated cone or a cylinder. It has been earlier mentioned that these latter
surfaces can be unwound on the plane, and on each other.

11.16. Angles
Elsewhere* I have extensively analysed the various concepts of angle. I started
with the mental objects in order to reach the mathematised concepts as fast as
* Mathematics as an Educational Task, pp. 476494.

FIGURES AND C O N F I G U R A T I O N S

323

possible. Though the present approach is a bit different, duplications would be

unavoidable if I were not allowed to call on the former exposition.
In agreement with the title of the present chapter, angles are dealt with
here as figures or configurations. Just as squares differ from oblongs and the
same square or oblong is possible at various places, so angles will be distinguished as different or equal before measuring angles is discussed.
Angles present themselves in a variety of ways early or late in the
development. If these presentations shall be classified phenomenologically,
it is quite natural to look at the components that contribute to the angle
concept, and at the degree in which they are.

concretised,
made explicit,
indicated,
suggested,
imagined.

By the components I mean

the sides of the angles,
the planar or spatial part enclosed by them

if the angle is a static datum, and

the turn
if kinematics comes in.
I introduced angle concepts in the plural, because there are indeed several
ones; various phenomenological approaches lead to various concepts though they
may be closely connected. Even in practice one single version does not suffice.*
This is witnessed as early as Euclids Elements; in Book I after the definition
VIII of the angle as the mutual inclination of two lines, definition IX speaks of
the angle enclosed by two lines, and though by definition angles are less than
180 this does not prevent Euclid from adding angles beyond 180.
The most concrete though also the most misleading phenomenon of the angle
is
the concrete part of the plane or space;
one is expected to restrict ones attention to the neighborhood of one vertex
or one edge of an angular sector or an edged body, while this local view is
disturbed by the actual limitation as to extension and shape of the sides and
of the figures as a whole two angles are equal if they
fit each other locally.
that is,
if the one fits locally in the hole of the other.
* Mathematics as an Educational Task, pp. 476494.

324

CHAPTER 11

The local character implies that one is allowed to anyhow break off an angular
corner in order to seize the angle (Figure 132); the
nondescriptly broken off corner

then is a means to suggest what is meant by an angle. Whoever has observed

children working with jigsaw puzzles of polygonal pieces will have noticed that
they are familiar with this mental object of angle, though there are 78-year

olds who fail in that matter for lack of former experience.

The stress laid on the enclosed planar or spatial part is weakened if the
enclosure is removed as much as possible, as happens with instruments like
drawing triangles,
where the inner border suggests the same angle, and suchlike moulds. Even
leaner concretisations are possible the sides only concretised,
as legs of a pair of compasses,
as two hands on a clock dial,
as sides of an angle folded from a sheet of paper,
as main and side branch of a tree,
as two bones or staffs in a joint,
as sheath and blade of a clasp-knife,
as two canvas walls of a tent,
as a telephone wire that changes direction at a pole.

The enclosed planar or spatial part can still be suggested by the sides as a
skeleton:
the steel wires of a fence, or wire netting
form angles of squares or hexagons,
the wings of a folding screen
hide a part of space, the

edges of a wire cube

F I G U R E S AND C O N F I G U R A T I O N S

325

form angles and at the same time squares. The leanest representation of an angle
is a drawn pair of
line-segments from one point,

which are supposed to represent line rays, with possibly an arc or a system of
arcs between the sides added, which however does not represent the enclosed
part of the plane, but rather indicates that the figure means an angle, and well
the same angle if the same arc or system of arcs is repeated.
But it is even possible that the two sides are not or are not both concretised, made explicit, indicated but at most suggested. This happens with the
angle as
deviation from a rectangular or flat continuation,
branching off, breaking
when one of the sides is not concretised but imagined or imaginary as the
continuation of a concrete line or plane the dotted feature in Figure 133
that is an

adjacent angle of a concretely given angle.

A polygonal disc or prism shows concretely a number of interior angles; if they
are walked around, the change of direction is expressed by the exterior angles
a subject usually neglected in instruction.
Subjects in Piagets experiments describe the reflection of a ball at a wall by means of the
angles between the wall and the direction of the movement, which is quite natural the
deviation of concrete legs. Piaget, however, accepts as a proof of full understanding only
the formulation by means of the so-called angles of incidence and reflection (that is with
the perpendicular on the wall), which is the way it is taught in textbooks.

Of the pair of sides of an angle there can be

one explicit and the other implicit,

for instance in a topographical context: in the terms

upright, oblique, lopsided, leaning.

326

CHAPTER 11

A line or a plane deviates from something that is considered as normal:

an oblique roof,
an upright wall,
an oblique ladder,
a painting hanging askew
a steep road, ladder, mountain,
an inclined plane,
a smoothly sloping landscape
aim at an angle with respect to the horizontal plane, whereas
crooked growth, out of plumb, slanting in the curve
may rather mean an angle with a vertical line or plane. In

slanting letters, sloping line, oblique square

the margin line of the sheet, the lines or squares on the sheet may be the norm
with respect to which the deviation is stated.

An oblique nail
forms an angle with an imaginary nail that would be right in the plank.
So far we mentioned cases where the other side of the angle is apparent.
In the case of the angle between
line and plane
the other side is not as obvious. If they are not orthogonal, it requires a profound analysis to understand that the other side is obtained by projecting the
line on the plane.

The angle is even more stripped of its concrete elements if both sides are
imaginary, as is the case with the
vision angle

under which an object or an (as imaginary) line appears. The sides of the angle
are vision lines, imaginary lines that connect more or less concretised points. In
the case of the
elevation of the pole and the angular distance of stars
vision lines provide the sides of the angle, which afterwards are interpreted as
radii of a sphere, the celestial globe. In the case of angular distance on the
Earth, such as

geographic longitude and latitude

the sides of the angle are entirely imaginary lines, drawn from the centre of
the Earth, and extended to the zeniths if the measurement is performed by

astronomical means.

FIGURES AND CONFIGURATIONS

327

If an
arc is interpreted as an angle,
as happens in geometry and trigonometry, the sides of the angle have entirely

disappeared, though this angle can be interpreted as

change of direction
(Figure 134) continuous rather than instant change.

We have now arrived at the

angle as turn,
that is, neither the static pair of sides nor the enclosed planar or spatial part,
but
the process of change of direction,

the transition from side to side, where the enclosed piece is swept out. It is my
experience that nobody except the teacher, who exerts himself to explain it,
has any trouble with this aspect of angle.
The key in the key-hole,
the tap,
the hands of the clock,

the knob of the watch,

the handle of the door,
the screw cap,
screw and nut,
the cork-screw
in all cases one has quite precise kinesthetic ideas about the meaning and consequences of a certain turn, how far one can and may turn: a quarter or a third
turn of the key if the door is simply closed, and one of two turns more if it is
bolted, a quarter of a turn of the tap for washing ones hands, and many turns
if a bath is to be filled; angles at clocks tell you the time even if there are no
figures on the dial, and each of the familiar turning instruments possesses characteristics that tell you how far you can and may go.
Benjamin (7;5) is shooting with a dart folded from paper called a Concorde at his
sister. Twice the dart lands at the same spot at her right. The third time he aims while taking
the deviation angle into account, and hits.

328

CHAPTER 11

When starting this subject, I said that only people who try to explain it to
others, have trouble with it. The reason why conceptualisation of this mental
object requires so much effort may be the fact that it is the most natural, the
most instinctive aspect of angle.
I refer to the earlier quoted exposition in order to stress that the mental
object angle is conceptualised in various ways, depending on the approach,
and the possibilities and needs of application. Euclids angle finishes at 180,
as does the usual protractor. The angle of trigonometry is represented on the full
circle protractor. The pair of sides can be considered as ordered or it is understood that the order does not matter. For the turn angle it can be essential
whether after a full turn one counts further or starts anew, and whether it is a
right or a left turn. Whether this or that matters, this depends on the context.
The mental object is being constituted in a context.
The famous joke of one of our TV entertainers a few years ago in a political
context that three turns to the left is the same as one to the right, rests on
intermixing two contexts. In the gymnastics lesson both of them give the same
result, but if a quarter of a key turned to the right opens a door, three quarters
to the left will not do on the contrary the door gets even more closed. Whether
a gymnastics lesson or opening a door is a better model for politics, fortunately
need not be discussed here.
But we may ask why what is true in the gymnasium does not hold for the
door lock. The action of the key in the door is: bolting and unbolting. A rough

sketch (Figure 135): a toothed wheel moving a toothed bar. When the wheel has

made a full left turn, the bar has moved a full circumference of the wheel
to the right. If the bar were long enough, one could move as far as you like it
by means of the wheel, which itself stays on its place as far as the wheel is
concerned, it may look as though nothing has happened. But the converse can
also be realised, that is with a rack-and-pinion railway, where the moving toothed
wheel of the train steps into the fixed toothed bar between the rails. Or, without
teeth: a circle wheeled on a straight line.

FIGURES AND CONFIGURATIONS

329

In all these cases turn angles that count unlimitedly, are tied to angles where
a full turn is counted as though it were nothing two mental objects of angle,
the relation between which is clear enough as long as one stays away from
conceptualising. In order to pass from the turn angle to the angle determined

by two sides, one must abstract from consider as inessential that what
happens in the meantime when one side is being turned into the other. Making
this explicit can be difficult, indeed. It can obscure the insight while its strength
may be spared if it is left implicit. This then may be the reason why teachers
meet difficulties when they try to explain the turn angle. I did not investigate
this problem but it would be worthwhile trying it.

A heptagon (Figure 136) has the sum of angles 5 180; 5 times that of a
triangle. What kind of angle? If I walk around the heptagon according to the
arrows, the change of direction in each vertex corresponds to the exterior
angle, which is after a whole walk a full left turn (360). This then is the sum
of the exterior angles. Interior and exterior angles together are 7 times 180,
which leaves the interior ones together with (72) times 180. How does
this angle arise? I am turning the line 71 around 1 clockwise until it is lying
along 21, then 12 around 2 clockwise until lying along 32, and so on, until

finally by a turn of 67 around 7 I come back to 17. That is 5 times 180 turned
clockwise.

CHAPTER 12

GEOMETRICAL MAPPINGS

12.1. From Figures to Mappings

In the original version I slipped at this point straight from the figures into
the mappings. I slipped though geometrical mappings as cultivated now are
phenomenologically all but selfevident. Historically they were a late phenomenon
this will be discussed later. I doubt whether in the individual development
there might be something like a spontaneous genesis of geometrical mappings
as mental objects; didactically viewed the most urgent problems of geometrical
mappings have not yet been reconnoitred.
For the many people, who have not yet conquered Aristotelism, concept
formation, even in developmental psychology, is synonymous with classification.

In our last chapter the titles of the sections may evoke classificatory associations;
those of the present chapter will rather conjure up ideas of relating objects with
each other, which indeed is the germ of mapping. A mapping requires three
components: the thing to be mapped, the image, and the process of mapping.
In the course of the present chapter stress will shift from figures as the objects to
be related with each other via the mapping as mental operation to the mapping
as mental object. To be sure, from the beginning onwards mappings will be
considered, though within a restricted context: as a means to compare figures,
to recognise them as the same, to distinguish them, to structure them, according
to varying criteria.

12.2. Comparing Figures

Let us start with the last examples of the preceding chapter on angles.
I face a planar angle. I fix one side and move the other I map it from one
position into another. The angle shrinks or swells I mean the part of the plane
enclosed. Or I follow with the second side the movement of the first such that
they move as fast. The part of the plane enclosed changes its place though not
its shape. I can move an angle over the whole plane, pass with it into space, the
sides rigidly coupled to preserve the angle or hinging in a joint such that the
object remains an angle.
Mathematicians might take offence at this exposition. What remains the
same, if the angle is rigidly displaced, is the angular measure rather than the
angle. It is the same as with the line-segment, where displacing preserves the
length only. We learned, indeed, to avoid sloppy terminology. We indicate two
line-segments, two angles, two triangles, the one a reproduction of the other
at another place, congruent rather than equal.
330

GEOMETRICAL MAPPINGS

331

Right, we act this way as soon as we have put objects into a geometrical
context, or rather if we have done it so consciously that it is a logical geometrical
context.
On my writing desk I am facing two cylindrical plastic little barrels, exactly
the same, that is so exactly that at a glance I can be mistaken about which
is which. One contains pills, the other is empty. If I lift and possibly shake
them, I know which is which, I can distinguish them. No, I can distinguish
them right now: the left one and the right one. If I interchange them, I will
have interchanged these attributes. But the other thing, their substance, has
not been interchanged. I can move, turn each of the barrels, put them in my
pocket, and they remain the same. Moreover, they remain the same as geometric
figures, perpetually congruent. The one Peugeot 404 (1972) is congruent with
the other, and each with itself, but with itself it entertains a closer relation than
congruence only.
A talk with a group of 10 11-year olds: a sum 3 with two dice how is it possible? The
dice are lying on the table. They manipulate the left one to show 2, the right one 1. But it
can be done differently: to show it they have both dice change places. A desperate discussion. The aim to have the left die manipulated in 1 and the right one in 2 is unfeasible,

because it is much easier to have them change places. With the best of intentions they
do not succeed because the concreteness of the material dice prevents them from forming
the mental objects die 1 and die 2. Dice of different colour would have been more
useful.

What is wrong here? Of two somehow equivalent objects

substantial and local exchange

are confused. Elsewhere* I have explained how much trouble textbook authors
cause themselves and their pupils with this confusion; though I hardly scored
any success with this exposition, I am weary of repeating it once more. It is
clear what is wrong here. In order to have two dice function as an ordered pair,
I must be able to distinguish them as actualisations of one die, independently
of their localisation. Speaking about figures I neglect these elements. At a glance
both barrels are somehow cylinders. This is their figure. There are different
kinds of cylinders, thick ones and long ones, but both of these ones have the
same figure. It is one and the same figure, twice actualised. I do not say: embodied. There is one natural number 5.I can embody it by the fingers of my left
hand, of my right hand, by the five continents, or by any imaginary set of five
things. Likewise I can embody a cylinder in many ways, and this happened here
by the pair of barrels. But I can also take one of them and place it differently
in the space. Both barrels are not only different embodiments but they are also
together in an order of coexistence.
The same means various things in colloquial speech. Fortunately! A profusion of words can be annoying. But a profusion of meanings of the same word
* Mathematics as an Educational Task, pp. 377387.

332

CHAPTER 12

is no evil as long as the meaning is determined by the context. Look at this pair
of cubes, and say what is different. Colour no, we do not want to pay attention to colour. Plastic or wood no, the material does not matter. There is a dot
on one of them it does not count. A scratch as little. The place? They are
at different places. Does it make a difference? If the one faces you with a side
and the other with an edge, it might be considered as a difference, which,
however, might be reasoned away. It is a pair of cubes, which can produce a new
figure by putting one upon the other, as one can add up the same two fives.
How long can this terminology be maintained? When does the need emerge
for the word congruent? I would say, in such a context as congruence
theorems. Two cubes are the same if their edges are the same this can still
pass. But: two triangles are the same if their corresponding sides are the same
this walks with a limp. Is it that much better to say two triangles are congruent
if corresponding sides are congruent? Well, let us say: Triangles ABC and
ABC are congruent if AB, BC, CA are congruent with AB, BC, CA respectively. But be cautious! A triangle is more than a triple of points and a triple
of line-segments. Triangle then includes also an order of the vertices (a linear,
not a cyclic one), because arranged in another way they may no longer fulfill
this condition. Well, I could also say: Two triangles are congruent, if I can
assign their vertices in such a way, A to A, B to B, C to C that the corresponding sides are congruent.
It appears from this formulation how far we are advanced in a logical geometrical context. Being congruent means here the existence of a congruencemapping, which as such has been eliminated by the existential quantor. Are
these superfluous sophistications? Yes and no. One can come into situations
where a pair of figures is congruent in various ways and the way in which, that
is the particular congruence, does count for this or that reason. An isosceles
triangle is congruent with itself as is each figure, but it is so in a non-trivial
way, to wit by interchanging the vertices at the basis; an equilateral triangle
is self-congruent in as many as six ways, a square in eight ways I will deal
with symmetries later on.
The phenomenon signalled in the last paragraph is quite frequent in mathematics. Two mathematical objects are the same thanks to a certain equivalence
but this equivalence may have been drawn from a stock of equivalences, and
at a certain moment it may matter under which one the objects are the same:*
Third graders are building four cubes houses. All of them should be different. Houses that

can be turned into each other, are the same, but minor images are being considered as
different,
and
are mirror images, but at the same time they are turn images
of each other. Thus, mirror images will be considered as different if they are not at the same
time turn images.

The cubes of the quoted experiment were differently coloured but nobody
paid attention to this fact. They were of the same size, thus congruent; they
* Wiskobas Bulletin 6, nr. 2.

G E O M E T R I C A L MAPPINGS

333

must touch each other along full faces. It was tacitly accepted that they were all
the same cubes, interchangeable. But another stock of mutually congruent
cubes, of another size, would as tacitly have been accepted as the same
material. Comparing figures as meant in the title of this section need not be
restricted to stating congruence.

If some class of figures is recognised as squares, rectangles, cubes, planks,

prisms, pyramids, cylinders, cones, spheres, a comparison takes place in order to
establish agreements and differences. Comparing can have the character of pointby-point mapping, but it can also be of a more global, combinatoric kind. If two
rectangles are recognised as such and compared, the attention can be restricted
as far as the common features are concerned to the fact that all angles are right
and opposite sides are equal, and the sides of both of them can be compared with
each other, but this comparison can also be extended to a not necessarily
congruent mapping of the one upon the other, possibly embodied by the
deformation of elastic material. In the case of parallelograms, which as such
are recognisable by the parallelism of opposite sides, more characteristics can
diverge: an angle and the sides around it. In this case, too, the mapping an
affine one can be extended by a systematic construction of parallelograms
(Figure 137): in the case of rhombuses with constant sides and variable angle
this can be embodied by a frame of threads. If a pair of triangles is given it
is natural to extend the assignments of vertices to each other as an affine
(barycentric) mapping. Prisms, and likewise pyramids, are distinguished by
base and height; as to the base the kind of polygon matters; a quadrangular
prism can be divided into two triangular ones or two standing faces can be
turned with respect to each other in order to produce a triangular one.

I close the present section with these examples in order to stress that
primarily, comparing figures does not mean pointwise mapping though comparisons can be unified under the aspect of mapping.

12.3. Distinguishing Figures

Comparing and distinguishing are two aspects of the same activity, differently
shaded. Equality does not mean much without inequality; it matters what is
stressed. The means to distinguish figures, specified in the preceding section,
were place and size. In elementary geometrical material, but also in drawings

334

CHAPTER 12

on the blackboard, colour can be used as a means to distinguish: the red and
the green triangle, the red lines and the green lines of course I could have
distinguished them also by letters or numbers. The faces of a die are distinguished
by primitive number symbols. The crystal lattice of household salt shows more
structure than the ordinary cubic lattice: by chemistry the spot of the Na atom
is distinguished from that of the Cl atom. Does this still belong to the geometrical
context? Yes, and no. We can pick it up, as the colour of triangles and straight
lines, in order to refine a structure, to compare more sharply, and in this sense
it is still geometry though not expressed in purely geometrical terms.
As a means of distinguishing, new kinds of characteristics are added to the
purely geometrical ones, not only in order to stress certain geometrical differences by material and colour factors, as assumed in the beginning of this
section, but to create new structures. Triangular pyramids that are congruent
when viewed as tetrahedra, are distinguished as pyramids by extending to one
of their vertices for each of them another one the predicate of top (and to
the opposite face, the predicate of base). Are these new characteristics of a
geometrical nature? Base and top seem somehow related to the terrestrial
horizon and the plumb, but this is not the intention; pyramids with 4, 5, 6
side faces are recognised as such however they are situated in space; tops may
show downwards and side faces lie horizontally. If, however, a tetrahedron is to
be considered as a triangular pyramid the top is not determined by internal data
but by a stigma: this shall be the top. If now I am going to compare triangular
pyramids with each other as was done in the preceding section whether they
are equal or different according to this or that characteristic, mappings play
a part: assigning vertices to vertices, and of course top to top. This can lead to
distinguishing figures by extra structure.
Two planar partitions into equilateral triangles or two regular partitions of
space into cubes are the same in the sense of similarity. But by painting
the plane or space by means of triangles or cubes of various colours, I can
create diverging configurations, which are distinguished by some extra structure,
which is not determined by geometrical figures but by stigmata.
The most striking example in this respect is the cube transformed into a die.
As a matter of fact, I could interpret the familiar number symbols of the die or
the domino as geometric spot figures, but this would be a preposterous way to
save the geometrical context. Indeed, I do not make a difference between say
and
which shows that the pictures on the faces of the die are arithmetical symbols rather than geometrical figures.
The faces of a cube are numbered to transform it into a die. It is not done
arbitrarily. It is a convention that opposite faces add up to 7. In front of me
on my desk there are two dice. I manipulate them to get the 1 at the top. Then
I turn them to get the 2 in front of me. Where is the 3? For the left die it is at
the right, and for the right die at the left. Thus they are not the same, they
are a different manufacture, printed in different moulds, mirror images of each
other. Well, as cubes they are the same, and not until they were marked with

GEOMETRICAL MAPPINGS

335

different stigmata, did they differ from each other; as dice they are different.
As cubes they could be mapped on each other but by the stigmata the possibilities of mapping were restricted.
This kind of example can be diversified ad lib: A cube is different from the
open box in which it fits; one of the faces has been provided with the structuring
stigma of lacking. A cylinder differs from a tube which is open at one end or
at both of them. A rectangular sheet can be folded in many ways, accordeon-

like, that is alternating to the right and the left, or always in the same sense,
or first threefold in one direction and accordeon-like or otherwise in the other
direction the inventiveness of map and folder producers is unlimited. All
these objects are the same if viewed as rectangles, but as folded figures they
differ by the extra structure, which looks accidental though it can reasonably
be put into a geometric context. Which among those structures are the same and
which are different, is again determined by the existence or non-existence of
mappings respecting the extra structure. It may happen that I am in no way
concerned about the length and width of the rectangle and that all that matters
is the fashion of folding; then it depends on the existence of a mapping that
respects the folding stigmata whether two folding structures are the same or
not. This too is geometry, albeit not the traditional brand, and of a heavily
combinatoric kind.
12.4. Structuring Figures

In our badly formalised and for just this reason practical colloquial language,
plurals of nouns may mean many things. Comparing figures and Distinguishing figures, which was discussed in the preceding sections, means comparing
the one with the other (perhaps within a given stock of figures). Structuring
figures means a plural of structuring a single figure, once this one, and
another time that one maybe in order to state that the first admits other
structures than the second (which is again a way of distinguishing) or that
certain structures are common to both (in which sense they would appear as the
same). Yet as a principle structuring as meant here will be a way of comparing
a figure as it were with itself. Comparing and distinguishing as discussed in the
preceding sections involved mapping. Since this plays also a role in structuring,
the present section is a counterpart to the preceding one.
I cast another glance at the little barrels on my desk. As we know they
are the same. In fact I can dispense with one of them. I move the one that is
left, not by displacing it, but by turning it around its axis. It remains the same
on its place though not in all details. Rather than with another figure it is
compared with itself. In fact to get this done I need not turn it, I can rather
view it from the other side to compare it with itself. Or, transferring the ikonic
and enactive to the mental mode, I can imagine the barrel mapped upon itself
by a rotation or a reflection, while it remains the same. The same because
I have neglected the labels and the print on the labels, which do not belong

336

CHAPTER 12

to the geometrical context, in which my barrels are cylinder-like figures. How

about the screwcap? Does it belong to the geometric context? It prevents
me from turning the barrel upside down a stigma as I called it. But I have
got on my desk also a long plastic tube closed at both sides with a screwcap,
which I can turn upside down on its place without changing it. Of course instead
of this I can look at it from above and below to get the same result and finally
I can imagine it done, say by reflecting the tube at a transverse plane cutting
it in halves, in order to state that the mirror image is the same.
Structuring figures as it happened here, is done by mapping the figure on
itself properly said, we just look for structure phenomena expressed by
mappings. A line can be articulated to become a garland of beads, a closed curve
to become a chain of beads the look jumping from the one bead to the next
produces something like a mapping. The more striking structuring feature of a
figure in mirror symmetry, often coarse and only global, sometimes restricted to
parts of a figure or configuration, sometimes disturbed by stigmata of geometric
or non-geometric character. Leaves, simple or composite ones, often display
a plan of symmetry, with sometimes systematic deviations, but the assignment

to complete half a leaf is by young children well understood as a construction

of symmetry. Star-fish display a five-fold symmetry, flowers according to
their species a three-, four-, five-, six-fold symmetry and a circular one, which
is not as precise as that of crystals or technical products, though its plan cannot
be mistaken. Rotational symmetry is approached by trunks of trees, pupils,
all kind of fruit, though as far as precision is concerned, the products of the
potters wheel and the lathe are more perfect: a vase, a cup (the symmetry of
which can be disturbed by a handle), a tube, a drill hole. The screw symmetry
of winding plants is surpassed by that of the screw-thread and the winding
stairs. For all that, we have the habit of putting the less perfect realisations of
symmetry into a geometrical context. Why? Because these sometimes global,
sometimes coarse, sometimes only partial mappings of the figure on itself
make it easier to grasp the figure structurally.
In front of me on my desk a button is lying, invariant under rotations, though
as becomes a button, pierced by four holes that is stigmata reducing the
rotation group of the button to a four-cyclic group. My desk-lamp is mirror
symmetric, with rotation invariant parts, such as the switch and the bulb no,
it is a bulb with a screw fitting, which in itself suggests screw invariance.
Suggests that is what I said. Let us be cautious. It is only a finite piece of
fitting. At some place the screw motion stops or the lamp drops out of the
fitting. The mappings I discussed so far, are lacking something. They serve to
compare, to distinguish, to structure figures and they are infected by the imperfections of these figures.
12.5. Comparing by Mapping

The stress is going to be shifted, from the objects that are being compared, to

GEOMETRICAL MAPPINGS

337

the act of comparing, in particular by means of mapping. Mapping indicates

more sharply what is meant than map would do the act rather than the
result. Mapping means a process while map suggests the result of the
process.
Mapping geometrically is kind of concretely or mentally superimposing.
It is easier with one- and two-dimensional figures than with three-dimensional
ones, where the act will in general be restricted to the mental reality. As far
as the objects to be compared are concrete, the possibility of comparing is
often guaranteed by the knowledge about the origin of the objects having
arisen from the same mould, a mass product, a mechanical or optical copy,
DIN format.
For rectangles only one term is available, if the limit case of the square and

vague expressions such as big, small, narrow, long are disregarded.

Triangles, however, exist and are named in a great variety: equilateral, isosceles,
right, acute, obtuse and some of these properties combined. The size does
not matter in this case; all of them are characteristics of similarity. Several times
I claimed that congruence and similarity of figures are mental objects formed
at an early age. Mathematically similarity can be defined by means of the invariance of ratios of distances, but this is an a posteriori statement, the result
of a logical analysis. It is a hard thing to establish similarity of two rectangles

at a glance the ratio of the sides is not perceived unless the divergence is big,
as long as ones attention is not drawn to it or one has to copy something.
Triangles are quite different: equilateral ones differ strikingly from right ones,
obtuse triangles from acute ones. If similarities are perceived and established,
factors other than distance play a part, in particular the angles are an important
factor. Unfortunately a rectangle has only right angles, which makes rectangles
strongly resemble each other. As soon as the diagonals are added, the image
changes; more visual structure arises, in particular because of the new, non-right
angles, which makes the recognition of similarity at a glance easier. The same
effect can be obtained by other means. Rectangles as pictures representing
something are with greater ease tested on similarity or established as similar than
mere frames (Figures 138140), and according to the increase of detail the
impression of similarity and dissimilarity gets stronger.

338

CHAPTER 12

12.6. Mental Constitution of Mappings

However, what matters now is not comparing figures but constituting mappings
mentally, a process that again precedes that of concept formation. The child
becomes acquainted earlier with images of objects and knows how to relate

the object to its image,

two images of the same object to each other.
They are

global relations,
headhead, neckneck, bellybelly, footfoot, doordoor,
roofroof, which are refined to

pointpoint-wise relations
connections
between different objects
or

between different images of an object

or

symmetries within an object or an image.

The connections can be given

to be discovered
or proposed

to be constructed,
such as if an object or an image is to be copied; or they can partially be given or
proposed in order
to be completed.
The connections can be fixed

in all detail,

or indicated or suggested
by some characters,

which is the usual procedure in geometric representations. It will depend on the

actual situation, how and how far these characters, indications, and suggestions
in fact determine the connection. Figure 141 explicitly relates the six vertices
of two planar hexagons to each other, and at the same time suggests an extension

GEOMETRICAL MAPPINGS

339

to the joining sides, which by preference would be linear, or rather affine, for
instance by starting with relating midpoints of line-segments to midpoints, an
affine interpolation.
Extending this mapping to the interior of the hexagons is less obvious. (In the
case of convex figures it could be done in a piecewise affine way.) As regards
the two circles of Figure 142, first of all the equally numbered points are related

to each other; the drawing suggests extension to the arcs, by preference linearly
with respect to the arc length. Two less regular arcs (Figure 143) can suggest

many things beyond the relation between the endpoints, linear extension according to the length, or somehow topological. The net of the cube (Figure 144)

340

CHAPTER 12

asks of course to be mapped on the cube by identification of equally numbered

vertices and edges. However (Figure 145), a cube can also be mapped on three

projection planes while the numbers at the projections correspond to vertices

of the cube and relations between the projections are created by means of
the number identification.
It is a mathematical fact that an affine mapping, say, of the plane on itself

is determined if of three non-collinear points a, b, c the (non-collinear) images

a, b, c are given (Figure 146). This can be established in various ways the

easiest being a barycentric construction: each point of the plane can be understood as the centre of gravity of three (not necessarily positive) masses at a, b, c,
which are transferred to a, b, c, respectively, in order there to yield another
centre of gravity. A more visual geometric method is that of constructing a net
from a, b, c and a similar one from a, b, c that covers the whole plane, and to
refine it gradually (Figure 147). Space can be dealt with in a similar way by
starting with four non-coplanar points.

GEOMETRICAL MAPPINGS

341

If the original triangles (tetrahedra) were mapped by congruence or similarity,

the mapping affinely extended to the whole plane (or space) is again a congruence
or similarity.
From Euclid onwards up to school geometry as taught until about the middle
of our century, statements about congruence and similarity were restricted to
figures in the plane or space often even to triangles, since statements about
other figures could be reduced to the corresponding ones about triangles.
Extending such congruences or similarities to the whole plane or space is an
idea that came rather late in history, properly said not until the importance of
other mappings affine and projective ones had been discovered. Projecting
entire pictures already bears witness to a broader view of the object to be
mapped than restricting it to a single figure. Regular planar pictures and in
particular symmetries as encountered from olden times in ornaments, suggest
something like mappings of the whole plane onto itself by means of the congruent displacement of the picture or by means of the interpretation of symmetry
as reflection.
It looks as though it requires a long step from comparing figures by mapping
the one on the other to mappings of the whole plane (or space), and indeed a
long step it was and still is for people educated in traditional geometry, as appears
from numerous mathematical and didactical mistakes in modern textbooks.
The habit of reducing the congruence of figures to congruence of triangles
was a historically determined method to guarantee the validity of statements
and proofs. The quadrilaterals of Figure 148 are flexibly equivalent, since

corresponding sides are congruent. However they are not congruent as quadrilaterals are understood by tradition. According to the traditional view, though
not at the heart of the matter, it is the difference of angles that disturbs the
congruence. It is more to the point that the congruent mapping of the union
of the line-segments does not extend as a congruence to the plane. An even
more drastic example is Figure 149, a parallelogram, which strongly suggests

342

CHAPTER 12

a congruence
that is a mirror symmetry a frequent mistake,
which is easily understood. This figure betrays a visually structural symmetry
which can be extended affinely, but not congruently, to the plane. Traditionally
one would explain this phenomenon by the acute and obtuse angles at a and a,
respectively, though it is more to the point to compare the diagonals, the short
and the long one.
The classical method of discovering congruences is to extend the figures to
nets of triangles an extension of the mapping from a given figure to a larger
one. Why has the ultimate step of extending the mapping to the largest possible
figure plane or space been taken so late in history? Though a phenomenological approach is not tied to any historical track, it should include an
explanation of the possibly diverging historical track. This analysis will
occupy us now.
Undoubtedly Greek geometers were familiar with symmetries; they did not
spurn symmetry as a heuristic tool, and even as appears from pre-euclidean
traditions as a deductive argument. As late as the euclidean revision of the
Elements, symmetries disappeared and for a theorem such as the equality of
the basis angles of the isosceles triangle, a complicated proof by means of
congruence properties was contrived. This artificial system of congruent triangles
has been canonised in the traditional school geometry. We cannot tell for sure
what moved Euclid to reorganise geometry that artificially, and his successors
to stick to this form of organisation, but we can guess that one argument, if not
the decisive one, against mappings might have been that of purity of method:
motions is a subject of mechanics rather than of geometry.
Of course this argument is not to the point. This insight, however, that it is
not to the point, requires a course of thought that, historically viewed, was not
at all obvious. I already pointed out that as mappings the projections precede
the motions, and that even affine and projective mappings drew the attention
before motions did so.
Motion is first of all something that occurs

to an object,
within space (or plane),
within time.
In order to view motions the way mathematicians are now used to, three steps
must be taken:

from the limited object to the total space (plane),

from within space (plane) to on space (plane),
from within time to at one blow.

In particular the last step is important: the mapping must be viewed as a relation
between the initial and the final state; there is no in between, or as far as it
exists, it must be neglected. It is quite usual to visualise a translation by a field
of equal and equally directed arrows (vectors) (see Figure 150); the misleading

GEOMETRICAL MAPPINGS

343

feature in this representation is the suggestion that the points move along these
arrows. A system of curved arrows, with the same heads and tails (Figure 151)
is as well to the point or is not.

(Of course a movement taking place in time can also be represented mathematically, though not as a mapping of the space R on itself: one has f to depend
on a parameter t such that is a mapping of the cartesian product of R and
a time interval T,

but this does not matter now.)

The didactical consequences are obvious:

either an approach to mappings where the dependence on time is a priori

lacking, or is not to the point, or can easily be eliminated,

or with any other approach,

stressing the irrelevance of the in between.
There is a danger lurking in the second approach. Stressing the irrelevance of
something includes drawing the attention to it in order to warn against it. In
the long run mappings that invite embedding into the course of time, such as
translations and rotations, cannot be avoided in any reasonable way. What one

can do about it, is to watch that they do enter in a way and at a moment where
the irrelevance of the in between need not be stressed any more, because it
has already become obvious.
The mental object mapping is constituted along the way of constituting
special mappings as mental objects. If this occurs by means of mappings that
are by their very nature in one blow, one may expect that the principle of one
blow becomes itself one of the constituting features of the mental object mapping, so that any warning against the misleading in between can be dispensed
with, or can at least be pronounced with less stress.

12.7. Mappings at One Blow

The most natural examples are
projections, that is mappings by means of light sources at a finite or
infinite distance,

344

CHAPTER 12

which produce
light and shadow images.
They include a wealth of geometric information, which may not be side-stepped.
It is a disadvantage that they are mappings of the space on a plane, or of one
plane on another, rather than of space on itself, or a plane on itself. Another

disadvantage is that the connection between image and original is too concrete
lightray, thread which may block mentalising.
A second kind of mapping at one blow
endomorphisms, isomorphisms, and automorphisms of discrete structures,
such as

grids and other patterns in 1, 2, 3 dimensions,

the ordered set of natural numbers,

the ordered set of integers,

a square lattice in the plane,

a cubic lattice in space,

a hexagonal lattice in the plane,
ornaments,
crystal structures,
mosaics,
thus structures that structure also the straight line, the plane, the space around
them, and determine structure conserving mappings, which extend to the surrounding medium. An example is shown by the rectangular pattern of Figure

152, which possesses a set of automorphisms, each determined by one lattice

point and its image a mapping of the plane performed at one blow.

Without relying on an extra discrete structure one gets mappings of the

straight line, the plane, the space on itself, performed at one blow, by means of
reflections,
in particular if the relation between point and image is performed by means of
a mirror. Original and image are simultaneously present, without any lag, at one
blow. The use of the mirror, however, as a producer of a mapping, should be

GEOMETRICAL MAPPINGS

345

fully exploited. One is easily seduced to place the mirror only vertically, which
is an unnecessary restriction. On the other hand the actual use of the mirror, if
maintained too long and too rigidly, can block the development of the mental
operation and the mental object reflection.
Reflection is first of all an
operation in space,
which relates each point to its mirror image with respect to a planar mirror. By
reduction to a table or the drawing plane one gets the reflection as an
operation in the plane,
with a straight line in the plane as its axis. Besides these, one should not neglect
point reflections in the plane or space
the first realisable by a half turn, but nevertheless a mapping at one blow
and the
spatial reflection at a line
again realisable as a half turn, around this same line as its axis.
All these reflections are involutions, that is mappings relating two points
mutually to each other as original and image, which reinforces their one blow
character.
Less pronounced, though still present is this one blow character in the case
of the so-called enlargements, and this is certainly so if the enlargement factor
is negative: enlargement with centre O and factor maps the point P on P
such that (in vector notation)
In traditional school mathematics this was the only admitted mapping again
an argument in favour of the necessary one blow character if mappings are to
be constituted as mental objects. Unfortunately in that instruction enlargements
were only a tool to study properties of figures and configurations rather than an
object of study as such. An example (Figure 153): The enlargement by the
factor
from the barycentre of the triangle ABC into the triangle ABC,
which served to reveal a large number of beautiful properties.
Even by the mappings at one blow of space or plane onto itself, it is first
of all figures and configurations which are being related to each other;the bare
plane and bare space are totally inappropriate opportunities to step in. But even
in this case reflections are privileged. A translation, a rotation, an affine mapping
requires in general, in order to be visualised, the production of a figure and
its image. For a point, line, or plane reflection it suffices to give the point, line,
or plane, which serve as mirrors at least if at the same time the mapping is
explicitly supposed to be a reflection. The converse problem, however, requires
a change of perspective: given a figure and its reflection image to find out the

346

CHAPTER 12

mirror that causes the mapping. This change of perspective can be hampered
by misleading cues. A well-known error: in an oblique parallelogram the lines

joining the midpoints of opposite sides are being considered as symmetry axes.

12.89. From a Mental Operation to a Mental Object and to Concept Formation

12.8. Of a figure or configuration I can make a visual or haptic image; of a

mapping I cannot, at least not directly. Except for the graph of a function, there
is in general no visual or haptic means to represent a mapping in its totality. All
that can be done is to relate a number of points or marks to each other and
then to extrapolate. The case of reflection is particularly favorable: one datum,
the quite simple set of invariant points suffices. In order to extrapolate from
such data, certain general rules are required. It is a situation quite similar to that
of acquiring the arithmetical operations: at each stage the learner has certain

ready data at his disposal, from which new relations are formed by means of
certain rules which initially are handled by insight in order to be algorithmised
in the course of learning.
Rules, as meant here, can initially be implicit in order to gradually be made
explicit, and as far as the available linguistic means allow, be formalised. The
learner is in particular confronted with the need to recognise a mapping as a
mental object and to describe it if wrong constructions ask for clearing up and
removing misunderstandings such as is the case if right and oblique
reflections are to be distinguished. Then explanations are required like
the image is as far behind the mirror as the object is in front of it,
or
the image is found by dropping a perpendicular on the mirror and setting

out behind the mirror a piece as long as that in front,

GEOMETRICAL MAPPINGS

347

or
object and image are lying on opposite sides of the mirror at the same
distance and straight across each other,
or
the line-segment joining a point and its image is perpendicularly halved by
the mirror,
or, with a change of perspective,

the mirror is the perpendicular bisector of the line-segment joining points

related to each other
or
is a reflection at the line (plane) S if and only if S is the perpendicular
for each

bisector (plane) of

The change of perspective also delivers as a byproduct an indication to answer

the question

what is the place of the mirror mapping p on q?

if the change of perspective and the reformulation of the description of reflection has not even been provoked by this question.
According to the progress of awareness about the characteristic properties
(of a reflection), concept formation progresses. But even without verbalising
what is becoming conscious about the properties of reflection, the character
of mental operation or mental object can be maintained, not only when mappings (reflections in the present case)

are being constructed from data,

or

are being recognised within data,

but also
when inverting them,
when composing them,
when stating that composition in a different order can yield different
results.
Starting from reflections (on lines in the plane or planes in space) and composing them, one gets
translations and rotations

348

CHAPTER 12

on the level of mental operations, or if subjected to closer analysis, of mental

objects. Such as reflections by a mirror, so are translations given by
one translation arrow as a model of all of them (original-image),

and the model rule

to move all points the same way,

sharpened to the prescription of

parallel moves of the translation arrow,
and refined by a prescription as to how to act with points lying in the extension
of the translation arrow.
In order to constitute rotations as mental objects two data are required,
the rotation centre (axis)

and a second, which may be

a model out of all pairs of points object-image
or (Figure 154)
a model out of all pairs of rays (half-planes) object-image

with the prescription that under a rotation

all points turn as strongly.

The precision required for as strongly cannot be obtained in the case of

rotations as readily as it can for translations. The tool needed here the substitute for parallel displacement of the translation arrow is the
turning angle,
that is the oriented one: all turns through the same angle.
In order to progress from here to the group of congruence transformations,

complete concept formation is required. In order to understand that the product

of two rotations is again a rotation (or a translation), and why it is necessary to
admit slide reflections and screw transformations, sophisticated reasonings are
required, which presuppose concept formation at a high level.

GEOMETRICAL MAPPINGS

349

Composing reflections (at a line in plane, or at a plane in space) is an activity

where mappings function at least as mental objects if not as conceptual constructs. It requires insight at a high level to pay attention to the fact that
invariance of distance
is inherited from the reflections by their composition results in general. It is

a change of perspective, characterising the activity of axiomatising, to ask for

all selfmappings of the plane (space) that leave distance invariant,

and a sophisticated mathematical organisation is required for the activity of
exhausting this sort of mapping by classification.
Similar considerations could be made for affine mappings (of plane or space).
As we mentioned earlier they can be fixed by and constructed from a model
triple of three non-collinear (quadruple of non-planar) points and their images;
it is quite a simple construction by extrapolation to any point. This constructive
process of taking possession of the affine mappings can evolve entirely on the
level of mental operation. For the conceptualisation, however, an axiomatic
change of perspective is required: an affine mapping defined by the preservation
of rectilinearity (and parallelism).
12.9. As regards less structured mappings concept formation is even further
away from mentalisation. If two simple closed plane curves are visually given,
a topological mapping of the one on the other intrudes itself with the same
visual force upon the viewer, and even a topological mapping of the interiors
upon each other. Without much ado one would replace the word visually
in this statement by mentally, and even by conceptually. However, in
order to pass from the mental to the conceptual closed curve, one has to prepare
a more or less explicit definition, which in turn presupposes the concept continuity of a mapping, thus certainly presupposing the concept of mapping,
of mapping a line-segment or a circle into the plane, and such a mapping as the
object of investigation. If, however, such conceptual definition of a simple
closed curve is available, the topological equivalence of two of them is a conceptual triviality and the existence of a topological mapping of the one on the
other is an object of an entirely conceptual status. Then one is staying on a
conceptual level where topological mappings (or continuous mappings in general)
can be concepts. Only in this state can the question of whether the interiors of
two simple, closed, planar curves really are topologically equivalent, as they
look visually, be asked on the conceptual level.
The gulf between topological and geometrical mappings congruences,
similarities, affinities, projectivities is obviously caused by the phenomenon
of continuity, with its unusually big distance between mental object and concept. The elementary mappings, too, are continuous, but this continuity, if it
ever draws attention to itself, does not ask to be made explicit and certainly

350

CHAPTER 12

not to be conceptualised. It is as it were, accidental, whereas for continuous

mappings it is the essence.
Nevertheless there is a non-negligible bridge between elementary geometric
and continuous mappings I mean the piecewise affine ones. Graphs can be
distinguished according to their combinatoric structure as equivalent or nonequivalent; if the graphs are built with rectilinear edges, then a combinatoric
equivalence can be embodied piecewise affinely. Two simple closed polygons
can be recognised visually and mentally as topologically equivalent, and this
equivalence can be established even constructively, by mapping constituting
line-segments on each other, which provides the mental topological equivalence
with a primitive operational status, that excedes the purely visual one. This
procedure can be extended from the polygons to their interior. By affine
means it can be understood that the interiors of two triangles are topologically
equivalent, and this can happen on a visual, mental, and conceptual level. A
quadrilateral can be split into triangles, and the one of them can be pulled to
form a larger triangle together with the first (Figure 155) a procedure used to
transform polygons into triangles of the same area.

If one agrees to restrict oneself to simple closed plane curves, built from
line-segments (simple closed polygons) it is not hard to pass from the visual
via the mental to the conceptual mode. It can be realised while still avoiding
continuity, which means however that the one element is lacking which is
required for a genuinely topological context. So as nice as it might be, this
procedure does not help to bring the concept formation of continuous and
topological mappings closer.

CHAPTER 13

MEASURING BY MEANS OF GEOMETRY

13.1. Geometry

Etymologically geometry means measuring the earth, geodesy. By the need, after
each inundation of the Nile, for restituting everybody the land he was entitled
to, Herodot explains the allegedly Egyptian origin of geometry. In every somewhat advanced society the need for measuring land is felt: as a basis for levying
taxes and rent, for delimiting and dividing land, distributing water and seed.
In order to plan buildings and to establish their capacity, mensuration and
geometry is required, as it is for the construction of roads, canals, tunnels,
temples, pyramids, fortifications, which as geometrical figures are designed
according to geometrical principles. Measuring distances by days journeys is
the basis of early cartography, measuring angles that of the measuring survey
of the sky. This is not the place to explain how these methods were ever refined
to trigonometry, spherical trigonometry, and differential geometry, in theories
which were to explain the structure and shape of the universe.
As I stressed before, there exists also non-measuring geometry. In affine
geometry line-segments must be parallel in order to be compared metrically.

Projective geometry works without measuring though the cross ratio can serve
as a substitute. In topology and combinatorics there is nothing left that reminds
one of measuring.
13.2. Measuring Length A long Straight Lines
Measuring length was already discussed at the end of Chapter I though there the
geometrical context was poor. In fact it was announced: Geometrical insight

leads to refined methods of measuring distance. Some of them are possible early.
We will come back to this point.
Not only was the geometric context of measuring length insufficient. Length
was a function of long objects, invariant under displacement (geometrically:
congruence mappings), flexions, breakmake transformations. Long objects
could be broken into pieces and again composed to form new long objects, with
an additive behaviour of length. Distances emerged as length of possibly
mental long objects: roads that could be straight or curved. The long objects
serving in primitive measuring rulers, straight ropes suggest straight lines,
but straight lines and line-segments were hardly discussed in Chapter 1. Meanwhile much attention has been paid to the straight line (see Section 11.6). The
geometric context in Chapter 1 is, however, insufficiently explicit; a closer
analysis has meanwhile taken place, the essentials of which are resumed here,
and which is continued now.
351

352

CHAPTER 13

Lengths are being measured rectilinearly; as far as curved objects are concerned, measuring is reduced to that of rectilinear ones. Distances are understood
by birds eye view, unless it is otherwise explained, or clear from the context.
In order to determine the distance along a broken path, the lengths of
the pieces are added, a curve is approached by broken paths or otherwise
straightened out. This rectilinear measuring first of all seems instrumentally
conditioned: the measuring instruments rulers, straightened ropes are
embodiments of straight line-segments, indeed. I said seems instrumentally
conditioned because I am not sure whether this is also true genetically. The
first spontaneous measuring acts of children I observed, were not instrumental,
but pacing and spanning (between thumb and forefinger), or by means of
the palm, or with parallel fingers in sand, or parallel hands at breast height,
thereby not using any instrument that would suggest rectilinearity. Nevertheless
rectilinearity plays a part even here: the steps are taken straight forwards, the
spans are prolongations of each other at least this is the conscious or half
conscious intention. So the straight line is mentally rather than instrumentally
present in such measuring acts.
In Section 11.6 we met the straight line in a large variety of phenomenological
contexts. In each of them the straight line gives occasion to measuring procedures.
A particularly interesting situation is that where the straight line along which
a distance is to be measured, is not present in the data but has still to be constructed mentally, for instance as a straight ahead or as a vision line. Indeed, if
the measuring staff (step, span, ruler, string) is too short and has to be applied
several times in succession, the rectilinearity of the measured mental object
has somehow to be guaranteed by the rectilinearity of extension,

by overlapping
by aiming along a vision line.

In Section 11.6 the straight line also occurred as the shortest line. Is it meaningful to posit that distances are being measured according to shortest lines, or is
this a vicious circle? Indeed, before deciding whether something is a shortest
line, I must know what shorter, hence what distance means. No, it is no
vicious circle. In order to be able to discuss shortest paths, I need not measure
lengths; I only have to compare them with respect to the
order relation
of longer and shorter and this length comparison according to order is first of
all and at an early age actualised by the insight that
the detour is longer,
there is more rope or rail needed to connect along a detour. Or to state it in
terms of the rectilinearity of the ruler: the rectilinear path from a to b (Figure

156) is shorter than the polygonal one composed of line-segments. Mathematically this fact is known as the

M E A S U R I N G BY M E A N S OF G E O M E T R Y

353

inequality of the triangle,

(Figure 157), which purely logically extends to the

inequality of the polygon.

There is, however, even more to say about measuring distances along shortest
paths. The possibly mental tools to measure distance first of all assure
rectilinearity locally. Going straight on, cutting straight through, is primordially
a local procedure; the global line arises from piecing together local segments;
special arrangements are needed to globalise the procedure overlapping or
aiming. The vision line looks like a more global tool, suggested by the imagination, the mental substratum, of the light ray. One notices a certain transitivity:
if for the eye point a covers point b, and point b covers point c, then a does so
with c. The origin of the straight line as a shortest path, however, is
a priori global.
It can be conceived for points which are arbitrarily far away from each other,
and realising it can progress from the local to the global domain with or without
the stretched rope as its dynamic embodiment. The shortest path is shortest
in all its parts, as a tight rope is tight in all its parts, and conversely. These are
two perspectives of shortestness:
from the global to the local,
from the local to the global.

They are of a quite different character. It is trivial that a globally shortest path
is also locally shortest; it simply follows from the additivity of length: if I can
shorten the path somewhere, I can apply this shortening to the whole.
But the converse is not at all trivial. It is not caused by general metric properties but by the specific structure of our space, as can be seen by comparing its
geometry with that of the spherical surface. On the sphere going straight ahead
and tightening strings produced arcs of great circles. The great circles and their
arcs are the straight-ahead-lines of the spherical surface. Small pieces of great
circles are also shortest lines, but this ceases to be true as soon as the arc gets
longer longer than the spherical distance between antipodes. In the plane and
in space the notions of straight-ahead-line and shortest line coincide; on the

354

CHAPTER 13

sphere (and on other surfaces and in spaces with positive curvature) they fall
apart: going straight ahead you reach a point that you could have reached from
the starting point on a shorter path it even happens that going straight ahead
one returns to ones starting point.
In the plane and in space an everywhere locally shortest line is also globally
the shortest line. The great circles on the surface of the sphere are locally,
though not globally, shortest paths a phenomenon that is characteristic
of spaces with a positive curvature. This, indeed, shows that with respect to
shortestness the inference
from the local to the global

is implied by the structure of space in a more profound way than that

from the global to the local.

13.3. Comparison of Length by Special Congruence
If length is measured along straight lines the factual rectilinearity of the line
to be measured is guaranteed by linear production, actualised by means of
overlapping or aiming. Often such procedures do not suffice: the path to be
measured may be inaccessible to these procedures; obstructions can block the
path between the pair of points the distance of which has to be measured. Then
the path to be measured must, at least partially, be replaced by another one
that on geometrical grounds may be supposed to be of the same length. In the
simplest way this happens if for instance when measuring a room a piece
of the path is being replaced by a parallel piece. The geometrical insight behind
this procedure,

translation preserves length,

can be formulated less globally,
in a rectangle, opposite sides are equal,

or more generally,
in a parallelogram, opposite sides are equal.
Geometrical insight in this context means knowledge about such mental
operations or mental objects as translation, rectangle, parallelogram a
knowledge that does not necessarily include geometrical deductions, nor the
conceptuality of these mental objects, not even the knowledge of their conventional names. By numerous things that suggest the shape of a rectangle the
existence of rectangles of arbitrary size
and their
most important properties,

M E A S U R I N G BY M E A N S OF G E O M E T R Y

355

among which
parallelism and equality of opposite sides
are visually and mentally being suggested. Though parallelograms are less frequent
among these concrete things, they can profit from those properties of the

rectangle that characterise them; rather than as concrete things they play an
important part as representatives of rectangles in parallel perspective drawings.
A change of perspective: the construction of rectangles and parallelograms by
means of long things which may be supposed to be equal because they are of the
same manufacture:
parallelograms made from straw, sticks and suchlike things,
rectangles by vertically planted poles,
larger parallelograms built from smaller ones of the same manufacture.

Rectangles and parallelograms can only serve for translating line-segments;

so they can guarantee length equality only for equally directed line-segments.
The most primitive means to ensure length equality of unequally directed linesegments actually or mentally is symmetry. By numerous symmetric things

symmetry as a mental operation

is suggested at an early age, and so are
the most important properties of symmetric figures,
in particular, the

length equality of corresponding line-segments.

Likewise

length equality of different directed line-segments

is being suggested by
isosceles, in particular equilateral, triangles,
squares, and more generally, rhombuses,
and, even more generally, by

regular figures.

A more global means to suggest equal lengths, is

patterns, such as mosaics and wall paper,
where
transitivity of length comparison
is an extra mental factor.

Comparing length by means of translation and symmetry is of course not

356

CHAPTER 13

restricted to line-segments. Curved arcs can be recognised as equally long by the

same procedures.
The figure that produces at one stroke line-segments of the same length in all
direct ions, is
the circle.

By means of a ruler or any other long object that is provided with marks, linesegments can be transferred to other places while length is preserved. The same
can be done with a span of fingers, hand, arms, or with the points of a pair of

compasses, although the compasses can perform more than this: producing
the whole set of points at the same distance of a given point, and that for all

distances within their scope.

13.4. Measuring Lengths by Congruence Properties

It is told about Thales that he had measured inaccessible distances (the distance
of a ship off the coast, the height of pyramids) by geometrical arguments. For
this reason it is the habit in some countries to call a certain theorem on similar
triangles (Figure 158) Thales theorem. This terminology cannot, historically,

be justified. In the most trustworthy source Thales method is linked to the

congruence theorem one side and two angles; according to this method,
which has explicitly been recorded in ancient sources, the inaccessible distance
AB (coast to ship) is replaced by the accessible distance CE (on the land) (Figure

159). The height of a pyramid is said to have been measured by Thales by means

M E A S U R I N G BY M E A N S OF G E O M E T R Y

357

of its shadow at the moment of the day when a verticle pole and its shadow were
equally long, thus with no appeal to similarity.
Both indirect procedures to measure length can- also be understood by symmetry arguments, and the same holds if the equality of the diagonals of a
rectangle is to be recognised. In many cases like this congruence theorems are
too heavy guns.
A congruence mapping f of the plane is almost uniquely determined if the
images a, b of two points a, b, respectively, are prescribed (of course such that
. The image
of an arbitrary point c then is
subjected to the equations

which have (at most) two solutions. The triangles abc and abc are congruent
by means of f (congruence theorem, three sides). Congruence, however, also
extends to the angles, which means that in the pair of triangles abc, abc the
corresponding angles are equal. After a change of perspective one can localise
the point c with respect to a and b (and correspondingly c with respect to a
and b) also by means of the angles at the side ab (and ab). If the line-segment
ab is given together with the angles under which the point c is seen from a and
b with respect to the line ab, the position of c is determined, but for a symmetry,

and in particular the distance of c from a and b.

For the practice of distance measuring the congruence properties are only
significant if related to similarity the didactical consequence will be dealt
with later on. It is, however, a matter of principle to know that a triangle is
determined by

three sides,
two sides and the enclosed angle
one side and two angles
as to shape and size,
and if two vertices are given by position, also

as to its position,

albeit but for a symmetry.

With this principle one can at an early age answer the question of how inaccessible distances the height of a tower, the distance of Moon and Sun
(with a terrestrial base line), the distance of a star (with a diameter of the
terrestrial orbit as a base line) can be measured.
A particularly important case is that of the right triangle, thanks to various
applications of the Pythagorean theorem.

358

CHAPTER 13

13.5. Measuring Lengths by Similarity Properties

In an explication of how distances in the universe are measured, a seven-year old accepts
without any hesitation a model drawing (Figure 160): Suppose this is 1 km and one

measures these two angles, then one knows how far away that point is lying. Unfortunately
the explainer was at a loss to answer the next question: Yet how can one figure out the
distance?

Several times I drew attention to the fact that from the picture books onwards
representation by similarity is accepted as a most natural thing, and for this
reason the representation by similarity is the practical realisation of what I called

a matter of principle in the preceding section. For the complete realisation

more is required: trigonometrical functions and tables in order to perform
calculations.
Similarity is readily accepted as a means of faithful representation if maps
and ground-plans are to be read. The scale explains and a drawing beside the
scale shows how many metres or kilometres in the reality correspond to a
centimetre on the map. It requires some insight into decimal fractions and the
metric system to draw the desired conclusions from these data. Or, the other
way round: insight into decimal fractions and the metric system can be acquired
by interpreting ground-plans and maps. Besides lengths the map also provides
information about angles, which in the open field is a way possibly aided by
a compass to take ones bearings.
Didactically instructive cases of measuring inaccessible or less accessible
lengths and distances is for instance

determining the height of a tower from certain data,

determining the spatial diagonal of a rectangular prism,
determining the height of a pyramid from its network.
Insight with no intervention of geometrical theory suffices to deal with such
problems by means of drawings in scale; from a didactical point of view they are
entitled to be solved in this way at a stage where no algebraic or trigonometrical
means are available. Surveyors have been satisfied with such methods for quite
a time. The trigonometrical functions were invented by the ancient astronomers
to solve spherical triangles; not until Snellius were they systematically used to
solve terrestrial triangles.
What is said in the preceding sections applies also to measuring polygonal and
curved distances and perimeters of figures and lengths of curves, which again

M E A S U R I N G BY M E A N S OF G E O M E T R Y

359

can be determined from reproductions on scale. Applying similarity gets to be

a matter of principle if figures are compared with each other:

all equilateral triangles,

all squares,
all regular n-gons,
all circles
are mutually similar. Similar figures with subdivisions that match each other,
produce
patterns

that are again similar. By similarity

division of length (of a path) in a certain ratio

is carried over to its image.

13.6. Measuring Lengths by Euclidicity

As a summary of Sections 13.35 I state a number of geometrical insights and
abilities that serve measuring of lengths:
the insight that certain lengths and angles are determined directly or indirectly by others,

the ability to use them in a productive way,

the faithful reproduction of lengths from reality on ground-plans, models,
networks and so on,
the interpretation of such reproductions,
the insight that the faithful reproduction of the internal ratios is equivalent
to the reproduction in scale (the constant scale factor),
the interpretation of the reproduction in scale as a similarity transformation,
the interpretation of similarity transformations as reproductions in scale,
the insight into the behaviour of lengths in reproductions in scale,
the insight into the behaviour of angles in reproductions in scale.
13.713.22. MEASURING ANGLES

13.7. The Greek equivalent of measuring angles is goniometry. This, however,

is not what is meant here. We rather mean it in an elementary way, with or
without an instrument made to measure angles. One would like to say that such
knowledge should precede the use of trigonometrical functions. Nevertheless it
is a fact in teaching practice that the mastery of trigonometrical functions and

360

CHAPTER 13

the ability to use their tables does not necessarily include any insight into what
is an angle and how it is measured.
Earlier on we have signalled various aspects of the angle (inclination of directions of straight lines; the part of the plane enclosed by half-lines, oriented
and non-oriented angles). Here we will stress measuring angles, and then the
primordial thing is to make clear to the learner what is measured and according
to which criteria he has to construct something of a prescribed measure.
13.8. The fundamental distinguishing feature in measuring lengths and angles is
the availability of a natural unit in the case of angles: the full angle, or if you
like it better, the stretched angle, or the right angle a full turn, a half turn,
a quarter of a turn. This natural unit is even the most natural access to the
mental object: the angle subjected to being measured at the same time a not
unusual approach to fractions. The most natural access this does not mean
the most familiar, the didactically most appropriate, and certainly not a unique
access.
First of all, it is a didactical misunderstanding to believe that measuring
anything should start with constituting, or even bringing about, a measure
anyway the existence of a logically natural unit in the case of angles, in contradistinction to that of lengths, is no argument to excuse it. There is a stage

preceding the use of measures, natural or conventional ones, in which the need
for measures can be developed. Acquaintance with length measures and instruments to measure lengths is preceded by a developmental stage, starting in the
early years, where the child becomes familiar with length and comparing lengths.
Didactics of angles starts differently, as a consequence of the fact that angles,
in contradistinction to lengths, are being introduced and made explicit in an
already heavily mathematised context. The angles of a room are without delay
used as models for geometrical figures, to wit for the angles of planar figures in
pictures and drawings, and very soon or even immediately for the bare angle,
represented by two sides and a little arc between them. This bare angle can
be attractively clothed by tart and clock dial divisions an approach that
is didactically sound and easily elaborated, though too narrow to be restricted
to these concretisations.
At an early age children start showing comparing and even measuring lengths
by means of the spaces between palms or fingertips or using strings or sticks
as measuring tools. Almost no didactic attention has been paid to the corresponding activity in the case of angles, probably because angles are too late
in instruction for such childish work. Yet it may be doubted whether it would
really be that childish, in view of the lack of understanding which even adults
may show with respect to magnitude of angles.
In the same way one can have children indicate with their hands how large,
how wide, how high a thing is, one can ask them to use their limbs to show or
to imitate how big is the inclination of a roof or a shed, how steep a ladder or
a slope is, how a ball is scattered, how acute or obtuse a crossing is, how far

M E A S U R I N G BY M E A N S OF G E O M E T R Y

361

a door is opened, that is to show or to imitate it by spreading their fingers, arms

or legs or by flexing the elbow.
In order to compare angles, one can use materials and tools, cardboard angles,
pen-knives, scissors, compasses and certainly also the clock-hands and wedges
of cake. Activities in or towards the open air can prevent a premature mathematisation of measuring angles. Whoever points to the window of a building, to
the top of a tower, to a bird in the air, to a star in the sky, defines an angle with

respect to the horizontal plane; at a certain moment the need will be felt to
view not only the shown object but also the slope of the indicating line and to
make its magnitude explicit. As one turns the head of a child that looks the
wrong way, into the right direction, as at a more advanced stage one has ones
finger pointing to one object describe an arc to point to another one, there may
be many opportunities to make the change of direction and its magnitude
explicit.
The examples I just mentioned are predominantly found in what I called the
topographical context. It is not that strange, or is it? In an advanced, more
technical stage of topography (think of surveying, cartography, cosmography),
angle is the most important object for measuring, much more comprehensive
than measuring lengths. Would it not be recommendable to pay much more
attention in the primitive topographical context we are didactically concerned
with, to measuring angles albeit in a primitive, more qualitative way?
At this stage measuring angles is perhaps a less practical concern than measuring lengths. In the practice measuring angles is rather an auxiliary means to
determine lengths. This possibility as such is created by the geometrical insight
discussed in Sections 13.4 5 a theoretical insight that can be present or made
conscious early.
The 7-year old about whom we reported in the beginning of Section 13.5 had not the

slightest difficulty with the term angle. He even described though indistinctly an
instrument to measure angles and finally brought a book to show the picture of a sextant.

13.9. Quite the opposite of the view I explained in the last section on measuring
angles as a developmental phenomenon is Piagets as it appears in an earlier
quoted work.* The chapter on measuring angles starts with and is almost
exhausted by the following test.
The subject is shown a drawing (see Figure 161) of two supplementary angles ADC, CDB,

and is asked to make another drawing exactly similar. He is not permitted to look at the
model while he is drawing, but he may study and measure it as often as he wishes while
not actually engaged on his own drawing. This requirement is met quite simply by having
the model behind the subject. The latter is provided with rulers, strips of paper, string,
cardboard triangles, compasses, etc., all of which may be used in measuring.

* J. Piaget, B. Inhelder, and A. Szeminska, La gomtrie spontane de lenfant, Paris,

1948, Chapter VII. Our quotation follows the English translation, The Childs Conception

of Geometry, London, 1960.

362

CHAPTER 13

The first superficial impression of the reader might be: a clever idea. But after
studying the behaviour of the subject, one notices that up to the highest levels
the problem is not at all interpreted as one of measuring angles, and indeed this
result had been intended by the authors, who confessed later on*
indeed the figure was shown deliberately so as to permit of its being considered as a problem
of angular measurement while not imposing that solution; we therefore rejected a figure
consisting of one angle in favour of two supplementary angles.

The first impression of how clever is very likely to be followed by a second

too clever. Only the oldest subjects (1011-year olds) can somehow bring
it off, essentially by doing something with the perpendicular distance of C
from AB. It has nothing to do with insight into angles nor can it be interpreted
as such in spite of what is claimed by the authors. Copying such a drawing
is a geometrical construction that requires a large amount of geometrical experience and instruction. Relating the task to angles and measuring angles

would bear witness to such a degree of geometric insight and skill that can hardly
be expected at the ages under consideration and should not matter at all in such
an investigation.
If one analyses the formulation of the text more closely, the first question
that arises is whether the letters A, B, C, D belong to the test data or have been
added later on. The next question is what the words and so on mean in the
list of the material made available to the subject. Probably nothing. In any case
no material from the phenomenological and didactical context of the concept
of angle no scissors to cut out something, no hinging pair of sticks, no paper
to be folded or to be used to copy. The pair of compasses the authors do mention
sounds like an acrimonous irony the great majority of subjects will never
before have used this instrument or be able to use it, and in fact nobody tries
it. The experimenters have succeeded excellently in concealing what they
had the intention to test, but at the same time they blocked the path to any
knowledge whatever about the mental object of angle and the mental measuring
of angles. In Section 13.8 I sketched the kind of experiments with which to
begin such investigations.
I have dealt with this test circumstantially because it reveals in a particularly
* The English is more explicit at this point than the original.

M E A S U R I N G BY M E A N S OF GEOMETRY

363

drastic way a certain weakness of Piagets method: in order to test insight

in a certain idea, the test itself is chosen at a rather high level and uniformly
administered to all ages, a procedure which guarantees the required spectrum of
levels: performing not at all, insufficiently, to some extent sufficiently, almost
sufficiently, sufficiently. In the present case the level was unfortunately chosen

so high that nobody succeeded. It looks like an investigation of how much

weight various people can lift, with weights that only professional weight lifters
can raise.
13.10. Of course the preceding criticism is not made against the extremely
clever test as such. Recognising angles in this problem and attacking it by angle
measurement represents a high level of geometric skill, which deserves maximum
attention.
In the sections on length measurement we repeatedly mentioned angles
angles as a means to measure lengths, thanks to congruence and similarity
theorems, which involve angles as well as lengths and length ratios. Measurement
of angles consequently plays a part as a means of comparing angles when

reproducing according to ratio (on scale),

checking reproductions,
reinterpreting the reproduction in the reality,
in general when
identifying and checking directions with respect to some basic direction
or basic plane.
13.11. The classical instrument to measure drawn angles and to draw angles of
a given measure is the protractor essentially half a circular ring, subdivided
by ray segments into 180 degrees. For reasons I was unable to find out, this
instrument has recently been superseded by an isosceles right triangle called
geo-triangle, solid, transparant, made of plastic with an angular division

radiating from the midpoint of the hypotenuse to the other sides. Well, inside
the triangle half a circle with the midpoint of the hypotenuse as its centre is
indicated, and from the position of the degree numbers it becomes clear that
it is the semicircle that really matters.
One is inclined to say an outrageously misleading instrument. An angular
division should suitably be constructed on a circle or its periphery, rather than
on the rim of an isosceles right triangle. Well, it is misleading but usefully misleading, a trap the learner should have walked into once in order to avoid it in
the future.
In order to measure angles, one has to subdivide angles. Subdividing angles
can be confused with subdividing lengths or areas. This source of misunderstanding should be uncovered and scotched as early as possible.
When a cake is divided into five equal wedges, equality of angle, area, and
arc coincide, at least if it is a circular cake.

364

CHAPTER 13

If it is a square or otherwise regular, division into equal wedges still means a

coincidence of equality of area and border segment.
If, however, the cake is rectangular, or otherwise irregular, the equality of
area and border arc cease to coincide.

Dividing, comparing and measuring angles is required in many figurative

contexts. If from the start onwards it is performed exclusively or one-sidedly in
the circular context, the learner is insufficiently prepared for other contexts,
which in fact are more frequent than the circular one. Among these other
contexts there are a great many where the equivalence of equality of angle,
area, and border arc holds approximately, such as triangles with a pronouncedly
acute angle where bisector and median are identified with each other by sight,
or rectangles that look like squares and where the diagonals are understood as
bisectors.
13.12. Perhaps it is a surprise that under the title of angle measurement so much
fuss is made of angle division. However, this is a most natural thing. For angles
there is a natural unit, and each angular measure is related as a part to this unit.

It is not farfetched to require that the learning process of measuring angles

properly starts with manufacturing a measuring scale, and there is hardly any
other way to do this than by subdividing.

As stressed earlier the most natural substratum for angular subdivision is the
circle (or the circular cake), which in fact are already instrumental in introducing
fractions. But as in the case of fractions the cake, so in that of measuring angles
by subdividing the circular disc is too poor an approach.
The simplest operation of dividing angles is halving and the most elementary
and concrete way to perform it is folding; dividing into three equal parts by
folding is more difficult to perform though it can reasonably be done; five equal
parts by folding is almost impossible. Continued halving, which means dividing
by a power of 2 is quite easy. Pleating an arbitrary sheet of paper produces an
angle of 180 at each point of the fold line as a vertex, that is half a full angle.
Folding once more such that the first fold is doubled produces a quarter of a
full angle, that is right angle; if the sheet of paper is opened, one notices four
right angles produced by the fold lines. Continuing the same way one can halve
all four right angles at the same time, and so one can go on. It is the way to
produce a compass card: the full angle divided into, say, 32 equal parts, points
in the nautical terminology:
1 point = 1/32 of the full angle = 1/8 of the right angle.
To my astonishment a seven-year old knows the four cardinal points and can indicate them
in the open field, he even knows the meaning of NE, SE, SW, NW, and immediately grasps
the meaning of NNE, and so on.

I started with halving angles because with appropriate material it is the

simplest operation that leads to measuring; as a point of departure I chose the

M E A S U R I N G BY M E A N S OF GEOMETRY

365

full angle. I can, however, apply the operation of halving to every angle, given
as a part of the plane bounded by two rectilinear sides. It does not matter what
the remainder of the boundary looks like, but the fact that it does not matter

is important enough to be illustrated by an explicit variety of examples, in

particular by folding sectors that are not a priori symmetric with respect to
the virtual fold line, in order to show clearly that
halving the angle does not imply halving the area,
halving the angle does not imply halving the border arc,
and conversely the even more important fact that
halving the area or the border arc does not lead to halving the angle.

13.13. Adding and subtracting angles occurs in this course as operations of

temporarily second order, where during the process of measuring the additivity
of the angular measure is tacitly assumed as was the case with the length
measure.
First of all, the addition of equal angles an operation so natural with a
view to measuring as long as there are no prestructured scales, that it is unbelievable that didactically it has been neglected, if not overlooked.
A ten-year old boy tells me that in the afternoon he had measured the shadow; it was 3
and 9 (probably decimetres). I ask him which was 3 and which was 9, and though a bit
surprised or annoyed by this query, he gives the correct answer. I ask him, if the stick
had been 1, 2, 5, how long the shadow would have been, and he answers correctly. He
tells me he would have liked to measure the angle too. I had neither a protractor nor tangent
tables at my disposal; I did not continue the talk satisfactorily a serious didactical failure.
I should have had the boy draw a triangle of 3 to 9, cut it out and multiply the angle by
turning it around five times to get with an incredible precision a right angle, and finally
as an unexpected crowning of the work, divide 90 by 5 to get the angle of 18, which
was the aim of the boys experiment.

If an angle is given, on paper or by means of show or vision lines, the first

step to estimate its size is

comparing it with a known angle, or

multiplying it a number of times to get approximately a full angle.
The first can be done

by superposition, the one on the other, concretely or mentally,

the second
by reproducing the angle a number of times cyclically cut out, by
means of turning, folding, or mentally.
If the sequence does not fit into the full angle
it is continued to two, three or more full angles.

366

CHAPTER 13

If

n copies of the angle complete approximately m full angles, the given

angle equals approximately
full angles.
13.14. One should fully realise that by this measuring procedure the concept
of angle behind the mental object, if not the mental object angle itself,
has changed. Angles are not somehow added anymore but adding angles has
progressed to
adding in a certain orientation,
and what is added in order to be measured, and arises as the result of the addition,is
rather than the angle as a part of the plane or
as inclination of two rays:
the turn.
Euclid did not have any trouble with this transition, as little as did the Babylonian
and Greek astronomers, who from the number of years and months elapsed
between two eclipses at the same place of the sky calculated revolution times
and angular velocities of celestial bodies. Earlier on* I explained how much is
required to attain concept formation in the broad field around angles I need
not repeat it now. I only summarise step by step what happens here in the
sphere of the mental objects:
adding in a certain sense,
meeting the natural unit, the full angle,
initially as an upper bound,
then to be trespassed by counting further,
exhausting (or approximately exhausting) a number of units,
interpreting a quotient as an angular measure.
13.15. Traditional didactics takes it easier. Nobody thinks about measuring
angles by means of ones limbs or primitive instruments nor about making
conscious the natural unit, the full angle, nor about measuring as exhausting
this unit or multiples of it. Angular measures are offered prefabricated on the
protractor. Miscomprehension, created in lower grades and not eradicated, can
persist up to the highest grades. It can start quite early.
A teacher who tries to have children find out what is a circle, gets as one of the definitions:
360.
* Mathematics as an Educational Task, pp. 476494.

M E A S U R I N G BY M E A N S OF G E O M E T R Y

367

To be sure, the prefabricated subdivision should come up some time, but

it matters whether this happens after the children have themselves learned
to make subdivisions in 32 points or in some other way.
Must the division in 360 parts, called degrees, be pressed upon the children
in order to be accepted straight away? There are deviations from the decimal
system pupils are familiar with, and have been so for quite a time: the division
of the day into
hours, the hour into 60 minutes. (Anglophone pupils
are acquainted with more deviations, for instance in the length measures.) The
origin of these deviations in the Babylonian sexagesimal arithmetics, and the
likewise Babylonian duodecimal zodiac corresponding to the twelve months
of the year, is worth being discerned as at least an approach to a discussion
about the difference between necessity and convention in mathematics I
would have liked to dedicate a whole chapter to this subject, and it is a pity
I cannot tackle it here.

13.16. One of the characteristic insights of what is traditionally geometry is:

transferring operations such as congruent displacements and subdivisions of linesegments from the context of measuring into the less instrumental and more
theoretical-looking context of congruence theorems. This insight also extends
to angles. Angles are being transferred by means of a pair of compasses, thanks
to the congruence theorem on three pairs of equal sides, and halved, thanks to
a symmetry property, which can be explained by the same congruence theorem.
Congruence, however, is only an a posteriori justification for a procedure of
transfer of angles, which is phenomenologically more easily described. First
of all the pair of compasses is used to delimit an unlimited angle (Figure 162).

This extra structure means a change of perspective from

the angle
to
the arc or/and the circle sector.

More strongly than with the use of the protractor, this construction stresses the
idea that
the angle is measured by the arc or the sector.
The object that is mentally transferred, is this extended figure. The triangle from
which the congruence property is to be read is only present explicitly by two of

368

CHAPTER 13

its sides; the third side, the chord, can be added mentally, but the arc impresses
itself more forcefully as the thing that is caught between the compasses and
transferred. It is, in fact, tacitly assumed that the result
does not depend on the radius of the auxiliary circle (Figure 163).

It is not at all obvious that the angle can be transferred by means of the
compasses, and the discovery of this fact looks like a surprise. It looks as though
the chord measures the angle; at least when transferring an angle one is allowed

to act as if this were true. But why, then, are chords inappropriate to measure
angles? The objection that the chord is shorter than the arc is not to the point.
Indeed one can measure the distance AmsterdamUtrecht on a map, where it
is albeit to scale unfathomably smaller than in reality. What really matters
is understanding that the relation between angle and chord is not ratio preserving;
the double angle is not matched by the double chord, nor half the angle by
half the chord. This should be made explicit, not only in its geometrical context,
but also as a paradigm of a non-linear relation.
The transfer of angles by means of compasses also serves to double, triple,
. . . angles. Pacing with the radius the circumference of the circle in order to
return after six times to the starting point, is a construction many children
learn to perform the construction of the regular hexagon long before this
property of closure motivates them to ask why?. I do not know whether
anybody ever tried to use this phenomenon in order to test in the development
of children stages such as
readiness to
need for
ability to

perform deductions. It is a particularly appropriate example because whatever

one tries one cannot manage it by means of symmetries. Some kind of knowledge about the sum of angles in a triangle is indispensable. I will shortly return
to the figure of the regular hexagon.

13.17. I started displacement of angles with the classical compasses construction.

A simpler, though less general one is sliding along a ruler, to say it phenomenologically. In the traditional geometric context this construction is justified by
theorems on pairs of parallel lines intersected by a third, and on the related
angles.

M E A S U R I N G BY M E A N S OF G E O M E T R Y

369

As a matter of fact,
by sliding the triangle along an appropriately chosen third line, a quite
special angle is being transferred,
the variable side of the triangle of the varying angle is considered to
remain parallel to itself while gliding,

the concrete triangle with a given angle is replaced by a mental one in

order to be mentally transferred,
which mentally produces the same system of parallel lines, from which by
change of perspective equality of angles is inferred

at least this would be a possible phenomenological course.

Yet similar effects can be reached by paving the plane
concretely, by drawing, mentally

with congruent triangles (Figure 164).

Two congruent triangles are juxtaposed to produce a parallelogram;

the figure is extended by a third triangle, which makes two sides extending
each other;
a fourth triangle completes the figure to a triangle, similar to the original
one;
and so it goes on,
producing three systems of parallel lines.
Why do the line-segments continue to produce lines? It can be a primitive
experience, not subjected to further analysis, but it can also be derived from the
displacement of triangles along rulers by change of perspective, which was
discussed earlier.
13.18. If the plane is paved as in Section 13.17, in any vertex the angles of the
triangle succeed each other cyclically, each twice. Thus the sum of the angles of
a triangle is half a full angle. The angles viewed here are the interior ones. Earlier
on we dealt with the exterior angles (even of an arbitrary polygon) as turn angles
in the trip around the figure, with the full angle as their sum. Of course there is
a close connection between both of these treatments. In a larger context the
statement that the turn angles of a simply closed path add up to a full turn is

370

CHAPTER 13

equivalent to that of the sum of the angles in a triangle being half a turn and to
that of equality of angles along parallel lines (the postulate on parallel lines).
In this connection I take up again the figure of the regular hexagon, which
I started discussing in Section 13.16. The construction by pacing the circle with
its radius can be interpreted as paving with equilateral triangles as soon as the
lacking sides are actually or mentally completed. By doing this one understands
why the construction closes, thanks to a change of perspective: the view
from the circle to the inscribed regular hexagon
is changed into that
from the regular hexagon to the circumscribed circle.

A more explicit deduction is required if the angular measure is called upon:

each angle of the equilateral triangle is

one third of half a full angle,

thus six of them add up around a vertex cyclically to produce a full angle (in
other words, three in succession, half a full angle, to produce the extension of

a side) (Figure 165).

13.19. The proportionality of

angular measure, arc length, sector area
of corresponding

angles, arcs, sectors

of a given circle is an intuitive fact and as such is easily accessible to conceptual

analysis as soon as the need for such an analysis is felt. It should be preceded by
an analysis of arc and area in general, obviously by means of the polygonal

approximation of circumferences and areas. If the possibility of such definitions

is accepted or understood, the additive invariance of the viewed relations and
consequently the proportionality of the measures becomes obvious.
If the full angle is agreed upon as the unit of angular measure, the proportionality factor is
for the arc:
for the area:

where r is the radius of the circle. In these expressions the factors r and r2,
respectively, are nothing but scale factors of the similarity which reduce the

M E A S U R I N G BY M E A N S OF G E O M E T R Y

371

circle to the unit circle. The construction of the regular hexagon proves that
It is well-known how by continued halving of angles and doubling of
numbers of sides this approximation can be improved. The step from
at

the arc length to at the sector area is made by considering the sector (as
regards its area) as a triangle with the arc as its base, and the radius as its height.
13.20. Though it is not my intention to deal here with trigonometric functions,
it is worthwhile indicating the part played in practice by the tangent of an
angle as a measure of inclination, especially in the topographical context. The
inclination is usually given in percentages. Most often the angles to which
this is applied are so small that the angle and the inclination are approximately
proportional.
13.21. In spite of the more general first approach, I restricted myself later on to
angles of lines or rays in the plane.
I just said the plane a typical formulation as though there were one
unique plane, the drawing plane where all things occur, horizontal or vertical
(the blackboard in the classroom), possibly a bit out of level, but then solidly
supported and certainly not floating in space. The straight lines and rays, the
angles of which were measured, were tightly bound to such a plane, and it
requires quite an effort to detach them from it.
The angle between the direction towards two arbitrary points in space (for
instance two stars), swept out by the stretched arm, moving from the one to
the other, can in the topographical context be a first opportunity to detach
angle measurement from horizontal and vertical planes. Models of solids, in
particular pyramids, but also cubes with their face and space diagonals are a
means in the geometrical context to abolish the confinement of angle measurement to the plane.
Though the angle between planes and that between a plane and a straight line
are, at least in the topographical context, as natural as that between lines, their
measurement is practically reduced to that of the mutual inclination of two lines
within a plane, to wit the plane that intersects the figure perpendicularly. Here
too it is important that the plane in which the angle is measured has been
experienced in many positions, not only horizontal and vertical ones.
Angles of skew lines are less farfetched than one would believe. A look out
of the window is an opportunity to discover such angles in a meaningful context,
for instance, the protuberances of TV aerials. There are many more examples:
roads, tubes, conduits that cross each other, though not on the level. They
define angles, measured after parallel displacement to one point or by turning
the one into the other.* Again it is important to detach the situation of skew
lines from the topographical context of horizontal and vertical (the TV aerials
moved out of their vertical position on the roof).
* Mathematics as an Educational Task, pp. 479480.

372

CHAPTER 13

13.22. Solid angles, delimited by more than two planes (or parts of planes) have
not been discussed here, though they are quite concretely suggested by the
corners of a box, a drawer, a room. Obviously three mutually perpendicular
planes delimit an octant, the eighth part of the space around the vertex. Yet
how to measure such angles of three or more planes in general?
Parts of the plane bounded by the sides of angles are firmly delimited by
circles, and then measured thanks to the proportionality of angular measure, arc
length, sector area on the circumference or inside the circle. This definition
extends to solid angles: A sphere is considered with its centre in the point where
the side planes meet, and the solid angle is defined as proportional to the part
cut out on the surface or within the sphere. In other words, with the full
angle as a unit corresponding to the total sphere area or volume: the solid angle
equals the part cut out from the surface or the volume of the sphere.
13.2335. Measuring Areas and Volumes
13. 23. So close to reality as a mental object and so profound as a mathematical
concept I do not know where to start with area (and volume) as a subject and
where to finish, and nevertheless to aspire to completeness. Nowhere is the
variety of levels so patent, the restriction to one level of understanding and
exactness so much in contradiction with the demand of mental growth and
development as it is in the case of area (volume) as a didactical subject. Shall
I first expose, or at least sketch, the mathematical theory, then take the phenomenological course, in order to finally put this phenomenology within a
didactical context? I tried this disposition with subjects that lent themselves to
it, whereas I dealt with others in a rather unstructured way. In the present
situation the most urgent seems to be the phenomenological approach, albeit
not towards area itself, but towards the current instruction of area.
The striking feature of both area and volume is their wealthy context in
nature, culture, and society on the one hand, and the extreme poverty of
the related instruction on the other hand. In education at least in primary
education, where attitudes are acquired and fixed area is emaciated to length
times width, replenished with a formula for the circle, which has neither of
them (or does it?); for volumes there are a few more formulae in primary and
secondary education; and Calculus boasts a machinery to compute the areas
and volumes of ad hoc fabricated figures. This is from the lowest to the
highest level a degree of emaciation as I think has not been the fate of any
other mathematisable subject in instruction.
There is even more to it: no other subject looks as little problematic with
mountains of problems hidden to the unskilled observer; none is afflicted by
a more rampant battle between the mathematical and the didactical conscience
than is area (and volume).

13.24. Area is a magnitude to measure objects of a variety that looks more

variegated than for any other magnitude:

M E A S U R I N G BY M E A N S OF G E O M E T R Y

373

a sheet or a roll of paper or cardboard,

a slab of wood or hardboard,
a roll or a cutting of fabric,
a hide of leather,
a football field or a swimming pool,
a plot of farmland or forest,
premises,
the territory of a state or parts of it according to their kind or use,
the tributary domain of a river,
the surface of a lake or sea,
a wall to be painted or plastered,
a floor to be covered with mats or parquet,
the floor or window surface of an office to be cleaned,
a street to be paved or asphalted,
a roof to be tiled,
a meadow to be mowed,
a field to be plowed, sown, planted,
a skin surface to judge its heat exchange,
a skin surface, to be clothed tightly or airily,
the leaved surface of a forest to judge its evaporation and gas exchange,
and then

areas of geometric figures

plane, developable, non-developable,
expressed, or not, by formulae,
interesting as such,
or as a means of expressing lengths as functions of other lengths
(as in the theorem of Pythagoras).
13.25. As a magnitude we already discussed length, a
function of long objects,

phenomenologically. In order to form a magnitude in a category of things

an equivalence relation,
an order relation,
an operation of composing
are required. In all these aspects area is more varied and more complex than
length. An object to which an area is assigned, must have both length, and width,
that is, in the phenomenological sense as meant by us it must be a

two-dimensional object.
In the case of length the constituting equivalence was generated by comparatively
simple operations, such as

374

CHAPTER 13

displacements, flexions, break-and-make transformations;

in the case of area the character and scope of these operations is much less
perspicuous this will be dealt with in Section 13.26. The operation of composing had a unique result in the case of long objects, that is, up to permutations
of parts; whereas the objects to which an area is assigned can be composed in a
multifarious way.
Comparing objects with respect to length without measuring them explicitly
is in principle a simple task. Indeed, longer and shorter are easily reduced

to containing and being contained. For surfaces, measuring looks in most

cases the most natural way of comparing.
To measure lengths one chooses some unit and determines how often the
unit fits into the object; if a possible remainder is worth being accounted for,
one switches to a smaller unit, from metres to decimetres, from decimetres to
centimetres.
As a matter of fact, this method of
measuring by exhausting with units

ultimately works for all magnitudes. In the case of

lengths, weights, times, quantities of liquid

one can leave it at that, but for

surfaces and solid contents

it is at least as a unique method unsatisfactory. A square as a unit for areas
(a cube for solid volumes) is a natural figure to exhaust with congruent (possibly
similar) copies
plane figures bounded by irregular curves,

spatial figures bounded by irregular curved surfaces,

but for
figures bounded by straight lines or regular plane figures,
figures bounded by planes or regular spatial figures,
this can be a perverse procedure. To show this by two extreme examples:
A square divided into two congruent parts, produces a partial figure, which
is directly (not by exhausting with little squares) recognisable as one half
half of the original figure (Figure 166). A circular disc is obviously better
exhausted by regular polygons than by little squares.

M E A S U R I N G BY MEANS OF GEOMETRY

375

For all magnitudes

fair sharing

can be reduced to numerically dividing by measuring the thing to be shared. It is

quite natural to do it this way if measuring is easy as is the case with
lengths, weights, times, quantities of liquid.

Measuring areas and solid volumes, however, is more complicated if not extremely
difficult. Under such circumstances it is most natural to benefit from a geometrical structure of the thing to be divided, if there is any. There are many
ways to divide a square, a circular disc, a cylindrical surface in three congruent
parts, and each of them produces a fair division of the area (Figure 167). Even
if the thing that is to be divided, shows less or no geometric structure, fair
sharing by estimate will mostly be preferred to involved measuring procedures.

13.26. Let us put this confrontation also in the context of mappings! Which
mappings leave length, and which leave area (volume) invariant? As regards
length, this could be discontinuous mappings (cutting), continuous mappings of
single figures (flexions), and continuous mappings of the whole plane, leaving
each length invariant, that is, congruences.
Congruences of the whole plane (space) leave of course area (volume) invariant but there are many more mappings that do so. An affine mapping of the
plane (space) transforms parallelograms (parallelepipeds) into the same kind
and a definition of area (volume) for this kind of figure suffices to understand
the behaviour of area (volume) under affine mappings. It appears that
an affine mapping of the plane (space) multiplies all areas (volumes) by the
same factor.

(This factor, a property of the mapping, is called its determinant.)

Then there are
special affine mappings

376

CHAPTER 13

(that is with determinant 1) which leave invariant the areas of all parallelograms

(the volumes of all parallelepipeds) and for that matter area (volume) at all.
There are still many more. A kind of area- (volume-) preserving mapping that
has drawn too little attention is
shearing.

Let us consider a plane figure as constituted of

linear layers
in order to have them

glide along each other

(a spatial figure as constituted of

plane layers
gliding (also twisting) along each other.
In mechanics, indeed, this is called shearing, and there I took the term from
(Figures 168170).

Shearing preserves areas (volumes). This is intuitively obvious: One ascribes

a thickness to the linear (plane) layers, which means considering them as having
been built from rectangular (plate shaped) layers, which glide on each other.
There the in variance of area (volume) is an elementary fact, which with the
thickness approaching zero is being taken along to the limit.
The invariance of area (volume) under shearing also follows from

Cavalieris principle:

M E A S U R I N G BY M E A N S OF G E O M E T R Y

377

Two plane (spatial) figures the sections of which at each height are correspondingly equal as to length (area), have the same area (volume). By this principle
one proves for instance that

pyramids with equal basis and height

have the same volume, whereas the equality of area of
triangles with equal basis and height
is ascertained by simpler methods.
(Cavalieris principle is also the basis of Archimedes best known proof for

the volume of the sphere. As far as volumes are concerned, the hemisphere is
considered as the difference between a cylinder and a cone (Figure 171); indeed
at any height the plane section of the hemisphere equals that of the difference
between cylinder and cone.)

Even shearings do not exhaust the area (volume) preserving mappings, which
of course form a group: two of them in succession or the inverse of one of them
have the same invariance property. Yet the product of two shearings according
to differently directed layers is in general not a shearing any more; it is just a
fresh area-preserving mapping. (The product of two spatial shearings according
to differently directed layers is a shearing along parallel lines; the product of
three spatial shearings according to planes that do not meet in a line, is again
a new volume-preserving mapping.)
The plane (spatial) shearings according to arbitrary layers generate a group of
area- (volume-) preserving mappings. I do not know whether all area-(volume-)
preserving mappings are obtained or at least approximated in this way.
13.27. It would be a serious shortcoming if I did not put area and volume in
the frame of the mathematics of the measure concept. Measure, as understood
in measure theory owes its existence to the mathematical need for an extension
of such ideas as length, area, volume originally ascribed only to geometrical
figures to sets in general. Axiomatic theories make clear how far this is
possible.
What is a measure ? One is concerned with

a substratum set E,
a set
of subsets of E,

378

CHAPTER 13

a function
the measure, which to every element of
negative real number, possibly also

ascribes a non-

For instance E can be the straight line, the plane, the space. Anyhow
will be subjected to certain postulates. The most important one is:

and

Additivity: The measure of a union of mutually disjoint sets equals the

sum of the measures of the particular summands.
However, in this form the postulate is untenable. We wish to assign to the unit
line -segment the (linear) measure 1. The line-segment is the union of all its

points (considered as one-point-sets), which should have all the same measure,
and 0 by preference. The postulate is defeated by this incompatibility. We
shall be more modest. One choice is:

Finite additivity: The measure of the union of a finite number of mutually

disjoint sets equals the sum of the measures of the particular summands.
Instead of a finite number of I may say two; step by step this extends to
any finite number. Another choice is

The measure of the union of countably many mutually

disjoint sets equals the sum of the measure of the particular summands.
There are mathematical reasons why -additivity should be preferred to finite
additivity. This is what actually happens in Lebesgue measure theory.
It would be nice if we could leave it at that. However, for a measure theory
with the line, the plane, or the space as E we would also require the
invariance of the measure under congruence transformations,
which means that

congruent sets have the same measure.

Unfortunately this postulate is defeated by
Let E be the circumference of the circle with radius 1, in other words the

set of complex numbers of absolute value 1 (or the set of the angles).
E is gifted with a group operation, the multiplication of complex numbers
(or addition of angles). Let be chosen
such that for no integer n
does
hold (
is irrational), generates an infinite cyclic group
F. Form a set V by taking exactly one element from each coset of F in
E. Then V is fully divided into the sets

which are mutually disjoint. However, they are also pairwise congruent. If
congruent sets
shall be assigned the same measure which is fair and E shall be assigned

a positive measure, say

, we run into trouble. The only way out is to

forbid the set V to have a measure.

M E A S U R I N G BY M E A N S OF G E O M E T R Y

379

Rather than E I could have chosen, with a slight variation, the straight
line, the plane or the space.
In the space even finite additivity runs aground:
The spherical surface can be divided into three sets
into two sets
all of which are pairwise congruent.

as well as

In order to arrive at a decent measure theory, one is obliged to restrict the

set
of sets gifted with a measure; pathological sets should be avoided. It
would take us too far afield to go into details. Our aim is not a measure theory
for sets but the connections between the various phenomenological approaches
to area (and volume) of figures. Even this will be difficult enough.
13.28. Intentionally we abstain from establishing precisely what a figure is.
Phenomenologically viewed even attempts at precision would bring us into the
same kind of difficulties as we encountered in the discussion of topological
concepts. It would be a relief if we could restrict ourselves to planar (spatial)
figures delimited by straight lines (planes), while neglecting circles, circle,
sectors, circle segments, ellipses, parabolic segments, cylindric, conic, spheric
surfaces, and parts of them (and similarly for the volumes). We cannot but
admit curved boundaries, yet as soon as we do so we must require a certain
smoothness, if we want to avoid trouble. What degree of smoothness and how
to define it? By means of differentiability conditions, and then of what order?
Piecewise differentiability and should we allow an infinite number of pieces?
Even this can cause pathologies.
Shall we in a planar (spatial) figure include its boundary, or are we allowed
to neglect it, totally or partially, if it is the area (volume) that matters? This is
the very question that goes to the heart of the matter. It is clear what the
phenomenological answer will be: The boundary may not matter if areas
(volumes) are to be compared or measured.
This then is the principal reason why the quest for area (volume) of figures
cannot be satisfied by a measure theory for sets. As mathematical concepts
dimension and measure are estranged from each other by phenomenologically
unexpected mathematical consequences.
As an example I choose a well-known 0-dimensional set (on the line)
with a positive linear measure:
From the line-segment [0, 1] the open mid-third
is deleted.
From the remaining pieces
and
again the open mid-thirds
ard deleted, and so it continues indefinitely. The remainder in the limit
is a closed set C (Cantors discontinuum), which contains no line-segment
a 0-dimensional set. What is finally left from the length? At every step
of the previous total length is left, that is,
after the nth step,
which fortunately converges to 0 with

380

CHAPTER 13

But let us now do it in a topologically identical, though metrically

different, way:
At the first step the open mid-third is deleted, at the second step the
open mid-sevenths, then the open mid-fifteenths, and so on. The remainder
is, as regards its total length,
at the 1 st step
of the previous length,
at the 2nd step
of the previous length,
at the 3rd step
of the previous length,
and so on, and finally at the limit

of the original an infinite product with the limit

The modification leads to something which topologically is again a
Cantor discontinuum, though according to our reasoning it should be
assigned a positive linear measure
A similar construction produces 0-dimensional sets in the plane (space)
with a positive planar (spatial) measure.
On the line the set C is all boundary. The boundary of a plane (spatial) set,
though of lower dimension, may possess a positive measure; a closed set with
no interior point can nevertheless have a positive measure.
On the other hand it seems to be an intuitive requirement that when comparing and measuring areas (volumes) the boundaries do not count a figure to
be measured is exhausted by squares or other measuring units, which touch
each other, and it does not hurt (or may not hurt) that some points are counted
twice or even more often. This is a new example of mental objects which,
sharpened to concepts, seem to show paradoxical features. Of course such a
paradox can always been scotched by ad hoc definitions in the present case
by restricting the definition of area (volume) to sets, the boundary of which
is smooth enough that its measure necessarily vanishes.
For our aims this would be a useless sophistication. In the sequel we will
uncover more profoundly rooted impediments than paradoxes of seemingly
pathological origin on the way from the mental object to the concept of area
(volume).
Hence I will not look for more precision with respect to figures and continue
to speak of their area (volumes) in a naive style.
13.29. The proper difficulty is the large variety of approaches to the mental
object area (volume) of which it is not a priori clear whether they form a consistent whole, though each of them separately and all together are being used
whenever they are needed.
Let us summarise them once more:

M E A S U R I N G BY M E A N S OF GEOMETRY

381

Planar figures can be compared with respect to their areas

directly, if one is a part of another,
indirectly, after
break-and-make transformations,
congruences and other area preserving mappings,
measuring.
Plane figures can be measured
by exhausting with a unit of area,
by interior or exterior approximation.

In this, use is made of

the additivity of area under the composition of plane figures
that up to the (one-dimensional) boundary are mutually disjoint,
or of convergence of areas under approximation.
Similarly surfaces in space can be evaluated with regard to area, as far as
they are composed of plane figures or can be approximated by such.
(A corresponding statement can be made with respect to spatial figures.)
13.30. It is not at all obvious that all these approaches lead to the same result.
On the contrary, the proofs require efforts that would surpass anything that
can be asked or realised, say, at the highest secondary grades, even if it would

be possible to excite any interest in this complex problematic. Nevertheless I

feel obliged to sketch this problematic, with the accompanying reasonings,
at least in order to explain how much is required to raise the perspicuous looking
mental object area (volume) to a proper mathematical concept.

13.31. The universal means of area measurement is exhaustion by units of area.

This means can be applied efficiently by
covering figure F by a square grid
The united squares
contained in F approximate F from the interior,

containing points of F approximate F from the exterior.

Their united area, which can be obtained by simply counting, is by definition,

respectively,
smaller than,

larger than

the area of F that is to be defined.

Refinement of the grid leads to better approximation, respectively,

from the interior,

from the exterior

by unions of grid squares.

The limit under unlimited refinement is, by definition, respectively

the inner area,

the outer area of F.

382

CHAPTER 13

Area of F has a meaning as soon as both of them are equal. A priori this equality
is not obvious for arbitrary F it would not even be true. Equality of inner
and outer area of F boils down to the property of F that
its boundary has an outer area 0.
It is rather easy to understand that the outer area of a line-segment of a

union of line-segments, and thus the boundary of a polygonal figure vanishes;

it is less clear which requirements a curve should fulfill to have a vanishing outer
area. We shall suppose implicitly that our mental object geometrical figure

satisfies them.
We required additivity of area. It is not obvious, though it is not difficult to
prove, that the area, such as defined a moment ago, does possess this property,
at least as far as finite additivity is meant.
Moreover invariance of area under congruence mappings was postulated. This
invariance is not at all a trivial concern. Suppose we take a square grid R with
its refinements fixed. Does it, with a view to area, matter how I put the figure
F upon the grid? Of course, it does not matter, but how to prove it?

The question can be formulated otherwise. Suppose the figure F in a fixed

position, and put the grid R with its refinements arbitrarily on F. Does it matter
how this is done?

Let us start with a rectangle F with sides parallel to the grid. By simply
counting and approximating from the interior and the exterior, I can verify the
formula

where as a unit of area one has chosen the grid square with side 1. What happens
if I subject F to a congruence transformation?
Under a translation the rectangle F keeps its sides parallel to the grid. So we
can apply the formula anew, which means invariance of its area under translations. This statement can immediately be extended to the case of an arbitrary
figure F, instead of a rectangle.
Reflections leaving the grid invariant do not cause any trouble either. We are
still left with rotations. Under a rotation D the rectangle F gets in a position
with respect to the grid R which seems to resist the direct evaluation of its area.
We will succeed by a trick (Figure 172).

Let us take as F a square with sides of length a, parallel to the grid. Whatever
F may be, DF is certainly contained in a square parallel to the grid with a side

M E A S U R I N G BY M E A N S OF G E O M E T R Y

383

length equal that of the diagonal of F, thus with an area

and on
the other side DF certainly contains a square parallel to the grid with a side
length at least half that of F, thus with an area
It has been shown that under a rotation D the area of a square parallel with
the grid is multiplied by a factor r depending on D such that

It is not difficult to understand that this factor is the same for each grid
square. The area of an arbitrary figure, however, is defined by means of square
grids. Consequently under the rotation D the area of each figure is multiplied
by the same factor r.

This holds in particular for the figure DF, where F is again a square parallel
to the grid R. Thus
One can continue applying D:
Dn, however, is again a rotation, with now the multiplication factor rn.

Thus

as previously proved, and this should hold for arbitrarily large n, which it is
impossible unless
Hence the area of a square parallel to the grid R is invariant under rotation, and
thus under all congruence mappings, from which the same is easily derived for
the area of any figure.
I could have proved this more elementarily but also more cumbrously
by breaking and making. However, the proof I gave runs almost the same way
in space (and even in arbitrary dimensions). My proper intention was to make
clear how much is required to make sure a seemingly obvious mathematical
fact.
13.32. The same holds if we choose a more elementary approach to area. Let
us now restrict ourselves to rectilinearly bounded figures, and let us start with
rectangles. The formula

can be obtained without lattices and approximations, by means of

break and make operations,
and similar triangles.
Indeed (Figure 173) by subtracting from the large congruent triangles PQC and

384

CHAPTER 13

PQC the pairs of small congruent triangles POA, POB, QOB, QOA, I get for the
remaining rectangles

On the other hand, by similarity

thus
Conversely, starting with this formula, one can reconstruct Figure 173 and the

rectangles, from which it follows:

Two rectangles have the same area if and only if they agree in the products
of their side lengths.
Thus the product of the side lengths characterises the area of rectangles. It is
additive if rectangles are stuck together to produce another rectangle.

One might remark that a similar argument holds if rectangles are replaced
by parallelograms, whereas it seems that the final conclusion ceases to
hold. In fact:
Two parallelograms with correspondingly equal angles have equal areas
if and only if they agree in the product of adjacent sides (Figure 174).

Thus the product of adjacent sides of parallelograms with the same

angle characterises their area.
One prefers rectangles and in particular assigns the unit area to the unit
square. One could as well choose as such the area of the rhombus with
side 1 and angle 60, or any other parallelogram. The whole difference is
a fixed factor, depending on this choice. It should be stressed that area
(volume) measurements presuppose the choice of a unit as does length
measurement. In principle these choices are independent of each other.

Even in practical metrology before the metric system the various units
(also non-geometrical ones) were not connected to each other.

M E A S U R I N G BY M E A N S OF G E O M E T R Y

385

By break and make transformations arbitrary polygons can be reduced to

rectangles as far as their area is concerned.
First of all, the parallelograms: Figure 175 shows how the parallelogram
ABCD is transformed into the rectangle ABCD by sticking on and cutting
off congruent triangles.

Then the triangles (Figure 176): ABC transformed into parallelogram ABMN
by cutting off and sticking on a triangle. Moreover it appears that any triangle
can be transformed in any other with the same base and height by a break and
make procedure.

An n-gon (ABCDE, Figure 177) can be reduced to an (n l)-gon (ABCE)

in order finally to arrive at a triangle.

The most noteworthy outcome of this discussion is that for polygons equality
and comparison of area can be defined by break and make procedures with no
intervention of approximations nor use of algebraic formulae. By break and
make procedures each polygon can be transformed into a rectangle with a
prescribed base and, as to their area, polygons can be compared by comparing
the heights of the corresponding rectangles.

386

CHAPTER 13

A nicety I may not pass over: Actually one can restrict oneself to cutting off
and glueing on triangles, and concerning the transport of such pieces translations
and point reflections suffice, thus (Figure 178) a triangle I displaced in positions
like 2 and 3 more general congruences are not required.

Yet new problems arise. If I insist on this elementary definition of area by

means of breaking and making and avoid any appeal to the former definition
by approximation grids, am I sure that I get a uniquely determined result?
Would it not be possible by means of different break and make procedures to
transform a polygon in rectangles with the same base and different height?
Or to say it another way: Would it not be possible to transform a rectangle by
means of a break and make procedure into a smaller one?
It is not possible, but this is not a trivial fact. A direct proof independent
of approximation procedures requires non-obvious tricks. Only after a number
of editions did Hilberts Grundlagen der Geometrie manage to produce a reasonable proof, and even this proof is not easy. 1 will try to give the reader an idea of
the proof.
Suppose a figure F is transformed by break and make procedures into a
smaller one, F. The pieces that are cut off and glued on are polygons, which
I may suppose are each divided into triangles. So the pieces may be supposed
to be triangular from the start onwards. I pack the figure F and all the auxiliary
triangles into a large triangle
I extend all line-segments in this figure to
produce a division of into little triangles, say
The problem can
now be reworded:
Is it possible by rearrangement of the
to cover only a part of
Of course it is not, and in order to prove it, I use the defining expressions
for the area of a triangle
base times height.
I shall show
the area of the whole triangle equals the sum of the area of the partial triangles. This, if proven, means that by rearrangement I cannot manage to cover
only a part of
So we shall prove (1). This formula is certainly correct for a subdivision of
a triangle in two, thus subsequently in the cases of Figures 179182.

M E A S U R I N G BY M E A N S OF G E O M E T R Y

387

Thus

This formula still holds if O is outside the triangle ABC, if it is supposed that I
allow negative areas and for instance in the case of Figure 183 I am counting
the area OAB negative. The general rule runs as follows:

Suppose the plane is oriented, that is provided with a sense of left turn,*
consider the triangle XYZ having walked around in the sense XYZX and count
its area positive or negative according to whether this circulation matches the
left turn of the plane or not.
If interpreted this way, (2) holds independently of the position of O.
Another remark: Suppose that the triangles ABC and ABC are on different
sides of AB (Figure 184) thereby having no common points other than those
of the segment AB. Then the triangles ABC and ABC have opposite circulations

Now back to the big triangle , subdivided into

We may suppose, if need be, after supplementary subdivisions that partial triangles meet
pairwise either in a common side, or in a vertex, or not at all (Figure 185).
* Orientation will be dealt with later on.

388

CHAPTER 13

Let O be some point, say, outside

I walk around the triangle
according the left turn of the plane, thus

If I add all these equations

then on the right hand side of the
equality all contributions area OPQ arising from inner line-segments PQ
cancel, whereas the little line-segments on the boundary of A just produce
together the sides of Thus

which is the formula we wished to prove. Thus it is impossible to rearrange the

partial triangles
such that is only partially covered.
It is a tough job, this proof, even though I touched on some points (such
as orientation) only superficially.
13.33. The area of rectilinearly bounded planar figures can be defined elementarily that is by using break and make transformations rather than approximations though justifying this procedure is not easy. The volume of planarly
bounded spatial figures is a different thing. The equality of two pyramids with
equal basis and height cannot be obtained by break and make procedures. To

obtain this equality approximations such as Cavalieris principle are required.

So it is less elementary than the case of triangles.
13.34. The preceding exposition shows clearly the gulf between the mental
object and the concept as far as areas (volumes) are concerned: obvious looking
properties and relations require sophisticated arguments to corroborate related
statements. It is a matter of course that at school, even at the highest level, no
more than a few little steps can be taken toward concept formation. Didacticians
of mathematics who consider concept attainment as the very objective of

mathematics education, are very likely to pass over the subject area (volume)
or to admit it only in the state of extreme emaciation. Whoever is convinced
of the importance of mental objects for mathematical thought, will not shun
it.

M E A S U R I N G BY M E A N S OF G E O M E T R Y

389

As early as with 8-year olds one can notice a well-founded visual idea of area.
With no hesitation they perform such tasks as

colouring half (a third of) an area red

as soon as the figure shows some suitable regularity, such as a rectangle, square,
circle, circle sector; they even can produce various solutions of such problems.

Even with irregular figures they succeed in solving them by reasonable estimates.
With ten-year olds
comparing and measuring areas by grid covers
proves that there exists a more profound insight into area. The observations bear
witness to the presence of a rather sophisticated mental object. Not until the

formula
length times width

appears, does the pauperisation start, which characterises the upper grades of

the primary school and where secondary instruction rarely or never adds any
enrichment. If then mathematics of area (volume) is taught at all, approach,
method and subject matter are estranged from the corresponding mental objects
by which the phenomenon area (volume) is understood in everyday life and
mathematical applications.
This didactical situation is not at all unique; it is only more involved and for
this reason looks less perspicuous than similar mathematical didactical situations
do. In Mathematics as an Educational Task* (where I did not yet use terms
like mental object), at the opportunity presented by the system of real
numbers, I asked the question describing or creating concepts analysis or
synthesis? In my present terminology I would adapt the same idea as follows:
the number line is a mental object, which is gradually and step by step learned
to be seized upon, that is by localising on it the natural numbers, their negatives,
the rational numbers, certain irrational ones and finally all the real numbers.
The number line is being described by the real numbers but not created by
some definition of real number this is a didactics which fits the needs and
possibilities at school and at the start of university studies.
With areas (volumes) the situation is similar insofar as they are constituted
and accepted as mental objects, and a posteriori, if need be, analysed in order
to arrive at gradual concept attainment. The situation differs by the greater
phenomenological wealth of the mental object area (volume), the variety of
approaches, the greater profundity and sophistication required for concept
formation as well as the built-in deceptive features, which may not be underestimated.
* Pp. 212214.

390

CHAPTER 13

I summarise the approaches:

fair sharing
by profiting from regularities,
by estimating,
by measuring;
comparing and reproducing (in another shape)
by inclusion,
by break and make transformations,
by estimating,
by measuring,
by means of mappings,
that is, congruences, affinities, shearings;
measuring
by exhausting with units
with even finer subunits,
by approximations from inside and outside
with fixed grids
with adapted figures,
by converting break and make transformations,
by means of general geometric relations,
by means of general formulae,
by means of principles, as Cavalieris,
by means of mappings,
that is, congruences, affinities, shearings.
Didactically all of them are acceptable, albeit with various weights. Being
restricted to one of them in order to pursue purity of method, is bad didactics.
On the contrary one is advised to profit from this wealth notwithstanding the
didactical impossibility of

a more than local organisation of this field of concepts,

that is, the didactical impossibility of

the question of the internal and external consistency,

the delimitation of domains of validity of the approaches,
the logical justification of the approaches in their mutual connection.
13.35. This does not imply an entirely nave and uncritical activity in this field,
which is a ground of traps and pitfalls. As soon as magnitudes are concerned,
confusions can arise: distances, times, speeds, metres per second, and seconds
per metre, gasoline per 100 km or the distance per litre are examples. For the
constitution of magnitudes as mental objects it is also required that they are
confronted with others. In the case of area (volume) the big misleader is perimeter (surface area). At the same time as children give evidence to be able to
divide an area fairly, they succumb by appropriately misleading examples to

M E A S U R I N G BY M E A N S OF G E O M E T R Y

391

judge the area of a figure according to its linear dimensions. Figure 186 shows
a situation where some assign a larger area to the righthand figure because of
its larger perimeter whereas others are misled by the narrowness of the figure
measured by the distance between the oblique parallels to draw the opposite
conclusion. Plato thought it worth stressing that the area of the Peloponnesos
cannot be measured by sailing around it. Galileos sack, sown up at different
sides is another historical example that might be cited here.

For the constitution of the mental object area (volume) one needs also
examples of figures that in spite of misleading deviations in the linear
dimensions have the same area,
such as the parallelograms with equal base and height, as well as
examples of figures that in spite of misleading agreements in the linear
dimensions, have different areas,
such as the rhombuses arising from a square by flexion (Figure 187).

Extreme cases are especially instructive: in Figure 186 where the parallelogram
gets ever narrower and nevertheless keeps its area; in Figure 187 where the
rhombus finally collapses to include no area at all.
With hindsight it is no problem to enclose arbitrarily little areas by a given
perimeter. In the constitution of the mental object area the impossibility of
enclosing arbitrarily much area by a given perimeter obtrudes itself. There is an
upper bound to the area enclosed by a given perimeter, which depends on the
kind of competing figures. This leads to the so-called
isoperimetric problem:
to find within a class of figures with a given perimeter the one with the largest
area. It is rather easy to prove that among the rectangles with the same perimeter
the square has the largest area. It looks obvious that among the planar n-gons
with the same perimeter the regular is the biggest, but except for the case
it is not that easy to prove. It looks just as evident that among the planar figures
the circle solves the isoperimetric problem but proving it requires quite a lot of
mathematics.

392

CHAPTER 13

Perimeter and area are to a certain degree, yet not totally, independent:
the perimeter imposes an upper bound to the area. The analogous feature one
dimension higher in the relation between surface area and volume has as regards
both aspects a great importance as a principle of explanation and control in
nature and technique. In order to aid the exchange of heat or matter with the
environment, nature and man, whenever it is-needed, create excessively big
surfaces in a small space; in other cases in order to minimalise exchange, shapes
are assumed or created which show as little surface as is compatible with the
other data.
The problem in three dimensions corresponding to the isoperimetric problem,
that is to enclose as much volume as possible by a given surface area, is mathematically even harder to solve than the isoperimetric problem, though here too
it looks obvious that more regularity, in particular that of the sphere, guarantees
more volume to the same superficies.

13.3646. Measuring Volumes

13.36. So far we have dealt with area and volume together the latter within
parentheses. There was one exception: for the elementary theory of area
(Section 13.32) there is no mathematical analogue in higher dimensions. On the
other hand it will appear that volume can phenomenologically and didactically
be approached along ways and dealt with by methods which have no analogues
in the approach to area.
Compared with area, volume shows besides the third spatial dimension
a large number of extra dimensions, which so far could not be done justice.
Concrete manipulations with volumes show a broader spectrum, moreover
volumes are less exposed to phenomenological impoverishment than areas.

There are two reasons why:

the double aspect of content and volume,
the concrete character of the volume preserving operations.
13.37. There are good reasons why the metric system in three dimensions
knows two kinds of measures:

cubic centimetres, decimetres, metres, and so on,

and
litres, hectolitres.
In principle one of them could be dispensed with. In fact they are closely
connected:

In traditional measure systems this was different: stacks of wood and big luggage
on passenger steamers were measured with different measures from oil, beer,

M E A S U R I N G BY MEANS OF G E O M E T R Y

393

and wine. The term content reminds one of a barrel to put things in, whereas
volume rather suggests a thing that claims space. A container has a content,
which tells you how much it can hold, but it has also a volume a bit larger
than the content which matters if a number of containers of a certain kind
are to be placed in a ship. So

content as creating space,

volume as claiming space
shall be understood
identified, and if need be
distinguished.
The relation between content and volume is complicated by what I will call the
pack and stow effect,
which I take the liberty to introduce by means of the paradoxical inclusion
syllogism

I fit into my suit,

my suit fits into my bag,
so I fit into my bag.
Or, more serious situations, which everybody will be able to locate:
it is a whopper but you cannot put anything in it,

the trunk packs poorly,

it is a piece of nothing but one cannot store it,
if it has in the length what it lacks in the height, it would be possible.
Content measures serve to measure liquids, powder-like grained matter
(provided it is not too coarse) things which can be stored in barrels. But
the content terminology reaches farther: to packing and stowing bags, containers,

carriages, vessels, though ultimately less standardised measures are used to

describe contents:
28 standing and 12 sitting places,
80 beds,
200 chairs,
to indicate contents of buses, hotels, theatres (or barber shops).
Later on I will deal with the need
to understand, and
to recognise the pack and stow effect,
and to eliminate it in the relation between content and volume.

Meanwhile, after now having stressed the phenomenological relation and difference between content and volume, I will, in general, in order not to complicate

the language, use one term: volume. The reader may add the word content,
where it is needed, in parentheses.

394

CHAPTER 13

13.38. In the course of constitution of the mental objects length, area, volume
an important part is meted out to procedures preserving these magnitudes:
discontinuous ones such as break and make procedures, and continuous ones,
the geometric correlate of which are mappings. It looks like a vicious circle to
claim in the constitution of magnitudes an essential role for procedures leaving
their magnitudes invariant, but as we know, it is not.
Indeed, in Sections 1.16 and 1.1820 I linked the constitution of
rigid body,
flexible body,
to processes of

displacing in a gentle way (not badly belabouring),

reversible plying and bending with negligible effort,
rolling off with no skidding
as physical correlates of
length-preserving transformations.

The physical context gentle treatment, negligible force, and non-skidding

breaks the circle, which would be a vicious one, if I would have length constituted
by the principle of length-preserving transformations.
The length-preserving mappings of the plane (and space), that is the congruence mappings, play also a part in constituting, comparing, and measuring
area (volume) because they preserve area (volume) too; together with the break
and make transformation they would even logically suffice for the constitution

of the concepts area and volume. I recall furthermore the

shearings,

a geometrical rather than physical kind of transformation. As regards area I

could add the
flexions,

a rather restricted means to compare curved surfaces with each other and with
parts of the plane as far as area is concerned. As regards volume, however, I

have more powerful transformations of a physical nature at my command, the

pour and mould transformations,
which in the course of constituting the mental object volume certainly account
for an important contribution. A liquid, poured forth and back, assumes, if it is
wanted, the same, that is a congruent, shape, although the mapping, as far as the
single particles are concerned, need not at all be a congruence. Plastic material
can after moulding be brought back into its initial state to fill a congruent part

of space but the pointwise mapping from initial to final state is not a congruence.
The excavated soil fills again the pit (unless it is too forcefully stamped), but the
soil grains will have been displaced in an unaccountable way.

M E A S U R I N G BY M E A N S OF G E O M E T R Y

395

This preservation of filling space then is interpolated to get a preservation of

volume: in between the initial and the final state the volume has remained the
same, albeit in another space. Of course this interpolating conclusion is only
accepted if the deformed substance has not been too badly belaboured. A rigid
body, molten or even gasified, can when resolidified take the same shape,
although meanwhile it has expanded its volume. A gas can at equal temperature
and pressure take place in the same container though if these parameters are
changed it can expand or contract. But for all these volume changing transformations there are still kinds of
pour and mould transformations

which by their physical appearance suggest an invariance that no doubt plays

a part in the constitution of the mental object volume.
Pour transformations are a well-known method of practically measuring
volumes of liquids or liquid-like matter. The object to be measured is brought
into a special shape, most often a cylinder with prescribed base such that the
volume is proportional to the height. This is how gauged litre measures, calibrated
measuring-glasses, bushels (to measure volumes of flour, grain, potatoes) are
used. Gauged conic funnels are also not unusual.
Measuring glasses are also used to determine the volume of solids: the substance is immersed into a liquid and the rise of the liquid column is observed.
The solid is as it were re-created into a liquid cylinder a congruent displacement procedure followed by a deforming pour transformation. Let us call it
by a specific name:

immersion transformation.
Altogether this is a rich variety of transformations that are relevant to volume.
Let us add the remark that for areas there is nothing like this. The area

preserving mappings are geometric constructs rather than correlates of physical

operations. That is why conservation of volume has drawn more of the
attention of psychologists than has conservation of area unfortunately in
a way that other invariances, in particular break and make transformations, have
been disregarded.
Pour and mould transformations are as it were break and make transformations made continuous. This continuation of the break and make procedure
need not be advantageous for the constitution of the mental object volume;
one can even imagine that it would cause troubles. The fact that the same
blocks fill again the same box can, because of the discreteness of the packing
activity, be a more important contribution to the constitution of the mental
object volume than the stream of pouring, gentle moulding, or immersing, which
cannot be followed in detail. Experiences with malleable and liquid substances
are less easily interpreted than those with solids.
Anyway, if conservation of volume is discussed, it is useful to specify the

396

CHAPTER 13

transformations with respect to which this conservation is meant. In Section

4.15 I formulated such general invariance principles as invariance
in the course of time,
under change of standpoint;
or

if something is added, it becomes more,

if something is taken away, it becomes less,
if nothing happens, it remains the same.

With regard to volume I can now add

shake transformations,
shearings,
break and make transformations,
pour and mould transformations,
immersion transformations
as physical transformations, and of course as geometrical physical transformations
displacement and other geometrically defined mappings.
13.39. In Section 4.14 I stated:
Up to the present day nobody can say whether the stress Piaget put on

conservation was justified, whether indeed certain invariances characterise

the constitution of certain mathematical objects, whether this might be

true of all mathematical objects for which Piaget developed conservation

criteria, for some of them, or for none. I guess that in principle Piaget
chose the right way, but I believe he deserved to be followed more critically
on this way in every detail than he has been in fact.
If I reread this and try to apply it to the mental object volume, I feel
undecided about whether to take a stand about it: I am hesitating between a
decided yes and as decided a no. One of the reasons is the lack of phenomenological clearness in Piagets work and even more in the research of others
that continued it.
In those circles people generally distinguish between invariance of
quantity (also called substance),
weight,
content (or volume).

As a matter of fact, for the same substance under the same conditions these
three magnitudes are physically identical I mean as seen from the viewpoint
of adult physical concepts. The researchers are right not to accept this identification, but they are wrong to deal with these three mental objects as separate.

M E A S U R I N G BY MEANS OF GEOMETRY

397

Conservation of volume is attributed by them to an even higher developmental

age than that of quantity and weight, at ages that even exceed the high age
stated by Piaget.

The difficulty is how to distinguish the three of them. By separate tests, or

if you do it by a single one, how to formulate the test questions? Weight looks
to be no problem it is explained by the adjective heavy. For quantity and
volume the questions are of the kind is it the same quantity? or does it take
the same space? in particular the second question presupposes much of
the linguistic development of the subject and of the linguistic ability of the
interviewer.
13.40. Let me first of all analyse the concepts at issue from our own viewpoint.
Quantity obviously means something by which only specimens of the same
matter can be compared; it does not make sense to ask which of two pieces,
one of clay and the other of iron, is more. Quantity is truly the thing that can
be said to remain the same as long as nothing is added or taken away.
The same quantity of something also has the same weight I disregard the
hardly relevant difference between mass and weight (in the sense of force). But
by weight different matter can be compared. Weight applies more universally.
The same quantity of a matter can under varying circumstances occupy
varying parts of space in the sense of volumes. Metals if heated expand; many
materials shrink under pressure. These are facts of physical experience. One can
imagine that a subject with a clear view on volume reacts the wrong way on
tests because he is not acquainted with certain physical facts or nurtures questionable theories about them.
13.41. Conservation of content and volume presupposes gentle transformations,
not too hard nor too hot. In this respect pour transformations look the kindest.

Strangely enough nowhere in the literature I studied is pouring a liquid used as

a test of conservation of volume. The reason is probably historical: when Piaget

made his famous experiments on pouring liquids, he was interested in conservation of quantity and thus formulated his questions in this sense. This has slavishly
been imitated. To my view, besides break and make transformations, pouring
liquids is a particularly appropriate means to observe phenomena of conservation
of volume, at least qualitatively. The most neutral experiment, which to the
best of my knowledge does not occur in the literature, is the following:
Cylindric vessels of different width and height, one of which is filled with
water (powder, gravel) the subject is asked to predict the level after
pouring the content into another vessel.
Children from the age of four onwards with whom I tried it unsystematically
answered it in a reasonably satisfactory way.

Obviously this really tests something with respect to volume, whereas the
usual pouring tests are connected with questions like is it the same?

398

CHAPTER 13

13.42. In contradistinction to this, conservation of volume is usually tested by

mould transformations, with clay or plasticine as material: if a ball is deformed
into a sausage, there is a large percentage of subjects conserving quantity and
weight, but denying that of volume, among which are even 16-year olds, which
runs counter to Piagets general theory which attributes conservation of volume
to the period of formal operations.
First of all we should ask the question as to what our certainty is based on
that clay and plasticine do not change their volume when kneaded. The immersion experiment is verbally adduced as an argument but it is never performed.
I would not stake my head on the statement that kneading is really such a gentle
operation, and I can imagine that denying it has nothing to do with misgivings

about conservation of volume. Kneading and moulding need not be that gentle.
For rubber, if stretched, no conservation of volume holds, if I am not mistaken.
Bread-crumbs can be compressed to small sizes, and this happens irreversibly;
a ball of wool yields, though the operation is to a certain degree reversible. A
full suit-case can be filled even more. Ironed linen takes less space than unironed.
It is a rich variety of experiences in which clay and plasticine are singular cases.
Then, what does clay and plasticine put in that singular position? This cannot
be answered by geometry.
13.43. Difference of weight is unmistakable. One can feel it with ones muscles.
It is easily understood why a child answers questions about weight reasonably
though these answers need not prove much, and certainly not what is wanted,
about weight. For quite a time children believe that they can make themselves
heavier.
Children also know rather early that a bagful of paper is lighter than the same
filled with iron. A bagful what does it mean? I think something like a volume
measure, though a bag can be packed more or less tightly.

13.44. I am inclined to think that the development of the mental object volume
should be rather viewed within the context of packing and filling. In order to
test it, one should design experiments of quite a different kind from those which
are usually performed.
Take two bags of different shape, which can be filled with the same number
of marbles (of the same size): a spherical bag, and an oblong narrow one.
(Obviously the second has a larger content, though it is badly packed.)
Then ask what would happen if the two bags were filled with smaller marbles,
or sand, or flour.
From the answers one could draw true information about how far a child is
advanced in separating geometrical and physical properties.
I am sure it is no fabrication of mine to put it that an envelope is judged to
be smaller of content if it packs more poorly. A non-conserving subject of

MEASURING BY MEANS OF GEOMETRY

399

the age of 16; 5 motivates non-conservation of volume if a plasticine ball is

kneaded into a sausage in a way* that fits quite nicely my hypothesis:
The molecules may be more compressed in the ball than in the sausage; although the sausage
has the same number of molecules and the same weight, its volume is not the same.

13.45 I am now coming to the immersion transformations. The question as to

whether they matter for the constitution of the mental object volume is even
more critical than it is for the moulding transformations. Numerous experiments
show that subjects relate the rise of liquid to weight rather than to volume.
This is not surprising. The most striking behaviour of solid bodies in a liquid
is floating or sinking, and whether it is the one or the other, the child knows
quite early, depends on weight, whether they are lighter or heavier than the
liquid. Compared with this striking phenomenon the rise of a liquid as a consequence of displacement is an unspectacular phenomenon, never observed before
by many among the subjects in this kind of experiment. It is not at all a matter
of course to put immersion into a geometric context. To be sure, it is quite
an achievement if a child does perform it and even more if it happens spontaneously but I would hesitate to admit that it is essential for the constitution of the mental object volume. It is one application among many others,
which is not at all obvious. Adult researchers can be excused if they think that
it is a trifle; they do not realise that it is an ingenious method, allegedly invented
by Archimedes. In their mind it is the method, which they learned in the physics
lesson, to determine volumes of solid bodies. They simply forget what they owe
to physics instruction.
13.46. Piaget and most of his followers set out to study spontaneous
development. What, heavens, is spontaneity? Does the childs environment
consist only of objects (unconcerned adults included) or arent they taught, the
one more, the other less? And what do the tests reveal, is it really spontaneous
or elicited by the tests? All right, let us forget about these questions. As a
pedagogue one is fascinated by other things than a psychologist is. Not by stories
about Caspar Hausers and wolf children, but by instruction and education.
The psychological approach can, and sometimes must, be adidactical. The
educational experimenter wants to observe learning processes. Even if he renounces the steering of them consciously, his design will anticipate a desired,
perhaps spontaneous looking, learning process. Piagets and his followers
experiments are valuable for the educator who wants to learn how it should not
be done from an educational point of view. (It is a pity that nevertheless some
educators slavishly imitate the experiments.) Certainly it is worth a learning
process to understand that kneading clay or plasticine preserves the volume
(at least if it is true). Certainly a child should learn to detach the water displacement by an immersed body from the context of its weight. Certainly he should
* Quoted by J. G. Wallace, Concept growth and the education of the child, Nat. Found.
Ed. Res. In England and Wales, Publ. 13 (1965), p. 111.

400

CHAPTER 13

learn to eliminate disturbing physical factors to arrive at the geometrical volume.

For all these, learning processes are required, which no doubt need to be made
conscious to teachers and learners, but in the didactical context they are located
at another place, far away from the first steps towards the constitution of the
mental object volume.
It should start with
break and make transformations,
building, converting, rebuilding, pulling down of buildings consisting of,
say, congruent cubes or simple parts of them,
their volume and its additivity,
(which includes subtractivity) with a stress on
the difference between volume and surface area.
It could continue with
the equivalence of content of open barrels and volume of solid bodies.
Further with
practical pour transformations to compare contents.
Beside this mathematically structured material one should deal with
less structured material that with regard to volume is compared by
estimate.
Moreover one should
connect the packing effect to, and if need be, eliminate from, the
content.
One should tackle
volume-preserving and non-preserving transformations,
though not in the petty context of clay and plasticine. The
volume determination by immersion
could be in a sense the coping stone yet not until many other things, such as
weight and force, have been dealt with in the course of instruction.
In the foregoing list I skipped a few items, which I kept for the next sections.

13.4750. Measuring in One, Two, Three Dimensions

13.47. It looks so simple, the relation between measuring in one, two, three
dimensions. A length unit determines an area unit, distinguished by the exponent
2 at the length unit, and a volume unit, with exponent 3. Area is length times

M E A S U R I N G BY M E A N S OF G E O M E T R Y

401

width, volume is length times width times height, where width and height are
only synonyms of length. Length and width are measured horizontally, and
height vertically, though to make things worse, there is a horizontal and a
vertical direction in the plane, too.
As far as mappings are concerned, the stress has been up to now on those that
preserve length, area, volume. When I discussed the mental constitution of rigid
body, I ventured to claim a larger part played by similarities than is usually
admitted. I take up this thread to look at what it means for length, area, volume.
Congruence mappings preserve each of them, similarities multiply their values
by a certain factor, which depends on the magnitude but not on the particular
figure. Each similarity can be composed from a congruence, and a geometric
multiplication, which is solely responsible for the special behaviour of the
magnitudes at issue.
A geometric multiplication by the factor multiplies
lengths by
areas
by
volumes by
There is hardly any insight that is as fundamental as this one for the constitution
of these mathematical objects and their measurement. It can be acquired
paradigmatically

for

line-segments, squares, cubes

with simple factors like

and is transferred to other figures

initially straight away,

and subsequently argued by means of
break and make procedures,
grid coverings,

approximations.
This principle deserves, as far as the moment of constitution and the stress
are concerned, priority above algorithmic computations and applications of
formulae because it deepens the insight and the rich context in the naive, scientific, and social reality where it operates.

402

CHAPTER 13

Formulae for perimeter and area of the circle, for area and volume of the
sphere and other solids are didactically and practically overshadowed by the
knowledge about their behaviour under enlargement and reduction, which
applies in a large field, not covered by formulae. The world literature from
Plato onwards is full of misapprehensions and warnings against misapprehensions
regarding the different behaviour of one-, two- and three-dimensional figures
under enlarging and reducing operations. The warnings are not superfluous.
Reason cannot be dispensed with to correct the eye, if the eye does not agree
to accept that of two vessels which at sight do not differ that much, the one
contains 1 cc and the other 10 cc.
What is even more important is what I called the rich context. Both Lilliput
and Brobdingnag are impossible for physical and biological reasons. Similarities
can physically be realised only within a narrow range and only macroscopically.
If miniature models of the tides in the North Sea are to be constructed, sophisticated measures must be taken to account for the fact that gravitation and size
of sand grains cannot be miniaturised. Even more fundamental are the limits put
on physical similarities by the physical constants, charge and mass of elementary
particles, speed of light, and Plancks quantum, which are not susceptible to
change. The size of cells is not as constant, but still not variable enough to allow
similarities on a large scale in biology. A more mathematical feature is the
restrictions imposed on similarities by the geometrical dimensions: forces
and interactions, determined by one, two, three dimensions, which must be
equilibrated, stress and strain and speed of flux in one dimension, exchange of
matter and heat through surfaces in two dimensions, inertia and gravity in three
dimensions. How can an elephant twice the normal size get its food and get
rid of its heat? How much should a mouse half the normal size eat to protect
itself against the loss of heat? How should the bones and muscles of a double
size horse be built to prevent its knees giving way and to allow it to move its
limbs? In Brobdingnag and Lilliput, how long, big, thick are the telegraph wires,
the water conduits, the antennae, the rain drops and the hail stones?

13.48. We already know that

area = length times width
volume = length times width times height

are too simplistic formulae. There are, however, shades of simplicity and complexity. First of all, the formulae
area = base times height
volume = base times height

for figures which measure the same at each height. Then for triangles and
pyramids, cones,

M E A S U R I N G BY M E A N S OF G E O M E T R Y

403

Where do these factors

come from? This question can be answered on
various levels.
The simplest paradigmatic:
a square with side a, thus area , divided from the centre into four triangles
with base a and height
(Figure 188), which consequently have each an area

a cube with side a, thus volume

pyramids with base
and height
each a volume

divided from the centre into six right

(Figure 189), which consequently have

A high level almost calculus:

a triangle with base b and height h is approximated by the unions of n rectangular strips (Figure 190) with
base

and height

together an area

which with

converges to

404

CHAPTER 13

a pyramid-cone with base area B and height h approximated by a union of n

prismatic discs (Figure 191 with base areas
and height

together a volume

which for

converges to

Finally in Calculus:
Take similar triangles with their areas as a function f of their variable
height h,

with fixed
the difference

is the area of a strip of height

on the base, the difference quotient

approximately equals the base, and the differential quotient

equals the base,

Thus indeed

M E A S U R I N G BY M E A N S OF G E O M E T R Y

405

Take similar pyramids-cones with their volumes F as a function of their

variable height h. Thus

with fixed

is the volume of a disc of height

on the base, the difference quotient

approximately equals the base, and the differential quotient

equals the base

Thus, indeed

13.49. The foregoing allows for a number of generalisations.

Consider a polygonal track L (closed or not) consisting of tangential segments
of a fixed circle with midpoint M and radius h, and the planar piece spanned by
M and L. It is a union of triangles, which can be considered as one figure with
base L and height h (Figure 192). So it again fulfills

Let now all tangential segments or some of them shrink such that the base
becomes a circular arc or composed of tangential segments and circular arcs, or
the whole circumference. Still

406

CHAPTER 13

with the special cases

Consider a polyhedron P (closed or not) consisting of tangential planar pieces

of a fixed sphere with midpoint M and radius h, and the space spanned by M and
P. It is a union of pyramids, which can be considered as one pyramid with base
P and height h. So it again fulfills

Let now all tangential planar pieces or some of them shrink, such that the
base becomes a spherical cap or composed of tangential planar pieces and
spherical cap or composed of tangential planar pieces and spherical caps or
the whole surface of the sphere. Still

with the special cases

These results can also be derived by the method exposed at the end of Section
13.48, that is, by considering similar figures as functions of the variable radius.

CHAPTER 14

TOPOGRAPHY WITH GEOMETRY

14.12. Localisation
14.1. The primordial parameter system to localise a point, consists of
standpoint

distance (of the point to be localised)

direction.

The standpoint need not be an

actual standpoint.
Indeed, if somebody tells somebody else how to go from P to Q, the point P
is an

imagined standpoint.
The distance can be defined by true

length measures
or for instance by required
travelling times
travels on foot, horseback, bicycle, carriage, car, train, plane, or by sound or
light. The direction can be a
looking or showing direction,

sound direction,
yah, there the sounds come from
following a street, road, river, edge of a wood, and so on,
or more objectively determined by the
point of the compass

or by the
angle with a fixed line,
angle with a fixed plane.

Another quite usual localisation of a point is as a

crossroads,
407

408

CHAPTER 14

that the crossing of two paths, rectilinear or otherwise conventionally determined.

Less obvious and less usual as a parameter system for localisation is

two or more standpoints,
their distances (from the point to be localised),
where the choice of only a pair (triple) of points in the plane (space) causes an
ambiguity.
Much more usual is the choice of a
standing line or plane through two or more standpoints, the distance of
which is known,

the directions (towards the point to be localised) defined by the angles

with the standpoint line or plane,
that is, as intersection of two rectilinear paths, which in turn are given by
standpoint and direction (Figure 193).

Iterative application of this kind of localisation is the principle of surveying,

where starting with a certain base line the procedure continues with a long
chain of triangles and angle measurements.
Localising of a global trend takes place

on the terrestrial globe,

on the celestial globe,

where in order to do justice to history I should rather interchange the first two
items. A definition such as sometimes found in newspapers,
300 km South West of Gibraltar,
or in popular astronomy

3 South West of Aldebaran

are localisations as considered in the beginning, where on the celestial globe
distance is measured in degrees.
14.2. So far we considered local means of description albeit in a global context.
Localisation in a global frame is done by

coordinate systems.

T O P O G R A P H Y WITH G E O M E T R Y

409

The first example in history was the celestial globe, where a point is localised
by

two angular coordinates,

each of which is the angle of the vision line with a plane. The two planes can be

horizontal and meridional plane

of the observer; the relevant angles are

height and azimuth.
Or

the equatorial plane and the plane through celestial poles, and vernal point;

the relevant angles are

declination and rectascension.
Or with the equatorial plane replaced by the ecliptic, the angles are
ecliptic latitude and longitude.

The second system has its match on the terrestrial globe: the
equatorial plane
is a natural datum; the angle of a radius from the centre to the point P with this
plane is the

geographical latitude of P.
The other plane is conventional: the

meridional plane of Greenwich,

with respect to which the
geographical longitude of P
is measured (Figure 194). Whereas the geographical latitude is easily derived
from the elevation of the pole, the geographical longitude caused navigators
a lot of trouble in the past. The most natural way to find it was by means of
the difference between local time obtained by astronomic observations and
Greenwich time, mediated by a trustworthy clock. Meanwhile these methods of

localisation have been outdated and replaced with

radar bearings,

410

CHAPTER 14

which reminds one of the parameter system explained in Section 14.1. Big
planes, however, fly on a modern version of what seafarers called
dead reckoning

direction observed and speed estimated where thanks to modern instruments

and computer processing, estimation has yielded to calculation.
All these coordinatisations on spheres have in common
structuring by means of

two antipodic points,

the bundle of circles through them,
and their orthogonal trajectories.
Localising in the universe requires besides that on the celestial globe,
distance measurement in the universe,
which for short distances can now be done by radar bearings, and for long
distances thanks to the cooperation of all kind of phenomena in fundamental

theories a fascinating subject, which is here out of issue.

14.35. Cartesian Coordinatisation
The run-of-the-mill rectangular coordinate systems have not yet been mentioned.
Coordinatisation in geometry is a recent acquisition. Though already in antiquity
Apollonius described conics by a symptom, that is, an equation between
variable geometrical magnitudes related to the conic at issue, and Descartes
extended this method to more general curves, coordinatisation of the plane
(space) itself, that is, localising points by pairs (triples) of numbers, is an idea
that gradually gained ground only after Descartes, mainly to satisfy the needs
of analysis and analytical mechanics.
The

cartesian coordinatisation
of plane and space is now applied as a means of localising
on geographical maps,
on city maps,

T O P O G R A P H Y WITH GEOMETRY

411

on building plans,
on plans of buildings,
in the planning of building,
in city planning,
on a chess board.
Moreover it has become

a model for visual organisation

of material and information according to two or three characteristics:
Tables with two entries (rows and columns),
perhaps even stacked in a third dimension

is as it were the geometric image in the coordinatised plane (space) of the

pairs (triples) structure

introduced in the material.

The cartesian coordinatisation of the plane (space) implies
structuring by means of two (three) systems of parallel (perhaps mutually
orthogonal) lines (lines and/or planes).
14.4. Once the cartesian coordinatisation of plane and space had been invented,
as the aftermath of the functions of two and three variables in analysis and
analytical mechanics, functions of an arbitrary number of variables came into
being, for instance, a system of k particles, localised by the 3k coordinates of
its members, soon complemented by its 3k momentum coordinates. The term
n -dimensional cartesian space

for the system of ordered n -tuples is of a more recent date than the application
of geometric methods in this structure.
In our century

geometric terminology
and

geometric visualisation
have become a habit wherever from n systems

the
cartesian product
is formed, that is the system of ordered n-tuples

412

CHAPTER 14

Sometimes it is not much more than terminology, sometimes an illustrating

model, for instance around the

timepath graph,
when the cartesian product is formed from
time and path axis
and

uniformity of motion and rectilinearity of the graph

go together. In a more general way the cartesian product serves as a model to

combine
magnitudes in pairs, triples, and so on, and to represent connections
between them graphically,

among others
proportionality by linearity.

14.5. According to Piaget* the plane is mentally being constituted as a cartesian

product a view that is the consequence of a logical systematisation of which
the product of relations is an essential element. This view is supported by
experiments where the subject is not allowed to do anything but structure the
plane as a cartesian product. In another series of experiments** the authors,
in spite of themselves, deliver all the evidence one can think of to refute Piagets
thesis.
In these experiments the subject has to reproduce a point P given on a rectangular piece of paper ABCD, by its image P on a congruent rectangular paper
ABCD. It appears that young children already understand the essence of this
task and can contrive and sometimes even realise, intelligent methods to perform
it. Unfortunately they cannot meet with Piagets approval, because no solution
bears the characteristic of cartesian coordinatisation. Almost all of them localise
the point P (and accordingly P) by means of the first parameter system exposed
in Section 14.1. The subjects choose one of the corner, say C, as standpoint,
measure the distance CP and give evidence of the idea to carry over the direction
of the line CP. The experimenters, however, deny them all aids and appliances
to perform this, because according to Piagets theory directions and angles are
only constituted after this has happened to the plane, and then by cartesian
coordinatisation. One is astonished at the inventiveness displayed by some
subjects, for instance, Col (5; 8) whose attempt at solving virtually boils down
* J. Piaget and Brbel Inhelder, La reprsentation de lespace chez lenfant, Paris, 1948.
** J. Piaget, Brbel Inhelder and Alma Szeminska, La gomtric spontane de lenfant.
Paris, 1948. In particular Chapter VII.

TOPOGRAPHY W I T H G E O M E T R Y

413

to producing CP until it meets another side of the rectangle and to measure and
carry over the line-segment cut off on this side an entirely acceptable method,
though not based on the multiplication of relations and therefore rejected.
For the same reason solutions are rejected where the subject besides measuring
the distance CP measures that of P to one of the sides of the rectangle. The only
solutions evaluated as correct are those where P is localised by means of its
distances from two orthogonal sides of the rectangle. Older children succeed
in doing this.
What is wrong with these experiments from the start onwards is the reduction
of the plane, which is, or is to be, constituted, to a rectangle. In both works
I cited this is a bodice created by Piagets logical multiplication of relations.
Why does Piaget not supply the subjects with a sheet in the shape of a circle,
ellipse, triangle or some irregular figure to locate a point on it? It is clear that
cartesian coordinatisation then is doomed to fail, whereas coordinatisation
by other means would succeed. But this is not allowed because it would be
contrary to Piagets dogma on constitution of the plane as a coordinate plane.
In the case of the rectangle as a reduced image of the plane the cartesian
coordination is indeed the natural one that is, for subjects who have experienced rectangles sufficiently, and in particular more strongly structured ones
than the bare rectangles administered by Piaget. It is quite probable that the
older subjects who succeeded were accustomed to ruled or squared paper and

consequently in possession of a stronger mental structure of the rectangle,

which might explain their cartesian coordinatisation which just works with
a rectangular sheet. Anyhow it is remarkable how well the younger subjects,
though deprived of useful aids, do perform. If anything these experiments prove
that the plane is not constituted the way Piaget asserts.
Piaget also quite briefly investigated localisation in space, which in a way
appears to function better than in the plane. Piagets explanation is not worth
anything. The material of this experiment is a little box where a point is defined
by a pearl on a vertical stick. Because of this vertical stick the right angled
parallelepiped is overtly presented as the cartesian product of the horizontal
rectangle and a vertical height. No wonder it is interpreted this way by the
subjects.

14.67. Polar Coordinatisation

14.6. Piagets idea of mathematics is one picture of mathematics, unambiguously

structured and moreover such that its structure is an image of cognitive development. Mathematicians are less dogmatic. They know about many a way to build

mathematics and they choose according to needs. Piagets starting point is the
straight line as a model of the linear order; plane and space must be constituted
as products of two or three lines, even if these products are in no way prestructured. As he noticed this cannot succeed but by prestructuring.
Piaget has intensively studied the linear but never if I am not mistaken

414

CHAPTER 14

the cyclic order. There is no doubt that cyclic orders are early mental objects
and arranging cyclically is an early mental activity,
sitting around a table,
standing or dancing in a circle,
walking in a circle
counting out,
walking around a block,

and many objects suggests cyclic order,

rims of dishes and cups,
dials.

I have dealt with cyclic order earlier*, also in its relation to linear order,
at such a length that didacticallyphenomenologically I have little if anything

to add. I summarise the main ideas:

recognising,
causing,
inverting,
analysing
the
discrete cyclic order,
continuous cyclic order;
further,

embedding and mapping cyclic orders,

relating cyclic and linear orders by counting on and rolling off respectively.

14.7. The motive for this digression is the cyclic order displayed by the system
of directions at a point in the plane. This system, determined by the horizontal

movements of the head, the horizontal turns of the body, and the possibilities
of stepping forwards, must be counted among our first mental objects. It is
structured by the frontbehind and the rightleft in a way that of course
changes with the position of the body. The primordial and natural coordinatisation is the polar rather than the cartesian one, that is, as explained in Section
14.1, with the standpoint as the pole, and as coordinates the distance and
direction, mathematised by the angle with a fixed direction. As we noticed,
Piagets experiments rather offer evidence in favour of the primordiality of this
kind of localisation. Historically, too, polar coordinate systems precede cartesian
ones.
* Mathematics as an Educational Task, pp. 465494.

T O P O G R A P H Y WITH G E O M E T R Y

415

Mathematically the polar coordinate system means a structuring of the plane,

after deleting one point (the pole) as the
product of the ray of the (positive) distances

by the cyclic system of angles (a circle).

Mathematically there is no reason why this structuring of the plane should be
inferior to that by means of the product of two lines, which is the cartesian coordinatisation. It is true that if there is no need to do so, nobody will prefer the

polar coordinatisation; the need, however, is often felt enough to be accounted

for.
Earlier the coordinatisation of the spherical surface was mentioned, in particular that of the terrestrial surface. This coordinatisation can be considered as a
restriction of the polar coordinatisation of space. After a point (the pole) has
been deleted, space is structured as the
product of the ray of (positive) distances
by the system of directions,
which in turn can be structured, for instance, by meridians and parallel circles.
Another coordinatisation of space is by cylinder coordinates, that is, structur-

ing space as
product of a straight line
by a plane with a polar structure.
14.8. Algebraisation
The global localisation by means of coordinatisation leads to the algebraisation
of geometry. Whereas the polar coordinate systems used to describe the celestial
vault and terrestrial surface primordially served to systematise the localisation,
the cartesian coordinate systems became particularly effective for describing
geometric figures and mechanical movements, and later on, mappings in general.
A figure is thus algebraically translated into a relation between coordinates, a
movement into a function of time, a geometrical mapping into a system of

functions of a certain number of variables. In this course historically viewed

the theory of conies is algebraised and at the same time extended, Newtons
geometrical mechanics transformed into the more efficient analytical mechanics,
and the theory of mappings is designed a priori algebraicallyanalytically. This
will later be exposed more broadly.

14.923. Orientation
14.9. In the topographical context Section 103 dealt with polarities. There
are more polarities than the ones mentioned there. Quantified ones like gainloss,
assetsdebts, heavylight, hotcold and non-quantified ones like goodbad,

416

CHAPTER 14

beautifulugly. Since plus and minus signs, positive and negative entered
mathematics, we have got accustomed to mathematise polarities by plusminus,
positivenegative, to distinguish opposite forces by signs, as we call positive and
negative the two kinds of static electricity, human attitudes and values, and
in general use this model of polarity quantified and nonquantified.

This indication of oppositions in which ever sense by opposite signs is

indeed an important piece of model forming. We will study it more intensely
when we deal with negative numbers and directed magnitudes. At the present
opportunity of mathematising polarities from the topographic context, we
restrict ourselves to such polarities as need no further quantification: front
behind, leftright, belowabove, left turnright turn.
They are, as I explained at the former opportunity, closely connected to the
human body but have been detached from it already in the topographic context.
As far as mathematising means detaching from reality, the bonds with the
human body and topography must be untied, though in a way that it does not
obstruct restoring the ties.
Between the polarities just mentioned, there are relations; we need not repeat
circumstantially what has been said in Section 10.3 about them. These mutual
relations will show the way to, and the need for, mathematising.
The first three polarities we recalled are linear, while the fourth may be
called cyclic or circular. By this I mean that their obvious standard models are

the straight line, and the circle, respectively, both of them travelled in a certain
sense, and on the other hand that these polarities can be used to establish on
straight lines and circles respectively, a sense of travelling. Of course, in this
context the straight line may be replaced with any simple curve, open at both
ends, the circle by any simple closed curve ultimately also by a screw.

14.10. Let us first consider linear polarities. I step upon a line in order to look
along it and to mentally travel on it from behind to the front. Or I look upon
a line in front of me in order to mentally travel on it from the left to the right,
or from below to above.
Behindin front of, leftright, belowabove are linear polarities, on straight
lines which are situated in space or with respect to my body in a way that allow
me to indicate a mental travel upon them. By change of perspective I can establish
on each straight line in space a behindin front of, leftright, belowabove, just
as I like it or need it, in order to indicate a mental passing over the line. In this
process of mentalisation the spatial behindin front of, leftright, belowabove
is replaced with the temporal pastfuture.
It is profitable to choose a neutral term to indicate this polarity, and as such
positivenegative presents itself. The linear polarity can then be expressed as
follows:
Globally: The straight line possesses two opposite senses of travelling, which
can arbitrarily and according to the needs be distinguished as positive and
negative. The line provided with this sense, is called oriented.

T O P O G R A P H Y WITH G E O M E T R Y

417

Locally (p): A point p on the straight line has two sides it divides the line
into two rays , which can arbitrarily and according to needs be distinguished
as positive and negative.
Connection between Globally and Locally (p): Running from p positively
on the line, one gets the positive side of p.
Connection between Locally (p) and Locally (q): If q is on the positive
side of p, then the positive side of q is a fortiori on the positive side of p.
The ultimate mathematisation of these phenomena is the
Linear order: A set S with a binary relation such that for all
one
of the choices
is realised, and transitivity, that is

for all a, b, c, holds, is called ordered. The opposite order is the relation

with

The straight line is by preference ordered by two mutually opposite order

relations such that for the oriented line it holds that:

lies at the positive side of a.

14.11. Though the polarities in plane and space bear a circular or screw-like
character, when passing from the line to plane and space we are confronted with
the straight lines and their orientations as our primary objects of attention,
thanks to polarities, specified by the new situation, of behindin front of,
leftright, belowabove. These polarities can be
topographically influenced
as far as the belowabove is tied to the upright, they can mentally get detached
from topography by a change of perspective that nominates an arbitrary plane
to become the
mental standing plane with its own belowabove.
By change of perspective a

mental behindin front of,

mental leftright
can replace the polarities determined by the actual countenance of the body.
The polarities can be related to an actual or mental plane
on which one is standing,
in front of which one is standing
in which one is standing (lying).

418

CHAPTER 14

A change of perspective we have got so acquainted with in early childhood that

we deal with it unconsciously as an obvious one, is that between
the vertical plane from which, and
the horizontal plane on which one copies,
where

leftright is unchanged
and

belowabove is replaced by behindin front of,

such that finally
behindin front of is even indicated by belowabove.
If one renounces combining two polarities in a planar structure, space shows
three polarities at a time,

which in a first step of mathematisation are described by three oriented

(mutually orthogonal) lines through one point. Going on one can have an
arbitrary triple of lines

play this part, where at most the arrangement of the axes (x -axis, y -axis, z -axis)
and a sketchy figure reminds one of the original role distribution. Such a system
of axes is used to coordinatise space. In physical applications the relatedness of
the system of axes to the human body is felt. Physicists have the habit of distinguishing lefthand and righthand coordinate systems. It is not farfetched to
suppose at the background of this terminology an impossibility in principle
or in practice to detach the space used in physics, entirely from the topographical context influenced by the human body.
14.12. In order to deal with this problem let us return to two dimensions.
Linear polarities separated from each other do not suffice to grasp the situation.
On the plane where I am standing, I can make a right or a left turn. At the
plane which I look at, I can make a turn with, or counter to, the clockhands.
In the plane I am leaning against I can have my arms swing from above to in
front to below, to behind, and again to above, or the other way round. I can
impose a turning sense to each plane whatever it may be. I need not restrict
myself to horizontal and vertical planes to impose turning senses on them.
Each simple closed curve Jordan curve in such a plane with a sense of
circulating around it contaminates with its sense of circulation as it were all
others. If in a plane two Jordan curves are given, one of which with a circulation
sense, I can immediately at a glance fix what is the same sense on the other
one. How is this decided?
How can I see it this was my original question but possibly it is more

TOPOGRAPHY WITH G E O M E T R Y

419

feeling than seeing, a kinaesthetic rather than optical experience. Something

is turning if I perceive a Jordan curve with an arrow; mentally I am running
over the curve and while I do so, something is turning simultaneously, the
surroundings of the curve, the whole plane. Anyway, this holds for Jordan
curves that are circles or like circles. How does it happen?
I am fixing a point o, in the interior of the curve if it is a circle, by preference the centre. The ray from o to a point p on the curve moves with p

a turning movement that involves the whole plane. To be sure with more
complicated Jordan curves (Figure 195) it does not work that smoothly, the
ray can swing back and forth but finally a whole turn, in the one sense or the
other, will be accomplished.

Turns of the plane can take place around different centres. How do I know

whether they are the same sense or the opposite one? This, too, seems primordially to be decided kinaestheticalry. One carries with oneself a feeling for
the turning sense, and so it is carried from one place to another. Mathematically
this can be established more precisely by shifting the rotation that determines
the turning sense by means of a translation. One can also look far away where
it hardly matters any more which point is the centre of the turn.
This transport of the turning sense of the plane conversely takes with itself

the sense of running around a Jordan curve. But the relation between the senses
of two Jordan curves can be made in still another fashion: under a continuous
transformation in which the two Jordan curves appear as the first and last of
a continuous array of Jordan curves, the sense is preserved. The question as to
whether in such a process a Jordan curve cannot be transformed in the opposite
one, will not be asked in this context let alone be answered by a formal proof.
Let us summarise the phenomena discussed here and their mutual relations:

Ru(J): the two (left and right) senses of travelling a Jordan curve in the
plane,

Tu(o):

the two turning senses of the plane around o.

For two Jordan curves J , J and two points o, o:

Ru(J) and Ru(J) are related by continuous deformation transfer,

Tu(o) and Tu(o) are related by parallel transfer, view from far away, or
continuous transfer,

420

CHAPTER 14

Ru(J) and Tu(o) are related by assuming o in the interior of J and having
the ray from o to a variable point of J and the whole plane participate in
the same turn.
There are two mutually opposite senses at issue. The decision about
which one is called the left or the right, is related to the human body or somehow motivated by physical phenomena.
Just as the behindin front of, the leftright, and the belowabove on a line
are determined by choosing an at least mental position on or with respect to
the line, so is the left or right turn in the plane determined by an at least mental
position on or with respect to the plane.
When standing on the plane, a right turn means that the vision line turns
towards the right hand, that is, the direction that means front turns to the right
perhaps in order to return after four quarters of a turn to its original situation.
When standing in front of the plane (for instance a clock dial) a right turn
means that the above turns to the right (the clockhand from 12 towards 3)
again a quarter of a turn which when repeated four times restores the original
situation.
(When standing within the plane, the arm swinging from above to the front,
or the body falling on the face, determines something that can be called a left
or a right turn, depending on the side from which it is viewed. In other words:
sitting on a bicycle I cannot tell whether the wheels turn left or right.)
Just as on the line, there is a need in the plane for complete mathematisation
in the sense of detachment of the human body or the topographical reality of
which right and left turn are derivatives, and just as on the line, there is a need
in the plane for more neutral terms like
positivenegative

for the two mutually opposite senses. By assigning the predicate positive
to one of the senses the plane is transformed into an
oriented plane.
The phenomena and relations expressed by means of the abbreviations

Ru(J) and Tu(o)

are then to be read by replacing

leftright
with
positivenegative.

T O P O G R A P H Y WITH G E O M E T R Y

421

14.13. By the preceding the explanation of what is a left or a right turn is

reduced to the cooperation of two polarities in the terminology of standing
upon the plane:

at a left (right) turn the front swings to the left (right).

We could also say to the left (right) side. Indeed a straight line in the plane

has two sides (half planes). If the line in the plane I am standing upon (in front
of myself) is oriented, I can distinguish the sides as left and right; if the orientation of the line is reversed, left and right are interchanged.
This concerns the plane I am standing upon (and with the usual change of
perspective, the plane in front of me). If travelling a Jordan curve in the plane
or the turning sense of the plane is at stake, we can free ourselves from the bond
with the human body or with topography by assigning arbitrarily to one of the
senses the predicate positive and to the opposite one the predicate negative.
We can do likewise with our reformulation of what is a left and a right turn.

A point determines two sides on a line; one of which can arbitrarily be called
positive and the other negative. Similarly I can arbitrarily determine a

positive turning sense

in the plane by assigning

to an oriented line l

in the plane
its positive side:

under this positive sense, some p situated on l at the positive side of o turns to
the positive side of l (Figure 196).

It is intuitively obvious that this definition does not depend on the choice
of o and p on l (provided p is lying on the positive side of o).
By moving the oriented line l I can carry with it its orientation as well as the

definition of what is its positive and negative side. It is likewise intuitively

obvious that this does not influence the definition of the positive turning sense.

Summarising these phenomena and relations, I can say:

S(l): the two (positive and negative) sides of an oriented line l in the
plane,

422

CHAPTER 14

S(l) and S(l) related by continuous transfer,

S(l) related with Tu(o): the positive turning sense turns the positive
part of the oriented line l through o to the positive side of l.
Here we understand Tu(o) in the way that left and right are replaced with positive

and negative. It is, indeed, usual in mathematics to identify in drawings the

positive turn with the left turn. This is a consequence of the fact that one
draws the first axis horizontally and the second vertically; one is as it were
standing mentally on the horizontal axis, while viewing to the positive (right)
side; then the left side of this oriented line is lying above; the positive turn

happens from horizontally left towards vertically above, which is against the
clock and all that is called right turn in everyday life.
14.14. Like a straight line any curve extending from infinity to infinity has two
sides, which if the curve is oriented can be distinguished as left and right. Or
conversely: one assigns to one side the predicate positive and to the other the
predicate negative and by this way defines the positive turning sense of the
plane.

I will take this variant into account by allowing l in S(l) to be a curve extending from infinity to infinity.

14.15. An even more striking case is the simple closed curve the Jordan curve
which moreover has an interior and an exterior. Just as the positive (left)
sense of J was fixed, so the interior is lying left of J. It is then quite natural to

assume that this interior of J is its positive side.

Summarising:

Int(J):

a Jordan curve determines in the plane an interior and an exterior;

if J is being travelled in the positive sense, the interior of J is

left of J.
If two Jordan curves J, J , lying outside each other, meet along an arc, then
while this arc is being travelled, one of its sides is interior to J and exterior to
J and the other side exterior to J and interior to J. Thus

Int(J) and Int(J) are under these circumstances related in such a way
that if both J and J are being travelled in the positive sense the common
arc is being travelled in opposite directions.
This leads to
Int(J) and Int(J) related by combinatoric transfer:
Between J and J one does not interpolate a continuous array for continuous
transfer but a chain of Jordan curves such that each has just an arc in common
with the next, and so transfers the sense from each to the next for instance,
between two oriented triangular circumferences a chain of triangles (Figure
197). In Section 13.27 this has been anticipated.

TOPOGRAPHY WITH GEOMETRY

423

14.16. Whereas on the line the

global idea of orientation

is easily given a precise meaning, in the plane it starts so vaguely that it requires
a profound analysis. We got acquainted with three ways to fix a positive orientation of the plane,
by defining a positive turning sense,
by defining a positive sense on a Jordan curve,

by defining the positive side of an oriented line.

This can be extended to polyhedra and curved surfaces, though it requires some
caution. Let us explain this with a spherical surface.
A rotation around o is at the same time one around its antipodic point o,

but for somebody standing on the surface at o the right turn at o is a left turn
at o. With a Jordan curve on the sphere there is no distinction between interior
and exterior, whereas in Ru(J) the point o must taken in the interior in order

to create a link with Tu(o), if an orientation is to be imposed on the surface.

There is no obstacle if one chooses the third way: a Jordan curve gifted with
a positive sense, the positive side of which determines an orientation of the
spherical surface. But if one passes to other kinds of surfaces, one has to be even

more cautious: on the torus there are Jordan curves that do not divide it and
consequently have one side only. As soon as one restricts oneself to small

Jordan curves, one can distinguish two sides, and can stick to the method
used on the sphere. Such difficulties arise with surfaces that globally differ

from the plane closed surfaces and surfaces on which Jordans theorem does
not hold. This is a reason for starting the business of orienting on such surfaces
locally:
A surface is considered
as flat in the neighborhood of a point o

424

CHAPTER 14

polyhedra are so in a general point and curved surfaces can be approximated

by flat pieces. The surface is oriented
by imposing an orientation on the planar piece,
which can be done in one of the three afore mentioned manners. By
continuous transfer to other points
the surface gets oriented at any point. According to Section 14.15 rather than
in a continuous way this transfer can be carried out

combinatorically.
For instance, for a polyhedron:
for each face a positive sense is established on its border in a way that
common segments of neighboring faces are travelled in opposite directions.

A curved surface can be dealt with similarly by division into small pieces.
There are, however, also
non-orientable surfaces,
as is the Mbius strip (Figure 198), a strip with the opposite short sides stuck
together after inversion. Here the above procedure fails: after a trip around the
strip, divided into triangles, one returns with a triangle bordered in the opposite
sense.

14.17. Up to now I have neglected a phenomenologically important source of

orientation: the reflections (planar ones if the plane is concerned). As a matter
of fact the need for orientation as a mental object is primordially or at least
to a high degree due to the existence of reflections. The d and b are the
same and yet not the same how can they be distinguished? The plate of Figure
199a does not fit into the hole of Figure 199b why not? Curls and spirals
are of two kinds what is the difference? The S is easily confused with its
mirror image. Space is even more intricate: the left shoe does not fit the right
foot; in twin houses the stairs turn differentely.

T O P O G R A P H Y WITH G E O M E T R Y

425

A reflection in one dimension, on the line, interchanges the two senses. A

reflection in the plane one is standing on or in front of you interchanges right
turn and left turn, thus both of its orientations however the plane is situated.
The planar congruences are of two kinds: motions that preserve orientation, and
slide reflections inverting it. Two figures that arise from each other by side
reflections do not differ from each other in their internal properties: it is perhaps
for this reason that the optical sense needs the support of the kinaesthetic one
to constitute the difference between left and right and between left and right
turn as such.

How this comes about, how and when things are made conscious that occurred
in our phenomenological analysis I cannot tell. I know that some children
have difficulties with left right, and with left right-turn, but I did not observe
learning processes related to it, and I do not know about people who did.
Neither geometrical propedeutics nor advanced geometrical education traditional or modern pay attention to this complex of phenomena and relations,
and so it is not to be wondered at that it has eluded the attention of psychologists and educationalists. This is just as true with respect to orientation of
space, which will be our next concern.
14.18. I am starting with two striking phenomena. The first is what I would like
to call the

stamp imprint effect,

or the
shop window inside outside effect
a right turn on a stamp is printed as a left turn, a right turn drawn inside looks
like a left turn from outside.

The second phenomenon is exemplified by the existence of

screws, cork-screws, winding stairs
of two possible kinds right and left ones.
Both phenomena are closely connected. Let us first consider the second,
which is the most solidly materialised. All cork-screws and most screws show

the same turn; it is quite natural and anyway usual to call it the right one.
A striking geometrical feature: if the screw is standing in front of you the
screw-thread runs from left below to right above, even though you view the

screw from the backside or upside down.

About winding stairs at least in old castles there is a tale that most of them are left

winding, for a good reason, as is claimed: so the defender of the castle who is facing downwards has more elbowroom for his right arm the sword arm than the assailant facing
upwards. The minority of right winding stairs it is claimed were built by left-handed
castle lords.

426

CHAPTER 14

Other examples of right screws: the hair crown of the human skull, the
hop stalk, most of the snail-shells. Left screws: the bean stalks.
The ordinary so-called clover leaf knots can be distinguished according
to right and left ones (Figure 200).

Running along the knot one has to consider how a new piece of the string
turns around one that has been passed before. Righthanded persons have the
habit of tying right knots.
Kinaesthetically the (right) cork-screw is characterised by a
synthesis of right turn and inwards,
which is also experienced with the screw-driver.
One can turn a right screw into a thin plate representing a plane from
one side or the other; this causes within the plane opposite turns depending on
the side from which the screw enters, although both turns are viewed as right

ones at the side where one is operating this then is the shop-window effect
or, closely connected the stamp effect.
The act of turning in or screwing in provides the turn or screw axis with a
sense, which quite naturally may be defined as the positive one. The plane
orthogonal on this axis has two sides half spaces; operating the instrument
I am looking from the negative to the positive side.
We notice the coupling between three phenomena:
a screw sense in space,
a turning sense in a plane,
a distinction between the positive and negative side of the plane.
Each pair of them determines the third:
given a screw (right or left) and an oriented plane if the screw is to be
drilled into the plane in the positive sense, it is determined from which
side it shall be done,
given a screw (right or left) and the side from which it is entered, the

turning sense in the plane is determined,

given a turning sense in the plane and the side of the plane from which
the screw is driven, it is determined whether it is a right or a left screw.

T O P O G R A P H Y WITH G E O M E T R Y

427

14.18. About a straight line or a plane in space I cannot tell any more what is
left and right, left turn and right turn as soon as they are detached from their
topographical context. I can arbitrarily call one orientation positive and the

other negative, with no regard to polarities depending on the human body and
topography. In order to state what is a right or a left screw, I can dispense with
the topographical context, but I must know what is my right and left hand.

If this knowledge is eliminated as not belonging to mathematics, then one is

still left with the existence of two kinds of screws and too screw senses

though the question which is right and which is left, becomes meaningless.
Among the methods of putting an orientation upon a straight line or a plane,
there was one, proposing:
a line is oriented by assigning to a point on it a positive side,

a plane is oriented by assigning to an oriented line in it a positive side.

This can be continued as follows:
the space is oriented by assigning to an oriented plane in it a positive side.

It should be noticed that:

if the positive side of one oriented plane is fixed, it is so by continuous

transfer of any oriented plane, in particular, if an oriented plane is transferred into its opposite, its positive sides are interchanged.
If the preceding definition is to be fed back to screws in real space, I have
to settle how the plane is to be oriented in reality and which is its positive side.
Then the orientation defined by the right screws is the negative one, just as for
the positive orientation I have to use the left screws.
Summarising I state the following phenomena and relations
Sc(P):

the space possesses two orientations (screw senses), which can

arbitrarily and according to needs be distinguished as positive and

negative. Fixing the positive orientation is equivalent to fixing

the positive orientation and the positive side of a plane P in space.
Sc(P) and
can be related by continuous transfer of the plane, its
orientation, and its positive side.
14.19. In order to impose an orientation on space, we do not need full planes,
rather we can be satisfied with little pieces. As a consequence we can replace
the plane in the preceding by any curved surface if it extends to infinity, as does

the plane. We can, however, do it also with an oriented closed surface and make
good use of the fact that it possesses an interior and an exterior. As a convention

we can consider its interior as the positive side, which then fixes an orientation
of space by means of an oriented closed surface:
C(S):

an oriented closed surface S determines an orientation of space

with its interior as its positive side.

C(S) and
can be related by continuous transfer.

428

CHAPTER 14

If two such surfaces S and S are touching each other from outside along something looking like a circular disc, one of its sides belongs to the interior of S
and the exterior of S, and the other to the interior of S and the exterior of
S. Thus
C(S) and
are under these circumstances related by the opposite
orientation borne by the common piece in S and S.
Moreover
C(S) and

can be related by combinatoric transfer:

Rather than by a continuous array and continuous transfer, one joins S and
S by a chain of closed oriented surfaces where each touches the next along

something like a circular disc, for instance, between two oriented tetrahedra,
a chain of tetrahedra (Figure 201) each of which touches the next from outside
along a whole face.

14.20. An oriented tetrahedron lends an orientation to each tetrahedron, by

continuous or combinatoric transfer. Before introducing the usual symbolism

for oriented tetrahedra, I shall return for a moment to lower dimensions.

The orientation of a line can be given by choosing two points a, b; the positive
sense of the line (and the line segment) is indicated by
ab.
The orientation of a Jordan curve can be given by choosing three points
a, b, c, such that the curve is travelled from a via b to c; the orientation of the
curve can be indicated by

a b c,

b c a,

cab

as you want it.

In particular the orientation of the border of a triangle can be indicated by
its vertices, placed according to the orientation
a b c,

b c a,

c a b.

The orientation of a tetrahedron with vertices a, b, c, d is determined by that

of a face, say that with vertices b, c, d. Let its orientation be
b c d (or equivalently c d b, d b c).

TOPOGRAPHY WITH GEOMETRY

429

Then by the rule of opposite senses of the common edge the orientation of the
other faces is

a d c (or equivalently d c a, c a d),

a b d (or equivalently b d a, d a b),
a c b (or equivalently c b a, b a c).

By means of the oriented tetrahedron, space is oriented: the interior of the

tetrahedron is its positive side. Thus: of each face the opposite vertex is on its
positive side.
If one prefers a notation for the orientation of the tetrahedron by means of
all of its four vertices, I might convene that the fourth vertex is added to the left
of the triple. Then the oriented tetrahedron is indicated in twelve ways:

abcd,
badc,
cabd,

acdb,
bdca,
cbda,

adcb,
bcad,
cdab,

dacb,

dcba,

dbac.

The 24 permutations of abcd fall into two classes, the even and the odd ones,
according to whether they are obtained by an even or an odd number of pairwise
exchanges. The above lists just the even permutations of abcd.
Thus, the orientation of a tetrahedron can be indicated by its vertices in
order, where

an even permutation describes the same orientation,

under the orientation abcd the faces are oriented as bcd, adc, abd, acb
and each vertex is lying on the positive side of its opposite face,
in other words,
a on the positive side of bcd,
b on the negative side of acd,
c on the positive side of abd,
d on the negative side of abc.
Feeding back to the space of our kinesthetic experience we can distinguish
a right and a left screw sense, respectively suggested by the left and the right
drawing of Figure 202.

430

CHAPTER 14

14.21. Starting from the orientation abcd of the tetrahedron (and thus of space),
one can arrive at another description of spatial orientation, that is by means of
an ordered triple of oriented lines
da,

db,

dc,

which can be identified with the triple of axes of a coordinate system: the third
axis shows to the positive side of the plane oriented by the first and second axis.
Fed back to the space tied to our body it is identifying the first axis with the
direction leftright, the second with backfront, the third with belowabove,
as is often suggested by drawings: the turn from the first to the second axis
combined with the progression along the third gives the left screw sense.
Another method to orient space is by means of a (non-ordered) pair of
oriented skew lines: assuming on each two points in the order of orientation

ab

and

cd,

and orienting the space by the oriented tetrahedron abcd. (The order of the
pair does not matter since cdab represents the same orientation.) One can
connect with it the kinaesthetic experience of screwing ab into cd.
Feeding back into physical space, it becomes meaningful to talk about two
oriented straight lines in terms of whether they cross each other left or right.
This applies also to clover leaf knots: two pieces of the knot are considered
as skew oriented lines; according to how they cross each other it is a left or a
right knot.
14.22. The existence of two orientations or screw senses of space is a mathematical fact; it depends on knowledge about our body which is called left and
which right we said. Does it really depend? The screw sense is applied in
physics: If electricity flows through a conductor above (or below) a magnetic
needle, the needle turns as though it tries to stand orthogonal to the current.
It is settled by the cork-screw rule in which sense it turns: One imagines the
(right) cork-screw with its axis in the conductor drilled into the direction of the
current; then the North pole of the magnetic needle moves as though it is hit
by the handle of the cork-screw.
It looks as though the right screw sense can be defined by pure physics.
Is this true?
The direction of the electric current and the North pole of the magnetic
needle are defined by convention. The electric current flows if it is a dry battery
from the zinc to the carbon this is an objective physical datum. The north
pole of the magnetic needle shows North, but the North is a topographical
datum, fettered to our Earth. So I would not be able to inform extra-terrestrial
beings by means of the cork-screw rule what right and left means among people
on Earth.

TOPOGRAPHY WITH G E O M E T R Y

431

However, there are other natural laws, related to the so-called weak force,
which do allow us to characterise right and left screws as such.*
14.23. What was didactical phenomenology in the sections on orientation?
Perhaps a few remarks at the end of Section 14.17, which bear witness to my
ignorance on genetic and learning processes with respect to orientation, left
right, left turnright turn; I can now add the screw senses to this list. I guess

that these objects are mentally constituted with kinaesthetic support. This
would be an interesting subject for research. Of none of the relations between
backfront, leftright, abovebelow, left turnright turn, left sideright side,
interiorexterior, left screwright screw, which I analysed phenomenologically,
would I be able to tell, when, how spontaneously, how directed, how explicitly
they can, may, must be learned, and how far the road from the mental object
to the concept should be travelled. Though important enough, I do not consider
it at present as a learning matter. I dealt with it circumstantially to stimulate
research in this field.

* See H. Freudenthal, Mathematics Observed, World University Library, 1967, Chapter VII.

CHAPTER 15

NEGATIVE NUMBERS AND DIRECTED MAGNITUDES*

15.1. An Apology
I have hesitated for a long time about which order I should deal with the subjects
indicated in the title. The chapters of this phenomenology are not logically
ordered; even a posteriori I do not see any possibility of introducing such an
order. I cannot do without anticipating and resuming. I do not write a mathematical treatise with a deductive structure that helps one to avoid logical circles.
In general the mathematical ideas under consideration are supposed to be
known. In the phenomenological approach I may sometimes suppose that the
reader is familiar with mathematical subject matter which phenomenologically
is delayed and dealt with later on.

15.23. History
15.2. Historically negative numbers are much earlier than directed magnitudes.
If precursors, as in Hindu mathematics, are disregarded, negative numbers arose
about 1500, though three centuries passed before they were wholeheartedly
accepted; the directed magnitudes are an invention of the 19th century. This,
however, does not say anything about their mutual relation.
The origin of negative numbers is of course the algebra of equations, such as
they are symbolically written. Methods of solving such equations were developed
and gradually automatised. As the automatisation progressed, one wanted to
extend their domain of validity. One tried to remove obstacles; solving equations
such as it had been automatised should go on under all circumstances. The
equation
should possess a solution x under all circumstances (of course provided
we would say at present, but in history it was the quadratic equations (and

equations reducible to quadratic ones) that created this need, say, something
like
or

* Directed magnitudes are dealt with in this chapter only in connection with negative
numbers, thus with the stress on directed rather than on magnitude.

432

N E G A T I V E N U M B E R S ET AL.

433

The algorithmic procedure led to two solutions, one of which was impossible but would nevertheless be admitted, at first hesitatingly but then with an
ever increasing degree of conviction. It is the same idea that led to imaginary
numbers, in particular when solving the cubic equation: pressing the solving
formula at any price. So it is not to be wondered at that the imaginary numbers
did not enter that much later than the negative ones. The resistance against both
of them lasted about equally long.
The first extension of the natural number concept, towards the fractions,
had been much less problematic. From the first mathematical documents onwards we meet with fractions. Heart-searching in this domain was of a much later
date, in Greek mathematics, where for philosophical reasons it was forbidden to
break the unit. The Greek mathematicians replaced fractions with ratio while the
calculators in commerce as well as in science continued to do it with fractions.
One can readily understand why the step from natural numbers to fractions
was unproblematic. Fractions have to do with and are required for magnitudes,
that is, wherever people are measuring and where continuous quantities are
divided, and from the first calculating activities onwards people had to perform
these activities.
Mathematics in the Greek sense is about numbers, and as far geometry is
concerned, about magnitudes a view that even mathematicians in more recent
times tried to share, at least in theory. The negative numbers had originated
from the formal algebraic need for the general validity of solving formulae,

but not until the algebraisation of geometry (the so-called analytic geometry of
former times) did they become effective I mean content effective.
The idea of algebraically describing geometric figures and solving geometric

problems is older than Descartes. We owe to Descartes the use of one coordinate
system (to express it in modern terms), independent of the figure and the
problem. Algebraisation and coordinate systems were dealt with in Chapter
14, where we tacitly presupposed negative numbers. Descartes had some trouble
with them; numbers were introduced as magnitudes; letters indicated magnitudes,
thus positive numbers. But those who applied Descartes method, could no
longer avoid having letters also mean negative numbers. If straight lines are
to be described algebraically in their totality, if curves are to be described
algebraically in any situation, one cannot but admit negative values for the
variables. The need for
general validity of algebraic solution methods

to which the negative numbers owed their existence, is reinforced from the 17th
century onward by the need for

general validity of descriptions of geometric relations.

The second need, more content directed than the formal algebraic one, is the
most natural and compelling. It is properly responsible for the success story
of negative (and also of complex) numbers.

434

CHAPTER 15

If the negative numbers are introduced, it does not suffice to claim their
existence this is often didactically overlooked, as it happens also with the

rational numbers. The negative numbers become operational by their use in

calculations, which obey certain laws. As I have explained elsewhere* at great
length, these calculation laws are uniquely determined as extensions of certain
laws governing the positive numbers. I gave the idea of extending while preserving
certain laws the name of
the algebraic principle.
It includes what I just called the

general validity of solution methods,

and virtually it is the same idea, albeit formulated in a broader view.
I recall a few examples of the algebraic principle.
is proved by starting with the definition equations for a,
adding them formally, while using commutativity and associativity, in
order to arrive at the definition equation

for (3 + 4).
Or:
Starting with the same definition equations, one proves

by multiplying distributively the first with 4 and the second with 3,

and subtracting them from each other.
Or: With
defined as the x making
by multiplying the definition equations

to get
Similarly if operations are to be extended,

because both terms have the same nth power.

* Mathematics as an Educational Task, pp. 224, 275 sq.

one gets

N E G A T I V E N U M B E R S ET AL.

435

With hindsight the term algebraic principle looks to me too colourless.

In the sequel it will be the

algebraic permanence* principle,
if I mean the in general unique extension of algebraic structures with
preservation of certain fundamental or fundamental looking or as desirable
considered properties. As far as negative numbers are concerned, the problem
is to introduce new objects beyond the positive numbers with operations such
that the impossibility of
with
is abrogated, while other laws, which can be specified, are preserved and such
that these acquirements of preservation imply the uniqueness of extension.
In history the negative numbers (and later on the complex ones) have been

invented according to the algebraic permanence principle. Initially people

worked with negative numbers in a naive formal way. Later on arguments of
content character were contrived to be adduced, which still subsist in newer
arithmetic and algebra textbooks, although some of them are not quite convincing (positive-negative as capitaldebt, gainloss, and so on).
15.3. I have discussed earlier the algebraic permanence principle and its
didactical implications at such a length that there is little to be added. I only

stress once more its didactical soundness and the didactical weakness of the
traditional objects against it.
I have propagated still another didactical means to introduce the negative
numbers, the

induction extrapolatory method,**

using tables like

If I do not deal with these methods now again, I do not mean to renounce them.
I would rather view them more closely connected than I formerly did to the

geometrical means of introducing negative number. Didactically I now see the

algebraic permanence principle rather as a
geometricalgebraical permanence principle.
* Recalling Hankels permanence of the calculation laws.
** Mathematics as an Educational Task, p. 281.

436

CHAPTER 15

This fits better the historical course; I already mentioned that the negative
numbers did not become really important until they appeared to be indispensable
for the permanence of expressions, equations, formulae in the analytic geometry. What I called the inductive extrapolatory method also fits into the
geometricalgebraical context (in particular of mappings), in which justice will
be done by the analysis I intend to carry out.

15.4. Old Models

Long before analytic geometry, numbers were used in geometry. Lengths of line
segments were measured by numbers of a unit. This then was a way to introduce
numbers and operations on numbers in a geometrically-based building of algebra.
Numbers as coordinates were used in astronomy with characteristic nonnumerical extra data such as East and West, South and North.
Coordinatising a road by mile stones has already been exercised in antiquity.
Each path or line can be measured in this way: measuring distances from a fixed
point the origin and noting them down. Roads start somewhere (in antiquity
of course in Rome, whereas in later times all of them led to Rome); even after
arbitrary extensions they remain half lines. If the origin (say Rome) is passed,
the half line can virtually become a whole one. Two points on such a track
with the same distance from the origin can be distinguished as back and front,
and in order to measure distances on such a path one has to distinguish pairs
of points at the same side of the origin from those on different sides.
To make it more precise let us view a horizontally drawn line. The points
can be determined by their distance, measured in a certain unit, from the
origin, with the extra datum left or right. If the mutual distance between
two points is to be expressed by means of such data, three cases are to be
distinguished: both points left, both points right, one left and one right.
Once negative numbers are admitted, it is not a farfetched idea to count
further on the number line beyond 0 with negative numbers, stepwise subtracting
a unit, or straightaway placing a at the mirror image of a with respect to the
origin. With capital and debts the term is black and red, reminding one of the
old table computers and the print-outs of more modern ones. Or the centigrade
thermometer they used to call it heat and cold and indicate it red and blue, but
in meteorological surveys it is now plus and minus.
Lodging the negative numbers on the number line is of course not the thing
that matters. Negative numbers are not only introduced; they are to be operated
on, added, subtracted, and later on, also multiplied and divided, to be put into
exponents and in all kind of functions. Let us first view adding and subtracting.
Models in the older literature to introduce negative numbers and their operations, are

the game of gain and loss, also with debts,

the stairs up and down, also into the cellar,
the thermometer up and down, also below the freezing point.

N E G A T I V E N U M B E R S ET AL.

437

The second and third are similar to the number line, and the first can also be
illustrated on the number line. The arithmetical operations are in all cases
related to events, the result of a game, a movement on the stairs, a change of
temperature. The same holds with the model that is now preferred,
the number line, and jumping on it.

The negative numbers are reached as soon as

one loses more than one possesses,

one descends more steps than one was above street level,
the temperature drops more than it was.
Or with the number line as path: if
one jumps back more than one had advanced.

These models make plausible

adding a positive number a to any given one,

subtracting a positive number a from a given one.
With the given point as a variable x, it means concretising the functions

with a positive a.
That is all these models can grant. In the older textbook literature more was
tried, but the best they could do, was to decree honestly but with a brazen face:
adding (subtracting) a negative number is the same as
subtracting (adding) its opposite,
thus

for positive a.
This decree can hardly be justified by the models used. The most satisfactory
is perhaps

negative gain is positive loss,

negative loss is positive gain,

but negatively the stairs up and down and negative temperature gains and
losses are unworthy of belief.

438

CHAPTER 15

Of course by formal algebra it can be justified. From

it follows, if commutativity is granted,

and from this by associativity

and even more involved,

But this is a bit too much.

The fault of the models, dealt with so far, is the didactical asymmetry between
positive and negative numbers. The positive numbers are more concrete in
the sense of greater originality; so one can operate with them; the negative
numbers are secondary, introduced as results of operations, which formerly were
impossible, fit to be operated on if need be, but unfit to have operations performed with them. In other words, the positive numbers are active, the negative
numbers only passive.
If rather than asking for a model, one is satisfied with the formalism of what
I called the algebraic permanence principle, this difference is non-existent;
as soon as one has decided about extending, the negative numbers have the same
legal status as the positive ones; operating with negative numbers is formally

justified and in no way distinguished from that with positive ones.

15.58. New Models
15.5. If rather than being satisfied with the algebraic permanence principle, one
looks for more satisfying models than those dealt with so far, it is now clear
that positive and negative numbers shall be given the same opportunity. The
former models admitted a symmetry between
adding and subtracting

as inverses of each other

the one undoes the other.
What is asked for is an equal status for

positive and negative numbers,

such that the

operations can be performed by means of this equal status.

N E G A T I V E N U M B E R S ET AL.

439

A non-geometric example are

positive and negative counters,

a method proposed, if I am not mistaken, by Gattegno; a geometric model is

the
reinterpretation of points on the number line as arrows.

15.6. One works with, say black and red, counters such that

a black and red one can annihilate each other,

or conversely that
a blackred pair can come into being from nothing.

It does not matter which colour is identified with positive and which with
negative; only where it is needed for the notation we will identify black with
positive. The annihilation rule allows us to identify with each other for instance

7 black and 3 red ones,

6 black and 2 red ones,
5 black and 1 red one,

4 black and 0 red ones.

Of course also

3 black and 7 red ones,

2 black and 6 red ones,
1 black and 5 red ones,
0 black and 4 red ones.
As a matter of fact it means considering an integer as

an ordered pair of natural numbers

with the equivalence relation

thus as an equivalence class, or at least operating with them in a way that mathematically conceptualised is known as equivalence class formation. The additions
of type

can be performed with the black and red counters directly by

taking together.

440

CHAPTER 15

require
annihilation

of three black and red ones. The subtractions

73

7(3)

can be performed by

taking away,
but

7(3)

(7)3

would require taking away three red (black) ones from a pile that contains black

(red) ones only. This is the first opportunity to apply

creating from nothing:
in order 3 red (black) ones to be taken away which are non-existent, three red
black pairs are added to the present stock:

(10 black and 3 red ones) minus 3 red ones

(10 red and 3 black ones) minus 3 black ones,

thus

Finally one has the types

37

(3)(7),

when 7 black (red) are to be subtracted from too small a quantity. So one
completes the insufficient quantity by pair forming such that subtracting becomes feasible

(7 black and 4 red ones) minus 7 black ones

(7 red and 4 black ones) minus 7 red ones,
thus

These are various cases the enumeration is complete but all of them are
solved according to the same principle:

simplifying by annihilation,
pair forming if it cant be done.

N E G A T I V E N U M B E R S ET AL.

441

It recalls artifices in column arithmetic,

replacing 10 units by a ten,

dissolving a ten into 10 units,
and this analogy with

pair annihilation,
pair forming
is not at all superficial.
Multiplying integers cannot be dealt with in this model unless it is decreed
by brute force that
black times black = red times red = black

black times red = red.

15.7. The geometric model I announced in Section 15.5 is that of arrows. A
point a on the number line is at the same time interpreted as an arrow from
0 to a, that is, as an arbitrarily movable arrow, a vector. The objects are, mathematically understood,
equivalence classes of arrows
with equality of length and direction as equivalence relation. Or if the arrow is

hollowed out to become the pair tailhead, an

equivalence of pairs of points

with equality of distance and order as the equivalence relation.

A warning: Arrows most often curved ones are also used with the model
of the number line such as mentioned in 15.2 to indicate operations; then an
arrow of length a means a jump of width a along the arrow (to the left or right
according to the arrow direction), that is, the operation
or

As explained in Section 15.2 this leads no further than adding and subtracting
positive numbers. It is true there are textbook authors who tacitly switch from
that interpretation to this new view on arrow classes as numbers and by this way
suggest more than the first interpretation can yield a not unusual kind of
intimidation.

In order to avoid this misunderstanding it is worth recommending that one

puts the name of the arrow nor above but as its head (Figure 203). Moreover
adding arrows must be defined explicitly:

442

C H A P T E R 15

Adding arrows means putting them tail upon head, while using freely the
movability of arrows (Figure 204).

How to define subtraction? As the converse of adding? It is possible, but it

easily leads into the temptation we warned against of interpreting the arrow as
an operation rather than a number. It is better, and even easier, to define a b
by
So one gets

a b as the arrow from the head of b to that of a

provided the arrows a, b are joined at their tails (Figure 205).

It is perhaps surprising that in this model the subtraction is better visualised

than the addition. It is visually also obvious that 0 a is the vector opposite to
a. If the notation
0 a is to be replaced by a,

this should be done explicitly. The definitions imply that

and

but as such
( 3)

is meaningless

unless it is defined, that is by

0(3).

As soon as this definition has been accepted

is a meaningful notation for the mapping that carries each arrow in its opposite,
and one gets

NEGATIVE N U M B E R S ET AL.

443

Again it is clear that

adding and subtracting are opposites of each other.
The associativity of addition is conceptually obvious in this model, which
cannot be asserted as confidently of the commutativity, but also the visualisation
of commutativity is poor as shown by Figures 206 and 207.

Not only of commutativity. The authors of various textbooks are wrestling,

each individually and anew without benefitting from the lessons taught by
former failures, with the visualisation problems for this model of operations on
integers. Properly the arrows should be placed on the number line, but then
they fall together and cannot be distinguished. A way out is to draw them at

a little distance parallel to each other or to the number line, another one is to
replace them with arcs from tail to head it is not elegant but at least practical.
It might work as long as one restricts oneself to the operations of adding and
subtracting positive, in particular natural, numbers, and perhaps even in the
general case. But a convincing visualisation may hardly be expected of this
method.
After adding and subtracting, the next issues are multiplying and dividing.
Let us restrict ourselves to multiplying.

is meaningful and visually recognisable as a

dilatation
(and similarly for other factors). This mapping can be continued on the negative

side as a dilatation, which yields for instance

Or otherwise: it is visually obvious that a dilatation behaves symmetrically on

both sides of 0, thus

444

CHAPTER 15

Or still differently:

In general
which can be considered as meaningful and true for positive a.
Again this is extended by recognising

as a reversing dilatation of the positive ray and as such continued on the negative
side. Thus in general,

and similarly

Of course there are fewer words and formulae needed to visualise this. This
leads to a decent motivation of

Commutativity and associativity of multiplication are almost obvious. Distributivity, however, requires either distinguishing a number of cases or a
visualisation that is not very convincing though.

15.8. Translating both models of Sections 15.45, counters and arrows, with
each other, one can parallelise
red counters
black counters
black-red pair
annihilation and
creation
equivalence classes
of pairs of counters

arrow tails
arrow heads
opposite arrows
displacing of arrows
equivalence classes of arrows.

The two models differ in that

counters represent

arrows represent arbitrary numbers

integers
a counter is either red each point can be both head and tail;
or black

N E G A T I V E N U M B E R S ET AL.

445

multiplication is
an artificial

a natural

operation.

15.912. Directed Magnitudes*

15.9. As P. M. van Hiele has rightly remarked in order to draw the didactical
consequences the didactical malfunction of the arrow model is caused by its
one-dimensional character. For a didactics of directed magnitudes as arrows
are the dimension 1 is the most inappropriate, because in one dimension these
objects and their operations lack badly needed relief.
Directed magnitudes (such as speed, force, acceleration) occurred earlier in
mechanics than they did in geometry. The mechanical principle by which these
magnitudes become operational is the so-called
parallelogram of speeds, forces, accelerations,
as an expression of the way to

compose and decompose

such magnitudes.
Adding vectors in geometry starts with another pattern: by bringing one vector with its tail to the head of the other, one indeed sweeps out a parallelogram,
which moreover demonstrates the commutativity of the addition (Figure 208).
As a matter of fact, such a figure can degenerate, if the two vectors fall into
one line and the parallelogram collapses a degeneration, which should not
be neglected didactically though there is little need to exaggerate its importance.

The most comprehensive concept to fit these phenomena is vector space

or linear space (as called in linear analysis) over a field, say that of real numbers.
A brief definition: V is a vector space means:

V is a commutative group with respect to an addition +, thus

* See the note at the title of the chapter.

446

CHAPTER 15

for

and V allows for multiplication

with real numbers

thus for

such that

These axioms can be fulfilled by a visual geometrical model with

as elements of V the arrows from a point o,

as addition the tail by head composition after parallel displacement,

as

the geometric multiplication from o.

A mathematical model fulfilling the axioms is n-dimensional number space

the set of ordered n-tuples of real numbers with the operations + and
such that

This model is in a sense exemplary. If the vector space R has a basis

each vector

can unambiguously be represented by

Assigning to each x the system of its coordinates with respect to the basis

effects an isomorphic mapping of R onto

15.10. The axiomatic version of vector space is the final result of a historical

process. In a mathematical system it is rightly a starting point. The ndimensional number space
is a richer structure than the n-dimensional
vector space R, which in order to yield
must be enriched by a chosen basis.
Progressing from poorer to richer structures is a sound principle of mathematical
systematism. It depends on circumstances whether it is didactically as sound a
principle.

If one asks for didactical starting points, there are a few possibilities.
1. Arrows in plane or space with a point o as their tail, and as addition the
tail by head composition after displacement of one of the summands along the
other.
2. Equivalence classes of equal and equally directed arrows in plane or space

with as addition the tail by head composition of appropriate class representatives.

N E G A T I V E N U M B E R S ET AL.

447

3. The translations of plane or space with as addition the successive performance.

For 1 it is required that plane or space are first being enriched by an arbitrary
point o which finally does not matter much; this can be dispensed with by 2
and 3.
Connections between
1 and 2: Each equivalence class according to 2 is unambiguously represented
by an arrow starting at o;
2 and 3: An equivalence class of arrows represents a translation
1 and 3: The translations according to 3 have the form
to be
understood according to 1.
No doubt 3 is the most elegant and at the same time the least didactical
approach. A grave objection against 3 is that the
of a translation does
not possess a natural interpretation until a high level is reached.
An objection against 1 is the geometrically unmotivated choice of the point
o as tail of all vectors. This objection, however, can be disregarded if the intention
prevails that plane or space will anyway be pictured with a coordinate system
of axes intersecting in o.
To 2 the general objection applies which holds against the explicit use of such
logical complexities as equivalence classes. If such a terminology is avoided
as was done here in Section 15.7 this objection may be dropped. The best

choice seems a compromise between 1 and 2:

One considers arrows in plane or space, agrees on considering equal and
equally directed arrows as the same and givens and draws them by
preference with the tail at o.
This is what I called* the intentional versus the extensional concept
formation.

15.11. One can go a long way in mathematics with vector spaces before introducing a basis and coordinates; in the theory, in particular if it is functional
analysis, it is often unsuitable or even obnoxious to pass to special coordinate
systems. In geometrical vector spaces, however, the coordinate free approach can
degenerate into a hobby with respect to content as well as to didactics.
Certain figures ask, in order to be represented algebraically, for special coordinate
systems in this way Apollonius already used coordinates. In order to understand linear equations and solving them in the context of vector spaces, I am
compelled to enrich the space with a basis. Determinants as expressions for
volumes do not arise unless I choose a basis in order to substitute the coordinates
with respect to the basis in the expression for the determinant. To be sure,
this use of bases and coordinate systems can be wrapped into an unimpeachable
garment that is adapted to the system: a coordinatisation of an n-dimensional
* Mathematics as an Educational Task, pp. 3032.

448

CHAPTER 15

vector space R is nothing but a linear mapping of R on the n-dimensional number

space
indeed. But unless we are obsessed by formalising, we may happily
forget about it in the present context.
We were induced into the foregoing phenomenological analysis of vector
spaces by certain didactics of negative numbers. In Section 15.9 we took a
standpoint where negative numbers were already a didactical acquisition; whereas
in Sections 15.28 we were concerned about acquiring negative numbers, with
the final conclusion that directed magnitudes was a good approach, were it not
that if restricted to one dimension they create more problems than they are
called on to solve. The idea of passing to the vector plane was primarily suggested
here by a need to algebraise not the plane, but the line in order to yield the
number line. Primordially the plane will be an auxiliary means; if it can be more,
it is a bonus; even if it should dominate, it could be accepted since sometime
the plane will come about anyway. But if it will, can, must come up here intentionally, it has to do so as an algebraic, coordinatised plane. Not as an axiomatic
plane that is afterwards coordinatised and exemplified by an algebraic model,
but as a geometric object, as the drawing plane that will be as naively provided

with an origin, axes, and coordinates, as happened in history. The historically

more recent turn towards the axiomatic vector space, the inversion from the
numerical to the axiomatic vector space should also didactically be of a later
date after the mastery of negative numbers and algebra rather than as a
certainly didactically insufficient precondition for it.
15.12. In P. M. van Hieles newest approach negative numbers arise in a twodimensional frame. A number pair

means 3 steps to the right,

means 3 steps to the left,
means 3 steps to the right,
means 3 steps to the left,

4 steps upwards
4 steps upwards
4 steps downwards
4 steps downwards.

The leftright, updown are those of the drawing plane, with horizontal and
vertical axes on which the numbers of steps can be read in units.
Performing such operations in succession one describes or prescribes rectilinearly constructed drawings in the plane. Adding these vectors is nothing but
performing these operations in succession.
arises in a natural way and defines as naturally what shall be
and

Let us consider a translation from his textbook*:

In Figure 209 a cats head has been partially drawn. You should finish it.
* P. M. van Hiele et al., Van A tot Z, 1a, Purmerend, 1976, pp. 7475.

NEGATIVE N U M B E R S ET AL.

a
b
c

449

Draw Figure 209 on squared paper.

Go from the fat point one step to the left and one upwards.
This is indicated by
Draw then a line segment going 0 steps to the left and 2 up. We write
it
Finish the picture by drawing one after the other the line segments

Take a piece of squared paper, 9 steps wide and 6 steps high. Start
at the point that is 4 steps from the left and 1 from below. Make
a maze by drawing from this point as a start one after the other the
line segments

Take a piece of squared paper, 8 steps wide and 8 steps high. Start
at the point exactly in the centre. Make a mill by drawing from
this point one after the other the line segments

In the two-dimensional model both the associative and the commutative law
are visually obvious; the same is true of subtraction as adding the opposite.
Likewise the multiplication is well visualised by a geometrical multiplication as
is the distributive law. Gradually, in order to attain the algorithmic automatisation, the bonds with the visual model are loosened. Meanwhile the visually
acquired insight into calculating in the extended number domain is again visually
used to extend functions formerly introduced by tables or else imperfectly such
as

It is an intentionally many-sided approach to calculations in the extended

number system.

450

CHAPTER 15

15.1315. The Geometrical-Algebraical Permanence

15.13. I take up the thread where I left it in Section 15.2. The negative numbers
were invented to lend permanence to algebraic solving methods, and the operations with them were defined to grant permanence to their properties. Afterwards
they entered geometry, which as far as it was algebraic, had from olden times
been concerned with magnitudes rather than with directed magnitudes. Soon
their indispensability in geometry became clear, that is, in an algebraised geometry that developed after Descartes. The negative numbers would have remained
a nice plaything, and the operations, motivated by algebraic permanence, rules
of a game, which could have been fixed in another way, were it not that geometry
had seized upon them. The negative numbers are needed to describe the whole
plane by coordinates and planar figures in their whole extension by equations.
The simplest figures in the plane, lines, are then translated by the simplest equations, those of the first degree, called linear because of their relation to straight
lines; circles and other conics are fitted by second degree equations. I think
that both in phenomenological analysis and didactics too little emphasis is
laid onto this fact:
the justification of the numerical operations and their laws by the simplicity
of the algebraic description of geometrical figures and relations.

Briefly said: algebra is valid because it functions in geometry.

It is strange that so far this insight has not or not strongly enough been pronounced. In my view it is one of the objectives of algebra instruction to convince
the learner of the validity of the operations and their property so forcefully that
he cannot but accept them. The most convincing argument is to show him the
operationality of algebra in geometry. This, I believe, should be our policy with
teaching negative numbers.
Here geometry does not mean an axiomatic structure but what is visually
obvious or conceptually follows from what is visually obvious a visuality
that neither requires involved explanations in the vernacular nor sophisticated
replenishments. The one-dimensional medium, the straight line, has not enough
visual structure, two dimensions is the minimum that is required, and with
a view to the graphic possibilities the most appropriate medium.

15.14. I am going to elucidate the preceding claims, and I will do so by the way
of a didactical sketch.
Functions like

NEGATIVE NUMBERS ET AL.

451

restricted to the positive realm, show when put into a graph, the image of a
part of a straight line. We wish their extension to be prescribed not by algebraic
but by geometric algebraic permanence. Let us start with
which before the introduction of negative numbers is meaningful only for

where it is represented rectilinearly (Figure 210). The rule of extension is

obvious: rectilinearity. This means geometrically and in a table (recall the
inductive extrapolation)

where the negative numbers are being put on the axes, albeit as abbreviations of
0 1, 0 2 , ... . Of course there is no need restricting oneself to integers.
On the contrary, the replenishment with

can add much to understanding.

So far negative numbers occurred as results only. If
is considered (Figure 211), geometric algebraic permanence requires negative x
to be admitted to fit the algebraic expression to the geometric image

452

CHAPTER 15

In fact the same step is required with

as soon as the graph is to be

extended left of the vertical axis (Figure 212). In

there is a need for extension on both sides since the graphical image is only a line
segment unless negative numbers are being introduced. Maybe by this very fact
this function if especially appropriate to be used to start with:

on the one side and on the other

By the preceding exposition addition and subtraction in the extended realm

are completely defined, at least if the commutativity of addition is assumed. The

introduction looks somewhat disintegrated but the operations can be integrated

in tables:

N E G A T I V E N U M B E R S ET AL.

453

(In the case of x y it is recommended to place no numbers upon the axes.)

From this representation it is a small step to arrive at nomograms for x + y
and x y (Figure 213) which with necessary interpolations define addition and
subtraction completely.

454

CHAPTER 15

The relation

is illustrated by the opposite slope of the nomograms for

The multiplication can be dealt with in a similar way.

and x y.

and

initially are defined only for positive x and thus graphically represented by half
lines. Geometric algebraic permanence requires that the representations is extended (Figures 214215) such that

N E G A T I V E N U M B E R S ET AL.

455

or in general, for positive a and b,

This leads to a nomogram of the multiplication by positive numbers, which
again according to geometric algebraic permanence can be filled up with lines
in the other quadrants.

This then yields the general validity of

What about the insight into the arithmetical laws in this approach? The
commutativity of the addition is visually obvious, the associativity appears from
the nomogram of
:

substituting y by

means

shifting the lines over z,

of course specified by numerical values of z.

Associativity and commutativity of multiplication, once supposed for positive
factors, cause no problem. The commutativity is illustrated anew by the symmetry of the nomogram for
(Figure 216).

15.15. Rather than leaving it here, I would immediately put the rectilinearity of
the graphs of

(for paradigmatic numerical values of a and b) into this context and deal with
the continued harmony between geometry and algebra as a teaching matter as
well as a means of motivation. I would do so each time in the sequel if a simple
figure is graphically represented by a simple equation.

456

CHAPTER 15

15.16. Order

So far in this chapter I have neglected the order relation. Without a model, say
can reasonably be argued,
is more difficult unless it is based on
that is requiring in general

In each geometric model the algebraic order is satisfactorily motivated by the

geometric order of belowabove (on the thermometer or in the staircase) or
leftright (on the number line). Likewise the behaviour of the order relation
under adding and multiplying
(1)
(2)

for

(3)

for

is visually obvious.
In the graphic context, order is tied to monotonicity, (1) and (2) express that
the functions

are monotonically increasing, (3) that

is decreasing. In general, each non-constant linear function is discovered to be

monotonic.
If graphs and nomograms are introduced early, as suggested by Sections
15.1315, there is another opportunity to demonstrate the algebraic-geometric
harmony, that is defining sets by inequalities: besides the line represented by
the equation
the half planes represented by the inequalities

N E G A T I V E N U M B E R S ET AL.

457

15.17. Reasoning Back from the Goal

After the preceding analysis we should keep in mind that the instruction on
negative numbers, if successful, should lead to automatisms. How does a person
who has become familiar with negative numbers, manipulate them, how does
he solve problems? I think the answer is so obvious and so unambiguous that I
may draw general conclusions from my own procedures to cover those of other
people and even of computers.
Let us first classify the possible additions and subtractions paradigmatically.

The first line consists of problems which, except for the last, are already meaningful before the extension of the arithmetical domain; the next lines contain
systematic variations, which become meaningful only after the extension with
negative numbers. The transformation rules I apply here, are

with the common pattern

And of course for the multiplication

These are not excessively many, nor complicated, transformation rules,

particularly if seen in the common pattern of the quartet of addition and subtraction. It is almost nothing compared with the rules a child must learn in order
to master column arithmetic. What does this mean for the didactics of this field?
That after all it would be the most simple and efficient way to programme
the learner with these six rules? This is all he needs, or isnt it? It is a question
I could have asked at many opportunities before and could ask as often in
the sequel. Though, there are reasons why it is asked right now. Elementary
arithmetic is blessed with a cornucopia of such transformation rules, from
arithmetic below 10 with all its connections between additions, subtractions,
and equations via column arithmetic to the various operations with fractions.

458

CHAPTER 15

Some of them are of the kind that every child not only the bright learner
can discover them if he is allowed to do so. Other rules are adapted to more or
less steered reinvention. But among them there are also a few that depend to
such a degree on appropriate and appropriately motivated definitions that
they cannot be understood unless their definitory character is made explicit,
with or without motivation. This holds in particular for multiplying and dividing
fractions.
It goes without saying that we strive for automatisation where it is required. The problem of whether automatisms are better pursued by goal directed
programming or as a result of learning by insight, is not at all original. There
are plenty of answers to this question, by way of theoretical expositions, of
textbooks, of teacher guides to textbooks, and by the ways individual teachers
practice teaching. Theoretical expositions and teacher guides no doubt emphasise
learning by insight even if as it happens sometimes they are belied by the
underlying textbooks. The actual policies in the classroom are certainly covering
a broad scale, from downright programming to learning by insight to attain
automatisation. First doing then understanding is the catchword of one of
our textbooks for secondary mathematics. To put it the other way round may
be a honest intention, or a cheap philosophy, or a swindle. First doing then
understanding can mean that some never achieve more than imperfect
doing, while understanding is an objective reserved for the privileged ones.
First understanding then doing can amount to verbalistic understanding that
does not lead and cannot lead to doing.
It is a good thing to fix the attention on three points while discussing the
learning of automatisms:
first, are the intended automatisms
worth being learned,
second, once learned,
how well do they function,
third,

how much value is the process of learning them?

These three questions are closely connected with each other. Clearly it is no
use learning automatisms
that are seldom applied and soon forgotten,
that are too complicated to be safely manipulated,
that can easily be confused with each other
unless the way of the check by insight of those automatisms remains open. On
the other hand too much insight during the learning process can block the
automatisation.
Is it in the era of the handheld calculators still meaningful to learn column
arithmetic? Well, if at the critical moment the calculator fails, what is one to

NEGATIVE NUMBERS ET AL.

459

do? Then automatisms learned by insight might help better than programmed

ones.
Tables of multiplications are most often learned by insight: the learners are
taught to build them by successive additions. After memorising, however, the
insight that multiplying comes down to successively adding, fades away. Why?
Because multiplication as repeated addition is only used to build up the tables,
in order to be repressed afterwards if column multiplications are learned and

exercised.

The consequences of column multiplications learned by being programmed

rather than by insight are particularly striking with, say 100, as first or
second factor. But the same features can be observed if multiplying was learned
by insight and has been repressed in the process of automatising.
Learning by insight, if restricted to an introduction by insight, does not
deserve this name; the didactic of teaching automatisms should rather be turned
to not repressing. This requires a more broadly flowing process rather than the
narrow one of learning algorithms. In the process that finally leads to acquiring
algorithms, more and other things should be learned than only that which
can aid the acquisition of automatisms.
Once more, why are these questions asked here? According to the tradition
negative numbers are not part of the teaching of arithmetic. Arithmetic is
usually taught and learned in narrow goal directed learning processes, though
sometimes a bit broadened by again too narrow applications. A pupil who
has gone successfully through these learning processes, may not, in general,
be supposed to have experienced them consciously enough to be able to reconstruct them. This kind of learning arithmetic has created a mentality where being
programmed dominates insight. It takes enormous efforts to break open the
automatism to have it understood by insight. Some textbooks at the secondary
level try it; at teacher training institutions one exerts oneself to have the future
primary school teacher experience consciously the processes that can promote
learning by insight.
After this long introduction, what can be said about the pros and cons of
learning arithmetic of negative numbers by being programmed or by insight?
The answer depends on the place negative numbers occupy in the total learning
process. If, after years of instruction in arithmetic negative numbers is the first
opportunity to learn meaningful, not yet algorithmised mathematics which
is the normal situation the answer must be different from what it might be
if the learner has already acquired a mathematical attitude.
The operations on natural numbers are usually learned by insight though it is
a fact that often because of the lack of relations with reality the original insight
is suffocated by automatisms. Fractions is an opportunity to justify the operations and the rules governing them, which, however, is most often disregarded.
Mathematics at least algebra is too often presented as a system of working
according to prefabricated and then imposed rules. There is everything to be
said in favour of preventing this impression as far as the learner is concerned.

460

CHAPTER 15

Certainly one is allowed to interpret mathematics also in this way, as a formal

system, as a game played according to certain rules, or even as a language constructed by arbitrary syntactic rules. This is a view of mathematics that under
certain circumstances foundations of mathematics may be useful. It is a
view worth being noted by somebody who masters mathematics at a high level.
It is a view of mathematics, yet it is no mathematics, and certainly it is no
mathematics to start with as a learner.
However one proceeds in extending the number concept, it is a necessity that
the fact and the mental process of extending are made conscious, that it is made
conscious why one has the extension take place in this, and in no other way,
and that as a background of the compulsory, the arbitrary in the definitions
becomes clear. To get this done I think negative numbers offers a good
opportunity, at least if others have been missed. Indeed, the rules of operating
on negative numbers are on the one hand relatively simple, wheras the constraint
to work with negative numbers just this way, is remarkably strong, provided this
rather narrow learning process is embedded in a broader current, like the one I
have sketched before.
With these remarks I conclude this chapter. Again I apologise that I dealt
with a question of mathematical principle not in a general context but at the
opportunity of some special subject.*

* Compare Section 6.19.

CHAPTER 16

THE ALGEBRAIC LANGUAGE

16.116.4. LANGUAGE IN THE BROAD SENSE

16.1. Language, Languages, Translation

The singular language is to obscure the fact that there are no two people
speaking the same language. Even one individual can use various languages
not only if he masters foreign tongues, but also depending on with whom and
how he communicates, orally or in writing, talking, calling, crying, whispering,
singing. Of course that individual will also listen to various languages. Learning
ones mother language is communicating with ones environment in two widely
differing languages, with adaptations which in the course of the time increase
at the side of the learner and decrease at that of the people he communicates
with.
Learning to read means initially becoming acquainted with a language cut
into letters, syllables, sentences, which compared with the spoken and heard
language is extremely poor with regard to form and content, until the printed

text starts carrying on words and constructions unknown and unheard in the
spoken language: as far as the level is concerned, the printed language overtakes
the vocal one.
When I put it a bit provocatively that no two persons speak the same
language, I used the word language in an improper sense, that is, not as I do
if I oppose, say, English and French. Between these two meanings of language
there is a scale of gradations depending on the means of expression spoken
and written language or on the environment dialect and educated language,
childrens language, boys language, girls language, bookish language, teachers
language, church language, thieves language, secret languages and on the
subject matter lawyers language, chemistry language, mathematics language.
Rendering the same content in another language is called translating, though
in the relation between spoken and written language it is called writing and
reading, in that between a language and a crypto language it is called coding and
decoding, and in the relation between people speaking the same language
it is transforming. Transforming is partly a lexicographic concern a word
or combination of words replaced with a synonymous one. Partly it is structural
and this is where so-called transformation grammar comes in.

16.2. Transformation Rules

In the course of individual development the first linguistic transformations are
determined by what I have called change of perspective. The example
461

462

CHAPTER 16

to the question where do you live? a child (2; 8) answers I lives [sic] there (showing
with her finger)

demonstrates a transformation of person description: you

These transformations form a whole bunch: I, me, my he, him, his are examples, which must

be replenished with proper names, other pronouns, personal descriptions by

means of appellatives, or of a relative structure like
the thief,
the man who has stolen the bike.
The first transformations by change of perspective are as such formal, the more
advanced ones show at least some formal structure.

By change of perspective
here and there,

inside and outside,

above and below

change their meaning according to who pronounces them or is considered to pronounce them, and correspondingly they are to be interchanged in communicating.
Of course these are not the only transformation phenomena: parts of clauses,
replaced with others, singulars with plurals and conversely, present tense with
past tense, active with passive, creating comparatives and superlatives, transforming verbs into nouns, and so on these are just a few examples, which
could be multiplied ad lib. In particular the transformation into the interrogative
is quite productive at a certain age: it can be a pleasant game to have every
sentence of the partner followed by one that starts with why, how, what,
when.
Our languages are so irregularly constructed that formal transformation rules

cannot be purely formal no rule without an exception holds in particular

for this kind of rule. It is a mystery as to how the learner finds his way in this
labyrinth, how he learns to speak well and judge a language at least if one does
not estimate the influence of formal rules on learning to speak (and to act) at its
true rate, if one does not consider the rules within their proper frame. Languages
are being learned within a factual context, in close connection to the facts of
the human, objective, and literary environment. This is a rich context, and it is
particularly so in the course of learning ones mother language; when learning
new languages, in whatever of the earlier senses, the learner can rely, by
translation or transformation, on previously learned languages. The wealth of
context can extend from the human and objective environment to the literary
one. A more complex language more complex with regard to vocabulary
or syntax or transformation rules is required to be learned in a richer context
than a simple language. The more formal a language is the poorer the context
can be, were it not that a large number of transformation rules demands, to
be handled, a reorientation towards the factual context.

THE A L G E B R A I C L A N G U A G E

463

16.3. Formal Languages

According to the wealth and kind of structure there is a broad scale of languages.
The language of pictograms such as used by, say, the Netherlands Railways on
the stations does not exhibit any structure; the language of traffic signs contains

a few structural elements, for instance the speed signs, by combining a general
pattern with a particular number; flections, conjunctions, sentence structure
are structural devices of what is commonly called language, with moreover

the punctuations in the written, and the intonations and pauses in the spoken
language. At the other end of the scale are the totally formalised languages,
which will be dealt with later on.
Pictograms are symbols which by their exterior betray what they mean (or

at least by the designer are judged to do so). In order to grasp the meaning of
words and sentences one needs linguistic experience; structural elements in the
language make it possible to understand linguistic utterances, words, clauses,
sentences, never met with before. On the other hand useful reproduction,

production and creation of linguistic utterances is only possible if one understands what the meaningful linguistic elements mean and masters the functioning
of the structuring elements. It is true, however, that while observing the structural requirements one can produce linguistic utterances that do not mean
anything nor are intended to do so, and this is the easier the more structure

the language possesses.

A language is purely formal if its utterances can be handled, imitated and

tested to see whether they are correct (that is, exhibit the required regularity),
without paying attention to their meaning, which is perhaps even absurd. Under
this label of formal language much can be comprised that in the usual terminology

does not deserve the predicate of language: for instance the game of chess with
the possible chess positions as lingustic utterances and the rules of the games as
transformation rules, which in each thinkable position define what is a new
admissible position obtainable from the given one by a move. The rules do not
tell what is a good move in a given position as little as the rules of grammar
give you any information about the value of a linguistic utterance.

There is a lot of theory in formal languages. Rules describe how words and
sentences are built from elementary signs, how from a sentence accepted as
true (or otherwise accepted) new true (or accepted) sentences are derived. No
concepts need be associated with the sentences; the mere form of signs, words,

sentences determines what can and may be done with them a work that can
be performed by a computer.

16.4. Arithmetical Language

This reminds one of the arithmetical work done by handheld calculators or, if
you prefer it, by well-trained human calculators. Arithmetic is to a certain

464

CHAPTER 16

extent, which depends on the calculator, such a formal language. Problems have
to obey certain rules.

is admissible while

is not. The counting sequence is a formal system as it were a story where each
word produces the next by necessity. There are formal rules to tell you how
to translate a vernacular numeral into an algorithmic one, an Arabic into
a Roman one, and conversely at least for not too large numbers.
But there is more to it. Beyond the fact that arithmetical problems are formal
data the form of which can be tested, solving such a problem is a formal business,

that is, one that goes on according to certain rules. Of course not from the
start onwards.

is acquired by insight and finally engraved on the memory to such a degree that
it becomes an automatic linguistic utterance such as continuing I pledge a ...
with llegiance. And so it happens with the tables. To be sure

is another case. The units and tens are added according to rules acquired by
insight or decree until these rules too are engraved on the memory. Solving
has then become a purely formal process. This extends similarly to column
arithmetic. In order to calculate safely and quickly one is advised not to attach
any meaning to symbols and operations. According to fixed rules well-shaped
utterances of arithmetical language are transformed into others. A lot of training
is required to attain this goal. The rules are not that simple in particular for
long division and fractions the insight can be of great help to reconstruct,
distinguish and correctly apply them. The computer can do it without this kind
of insight. Because of their small number its switches function better than the
associations in the human brain; moreover unlike the human brain it is intentionally designed to perform certain programs.
The human calculator, however, is prepared for tasks that are not the
electronic ones virtue, that is, use opportunities to calculate handily, and
though computers can be programmed to perform the same kind of tricks,
one will in general refrain from it, because it is unnecessary or unnecessarily
expensive. According to their size and means of being programmed, computers
have a restricted capacity and however this is increased, one can always create
situations where man uses better his likewise restricted capacity. A special
case is word problems; even if a whole dictionary is stored into the memory of
the computer, one can dress the simplest arithmetic problems in a fashion that

THE A L G E B R A I C L A N G U A G E

465

the computer cannot handle them. As soon as applications are at issue, arithmetic
starts ceasing to be a formal language.
Many one will not agree if I interpret figuring out such problems as transforming the given expression into its result. An oldfashioned view many one
will say. Of course I know and I stressed it before that the equality sign

is understood to mean an identity: at its left and right are names of the same
thing. It is just the same whether I say
or

It is still the same if I say

or

and so on.

Well, it is an objective fact that

and 12 are the same thing, as for Edgar
Alien Poe it was an objective fact that the killer of the rue de la Morgue was

an ape. The only thing that mattered was to discover this objective fact. Someone who does not yet know that
and 12 are the same thing, understands
the equality sign in another way, to wit, as a command to write behind
something that, for more or less intelligible reasons, will be approved.
will
certainly not, though it is correct, and 6+6 will not either. This then being

approved is the meaning of the equality sign in this kind of arithmetical

language. On this level arithmetic is not yet purely formal. Formally one could
write after the equality sign
; it yields a correct equation. In the arithmetical
language the equality sign invites a transformation procedure, and one has to
know which one. Right of the equality sign is the place of the result. But what
does it mean: result? In
the result is 5, whereas the figure behind the equality sign is 12. The fact that

a few pupils are able to grasp this, may not obscure the even more important
fact that the majority is not, at least at the age where it is being taught.
For many pupils the result of a division is the remainder. As late as the seventh
grade I came across pupils that answered 8 4 by 0. No doubt this is a domain
where the traditional arithmetical language shows serious defects.
I will try to sharply outline the problematic of arithmetical language.
16.516.8. L A N G U A G E AS ACTION

16.5. Performing Tasks Answering Questions

The learner performs tasks he is given by others or by himself. I order him to
count, and the oral or written number sequence means performing the task,

the activity of counting is documented by its linguistic result.

466

CHAPTER 16

I let him count something. The task can be formulated either count the
marbles, or how many marbles is this? In the second case the task is implicit,
at least if the child knows that in order to answer the question, he must count.
Anyway the task is performed by counting. The linguistic document can show
variants: the produced counting sequence or the number of marbles counted.
Operations is a similar case. A task
add 3 to 4,

take away 4 from 7,

or the question

how much is

how much is 7 4?,

most often abridged to

with, if it is asked orally, in the equals the intonation of interrogating. One
should expect

after the task: the performance,

after the question: the answer.
In the first case the command add or take away should indeed stimulate the
activities of adding and subtracting. But the performance of the task can also

be documented by the result of the problem, and in general the one who sets
the task will be satisfied unless there are special reasons why he would ask the
pupil to perform the task explicitly, for instance if the alleged result is wrong
or if he would like to check whether the pupil has not guessed it, or if he is for
theoretical reasons interested in the pupils procedures.
It depends on the pupils level what follows on the question 4 + 3 , 7 4.
If he has not yet memorised the problems, he is obliged to instruct himself to
add 3 to 4, to take away 4 from 7 with concrete material, or mentally.
A six-year old who had trouble with arithmetic, communicated at a certain moment to
her father the trick she had found out to do

counting further from 4: 5, 6, 7. She

considered it as a ruse, since she pretended she knew it whereas in fact she figured it out.

All that we demonstrated with counting, counting something, and simple

arithmetic, repeats itself again and again at every higher level and with more
complicated material. The task that is given or that is performed to answer a
question can be composed of more or less explicit partial tasks; which in turn
can be subdivided a long division with a divisor of more than one digit is an
example of an extremely involved task, with branchings depending on the
circumstances. At present such tasks are schematised by flow diagrams, for
instance in order to program people or computers.
But let us return to our starting point: the double dichotomy
task
performance
question answer.

THE A L G E B R A I C L A N G U A G E

467

is the most primitive relation. The task will mostly be

given by a linguistic utterance though it may also be a gesture, whereas the
performance need not contain a linguistic element other than the comprehension
of the task. If, however, this linguistic element is present, it may consist of the
documentation of the performance by means of the answer to a question substituted for the task.
can often be a direct reaction, for instance the answer to
the question How old are you?, How much is 10 10 ?, How many days
does the week have?. The questioner may mean a task May I get 10 postcards? or he is sure that the question illicits a task for instance in the
arithmetic question How old were you, when the war began? or How much
is
?. The question can be understood as a question, requiring an answer

obtained via a task and its performance; it can also happen that the question
is interpreted as a task to be performed while the result of the performance is
not translated into an answer a well-known phenomenon, if after the question
How many marbles? the child has counted without formulating an answer.
16.6. Knowledge of Facts and of Procedures
No cognitive teaching area is less concerned with the problem of what is called

ready knowledge in instruction than is mathematics. Questions about a phone

number, about the way to go to some place, about the spelling of a word, about

a year, about the name of a plant, about the meaning or translation of a text,
can more often be answered by looking up, consulting sources of information
than is the case with questions of mathematical origin or in which mathematics
is involved. Well, even in such cases it may be an art to know where to look up,
which sources of information to consult, how to use the means of information,
what to do with the information retrieved and how to know whether you can
trust it. On the other hand mathematics too cannot dispense with ready knowledge or knowledge from sources of information the elementary additions,
the tables of multiplication, formulae, tables of all kind of functions, and
computer solutions. But as far as this is ready knowledge, it is easy to be tested,

and as far as sources of information are concerned, reliability and standardisation are so strong that consulting plays a minor part in the process of problem
solving.
As a matter of fact, if in mathematics ready knowledge is called upon, it is
the solution procedures that really matter. Though such things are difficult to
estimate, I would guess that outside mathematics and its applications ready
knowledge of facts has by far the upper hand of knowledge of procedures, while
in mathematics the procedures look relatively simple, though one is easily
inclined to underestimate their length and depth.
16.7. Procedures as Linguistic Transformations

In Section 16.3 we already discussed a peculiarity of mathematics as compared to

468

CHAPTER 16

any other subject. Mathematics can be more adequately formulated even in

linguistic respect than any other knowledge. Of no physical or mental objects
can such a structurally simple and precise description be given, which moreover
disregards the diversity of languages, as can be done for the mathematical
objects; nowhere can operations on objects as adequately rendered linguistically
as those on numbers. Even the mental objects of geometry can, though structurally not as simply, be described with all desired precision by the conventional

terms used to indicate them in the various languages. This is one of the reasons
why mathematics is very often identified with its linguistic expression. Both
understanding and misunderstanding about what is mathematics, feelings of
power as well as of helplessness can be the consequence.
If a child has understood elementary arithmetic, it enjoys mastering it; the
ease of answering questions invites the desire to have them posed. He soon

forgets how much trouble the acquisition of this ability has cost. Meanwhile the
subject matter is extended: ever more experience is gathered about transformation rules that turn queries into tasks, by which tasks are split into partial

tasks, to be performed, in order finally to answer the questions. The data are of
linguistic character and gradually the whole solving procedure gets the character
of a linguistic transformation or a sequence of linguistic transformations, which
are mostly simpler and anyway more formal and regular than the transformation
rules of the vernacular.

16.8. Formalising as Means and Aim

The ease of formalising is a striking, though at the same time misleading feature

of mathematics, a virtue that, as with other virtues, can turn into a vice. The
ease of formalising gives us a feeling of power that can hardly be overestimated
and certainly not be disregarded with impunity, though it is one that can turn
against mathematics. For the majority who have got into contact with mathematics, it is mastering (or in fact not being able to master) formal rules. What to
do about it? Desisting from teaching mastery? This would be a preposterous
solution.

The rule of multiplying by 100 by appending two zeroes (which has to

be modified if decimals are concerned), the rule of so many places after the
decimal point if decimal fractions are multiplied, which competes with that
for adding decimal fractions, the mechanism of long division with all its ins and
outs shouldnt one be glad if this functions well? Whoever was able to learn
and apply this correctly, is certainly able to find out where these rules come
from and how they can be justified. Let us be cautious: maybe he would be
able to. Yet even the faintest idea that such a question admits of, let alone,
asks for a rational answer, is absent in the mind of the great majority of those
who master these formalisms. Even student teachers who are expected in the
future to teach such formalisms often do not grasp the aim of racking ones
brain about formalisms one masters perfectly (or thinks so). After a long division,

THE ALGEBRAIC LANGUAGE

469

performed in all detail, the question what is a division, is shelved as irrelevant.

The usual explanation is that they have learned it as a mere trick and that is all
they know about it. Is it really true that all who react this way have learned the
formalisms as mere tricks?
This is a rhetorical question. I have observed, not only with other people but
also with myself I mentioned it in Section 15.17 that sources of insight can
be clogged by automatisms. One finally masters an activity so perfectly that
the question of how and why is not asked any more, cannot be asked any more,
and is not even understood any more as a meaningful and relevant question.
This looks a natural course of affairs, and nevertheless one would like somehow
to change it. It is a problem that does not concern only arithmetic. Somebody
once said: When calculating starts, thinking finishes. No, it need not be true,
unless one blocks the way back to insight.
The didactical mistake resides in the principle of once learning by insight, and
then irrevocably going forward to the automatisms.* A variant is: back to the
start by insight as soon as something goes wrong. This variant is better, though
still unsatisfactory. Even if the formalism functions reasonably, the teacher or
the one who defines the instruction should avail himself of each opportunity
to return to the source of insight. Such opportunities exist on each level, for
instance. As regards column arithmetic, if procedures of measuring are discussed
or (in algebra) if powers are dealt with, or if the technique of brackets is taught,
or if arithmetical laws are formulated: write numbers and operations in the
form

There is no need to pass to other number systems in order to restore preformalist

insight. Other number systems are ad hoc creations, which can be viewed by
children as irrelevant games or dead ends rather than as an intention to lead
them back to the sources of insight.
The didactical necessity, which I stressed, is a consequence of the high degree
of formalising of the language of arithmetic. It is as strongly felt in the didactical
phenomenology of other formalised languages. In the normal use of less formalised languages, such as mother language and foreign languages, the form is
so strongly attached to the content that it can permanently be checked by
means of the content; in general it would even be impossible to handle the
linguistic form without content support.
Here I disregard cases of autonomy of the linguistic form, such as cultivated
in poetry and philosophy as phenomena of a use of language which does not
aim at, or is not appropriate for, unambiguous communication. I also exclude
as an aim cases of formalising, where language itself becomes a matter of study.
I restrict myself intentionally to didactical cases where learning formalising
* See also Sections 6.19 and 15.17.

470

CHAPTER 16

and formalisms is a didactic necessity, though a necessity among, and for the
benefit of, others.
A characteristic feature of mathematics is a line of progressive formalisation.
Transformation rules which have, or have not, been acquired by insight, are
generalised in order to solve problems with a greater efficiency while using the
acquired apparatus, while by this activity new more complex transformation
rules are formed and again generalised, which continues the same way; progressive formalisation leading to ever and ever more radical shortcuts.
16.916.15. C H A R A C T E R I S T I C S OF THE L A N G U A G E
OF MATHEMATICS

16.9. Algorithmic Construction of Proper Names

The most important source of progressive formalisation is the algorithmic
construction of vocabulary, the first and foremost example of which is the
construction of proper names of natural numbers.

Algorithmic features are not unusual in the vocabulary and the syntax of
whatever language, though they are incidental and unsystematic: plurals, past
tenses, comparatives, word compositions, sentence patterns to be filled out.
None of them approaches remotely the systematic structure of the numerals.
Certain historical remnants of irregularity mostly with respect to small
numbers have wholly been eliminated in the digital language of the decimal
system. Already the pre-positional systems knew a decent regularity; in a positional system one can, starting with a small stock of digital symbols represent
all natural numbers according to strict algorithmic rules. This trend continues
if the number concept is extended; synonymy of fractions is tied to rigid
transformation rules.
Likewise for the names of tasks or statements such as add three to four
(in the form 4 + 3 ) or four is greater than three
the algorithmic
construction dominates, though there may be a variety of names for the same
task or statement.
According to how complex objects, tasks, statements are, the names assigned
to them need a certain structure, which will now be discussed.

16.10. Punctuations
The natural languages have developed a large number of structuring devices such
as prepositions, conjunctions, affixes, suffixes, subordination of clauses, and so
on. Moreover spoken language uses for structuring such devices as pauses and
intonation, whereas in the written version structure is more or less adequately
indicated by punctuations. The most explicit structuring element in the language
of mathematics are brackets of various kinds. Moreover there is a lot of implicit
structure: in performing a task or reading a statement some operations take
precedence over others multiplication over addition.

THE ALGEBRAIC LANGUAGE

471

Explicit punctuations can usually be dispensed with in the vernacular. Often

criteria of content rather than formal ones decide the structure. In the sentence
There were aged ladies and children in the bus

the aged refers to the ladies only, whereas in the formally almost equivalent
sentence
There were aged ladies and gentlemen in the bus

the aged is very likely to include the gentlemen. (Spoken language is more

sophisticated than the written one; the two sentences are probably pronounced
with a different melody.) In
We visited Dutch towns and villages
and

We got lessons in Dutch history and mathematics,

the Dutch has different domains. A well-known example is
pretty little girls schools
which according to the places of the lacking brackets can have 17 different
meanings.
In mathematics more care is practised.

5 times ... 3 plus 7

must be distinguished from

5 times 3 ... plus 7

and this distinction is formalised by putting in the first case the 3 plus 7 between
brackets. I should do the same in the second case with the 5 times 3 were it
not that a convention says that multiplying takes precedence over adding and
subtracting. I already pointed out that spoken language can be more sophisticated
than written one. In the last two examples I tried to render pauses and intonation by dots; in the written language no similar devices are available. We must
rely on understanding the content in order to grasp the syntactical structure.
Rather than from the structure of the linguistic utterance, I grasp what aged
and Dutch include, from the content that has to be meaningful.
In the language of mathematics meaning is not a reliable criterion. In fact,

5 times ... 3 plus 7

is as meaningful as

5 times 3 ... plus 7;

472

CHAPTER 16

it must be perfectly clear which is meant if the expression is to be used. This is

attained by strict punctuation rules, strict indications as to how an expression
should be bracketed and read. If the mathematical rules were to be adopted,
one would write

(aged ladies) and children,

aged (ladies and gentlemen).
This, however, does not happen, which is one of the big differences between the
vernacular and formalised language.
I pointed out that the language of mathematics possesses more devices of

structuring than bundling brackets, in particular the precedence of some operations over others if a task is to be performed or a statement to be read. These
too are formal devices of structuring, in contradistinction to those of content
used in the vernacular. I do not explain details, which I actually have done
earlier.* What matters here is to map out the differences between vernacular
and the language of mathematics.
I must, however, stress that the system of structural rules in the language of
mathematics is not as simple as one would believe at first sight. Troubles that
learners experience with the language of mathematics, can be explained at

least partly by the lack of insight of textbook authors and teachers into the
complexity of the matter. As those who are responsible for the instruction of
this matter are not conscious enough of these details, they lack the insight into
the possible sources of mistakes.
It is a remarkable fact that children do learn their mother language from
people who have never thought about the structure of this language; no the-

oretical analysis of the mother language is required in the intercourse with

children who are to learn their mother language; it is even thinkable that an

adult does not become conscious about such features as, say, weak and strong
verbs until he hears children commit mistakes; that he is not puzzled by polarities
such as yesterdaytomorrow until he observes the uncertainty of children
about this pair.
The case of the language of mathematics looks a bit more like that of learning
foreign languages: in order to guide the learner, the teacher should master the
matter more consciously than the parents do with the vernacular that they are
to deliver this will be clear without specific examples.
This, however, does not mean that the highly involved system of mathematicallinguistic structure rules should be made conscious to the learner. In remedial

teaching this can become a necessity, in particular if strongly ingrained habits

have to be driven out. Primordially the rules for applying structuring devices (as

well as transformation rules) should be learned by using them consciousness

is required for fighting misuses.
Though I do not repeat my former phenomenological analysis of structuring
* Mathematics as an Educational Task, pp. 304311.

THE A L G E B R A I C L A N G U A G E

473

devices in mathematical language, I do not wish to have it disregarded. On the

contrary, I would urgently recommend textbook authors to pay attention to
it and to realise and show that learning this linguistic element requires a more
conscious didactics.
16.11. Variables in the Vernacular

The long step from arithmetic towards algebra (from primary to secondary
instruction in mathematics) is calculating with letters rather than numbers;
letters reckoning used to be a familiar term. It looks a superficial phenomenology to put it this way, but in phenomenology profundity starts at the
surface.
How letters entered mathematics, how half conscious of what they did, people
started calculating with letters, how the increasing consciousness influenced
didactics of mathematics in the present century, how these precious attainments
have been smashed to smithereens by the set theory rage of New Math and what
is required to clear away the rubbish I have told this in the past* though not
as a connected story, but dispersed over various places according to which detail
just mattered. So I have to start anew.
In the last few sections I used the term name in a somewhat broad sense:
I attributed names not only to objects but also to acts, tasks, statements, even
though such names could have the linguistic form of sentences. In traditional
linguistics one knows the term noun and nouns are distinguished as proper
names and generic names (appellatives). It is difficult to draw a sharp borderline
between both of them. John and Dad can be proper names in a family,
but it is easy to think of situations where they are used as appellatives. My
Peugeot and the lady across the street can be proper names, which, however,
show a more involved structure than simple nouns.
is the name of
a certain statement, but in the context it is as sure as two times two is four
it looks rather as the Tom in Tom, Dick and Harry.
Day by day we have to communicate about individual objects physical and
mental ones, about processes, acts, desires, which are objectivated, and in order
to do so we need names names which in fact, though they do not look this
way, are proper names. How can we invent ever and ever new proper names
for this incredible variety of objects? The answer is: by manipulating skilfully
with generic names, by tying generic names in changing situations to changing
individuals. For such names one knows in mathematics the term variable
and for fixing a variable in a given situation the term binding.
According to an old story Adams first task in the paradise was to name the
creatures. He himself had got the name Adam, that is man, and after him
his offspring were called Adam, that is man. He gave names to all animals that
passed in review, and that one he had called lion transmitted this name to his
* Mathematics as an Educational Task.

474

CHAPTER 16

offspring. So lion became a name with which you could name each particular
lion; in order to distinguish lions from each other, one could speak of this lion
and that lion, of the old and the young lion, the lion in the Amsterdam Zoo
and the lion in the London Zoo. It would not be feasible to invent brandnew
proper names for each mouse, chair, bicycle; we are familiar with the use of
ambiguous, or rather polyvalent, names one name for many objects. When my
daughter was at the age when children play the game of what does this mean?
and I asked her what is thing she answered: Thing is if you mean something
and you do not know what is its name. Thing is a name that fits an incredible
variety of objects, chairs, bicycles, trees, and so on. Words as here and now
are also polyvalent names: here as name of the place where one says here,
now for the moment when one says now. In
the cricket is an insect

cricket is a generic name, the name of a species, but with

each cricket is an insect
one can maintain that cricket is a polyvalent proper name that fits each

particular cricket. In La Fontaines La cigale et la fourmi cricket is the

unambiguous name of a fable animal. In
the mouse is a rodent
mouse is the name of a species, in

the mouse is in the trap,

pronounced in a quite special situation, mouse will mean a particular mouse
that by these special circumstances is supposed to be well-known and welldefined. In

we have again a mouse

or
we have mice,

the name mouse is given to one individual or to a few about which not much
more is known than the existence.
In order to function as proper names, variables must be bound. Variables
can be bound independently of any context, by linguistic logic devices, or in
dependence of a context. Logical means are

the universal quantifier

a mouse is a rodent for every x, if x is a mouse it is a rodent,
the existential quantifier
we have got a mouse there is an x such that x is a mouse and we
have got x,

THE ALGEBRAIC L A N G U A G E

475

the article
our mouse; the mouse we have the x such that x is a mouse and
we have got x,
the set former
our mice the set of x such that x is a mouse and we have got x,
the function or species former
the species mouse the property of being a mouse
the interrogative
which mouse? which x such that x is a mouse?
Context depending devices of binding are
the demonstratives
this mouse, that mouse, the mouse in the trap, the mouse I hear
rustling.
Many variables assume another form according to the way of binding, for
instance the variable of place

Or the variable of time

Similarly the variable indicating persons

476

CHAPTER 16

16.12. Variables in the Language of Mathematics

It is worthwhile keeping in mind this variegated use of variables in the vernacular
in order to confront it with the more regimented and at the same time emaciated
use of variables in mathematics.
The variable mouse can be used for mice only, a variable like here only
for places, a variable like I only for persons in a certain measure they are
already bound by content. In mathematics symbols for variables are usually
letters, sometimes combined with each other or with numbers
Such symbols are unburdened and the variables denoted by them are unrestricted
with respect to their domain or only ad hoc restricted.
The use of letters for variables stems from Greek geometry. In a period

of exclusively oral communication of knowledge one was very likely to reason

about a figure while speaking of this point and that point, thus demonstratively.
For written communication a more practical way of description was required:
in a figure one numbered the points under consideration with the letters of
the alphabet, which in numerical texts were indeed used as numerals. The

original terminology the point at A that is beside A was later shortened

to the point A. In a similar way lines were named, triangles, quadrilaterals,
by letters or by combinations of letters, which in turn indicated points.

There we are on the threshold from the demonstrative description to the use
of genuine variables. In fact the variables are bound demonstratively by the
figure to spots on the drawing material. On the other hand the points and

figures are arbitrary paradigms. All points are the same, statements on a triangle
mean all triangles, however drawn or indicated. By this fact the variables look
as though they are bound by logical rather than demonstrative means. Indeed
the figure may even be forgotten.
If ABC is a triangle the orthogonal bisectors of AB, BC, CA pass through
one point

is such a statement where A, B, C may be considered as point variables bound

by the universal quantifier.
In geometry the use of letters for variables did not lead to the birth of an

algorithmic language; except for the indication of line-segments by pairs of end

points, there is hardly any algorithmic vocabulary nor is there any formalisation

of operations, which at most takes place in the vernacular serving as metalanguage.

In algebra letter variables are even of a later date. Linear and quadratic

equations and general solving methods are as old as the oldest cuneiform texts,
but always in a numerically paradigmatic setting. In the oral transfer of knowledge the unknowns could have been named in an informal way thingummy

and this actually happens in the later development. In Greek the letters also
meant numbers, which fact may have prevented Greek mathematicians from
using them for unknowns. There was, however, one more impediment to transfer

THE ALGEBRAIC LANGUAGE

477

the geometrical use of letters for variables directly to algebra: whereas all points
are the same, numbers have a well -distinguished individuality. Finally, in

the Hellenistic period, in Diophants work, there is at least a symbol for the
unknown, an abbreviation of arithmos (number). In the middle-ages cosa (Ital.
thing) becomes the name of the unknown the cossists developed a whole
symbolism for powers of the unknown.
The decisive step toward a more useful algebraic notation was taken by Vieta
(about 1600), who indicated also the indeterminate magnitudes, the variables
in algebraic expressions, by letters. This notation is the proper start of the
development of an algebraic language, which gets more and more detached from
the vernacular. Letters are first used to indicate arbitrary numbers, but soon as
well for arbitrary functions. At present we use letters for all kind of mathematical objects sets, relations, propositions, spaces, metrics and all kind of
structures and if the need is felt, we take them from all kind of alphabets.
This should be kept in mind: letters in mathematics mean something: as
symbols they represent something. At other opportunities* I have signalled and
analysed the misapprehensions demonstrated by New Math at this point: letters
that mean nothing or themselves, and mathematics viewed as a meaningless game
with symbols. It is true that this latter aspect can be consequentially elaborated,
with the aim to do foundations of mathematics, rather than mathematics itself.
Then meaningful mathematics is taken as a subject to put its meaning as it were
between parentheses and pay attention to its form only the formalism.This,
however, happens with and for the benefit of a meaningful mathematics. As
early as the language of arithmetic I have discussed both the usefulness and the
didactical danger of this procedure if a mathematical attitude is to be developed.
No doubt in the case of the language of algebra usefulness and danger of rigid
formalisation are as great or even greater. Miscomprehension of what is mathematics is often generated by blind algebraic calculations with letters. New
Math has contributed to spread these misconceptions even among teachers.
16.13. The Equality Sign
If in Section 16.4 I stressed the transformation character of the language of
arithmetic, I feel now obliged to attenuate it:

is primordially read as a task or a question, I said. The equality sign looks

as if it were asymmetric. One side is given and the other is to be filled out.
The problem

* Mathematics as an Educational Task, Chapter XV.

478

CHAPTER 16

fits less into this picture for learners a reason to get confused. If numbers are
replaced with letters, the confusion is even bigger. Putting an equality sign in
what should it mean? Well, it can be meaningful in an exercise like
how much is

if

and

thus if numbers are to be substituted for a and b. In contradistinction to

one cannot consider

as something to figured out. It is rather the algebraic expression of a number the

value of which depends on those of a and b. The expression
does not
suggest a task or a question, but it names a number depending on a and b. The
notation of fractions was the point where this turn took place in history. With

there is nothing to be figured out. It is as such a new number represented in a

conventional way by means of the previously known number symbols 3 and 4.
Of course the expression
can occur with an equality sign behind it,
such as in

in the first case either to introduce a new symbol for

or to explain c
by means of a and b or to require a, b, c to assume only values that lead to
an equality, and in the second case to state something that is true for all a and
b.
By this turn the character of the equality sign of arithmetical instruction
has changed; it is symmetrised. It is the intention that left and right of the
equality sign the same thing occurs, and this intention can be actualised as a
fact or as a demand.
The symmetric is plays hardly any part in the vernacular, even not in
Amsterdam is the capital of the Netherlands,
which betrays another intention than
The capital of the Netherlands is Amsterdam,
answers as it were to different questions, not unlike those asked by the pair

479

THE ALGEBRAIC LANGUAGE

Even less is this equality related to the is in

Socrates is a man,
or with its plural in
Apes are mammals
in the first case it rather reminds one of the symbol
Again another thing is the is in

in the second case of

The apple is ripe,

namely part of the predicate being ripe, and so is the is in
There is time left
as part of the logical quantifier there being.
Traditional arithmetic knows equality signs in divisions with remainder like
If this were the symmetric equality sign, it would imply
German didacticians of arithmetic have paid much attention to the problematic
around this notation.* Is it worth the trouble? Yes and no. It is worth being
seriously discussed but is not worth turning arithmetical instruction upside
down.
I confessed earlier that mathematicians have convened to use the equality
sign in the way that things left and right of it are the same, albeit indicated by
different names.
is allowed because with more or less ease one can
verify that
and 7 are indeed the same things.
is allowed because by
definition the two fractions mean the same rational number. Actually equality
in mathematics is often a matter of definition. At a certain moment it can
happen that a set of things so far viewed as different might better be considered
as the same, and in order to do this with a good mathematical conscience, one
introduces a new thing, the class of the things to be considered as equal and
avails oneself of the names of these individual things, like
as various
names for the new thing, the class a convenient and therefore allowed linguistic
misuse.
But let us forget for a moment about this sophisticated mathematics. In fact
even mathematicians play it fast and loose with their rules about the equality
signs,
what does the righhand term mean? And now
Does this mean that
How about this? Well, one can object that this is applied mathe* See the excellent paper by H. Winter, Zur Division mit Rest in: Der Mathematikunterricht 4/78, pp. 3865.

480

CHAPTER 16

matics. But number theory is pure mathematics, where nevertheless one writes in
cold blood
mod 3. (If number theory had been invented later, one would
write 16 ~ 1 mod 3.) I agree this is not an equality sign, it is kind of equivalence
symbol. But which equivalence? This is told only afterwards (as is done with
the dots behind 3,14159 and the 0,003 at 9,81, which tell afterwards what
kind of equality sign was meant.)
If one decides to place an equality sign between two expressions, one must
first know what kind of things one has in mind to name by them. What kind of
thing is
in the division with remainder? In higher mathematics such a
colon is unusual. ( 1 6 : 3 can mean a ratio, but that is not what is meant here.)
What kind of thing is
? Of course no natural number. It is an ordered pair,
consisting of the quotient 5 and the remainder 1. Or if you like more
mathematical sophistication:

a pair consisting of the whole

and a residue class mod b, indicated in its
reduced form.
This is a highbrow interpretation. As a matter of fact in arithmetical instruction we do not meet
but
a task such as

As early as elementary arithmetic there is a need besides the meaning of

as a task to pave the way for the meaning of
in the sense of 4 + 3 being another expression for 7. In fact problems like

is the first opportunity. Would it not be reasonable to adhere to this interpretation from the start onwards? Traditional arithmetic possesses symbolic devices
to formulate tasks by means of dots. This use could systematically be extended
even to the type
Modern textbooks prefer squares,

THE ALGEBRAIC LANGUAGE

481

There are more symbolic devices to formulate tasks, for instance in the arrow
language

and so on. Here the equality sign is explicitly renounced and reserved for a more
symmetrical use.
Let us reconsider for a moment the division with remainder.

requires a lot of reinterpretation to be considered as containing a mathematically

meaningful equality sign, but in fact
is a task. If one sets out to switch
early from the equality sign as a signal for a task to its static meaning, one
can choose another pattern for the division with remainder, for instance a
form to be filled out like

where in plain language the same is told as we did before in obscure mathematical
language.
If one thinks that by the intervention of algebra the equality has been abolished as a signal to perform a certain task, one is mistaken. In the traditional
school algebra it returns in full bloom in problems like

or
where the pupil is expected to fill out behind the equality sign something
depending on a general assignment that is or is not made explicit. Algebra as
a system of linguistic transformation rules leads automatically to an asymmetric
interpretation of the equality sign as unilaterally directed towards a reduction.
Reducing is indeed a term characterising school algebra and, more general,
automatised mathematics. According to certain rules expressions are reduced
in one sense or the other, that is transformed into others, where the intention
that meanwhile they remain the same, may readily be forgotten. This pattern of
behaviour is reinforced by the application of reductions not only on algebraic
expressions but also on equations, in order to solve them. The sequence of
steps

482

CHAPTER 16

is assigned the same character as that of the problem of dissolving

where then the various steps are justified by equality signs:

One is tempted in the first example to make the connection between the subsequent lines explicit by a mental, vocal or written equality sign. There, however,
the thing that remains equal at the various steps is not a numerical but a truth
value, which fact is registered by a
sign, or in the asymmetric case, that is,
if the truth value increases by

16.14. Formal Substitution

Replacing in
Socrates is a man

Socrates by I is not allowed; one had simultaneously to change is into

am. Replacing in
It snows

the hidden time variable now with yesterday or tomorrow requires a

change of the verb tense into the past or the future. The substitution processes
that lead from
to

THE ALGEBRAIC LANGUAGE

483

or conversely, are more formal. By formally substituting 3 for x one can verify
whether
is satisfied by
Formal substitution, however, extends further. From

one obtains

by replacing

a with

b with

that is, variables in an expression are replaced with more complex expressions,
which can in turn involve variables.
A powerful device this formal substitution. It is a pity that it is not as
formal as one is inclined to believe, and this is one of the difficulties, perhaps the
main difficulty, in learning the language of algebra. On the one hand the learner
is made to believe that algebraic transformations take place purely formally,
on the other hand if he has to perform them, he is expected to understand their
meaning. If in

... b
I have to replace

b with

it becomes
not

but

The minus sign of b is expected to extend its activity to the whole of b + d.

This then is the inevitable consequence of the interpretation of
as

sum of b and d
rather than the task
add d to b,
which holds whether the b and d are genuine variables or already specified

constants. Thus
the difference of . . . and b

484

CHAPTER 16

becomes after substituting b + d for b

the difference of . . . and the sum of b and d.

The learner is expected to read formulae with understanding. He is allowed to
pronounce

as

a plus b,

a minus b,

a times b,

a square.

Yet he has to understand it as

sum of a and b,
square of a.

difference of a and b,

product of a and b,

The action suggested by the plus, minus, times, square and the linear reading
order must be disregarded. The algebraic expressions are to be interpreted
statically if the formal substitution is to function formally indeed.
The formalism of algebra could have been designed more rigidly by requiring
that in substitutions the substitute is only accepted within brackets; in the case
of substituting
for a and
for b in

this produces

where parentheses that we do not like can be eliminated afterwards, and in this
way we are likely to instruct a computer, indeed. But since in this way algebra
would become a wearisome business, we appeal to understanding, even if we
teach algebra. It is a fact of didactic experience that this appeal falls on deaf
ears at the same time asking for formalist acting and content directed understanding is too much. What to do about it didactically?
After the preceding phenomenological analysis the advice is easy. In order to
teach the language of algebra, in particular the formal substitution, we have to
make an appeal to intelligent reading, and this appeal should be well-aimed and
if necessary made explicit. It would be unnecessarily complicated to read

again and again as

product of the sum of a and b and the difference of a and b,

THE ALGEBRAIC LANGUAGE

485

but in order to act as though this is said, it can didactically be necessary to

recall it to consciousness. An intermediate stage, kind of flow diagram, would
be:

How and when this should be done, depends on the total organisation of formalising in the didactical sequence, in particular on the way how, and the degree
to which, thinking in functions is going to be formalised.
If the teacher can afford to be patient, the formal substitution may arise selfreliantly. A

personal experience with a 14-year old underachiever:

I allowed her to solve equations such as

as she liked, in two steps (times 4, divided by 3) until she had found out how to do it in
one step. I had her calculating products like

by writing each power as a product until she became tired. When with types like

my efforts to have her combing both mixed factors, fell flat, I switched to
where I succeeded, and then back to

, which became a success too. This was the

start of a slow progress on the road towards the formal substitution.

Patience is expensive; the teacher must know whether the price is worth paying.

The formal substitution is so important because its validity extends much

further than substituting into given expressions. For solving certain problems the
algebraist has rigidly formalised solving procedures at his disposal. Solving linear

and quadratic equations and systems of equations follows patterns which may
be numerically introduced, then generalised by formal substitution and again

applied in the generalised form by means of formal substitution. Numerous

algebraic strategies function thanks to the possibility of formal substitution in
strategic patterns.

486

CHAPTER 16

16.15. Idiom
In the last ten or twenty years it has become a fashion to assert that the language
of mathematics such as we know and use it shows the features of the natural
languages spoken and written by the creators of the language of mathematics
in the past. Mathematics instruction in countries where the language structurally
differs strongly from Western languages suffers from this fact, people tell you.
Statements like this are made by linguists who experience the language of
mathematics as a strange and insufficiently analysed phenomenon.
As a matter of fact the so-called Western languages differ greatly from each
other, but never have difficulties of transfer been observed in the use of the
common mathematical language. On closer sight it appears that the language
of mathematics also differs greatly from the vernacular used in the various
countries where it has been developed. I have already dealt with some divergences
of principle and I will add a few more.
Divergences between a natural language and the language of mathematics can
indeed cause learning difficulties. The most striking and most serious is probably
the deviation from the rules of order according to the positional system in the

construction of numerals in many languages up to 20, and in a few even up

to 100. We have got used to most of the other deviations to such a degree that
we are not anymore aware of it.
is, one closer inspection, a strange manner to write down the

task add three to four

and an almost absurd one for

the sum of four and three

as it should properly be read. Nevertheless everybody reads it cooly as
four plus three
which is not English as little as the French, German, Italian ... translations

of it are French, German, Italian.

Subtractions were up to the beginning of this century in German and certainly
in many other languages formulated according to the pattern
vier von sieben

(four from seven)

for 7 4. When pedagogues got fed up with it, they made it

sieben minus vier

for the schools of the elite and

sieben weniger vier

(seven less four)

THE ALGEBRAIC LANGUAGE

487

for the common people, both of which are not normal German a fact noticed
today by nobody. The Dutch terminology for
was and often still is
dividing three on twelve

which is normal Dutch, though now in favour of the international mathematical

language and without tears replaced with
twelve divided by three.
Never in former times has the equality sign been read as equals.

was read as
quatre et trois font sept,

or similarly with facit, macht and so on in other languages.

Three times can degenerate into one times and nought times both
of which are all but English. One reads in all languages similarly

as a square,
instead of
square of a
as required by decent linguistic habits. In expressions is
log a

the order of the natural language is at least preserved in the mathematical

language, but nobody complains about the lack of something like the word
of between the two components as far as it is required and possible in the
surrounding language.
Let me restrict myself to these examples, which have no other pretension
than to make clear how strongly the language of mathematics differs from the
languages in which it has come into being, and that on points which as a consequence of habituation we do not notice any more.
16.1625. Algebraic Strategies and Tactics

16.16. A more profound study than I have been able to undertake so far, should
be required to enumerate if not exhaustively then at least representatively,
and at the same time systematically, algebraic strategies and tactics. Let me
produce that which just crossed my mind and repeat the warning that it is
neither complete nor well-organised.

488

CHAPTER 16

16.17. The algebraic principle I used the term earlier,* but in Section 15.3 I
changed it into the more colourful
algebraic permanence principle,

with the shade (Sections 15.1315) of the

geometricalgebraic permanence principle.
It means the idea of extending operations and relations, wherever needed, in
a way that

certain appreciated properties are preserved.

As a creative principle as well in its didactic function it has been discussed so
often that no further explanations are necessary. I will only add some more
colour to the originally neglected geometric shade by repeating expressively:

the justification of the numerical operations and their laws by the simplicity
of the algebraic description of geometric figures and connections.
16.18. The (formal) substitution discussed as early as Section 16.14 has
numerous tactical aspects. It can yield

specialisation if variables are numerically fixed;

(example: verifying whether
is correct);
generalisation if numerical terms are replaced with variables;
specialisation if a solving schematism is applied;
generalisation is a solving schematism is extended;
structural simplification if in a given expression partial expressions are
replaced with variables
(example: replacing
the with y);
structural complication if in a given expression variables are replaced
with expressions
(example: when solving
replacing x with
in
order, by having 3uv+p vanishing, to later simplify the left side);
elimination of a variable involved in a substitution;
restructuration in order to recognise patterns that admit of applying
certain algebraic laws
(example: replacing in
;
* Mathematics as an Educational Task, pp . 228241.

THE ALGEBRAIC LANGUAGE

489

On a higher level substitution is involved

if operations and relations are replaced with their defining terms;

(examples: multiplication as repeated addition,

raising to a power as repeated multiplication,
subtraction as inverted addition, and so on,
and the converses)
16.19. The algebraic translation of properties, connections, and problems
in the most original example, characterised by the advice call the unknown x.
In order to describe a problem in its context algebraically,

knowns, unknowns, parameters and relations between them are tracked

down and translated by means of variables, equations and inequalities.
In order to describe properties (sets) algebraically,
variables are introduced

(example: odd number algebraically described by 2n 1 with

integral n ).
It is surprising that even pupils in the higher grades of the highschool solve
problems like

A fraction with value

where numerator and denominator differ by 45.

Mary is six years older than John, and eight years ago she was thrice Johns
age,
The length of a rectangle is

times its width and its surface is

by trying; they do not hit on the idea to do it with x .

16.20. Solving equations and inequalities

by systematic trials,
by transferring all the terms to one side,
by distributing the terms appropriately on both sides,
by applying the same operations to both sides,
by complication to prepare a simplification;
in systems with several unknowns
by substitutional elimination of unknowns, if
needed, prepared by other transformations.

16.21. Considering an expression as a function,

a function as composite of functions,
a function as the inverse of a function.

490

CHAPTER 16

16.22. Change of perspective,

by considering a datum as an unknown,
by considering an unknown as a datum,
by considering data as solutions of a system of conditions,
by replacing inequalities with equalities,
by replacing equalities with inequalities.

16.23. Exploiting symmetries

in expressions, equations, inequalities, functions.

16.24. Stating the positivity of some expression

by trying to regard it as a sum of squares.

16.25. Exploiting analogies

in an informal way;

in functional connections,

such as the analogy between addition and multiplication informally or by means

of exponential and logarithmic functions.

CHAPTER 17

FUNCTIONS

17.1. Variables
In Sections 16.1112 we subjected the variable as a linguistic phenomenon in
the vernacular and in mathematic language to a comparative analysis. Indeed
what we discussed there, was language, the need for names, proper names as
well as
polyvalent names,
which according to certain procedures may supply proper names. This need was
felt early on in mathematics: to distinguish this point and that point as A and
B, the three vertices of a drawn triangle as A, B, C, which at the same time
can name the vertices of any triangle and by this very fact occur as polyvalent
names for points, and combined, for triangles.
Polyvalent names are a means to formulate general statements, that is statements that hold for all objects they name:
the mouse is a rodent
applies to every mouse,

to every triple of points in the plane,

to every pair of numbers. Polyvalent names can, as we pointed out, be fixed

in various ways, bound in order to be used in statements on objects.
The mathematical habit of calling polyvalent names variables is of a rather
recent date. Originally variable meant something that really varies, something
in the
physical, social, mental
but as well in the
mathematical
world, that is
perceived, imagined, supposed
491

492

CHAPTER 17

as varying, that is in addition to

the time that passes,

the path that is covered,

the aim that changes,
the moon that is waxing,
the temperature that oscillates,
the wind that is changing,
the days that lengthen,
the mortality rate that decreases,
the progressive rate of income tax,
also the

variable mathematical objects

by which these phenomena are described. From the

variable physical, social, mental

phenomena one is led to

variable numbers, magnitudes, points, sets,

in general
variable mathematical objects.

Locutions like

the number approaches (converges to) 0,

the point P runs on the surface S,
the element x runs through the set S,
the number e is approached by the sequence

if n goes to infinity,

witness this kinematic aspect of the variable. It is true that in the course of,
say, the past half century such locutions have been outlawed by purists. Indeed
one can dispense with them,
converges to 0
can be written

and be defined, with no kinematics involved, by

for every

there is an

x runs through the set S

such that

FUNCTIONS

493

can very simply be written as

Well, one can dispense with that kind of kinematics provided one has once
been in its possession, learned to use and then to eliminate it this didactical
feature will be dealt with later on.
The effort to suppress the kinematics of the variable goes hand in hand with
the annexation of the term variable, stripped of its kinematic undertone, for
the benefit of what we just came across as polyvalent names. In our former
examples
the mouse is a rodent,

the mouse, and the various letters are primordially polyvalent names by means
of which statements can be made, which, if not restricted, hold for all bearers of
these names. Variable objects is not the original idea here as it is in the examples
of
the time that passes,
the number that approaches 0,
the point running on the surface.

Lumping concepts of various origin together, using one name for things which
stripped of mere frill boil down to the same, this is one of the characteristics of
mathematical activity. We have met here again such an example:

polyvalent name
and

variable object
are being related with each other in the term
variable,

but they pay for it with the loss of important phenomenological properties,
which mathematically look like mere frill, though phenomenologically they
cannot be dispensed with. What happens if they are renounced can be described
linguistically and ontologically:
Linguistically: the variable even ceases to be a name, it becomes a placeholder for names of a certain kind of object, which are to be put into the place
that is, in a certain context for the same variable at each occurrence the same
name, while one context may require different kinds of place holders.
Ontologically: The variable does not indicate an object but rather the opportunity to evoke a certain kind of objects a virtual rather than an actual
variability.

494

CHAPTER 17

Expellas naturam furca, tamen usque recurrit nature though driven out
with the fork, nevertheless returns. The mathematical purism of high value
within mathematics is a forced and less sufficient language as soon as one steps
out of mathematics. The abundance of variable objects of the half-way mathematised vernacular can be eliminated by linguistic sophistication but by this
linguistic measure they are not disposed of. And even more important in
order to eliminate them by linguistic tricks, one must once have experienced
them. Indeed this is the only way to guarantee that one is able to restore them

as soon as one falls back on the vernacular in order to do something with mathematics in the real world. The world is a realm of change, describing the world
is describing change, and to do this one creates variable objects physical,
social, mental, and finally mathematical ones. There exist many languages of
description, or rather many levels of describing. At a high level of formalisation
the variable mathematical objects can be forgotten, but at less formalised levels
they are a genetically and didactically indispensable link with the physical,
social, and mental variables, which for their part are indispensable.
17.2. Dependence (or Connection)

In the phenomenological approach the variable is more than an instrument of

formalised mathematical language and even more than something one uses by
way of speaking. This should be completely plain before I start dealing with
functions. Indeed, the very origin of the function is

stating,
postulating,
producing,
reproducing
dependence (or connection) between variables occurring in the

physical, social, mental world,

that is, in and between these worlds. Particularly important are

mathematical variables
mutually related or related with the others.

The dependences themselves can in turn be objectivated, that is given the status
of mental objects. On the way towards this objectivation such a dependence can
be
mentally experienced,
used,
provoked,
made conscious,
experienced as an object,
named as an object,
placed in larger contexts of dependences.

FUNCTIONS

495

The precision with which such a dependence is objectivated can vary from
classifying according to sorts of dependences
to relating orders with each other (the more this the more that)
to relating in a more or less precise, possibly numerical way with

each other,
and depending on this precision the dependence can be given

a generic name,
a proper name
of ad hoc character
of algorithmic character.
Examples:
A body falls. There is a dependence between witnessed time and place of the
body. The dependence is more or less consciously experienced, as free fall (of

the body), which is the sort in which this dependence is placed. The longer the
body falls the faster an order relation, it falls according to a formula, the law
of free fall, by which the dependence is placed within the larger context of
uniformly accelerated motions.
Two elastic bodies collide. In the game of billiards the dependence between
the pairs of velocity vectors before and after the collision is experienced, used,
provoked; it is consciously encountered in a larger complex of experiences, and
in its sort described by the term collision. The dependence is more precisely
described by a formula and a whole theory, which also accounts for the spin of
the billiard balls.
For dependences in which time plays a part, one knows a number of generic
names, such as
movement, growth, decay, process, course, trend,
some of which are also used metaphorically when time is replaced with

change of standpoint, direction of vision, and so on.

For numerous physical dependences a generic name is

causal connection,
made more precise by such terms as

by attraction, friction, heat transfer, oxydation, collision, refraction,

optical representation, and so on.
From this one should distinguish the
automatic (or programmed) connection,

496

CHAPTER 17

for instance
between touching a key on the piano or typewriter and the production
of a sound or a typographical sign,
between turning a switch and certain mechanical or electrical phenomena,
between aiming and hitting.
A three-year old draws with chalk on a blackboard. Doll growing bigger, chalk growing
smaller.

The dependence can be described by the generic name chalk consumption by

drawing and made more precise by a numerical connection between the size of
the drawing and the piece of chalk.

When the days begin to lengthen, the cold begins to strengthen a dependence between day length and cold (which holds only for a restricted period,
and even so in a stochastic way). Quality pays, the more they have the more
they want, the sooner the better, the longer it lasts the worse it gets many
sayings and locutions aim at dependences that are somehow described by the
relation between orders the the .

17.34. Function Terminology

17.3. The function is a special kind of dependence, that is, between variables
which are distinguished as

dependent and independent

an old-fashioned looking terminology, which, however, stresses the phenomenologically important element:

the directedness from something that varies freely to something that

varies under constraint.
Mathematised : the function from A to B as

act* that assigns to each element of A an element of B.

Functions are all around in mathematics and its applications, albeit labelled
in various ways:
mapping, transformation, permutation, operation, process, functional,
operator, sequence, morphism, functor, automaton, machine,

which are used according to needs and opportunities:

Function is preferred if the set of values is numerical,

mappings and transformations come from geometry but serve as well, with
certain attributes added in algebraic structures such as

morphisms, prefixed with certain prepositions or adjectives,

* I now prefer act to the more statical law, which I used formerly.

497

FUNCTIONS

functors, acting on morphisms;

permutation is the term for a one-to-one mapping on itself, in particular,
if studied in a group theory context;

operation or process is the term used with certain simple standard functions
(addition, root extraction).
If A itself is a set (or space) of functions, then in order to avoid repetitions or
ambiguities,* the terms

functional and operator are used for functions from A to B the first if
B is a set of numbers and the second if it is also a set of functions.
The term
sequence is usual, if not obligatory, for mappings from N.

Automaton or machine are didactical equivalents of function.

Only to mention it:

argument as a term for the independent variable in a function, which is

falling in disuse.
17.4. Functions can be considered as special relations. A relation from A to B
is any subset of the cartesian product
Such a relation f is called a
function from A to B if
for every

there is exactly** one

such that

This definition is logically equivalent with the former; phenomenologically

it is not, and didactically phenomenologically not at all. It obscures the essential
action of assigning directed from A to B, which in order to repeat the terms
at the start of Section 17.1 is of a
stating, postulating, producing, reproducing

character. One can oppose these two definitions to each other as

dynamic versus static

if one does not care to use outworn terms.

From the fact that all mathematics can be reduced to set theory, one may not
conclude that it should be done and even less that it is a didactic necessity and
possibility. Switching
from

towards

* In a similar fashion class, family, system are welcome synonyms for set.
** According to some authors: at most.

498

CHAPTER 17

is mental gymnastics for mathematically trained brains, which can be too much

for less advanced people.

I have discussed this question more profoundly in the past.* At that opportunity I have also refuted the fairy tale as though subset of
.
is to
be preferred above law of assignment,** because of its greater solidity.
Didacticians around 1960, who experienced set theory as a new revelation,
believed they had to impose this revelation to mathematical instruction, a
tyranny from which instruction has not yet wholly recovered.
17.59. Functions Terminology
17.5. A direct pattern to describe the connection between independent and
dependent variable is the

function table

with a progression of assignment suggested by the dots. A similar device is the

function graph
which suggests a connection

visually (Figure 217).

The function table can be used for a complete definition in the case of small
finite sets, for instance

* Mathematics as an Educational Task, pp. 387390.

** Now I would even say: act of assignment.

FUNCTIONS

499

or

for the permutation that interchanges 1 with 2, 3 with 4, and leaves 5 invariant.

A more general and more generally applicable device to indicate the connections between independent and dependent variable is the pattern
as a function of

Here the function can describe a factual connection such as in the given graph
the ordinate as a function of the abscissa,
the slope of the tangent as a function of the abscissa,
the area extending from a certain abscissa onwards;

or at a given curve
the curvature as a function of the arc length from a given point onwards,
the torsion as a function of the arc length from a given point onwards;

or on a given surface
the normal vector as a function of the point,
the total curvature as a function of the point;

or with spheres
the area as a function of the radius,
the volume as a function of the radius;

or in the movement of a mass point

the path as a function of the time,

the velocity as a function of the time;

or at a certain place
the temperature as a function of the time,

the atmospheric pressure as a function of the time;

or at a certain moment
the temperature as a function of the place,
the atmospheric pressure as a function of the place;
or combined
the pair

as a function of the pair

500

CHAPTER 17

or

the state of a system as a function of the time,

the expansion of the mercury as a function of the temperature,
the attractive force as a function of the distance,
the period of oscillation as a function of the length of the pendulum,
the intensity of some radiation as a function of wave-length and temperature ;
or in a certain national economy
the price index as a function of the time,
the income tax as a function of the income;

or in a certain society

the death rate as a function of the age,

the average income as a function of education,
the life expectance as a function of sex, social class, and so on,
the fertility as a function of the social class.
In all such cases there might be more causes of variability while one of them
is stressed. Sleeping variables which at some time can be wakened up are
termed
parameters

their significance will be dealt with later on.

17.7. The pattern ... as a function of. . . also serves to describe more formal
connections, like

with long objects

the length measure as a function of the measuring unit,
in the act of measuring
the precision as a function of the measuring procedure,
in the act of calculating
the precision as a function of the calculating procedure,
and purely formal connections, like
as a function of x,
as a function of x,
as a function of

FUNCTIONS

501

Shorter ways of notations are: the function f defined by

respectively, or

respectively, or

In two of these cases the function moreover depends on a parameter a or

it is a function that is itself a variable. If the dependence of this function on the
parameter is to be made explicit, then in the first notation the parameter must
be made explicit in naming the function
defined by
defined by

in order afterwards to define a function f by

respectively. According to the second kind of notation it would be

502

CHAPTER 17

though I never came across this notation. According to the third kind of notation
it is

In general, an expression
T( . . . x . . . ) as a function of x can be briefly written as the function f for which

or

or

If this expression involves parameters

then it becomes
the function for which
tion f for which

and the func-

respectively.
In all these cases the variability can be restricted by means of subscripts

17.8. Another pattern comes from indicating functions by means of

the of
and suchlike schemes in linguistic terms, the genitive or genitive-like constructions.
Examples to illustrate this:
the mother of
the birthday of
the address of
the value of
the rise of
the start of
the target of

(a person),
(a person),
(a person),
(a coin),
(a celestial body),
(a process),
(aiming),

503

FUNCTIONS

the moment of ...

the place of ...
the tip of ...

(a happening),
(an object),
(an iceberg);

or from mathematics
the midpoint of ...
the median point of ...
the interior of ...
the area of ...

(a line-segment),
(a triangle),
(a sphere),
(a plane figure);

the
the
the
the

(a magnitude),
(a magnitude, quantity),
(a number),
(a positive number);

half of ...
triple of ...
square of ...
square root of ...

the sine of ...

the maximum of ...

(an angle, arc),

(a function),

the limit of ...

(a sequence),
the greatest common divisor of ...
the union of ...

(a set of sets),

the complement of ...

(a set),

(a set of natural numbers),

the number of elements of ...

(a set),

the set of subsets of ...

(a set);

the sum of ...

(two numbers),
the set of divisors of ...
(a natural number),
the number of divisors of ...
(a natural number);
the degree of ...
(a polynomial);
the operation of multiplication of ...
the factor group of ... by ... ;

(a group),

the solution set of ...

the domain of ...

(a condition),
(a function),

the derivative of ...

(a function);

the closure of ...

(a set in a certain topology),

the topology of ...

(a linear space);

the function belonging to ...

the graph of ...

(a graph),
(a function);

the equation of ...

(a conic),

the curve belonging to ...

(an equation);

the truth value of ...

(a proposition),

the disjunction of ...

the system of well-formed formulae of ...

(a system of propositions),
(a formal language),

the system of true statements of ...

(a formal system).

504

CHAPTER 17

17.9. Some functions have fixed notational names in the mathematical language.
Usually it is a function symbol with on its right side between parentheses or

not the thing on which the function is acting. Examples:

sin, cos, tan, log,

, max, sup, lim, gcd.

Less usual
#:

number of,
set of subsets of

If a finite or infinite sequence is interpreted as a function acting on

{1 , . . . , n}

or

N,

then in
respectively)
a can be considered as a function symbol with the thing on which it acts put
lower down rather then level. As regards the exponential function

one could maintain that the variable has climbed, but the exponential function
as such is better indicated by
exp.
In the sum as a function of two summands the function symbol appears between
the variables,
instead of

A similar notation is that of

factor group of by . . .

... / ...

In the notation of
square of . . .

it is the function symbol that has climed (to the right), as well as in
derivative of ...

In
closure of ... ,

conjugate complex of ...

the function sign is floating above that on which it acts.

505

FUNCTIONS

In

absolute value of ...

the function sign split into two parts surrounds the variable acted on, as it is
the case in
whole part o f . . .

[...].

These, however, are mere interpretations which can hardly be maintained.

If there is a need for symbols of such functions, one is better served by

The differential operator is more favourably placed; in

the
can be regarded as a function (operator) symbol that describes a transformation of a function of x into a function of x.
The integral operator is more troublesome; in order to consider

as a function of f, and certainly with

as a function of f, descriptions like

g defined by

as a function of f

or

and

g defined by
or

respectively, are to be preferred.

as a function of f

506

CHAPTER 17

17.1014. Sorts of Functions Terminology

17.10. Beside explicitly given functions and functions indicated by algorithmic
or proper names, there is a need for ambiguous names for functions in order to

deal with functions affected with, or subjected to, variability. Such devices have
already repeatedly been used in the preceding sections. In most cases the function indicated by the letter f was specified, but in Section 17.9 letters such as
f were also used to indicate rather arbitrary functions. The need for ambiguous
names for functions is felt as soon as statements (definitions included) are to
be made about a whole class of functions, the need for variable functions as soon
as acts are performed on, or yield, functions the function as independent or
dependent variable.
In Section 17.3 we enumerated synonyms for the term function; according
to the kind of functions under consideration there is some preference for a
special synonym. The following expositions about terminology with regard to
sorts of functions somehow overlap that of Section 17.3.

17.11. One interprets

functions of, say, m variables
as simple functions by nominating

an ordered m-tuple of variables

as independent variable, thus
An
ordered system of n functions

with the same independent variable can be considered as a simple function by

nominating

an ordered n-tuple of variables

as dependent variable, thus

In this way we get for instance

mappings out of m-space into n-space,

Such mappings are also used to describe

curves, surfaces, and so on

paths, deformations,
movements.

FUNCTIONS

507

Used in this way, the independent variables are also called parameters a point
to which we will pay attention later on.
17.12. Other aspects according to which functions (mappings) from A to B
are distinguished are

peculiarities of A,
peculiarities of B,
peculiarities of original sets of elements of B, for instance a one-to-one

(injective) mapping,
peculiarities of the factual image, for instance a mapping on (surjective),
smoothness of the mapping, for instance continuity, differentiability,
and so on.
Some school textbooks distinguish between functions and mappings in the
way that functions from A to B need not be defined in all A, wheras mappings
from A to B are defined in all A . This diverging school text terminology is

not to be recommended. If one wishes to indicate functions (mappings) to B

without specifying the part of A in which they are defined, one may prefer the
terminology
function (mapping) out of A into B.

I used it before (Section 17.11).

17.13. Sorts of functions can be defined implicitly by imposing conditions, for
instance
operations on or in certain sets,

required to fulfill certain conditions, which lead to a certain structure,

group, ring, field, linear space, and so on.

Mappings of certain structures can be subjected to requirements of preservation of structure, the

morphisms

(with various kinds of prefixes).

17.14. If a function acting on N is considered a sequence, one gets in accordance
with the variability of the dependent variable

number, point, function, and so on, sequences.

A function acting on N must not necessarily be considered as a sequence. One
speaks of the

508

CHAPTER 17

sequence of natural numbers:

of even numbers :
of squares
of prime numbers :

prime number

yet of the

function of Euler
(that is the number

of numbers

relative prime to n), and the

function of Moebius
if n is a square, otherwise + 1 or 1 according to whether the number
of prime divisors is even or odd), though in the context of the function of s

one would be inclined to speak of the sequence of coefficients this expression

is as it were structured as the inner product of two sequences, that of the
and that of the
Similarly in

one speaks of the sequence of coefficients and the sequence of powers of x.

17.1516. Parameters

17.15. The term parameter as synonym for variable is used with three meanings.
First, a secondary as it were sleeping independent variable, which, if
need be, can be accounted for as it were wakened up for instance in order
to get a system

of functions by the variability of the parameter t.

Second, a variable that according to its origin is dependent but according to
its appearence is independent, and which serves to distinguish figures, structures,
and so on, from each other within their sort, for instance

the radius of the circle as a function of the circle, by means of which

circles are distinguished according to size,
the parameter p of the parabola (equation
by which noncongruent parabolae are distinguished this is the very origin of the term
parameter,
the halving time as a parameter of a decay process.

509

FUNCTIONS

17.16. As a third occurrence one knows the so-called

parameter representation
of curves, surfaces, and so on.
Examples:
The unit circle in the
plane (equation
by means of the arc length s from a fixed point,

is parametrised

The unit sphere in the

space (equation
) is parametrised by means of two parameters s and t (corresponding to geographical
longitude and latitude)

In general, what is phenomenologically experienced as curve or surface, is

parametrised by one parameter, or two, respectively. These parameters arise as
dependent variables (arc length, longitude and latitude), dependent on the point
of the curve and surface. A posteriori they are used as independent variables
in order to
represent the curve and surface parametrically

Then the curve (surface) appears as the image of a number (pairs of numbers)
set.
An important aspect is the liberty to change parameters,

the parameter transformation.

With the new parameters the curve (surface) is described by (Figure 218)

510

CHAPTER 17

Old and new parameters are coupled to each other by the points of the curve
(surface) they describe a one-to-one mapping, respectively:

where, respectively:

Afterwards by a change of perspective the mental object curve (surface) is

mathematised:

A curve (surface) in R is a sufficiently smooth mapping out of the onedimensional (two-dimensional) parameter space into R,
where mappings that arise from each other by parameter transformations are

considered as the same.

More precisely:

A curve (surface) in R is an equivalence class of (sufficiently smooth)

mappings out of one-dimensional (two-dimensional) space (the parameter
space) into R with the equivalence relation:
if and only if there
is a one-to-one mapping (out of the parameter space into itself) such
that
By this change of perspective it is made possible to consider curves (surfaces)
with multiple points (lines) really as curves (surfaces) and to count multiple
points as simple or multiple according to circumstances: parameters on the curve
(surface) are introduced which take different values in the multiple points
according to the direction one comes from, and by means of these parameters
the curve (surface) can be described as a map or a mapping in the map the
multiple point is counted simply, in the mapping multiply (Figure 219).

Of course one need not restrict oneself as we did here to one or two parameters. The extension to more parameters is obvious: k-dimensional hypersurfaces are described or defined by means of a k-tuple of parameters this
indication may suffice here.

FUNCTIONS

511

17.1721. Function What is their Meaning and Use?

17.17. An overwhelming phenomenological variety of what is called function!
But is it not the same with all mathematical concepts? Mathematical activity
often aspires at unification, with the consequence of a vision directed from the
sublime endpoint of the activity to the cloudy sources a vision that reveals
itself mathematically and mathematical-didactically.

So far, however, we have stuck to an enumeration of the phenomenological

appearances of the function. What do you do with functions? this question
has hardly been tackled, except perhaps in the last section where the phenomenon

parameter required some illustration. But such an enumeration does not mean
much as long as the function does not become functional. If this point is
arrived at, it may appear that what is unified under the term of function is not
that much a union.
Set too is ascribed a unifying task, structuring mathematics itself. The
universe of sets with its relations of intersection, union, complement, cartesian

product, subset, and so on, is a rich structure yet because of lack of interpretation
poor mathematics. The true mathematical wealth is created by the perspective
of function. We have displayed a rich variety of objects which are understood as

functions. But it is one thing to bestow on an object the title of function and
another to do something with it, to work with functions in a way that is charac-

teristic of functions. Is function a name I can attach to all that fulfills certain
requirements or rather a signal how to act in certain contexts? Does one call
a thing a function in order to do something with it, and if so, what? It will
appear that function, which by its pure definition looks a unifying concept,
is much too variegated as regards its operational virtue to be profoundly unifying.

Moreover it will appear that the need for objectivation that is promotion
to an object of the function is quite often no more than a need for unification.

The intellectual pleasure of unifying is certainly not a bad thing. There are
enough examples to prove that this pleasure can create precious mathematical
products. But beyond this pleasure to structure the phenomenon that is called
mathematics by unifying means, one would expect other incentives. Anyhow

if they exist, phenomenology is obliged to account for them and to analyse

them.

17.18. The question is this really worth being called a function? will less
often be asked of the examples 17.67 which illustrate the pattern
... as a function of ...
and more often of quite a few of the examples 17.8, which fill out the function
forming pattern
the ... of ...

512

CHAPTER 17

What kind of context should justify the part played by functions like

the
the
the
the

mother of ... (a person),

address of ... (a person),
birthday of ... (a person),
place of ...
(an object),

and what part do they actually play? It is perhaps possible to contrive such
context but as long as this has not been done, I prefer to affirm that they
have been invented only in order to give a few more examples of what can be
called a function.
The more mathematical examples joining that list are a bit different. In
contradistinction to the preceding list it may happen that in many, if not in all,
of these cases one meets or introduces some symbol for the function under
consideration, whereas function symbols for mother of . . . look farfetched.
Some usual function symbols, belonging to this list, have been mentioned in
Section 17.9. The list is not large; one can perhaps add some symbols but even
then it would not become imposing. Even algorithmic name giving is unusual for
many among these functions. The midpoint of a line-segment is readily called
M, and if it really matters one can add in parentheses the line-segment that is
actually meant, but does this promote M(. . .) to the rank of a function? And
the same holds for median point of . . ., interior of . . . and many more.
Giving the topology of a space R the name T(R) may be a pleasure that is
restricted to such contexts as by T(R) I mean a system of subsets of R with
the following properties . . . and to the trick of comparing two different
topologies
and
of the same R by making use of a possible inclusion
relation between
and
Yet does this lend T the status of a function?
For the closure of a set S there is even a fixed notation , but who then
imagines as a function of S (and of the topology that defines the closure)?
Though

or

is clearly recognisable as a function, nobody is aware of the fact that this function
has an established name,
sum of . . . and . . .
For the solving set of the condition F(x, y, z) one has the notation

but it requires quite special circumstances to consider this set as a function of

the condition F(x, y, z), for instance that of a differential equation where one
is interested to know how the solution depends on the initial conditions.

513

FUNCTIONS

The widely diverging degree of operationality of the concept of function

will be illustrated with three paradigms.
17.19. A quite usual notation is

C(M, r)
for

the circle with centre M and radius r.

Here C seems to be a function that produces from a pair

a
circle. But is it really a function? I do not think so, it is just a notation. Of
course, one knows that a circle in the plane is determined by its centre and
radius, but one's attention is not attracted by this dependence as a mental
object. It is just a notation such as PQ for the line determined by the points
P and Q. Nobody asks more profound questions. The C of circle is not properly
a function symbol. The notation is just a formal imitation of the genitive
construction
the sine of . . .
the logarithm of ... ,

an abbreviation by analogy of
the circle with centre M and radius r.

C awakes the impression to be a function symbol, but it is not operational

as such.
But let us now restrict the liberty of point M to a curve K and r to a fixed
value. Immediately the function character of C becomes patent. M on K determines a circle C(M, r) and if M runs along the curve K a nice system of circles
with an envelope by pure set theory* determined as the set

If moreover r is made variable and considered as the time t, then one can interpret
C(M, r) as a wave front

of a wave motion and the envelope of the circles

as

wave front G emitted by K.

G is now a true function symbol, that is for

the wave motion G,

which at the moment t produces
the wave front G(t).
* A few conditions must be fulfilled if indeed the envelope has to coincide with this set.

514

CHAPTER 17

What does this example mean? That which originally was introducted as a
pure notation, bears in itself the potency of a function, though at least in the
present case a potency that could be actualised without speaking of functions.
Indeed, I could have told the same story with the same or perhaps even more
convincing power without using the word function even once.
17.20. For the symmetric group of the n permutands 1, 2, ... , n one knows
the notation
Is this more than a notation or does it really mean (symmetric group) as a function of the number of permutands? The answer is of
course: it is not.

Yet as soon as one considers the symmetric group

of the (finite) set

A (the group of all permutations of A, that is of all one-to-one mappings of A

on itself), one heads for a more essential use of the function notation. For

instance: for the aim to prove that a one-to-one mapping of A in B induces an

isomorphism of
in
I took care to introduce
as the symmetric group of {1, ... , n}. The
usual terminology is

the symmetric group of n permutands,

as though the nature of the permutands did not matter. Indeed, often it does
not. Yet sometimes it does. For instance if I set out to state that two full permutation groups with the same number of permutands are isomorphic.
If
is interpreted this way, that is, as
the symmetric group of n permutands,

then as a function symbol does not assign one group to n but a whole class of
groups that are isomorphic with each other in a well-determined way. Then

G is a symmetric group of n permutands

should rather be written as

17.21. Let us dive more profoundly into this example. Let

again be
defined as was done in the beginning of the preceding section. As far as function
symbolism is concerned, the notation
does not count much if compared
with all that is required and advantageous for understanding permutations as
mappings. Well, I can dispense with function symbols as long as I deal with a
single permutation, say for
the permutation that replaces
1

4,

with

FUNCTIONS

515

respectively. On this level I can even perform given permutations in succession

and invert them and even arrive at some understanding of, say,
Of course
this is quite a job: considering a permutation of objects as an object, beside, and
cooperating with, other objects. Just because this is a hard thing, one feels the
need for a notation of functions and operations with functions. This then leads
us into the alley, or rather on the highway of formalising.
Let A and B be two equipotent sets. Their
and
then are isomorphic. More precisely, each one-to-one mapping from A on B produces an
isomorphism from
on
Or, more generally:
A one-to-one mapping

from A in B induces an isomorphism of

in

to wit according to the pattern

the permutation f of A determines a permutation g of B: for each

we put by definition

and for each

In order to express how this g depends on f we need some function symbolism,
say,

and only if this has been settled one can go further, for instance to state that

no profound fact, but anyway a fact that in order to be enunciated requires

the function as a mental object with the full function notation.

17.2236. The Historical View

17.22. From enumerating what is called function and how this is done, we were
led to the question why?, from stating facts to looking for causes causes
that, as it happens in mathematics and mathematical terminology, are necessities.
The rich variety we have displayed, emerged within a wholly mathematised
context. Phenomenological analysis should start earlier. With the question

516

CHAPTER 17

whether this earlier implies an earlier place in this chapter, as well as with the
total arrangement of the subject matter, I have unceasingly struggled the
present chapter has been written and rewritten five or six times.
The phenomenological analysis should start earlier, but this earlier must
be exposed later. Earlier can also be understood historically, and in such an
investigation, history can be a guide-line that is not to be despised, albeit one
with twists that need not be followed though in the past some people believed
that the individual repeats the history of mankind in fact education and
instruction take care that this neither does nor need happen.
17.23. How much can the present investigation profit from history depends on
the profundity allowed the historical look. One can simply state that the word
function emerges in the correspondence of Leibniz and the Bernoullis and
that not until at least Euler does the use of this word start resembling what we
are accustomed to. We can also state that Euler and dAlembert were the first
to use indeterminate function symbols, and then by preference the familiar f
and

We can signal concept and term mapping (Latin: representatio) in

Eulers paper (1777) where he studies mappings of the sphere on the plane
in a cartographical context. Such mappings are of course analytically described
by systems of functions of systems of variables (parameters), yet this perspec-

tive is changed in the beginning of the 19th century when one starts considering
more or less explicitly such systems of functions as mappings. Mappings, if
not produced analytically, are of geometrical origin. They become an important
device in set theory and an object of investigation in topology. Meanwhile
function theory real and complex has grown enormously. Not until the
20th century do these two streams function and mapping merge in one

bed.
17.24. In many respects this historical account is too simplistic. Two streams
are not enough, there are more of them which can historically and must phenomenologically be distinguished. Moreover they can be traced back to more

ancient sources. The history of a mathematical concept starts long before it

is given a name.
Tables often represent functions and about such tables we know from
Babylonian mathematics onwards; it is as the background of tables that the
function reveals itself most patently in astronomy tables well-known from
Babylonian and Greek astronomy, mathematically constructed from empirical
data. The Babylonians at least since the 5th century B.C. proceeded in this
work more phenomenologically than the Greeks. They observed periodicities
in the positions of celestial bodies and interpolated their course in between
linearly, or rather piecewise linearly, taking into account the observed periodical
variations of velocity. The functions by which they attempted to describe the
celestial motions and by means of which they calculated their tables, are consequently superpositions of periodic zig-zag functions (Figure 220). The Greek

FUNCTIONS

517

proceeded less phenomenologically. They used kinematic models: centric and

eccentric circular movements and epicycles, that is, circular movements dragged
on by others (Figure 221). Mathematically this means that they described the
celestial movements as superpositions of trigonometric functions a model
that maintained itself up to Kepler. Then circles were superseded by ellipses
the sun in a focus and in equal times equal surfaces swept out by the vector
from sun to planet.

Interpreting celestial motions in functions this is indeed a historically

natural invitation to mentally constitute functions and continuity. One observes
the rotation of the celestial globe and with it the motions of sun, moon, and
stars as function of the experienced time, making this subjective time more
precise by means of the more objective looking, suggested by the kinematics of
the celestial globe. One sees the monthly backward course of the moon outlined
according to longitude and latitude between the fixed stars and learns to express
it numerically by means of the objective time, the kinematics of the celestial
globe to be sure only pieces of the course are actually observed and the
remainder is smoothly filled out. Less easily is the yearly backward course of the
sun along the stars observed directly by the setting and rising of stars of the
zodiac in the morning and evening dusk, or indirectly by the full moon as the
opponent of the sun. One states inequalities in these motions and attempts to
describe them by superpositions of piecewise linear or trigonometric functions
and, as to the latter to explain them by mechanical models. The forward and
backward movements of the planets are as many invitations to constitute and
analyse functions continuous functions of time. There are more of them:
the shadow directions and lengths as functions of the time time objectivated
on the sundial the meridian altitude of the sun as function of the geographical
latitude and noon as function of the geographical longitude.

518

CHAPTER 17

17.25. So far we have functions of time linear zig-zag or trigonometric functions. But Greek astronomy knew trigonometric functions also under another
aspect. Place finding by numerical coordinates in the celestial globe requires
spherical trigonometry which in fact strangely enough as we might think
has preceded plane trigonometry. We cannot tell exactly when this had
happened, but in Ptolemaeus astronomical work, the Great Syntax we find
it in the state where it remained for better than a millenium. Spherical triangles
have sides and angles, as the plane triangles do, but their sides are circular
arcs and hence also measured by angles. Between sides and angles of a spherical
triangle there exist relations resembling those of plane trigonometry, for instance
the sine theorem of spherical trigonometry

Greek astronomers and mathematicians knew what we call trigonometric functions, though in another setting. In Ptolemaeus work we find chord tables,
the subtended chord as function of the angle in a circle, the radius of which is
divided in 60 and the circumference in 360 parts. The connection between
sines and chords is clear (Figure 222)

The Hindu word for chord, phonetically adapted by Arabs and in their language
interpreted as bosom, led to the Latin translation sinus (the Latin equivalent
of bosom).
17.26. Even older than the sine is the tangent, which can be traced back to
Babylonian mathematics. As a measure of inclination and as the shadow length
belonging to a given length or angle of incidence, the tangent is quite natural.

Trigonometric functions, in particular the tangent, are used in Archimedes

Circle Measurement, the calculation of
17 .27. I started this historical exposition with trigonometric rather than with
the more elementary and natural functions like linear, quadratic, cubic ones,

which in fact can also be traced back to Babylonian mathematics. The Greek
term for linear dependence means proportionality: two variable magnitudes,
most often of geometrical origin, behave according to a fixed ratio. The areas

FUNCTIONS

519

of circles, the volumes of spheres are in the ratios of the squares and cubes,
respectively, of the radius a terminology, which maintains itself up to Kepler,
and even longer: according to his third law the squares of the times of revolution
of the planets are to each other as are the cubes of the long axes of their orbits.
The Greek had even names for inhomogeneous linear functions: a magnitude
if compared with another is a given magnitude bigger (smaller) than in proportion* because of the lack of appropriate symbols a cumbersome verbalisation
of the relation between x and y:

Even more involved is the terminology to describe quadratic functions. It is a

geometric terminology which serves to formulate, classify quadratic equations

known before to the Babylonians the so-called application of areas:

applying a given area F as a rectangle to a line-segment a (Figure 223),

plainly, that is
such that a square falls short
such that a square exceeds

These three cases are distinguished with the Greek words for agreement, falling
short, and excess as
parabolic,

elliptic,

hyperbolic

application. This, then, is via Apollonius the origin of our terms for the conies:

if F is given as a square with side y, then the above equations become

* Euclids Data.

520

CHAPTER 17

respectively, which are indeed the equations of parabola, ellipse, hyperbola.

In this terminology a quadratic function in x is described as an area applied

to a given line-segment (the independent variable) with deficiency or excess.
This is the way Archimedes expresses himself* when he considers what we
would call the graph of a quadratic function in order to perform what we would
call an integration.
17.28. The Greeks were familiar with describing curves and surfaces by means
of equations; in the case of the conies the equations even preceded the curves.

The equations, however, were not written in an algebraic but in a geometric

language: addition and subtraction are operations on line-segments, areas, and
other geometric magnitudes; the product is obtained as a rectangle of linesegments or avoided by positing proportions between geometric magnitudes.
The Greek mathematicians had sound reasons to
express algebraic relations geometrically,
which I will not explain here. The change of perspective to
express geometric relations algebraically

was performed by Descartes (and simultaneously by Fermat); it became possible

by Vietas creation of letter algebra.
Indeed, it is first of all geometric and kinematic-geometric magnitudes and
relations (such as the cycloid) on which the predecessors and inventors of the
analysis test the strength of their method. If a point is running along a curve,
one perceives a great many magnitudes that vary in dependence of each other:
not only abscissa and ordinate, but also the piece cut out by the tangent or
normal in a variable point on a fixed line from a fixed point, or the piece of the
tangent or normal cut off by a fixed line, or the radius of curvature, or the
coordinates of the curvature centre, or the distance from a fixed point, or the
arc length from a fixed point on the curve all of them functions in the
original LeibnizBernoulli terminology. Magnitudes vary in dependence on each
other the picture present in Newtons mind is that all of them together depend
on something like time and these dependences can, but need not, be expressed
by formulae.
If u and v are such magnitudes, then thanks to the connection that ceteris
paribus exists between them, an infinitesimal growth of u by
corresponds
to an infinitesimal growth of v by dv while neither of them is indicated as the
dependent and independent variable. The quotient
yields the growth
* See for instance: E. J. Dijksterhuis, Archimedes, 1956, p. 124.

FUNCTIONS

521

ratio in Newtons notation the ratio of fluxions

If the connection
between u and v is specified,
can be more closely determined, for instance
if

then

thus
I have explained elsewhere* that this view has maintained itself in physics,
that is, in those parts of physics that have sprung from classical mechanics.
If a mass point moves, its place, velocity, acceleration are functions of time,

the velocity is the derivative of place by time, the acceleration the derivative of
the velocity,

but there is no need to make the functions explicit, as we have learned to do

with function symbols,
in order to express a special dependence between time and place, which implies
that of velocity and acceleration,

Function symbols are of a later date than functions in the LeibnizBernoulli

sense, but to date the confusion between symbols for magnitudes and for
functions has persisted I discussed it in the earlier mentioned place.
In the original approach there is nothing like an independent and dependent
variable. There is dependence of variables on each other, which can be turned
into linear dependence between differentials. According to the needs some of
the variables may be stamped as independent and the others as dependent.
The need can be prompted by necessity, and this happens as soon as one passes to differentials of higher order. From the second order onwards one knows the asymmetry
between numerator and denominator in such expressions as
The inventors of differential calculus had a hard time before they understood that second
differentials are not meaningful unless certain first differentials are kept constant, namely
those of the variable declared as independent.
* Mathematics as an Educational Task, pp. 553559.

522

CHAPTER 17

I apologise for this side-leap, which may be less intelligible to outsiders. I had somehow
to explain why the easy looking game with mutual dependence did not satisfy in the long
run; at certain moments one must decide which are dependent and which are independent
variables if successful analysis is at stake. It is strange that people who believe they can
replace calculus by non-standard analysis refuse to learn this from history a point that
should be discussed at greater length.

17.29. The original view that each variable (or each system of mutually independent variables) can be chosen as the independent one (ones) had momentous
consequences for the development of Calculus. One could work with

without specifying or at least symbolising the actual dependence and in the

same breath one could assert that

One could switch to and fro between the available variables, for instance simplify
integrals

by passing from the independent variable x to a more appropriate t with

(In fact such an integral transformation was the great aha-experience in

Leibniz discovery story.)
Knowing how x and y and how y and z depend on each other one could
state by elimination the kind of dependence of x and z, and conversely by
inserting between x and z an intermediary y one could break the mutual dependence of x and z into one between x and y and one between y and z, moreover
with regard to the differentials one could establish the useful fact that

17 .30. This is an appropriate place to uncover the roots of the explosive growth
of the analysis and in particular of the tremendous success that was meted out
to Leibniz notation no sooner than in the course of this phenomenological
contemplation have the ideas that I am going to explain become clear to me in
all their pregnancy.
In the function notation the transition from one variable t that is given as
independent to another u that is chosen as independent means a substitution
Deriving from the dependence between x and y and that between y and z the
one between x and z means in the function notation composing the function

f and g that describe the dependences

523

FUNCTIONS

to a function h

In the dependence between x and y passing from x to y as the independent

variable, means inverting the function f that maps x on y.
Up to the rise of the new analysis, algebra was dominated by the typically
algebraic operations of adding, subtracting, multiplying, dividing, raising to
power and extracting roots. All these operations can also be performed with

functions. But functions open a new perspective:

substituting into each other and inverting.

This created on the one hand a never before known wealth of new objects
functions as wild as one wants to contrive; on the other hand the opportunity
to break up and invert functions in order to handle them more easily, and for
all these operations simple rules to put differentials and integrals in relation
to each other.
In Sections 17.1721 we asked the critical question functions what is
their meaning and use? Partly this is answered in the last paragraph, albeit
in a historical perspective: the strength of the function concept is rooted in
the new operations composing and inverting functions which create new
possibilities.
17.31. Let us return to Section 1728 where we interrupted the course of the
exposition: the viewpoint where the dependence between variables was not
yet explicitly indicated but implicitly determined by the geometric, kinematic,
physical relation at the background. Sooner or later this relation itself is made
explicit; the dependence is described by an equation, say,

Considering f anew as a function, of three variables, is the next step, which by

itself creates the opportunity to consider arrays of dependences

according to the choice of the constant.

The dependence between variables, say,

can be given by a k-tuple of equations

and here too one gets the opportunity to consider

as functions and

524

CHAPTER 17

as a description of a system of dependences, according to the choice of the

On the other hand this becomes the description of a mapping out
of
-space into
-space; conversely such a system of

equations can be interpreted as a mapping.

17 .32. This does not interfere with the need, if an equation

or a system of equations

is given, not to be satisfied with the mutual dependence but to consider

one variable (a few variables) dependent from the others,
which means
solving with respect to one variable (a few variables)
the equation (the system of equations).
Equations most often have more than one solution. If it is the equation

one shall make explicit both of them, by

3. In the case

or

and certainly in the case

one will be inclined to be satisfied with the notation

to indicate both solutions root symbols with an ambiguous meaning.

With this kind of multivalued symbol one has indeed worked and struggled
during centuries in algebra. In fact one can work with them and even quite
smoothly as long as one takes care and sees to it that a multivalued symbol
(even as a variable) keeps the same meaning in the same context. It is true this
may be not that easy. For instance, if one uses the formula

in order to replace
with
one may not forget to check whether
did not occur earlier in the same context, and then perhaps in another meaning,

and this indeed may require too high an intensity of watchfulness.

FUNCTIONS

525

The knot has finally been cut, though afterwards it has lasted almost a
century until this gordian solution was generally accepted:
the by definition univalent function.
But blood is thicker than water: in order to globally save complex analytical
function such as the nth root or the logarithm, which cannot be dispensed with,
one invented ad hoc domains where they could behave univalently the Riemann
surfaces but this is a question I had better skip.

17 .33. Originally functions are given implicitly or explicitly by analytic

expressions. Yet as soon as he had to solve the differential equation of the
vibrating string, dAlembert came across arbitrary functions, the amplitude
of the string at a certain moment could arbitrarily be prescribed. Solutions
of partial differential equations in general depend on arbitrary functions. As a
matter of fact this phenomenon led to the function notation, the ambiguous
names f and for a function. During the 18th century one has continuously

but unsuccessfully struggled with the arbitrary functions. Even more confusion
was created by the phenomenon that the differential equation of the vibrating

string could be solved by means of arbitrary functions as well as by what we

now call Fourier series, which look analytically nice but can represent strange

functions, even with discontinuities.

This knot was cut by Dirichlet, when in 1829 he studied Fourier series: the
function, with no regard to its generation, which to each point (of an interval)
assigns an arbitrary number. This was so revolutionary an idea that still a century
later older mathematicians, if they meant an arbitrary function, for safety
reasons added in the sense of Dirichlet.
This is the essential feature of the function concept as it has evolved in history

univalent

and

arbitrary.

The arbitrariness is no longer limited one may imagine a function equal to

1 for each rational number and equal to 0 for each irrational number.
Well, if it is needed, one will restrict the arbitrariness. But even continuous

functions can be arbitrary. I could not find out who was the first to interpret
the daily course of temperature at a certain spot as a function (of time).
Which was earlier, the graph of an empirical happening or the locution that
this represents a function of time?
17.34. The adjective arbitrary on function aimed in this historical exposition at the kind of dependence that the function constitutes between the
variables, not at the variables themselves. The variables were numbers or otherwise continuous magnitudes real and complex ones, possibly bundled in
n-tuples, in order to describe points in geometry or systems in mechanics. If
at present we speak of a function from A to B we admit in principle arbitrary
sets A and B. The road to this goal was paved with the one-to-one mappings

526

CHAPTER 17

of finite sets in themselves or others, which were known under names like
permutations, combinations, substitutions. The finite sets concerned were
primordially with Lagrange, Ruffini, Abel, Galois the roots of algebraic
equations, and from the start composing permutations was an essential element

in these investigations. Long before group theory, groups of permutations

(performed in succession) were implicitly involved in these investigations.

Composing and inverting is even here the most relevant aspect of what appears
as a function to our view.
17 .35. From the self-mappings of finite sets the road leads to mappings, in
particular one-to-one mappings, of arbitrary sets, the instrument by which

Cantor compares infinite sets. And now all the fences are down. Functions,
under any name one can imagine, are called up to restructure old domains.
An instructive example is calculus of variations. From the dawn of analysis one

knew maximum and minimum problems for functions to be solved by calculus:

maximising or minimising a function depending on certain variables. Beside this
there were the variational problems: the curve of the fastest descent (between

two points in a vertical plane though not on a vertical line), the revolution body
with the least resistance, the shortest line between two points on a surface
are a few examples. Here the variables are not numbers or points, but curves,

surfaces, functions, and that what should be maximised or minimised by an

appropriate choice of the variable is a number depending on the variable curve,
surface, function a functional, as such a function, of a higher order as it were,
will be called later on.
And so it goes on: variables which are in fact functions, or operations on

functionals, operators, and so on do the job of functions. And of course variables

which in fact are operators, acted on by functions called by various names.
17 .36. Geometrical mappings, whether algebraically represented or not, were
the principal motor of geometry as it developed in the 19th century, and again
composing and inverting such mappings was the structuring feature, which

finally resulted in the explicit investigation of the implicitly much older geometrical groups. In topology, mappings continuous and topological ones
become both the instrument and object of investigation. Even if they were

not called functions the preferred letter bears witness to the function at
the background of the mapping. The unification is announced before it is
pronounced.
17.37. The present author has lived through and undergone a part of this
development, but it should require cumbersome heart searchings to find out
the dates when for the first time he sensed, accepted, applied the various mathe-

matical objects as functions. It would be worthwhile identifying their first

occurrence and their sporadic or regular or definitive use in the literature,

though of course literature is a belated witness of what occurred in minds.

527

FUNCTIONS

Sequences of numbers, points, functions as functions from N to a set of

numbers, points, functions when did it come up, when did it become conscious
to myself, when did I make it conscious to others? Coordinates as functions of

points (and bases in vector spaces), the i-th element of an ordered n-tuple as a
function of i, a predicate as a function from individuals to propositions, a binary
operation in G as a function from
a function of variables as

a function of an n-tuple of variables, the limit of a convergent sequence as a

function of convergent sequences I could repeat all the above questions
with respect to these subjects and I would as little be able to answer them.
The closure

of a set S as a function of S perhaps I saw it always this

way, perhaps it first crossed my mind when I saw Kuratowskis axiomatics of

topological space, which postulates for each set S an

such that

because not until (4) where the iteration of the closure emerges, does closure
become operational as a function.

Similarly mother of ..., father of ... become operational only in the

context of genealogy where functions of this kind are iterated and composed to
grandmother of ..., grandfather of ... and with eliminations to brother
of ..., sister of ... , uncle of ... , and so on.
17 .37a. Lessons of History

There were functions prior to the use of this name, and there are and will be
functions before they are named as such and before you have a name for them.
Systems have been developed to name functions algorithmically.
Variables sprung from two sources: variable object (physical, social, mental,
mathematical) and ambiguous name. Both sources, though covered by the modern

view of variable as placeholder are nevertheless still active.

Functions explicitly entered as relations between variable magnitudes,
the variability of which was infinitesimally compared.
The liberty to change variables, from dependent to independent, and between
independent ones, led to a new kind of operation with functions: composing
(substituting) and inverting (accompanied by eliminating).
This new operational wealth has caused the success of the function.
The necessity of distinguishing independent from dependent variables led
to stressing functions rather than relations.
Notwithstanding what is suggested by algebraic and analytic expressions, the
development tended towards the univalent function.
From the description of visual data by functions a change of perspective led
to visualisation of functions by graphs.
Though somewhat younger than the algebraically and analytically expressed
function, the arbitrary function arises as early as variational calculus and solving
partial differential equations.

528

CHAPTER 17

This arbitrary can mean the character of the functional relation as well
as that of the variables: numbers, number tuples, points, curves, functions,
permutands, elements of arbitrary sets.
The functions of analysis, the geometric transformations, the permutations
of finite sets and the mappings of arbitrary ones flow together, in order to
generate the general function concept.
This concept is used to comprise a great variety of things: algebraic operations, functionals, operators, even sequences, coordinates, logical predicates.
It can happen that what looks like a functional symbol is a mere notation,

which, however, can become operational as a functional symbol.

The operationality of what is called function expresses itself often in the
activities of composing and inverting.
17.3851. The Genetic View

17 .38. Variability as a Source of the Function

The function enters history explicitly rather late and then by means of new and
advanced mathematics Calculus but with a view to its much earlier virtual
presence, one may expect similar discrepancies in the individual developmental
rise of the function.
In mathematical instruction functions have moved downwards from Calculus,
via graphs and supported by equations to the primary school, even to its lower
grades, where they are concretised by imaginary machines and expressed and
symbolised in table and arrow language. The fact that this is feasible provides
a certain susceptibility for functions at a young age. What does this susceptibility
involve?
At a certain age children are susceptible to learning reading and writing; the
question of at what age they might be able to invent an alphabet is as meaningful
as the similar question we can ask about the wheel. We cannot check it because
it is imposed by the social environment, as is the case with language. Would they
be able to invent number selfreliantly? The numerals are imposed on them by
the environment and even if they refuse to accept them (see Bastiaan 4; 3*)
and pursue an autonomous development, words such as that many and how
many, blown in from the environment, are as many milestones along the
road to number, halfway or even nearer the goal. The early ability to operate
with magnitudes at least with lengths has to wait for many years to be
made explicit, if at all, in arithmetical instruction an irresponsible lag.
What about the function? For its constitution, we noticed a few components
are required:

variables,
mutual dependence,
independent and dependent variable,
* Weeding and Sowing, p. 281.

FUNCTIONS

529

where the variability may be

physical, social, mental
in order to be mathematised, or

mathematical,

and the same holds for the dependence, which can be

mentally experienced,
used,
provoked,
made conscious,
experienced as an object,
named as an object,
placed in larger contexts of dependences,

in particular in that of
composing and inverting.
Humans even animals are from the start of their development confronted
with variabilities and dependences: the mother reacts on crying, with the hand
in front one perceives a thing, while moving one pushes something away, one
collides with something, the place is changed, turning a switch causes light or
dark qualitative dependences, which are later made more precise by order
relations: the harder one cries, the stronger the sound, the harder one tries, the
higher one jumps, the more blocks the higher the tower and the louder the
noise if it is knocked down.

17.39. Qualitative Variability

The simplest case of mappings of sets is trivialities excluded that where both
sets have two elements, which for our convenience we will name 0 and 1.
Let A, B, C be three sets like this and let the functions from A to B be
indicated by
those from B to C by
and those from A to C by
such
that

530

CHAPTER 17

and similarly for

tion table

and

. Then there is a composi-

or if A, B, C are identified with each other

Physical models of this are systems with two states and appropriate couplings
to realise the functions.
Example: A: Two pushbuttons at a sun curtain with the signs and B:
curtain down and up ; : press
do nothing;
press or according to
what causes changes; press .
Example: A, B are two lamps with states on and off, a certain circuit.
The examples are artificial, it is not easy, indeed, to integrate all four functions in one system. The following pattern looks more natural:
Suppose with respect to some action (for instance stimulating somebody)
A:

performing or not (0 or 1)

and with respect to the effect

B:

taking place or not (0 or 1).

Then the meaning of the functions

is as follows:

not being able to cause,

being able to cause by doing,
being able to cause by not doing,
not being able to prevent.

FUNCTIONS

531

This scheme reflects certainly early experiences, but is it somehow experienced

as a system of functions? Asking this question means answering it in the negative.
Or it requires much empirical experience and a high level of experiencing functions in order to place this in the context of functions. The right place for
schemes like the preceding one is the end rather than the beginning of our
knowledge about functions.
I chose this pattern of functions as being the simplest possible. Yet logical
simplicity does not mean developmental primordiality.

17 .40. Richer Structure

Let us proceed to broader connections, structured in a richer way. As an example

we choose: fitting a set of objects into more or less predetermined holes; which
in a highly structured and completely determined way is realised by a jigsaw
puzzle. Children educated in our environment are quite early (from about four
years onwards) able to fit flat figures into characteristic holes and to complete
simple jigsaw puzzles; moreover after the first success they repeat it with the
same puzzle without any hesitation. The ever greater complication is conquered
by ever more exercise. What exercise means for acquiring this ability, is shown
when children who have grown up in greatly differing environments are con-

fronted with such tasks; they lag far behind and slowly, if at all, catch up.
The jigsaw puzzle pieces form a set with depending on colour and shape

characteristics a more or less clear connection structure; arranging the pieces

is indeed a mapping that assigns to each piece its correct place. The criterion
of the correct assignment is fitting; if it is done according to a model, it is

identification with the model.

The question is again: is this action experienced as a function? I would say:
what looks like an action, is in fact enforced; there is not much to be experienced
like the act of assignment that characterises the function.
But there is no reason to restrict ourselves to the rigid determinism of the
jigsaw puzzle. A less rigid situation is that of the circle, triangle, square shaped
holes and flat objects fitting them, not necessarily according to a one-to-one
assignment, but in a more general relation. The factual placing then produces
a one-to-one connection, albeit between part of the pieces and a part of the

holes a function arising within the relation of fitting. Is this experienced as

a function? It depends on the total context: are two functions of this kind
compared with each other, are they related with each other by permutations
(thus by composing with other functions), is there perhaps even a view on the
totality of these functions? One can imagine natural contexts to realise it and
didactical ones to promote it.

17 .41. Structured by Counting Sequence

The same can be asserted about functions arising from counting a set of objects,
or counting a required number of objects out of a given set of objects. The case

532

CHAPTER 17

looks like the preceding one, the freedom of choice may even be greater: the
allocation is replaced with the counting process, the holes with the numbers.
The act of assigning is entangled with that of structuring (for the benefit of
counting) the set of objects to be counted or counted out, which in this way is
functionally adapted to the counting sequence. Is this counting or counting
out experienced as a function? Counting and counting out are so much product
oriented, that is towards determining the number or producing a set of a given
number of objects, that the act is hardly experienced as such, unless this natural
and didactical context is stressed to make it conscious.
To a higher degree the function can be experienced in the equipartition
process (dealing cards, sharing sweets, and so on) which produces functions that
need not be any more one-to-one as in the preceding examples. But here too
much depends on the context as to whether the functional experience is made
possible or not.

17.42. Function and Cardinal Number

At an earlier opportunity* I have reported learning experiences with the drawer
principle and have asked the question which mental conditions must be fulfilled
in order that it works. It was the well-known question as to whether there exist
two (not bald) people on the world with the same number of hairs on their
heads, or rather a didactical sequence of this kind of question. The solution can
be mathematised by means of the function that assigns to each person the
number of his hairs; since the number of hairs on whichever head is certainly
smaller than that of people, this function cannot possibly be one-to-one. I linked
the successes of ten-year olds and the failures of eight-year olds to the ability

or non-ability to constitute certain sets and mappings. As soon as there is more

structure to profit from, this can be easier. On the same occasion** I reported
certain questions to test the mental possession of the cardinal number: A vase,
or its picture, decorated with a girdle of alternating suns and moons (only
partly visible) and the question which is more, suns or moons?. Or a more
sophisticated one about a wall-paper pattern, or a long open chain of beads,
alternating red and blue or its picture with the complication that the ends
can be equally and unequally coloured. Or the question of whether each layer
of a drawn wall contains the same number of bricks, or how a game should
be played so that each gets the same number of turns. Children in the first
grade can not only answer such questions by yes and no but also motivate
the answer meaningfully, that is, by arguments that reveal their experiences
with the relevant one-to-one mappings mappings that are inherent to the
given structure but not forced upon the user as happens with the jigsaw puzzle
and which differ from those occurring in counting and counting out by the
fact that they can be made conscious by the query why?.
* Weeding and Sowing, pp. 210214.
** P. 199.

FUNCTIONS

533

17 .43. Linguistic Structure as a Source of the Function

The possibility of experiencing functions cannot but be determined to a high
degree by the given structure of the underlying sets and of the functional relation
a structure that may be geometric or at least visual. Even linguistic structure
can be influential. The functional connection between units and tens by means
of ... ty is at least by some children experienced before they know written
numbers and after they become acquainted with written numbers before they
can identify this function with the function ten times .... For that matter
the language creates numerous not always regular enough opportunities

like this one: suffixes for comparatives and superlatives, tenses, plurals, action
and actor nouns, and so on none of them as convincing as the one we related
to 10 times ..., 100 times ..., and so on, though experienced earlier.
As early or even earlier can geometric or geometrically structured connections
be experienced as functions I will deal with them at a later stage, in order to

tap meanwhile a particularly rich source of experiencing functions.

17 .44. Geometric Mappings as a Source of the Function
There can be little doubt that geometrical mapping if picturally structured, in

particular faithful copies, but also mappings in scale, are experienced as functions
in Section 14.5 I cogently refuted Piagets postulate that this would require

the cartesian structuring of the plane. I have extensively dealt with the mental

constitution of all kind of geometric mappings in Section 12.6. All kinds of

that is, not only pointwise and geometrically heavily structured ones but also
objectwise ones as in the compository reproduction (Section 8.9) which
are geometrically less structured. It is an advanced stage, though with geometric

mappings more easily accessible than with others, that such mappings are not
only passively and actively experienced but also objectivated and named. A
rare case:
In his holiday resort, Bastiaan (7; 3) has made, along a piece of a brooklet, a model of
the North Sea coast of The Netherlands, Germany and Danmark, with islands and so on,
as far as known to him. He calls it a miniature [North Sea] a word he probably knows

from golf. His grandmother takes a picture of his work. He says: This becomes a double
miniature.

The mapping here is so strongly felt as a mental object that it expresses itself in
the composition of mappings, and the verbalisation of the activity of composing.

17 .45. The ... the ... as a Source of the Function

In Section 17 .38 and even earlier I placed the pattern the ... the ... into the
context of the function. Rightly, I think, since this relation in a way expresses
a property of functions, called
monotonicity:

534

CHAPTER 17

If A, B are supposed to be ordered sets, the function

monotonic increasing if

from A to B is called

and monotonic decreasing if

(A weaker form of monotonicity requires only

respectively, but this kind of sophistication does not matter now.)

Monotonicity is certainly experienced at an early age, but is this then a

monotonicity that presents itself as property of functions? One would be
inclined to say if monotonicity is a property exclusively of functions
then empirical and mental experience and use of monotonicity presupposes
that of the function. This, however, would be too rash a conclusion. In our
mathematical formulation of monotonicity A and B were ordered sets. One
may doubt whether in the monotonicity experienced in the pattern the ...
the ..., the underlying domains (the sets A and B) are really experienced

as ordered sets or even as sets.

In Section 3.12a I have discussed order structures and explained what is

required for the constitution of order in a set as a mental object. Comparatives

like those occurring in the pattern the ... the ... can be restricted to local
comparing; they need not indicate the mental presence of a set, let alone an
ordered set or even something like a scale. Doll growing bigger, chalk growing
smaller quoted in Section 17 .2, is a surprisingly early utterance, which seems
to reveal something about the constitution of variables in ordered sets and of
a monotonic dependence. Seems to reveal in fact it can be interpreted merely
locally, that is, as the gradual transfer of chalk on the drawing board.
Moreover one has to admit that with comparatives regarding spatial (or
perhaps temporal) dimensions in the pattern the ... the ... we are left in
a rather concrete sphere. Comparatives of intensities like
heavier,* hotter, sweeter, dearer, nicer
bear witness even less to the mental existence of sets or, let alone to the ordered
sets of the qualities they are supposed to order.
I would require for the ... the ... to bear witness to an experienced function that the order structures in both the-domains are more than only locally
experienced and that the same should hold for the relation connecting them.
Order and monotonicity are weak structures. In stronger structures the
function could possibly be more strongly experienced. This strengthening can
take place in quantities and magnitudes. Then the pattern the ... the ...
can acquire greater precision. This may happen in different ways:
* In the development heavy is initially indeed intensive, as at a very early stage are even
long, far, and so on.

FUNCTIONS

535

additive subtractive
multiplicative inverse.
The additivesubtractive variant could in a merely mathematical context easily
be neglected, though in the development it might be the original one. The
multiplicative kind gets its first expression in ratios and proportionality, not
only of magnitudes but also of their powers, positive and negative ones.
We pursue the exposition in both directions.

17 .46. The ... the ... AdditiveSubtractive

One of the earliest functional specifications of the pattern the ... the ...,
experienced by children, is the functional dependence

between ages in the course of time,

an additive dependence, which lends itself even for
functional composition and inversion.
x years older

has as its inverse

x years younger.
A is x years older than B and B is y years older than C

are composed to get

A is

years older than C.

Because of the different birthdays it is not a clean function, though

age is a (discontinuous) function of time,

which, however, is experienced as such not until a more advanced stage, as

will be the connection with the earlier mentioned functions.
Beside this additive specification there are perhaps not that early in the
development subtractive ones:
On the way from A to B,
In the course of time from A to B
the farther from A the nearer to B

made precise by the knowledge about spatial and temporal distances and their
additivity,
distance from B = distance AB distance from A,

536

CHAPTER 17

an equality which can more or less be experienced precisely. There are a lot of
realisations of the subtractive the the :

the string through a slit in the door,

the more inside the less outside,
pouring from one vessel in the other,
the more in the one the less in the other,
on the seesaw,
the higher the one the lower the other,
playing with marbles,
the more the one wins, the more the other loses.
Additive and subtractive specification of the pattern the the can
precede at least as far as magnitudes are concerned the additive and subtractive arithmetic. It is a didactical problem as to whether the experience of these
functions helps, or is killed by, arithmetical instruction something to be
recalled.

17 .47. The the in Proportion

Even less does the multiplicative specification of the ... the ... require the
support of arithmetical multiplication. Multiplicative is even not a good
term for what I actually mean. It is multiplicative if viewed from an already
constituted linear function; on the road towards this constitution the correct
term is ratio or proportion, which shows up in the title of the present
section.
I have dealt with ratio and proportion extensively in Chapter VI. This very
fact indicates that I reserve a particular place for ratio, not only in the development but also in didactics contrary to modern trends.
It is a commonplace that ratio and proportion have to do with the linear
function. In antiquity (and not long ago even at school) much was described

in the language of ratio, which is now done with linear functions if need
be with certain positive or negative powers attached. Nevertheless I have claimed
a special place for ratios and proportions as opposed to fractions and linear
functions, a kind of phenomenology, which must have descriptive consequences
in developmental and normative ones in didactical respect.
In the development, ratio comes prior to the linear function, perhaps even
to function at all, and this origin should leave didactical traces. I mention
relatively larger, smaller, sweeter, more expensive, and so on (Section 6.8),
which precede ratio as a mental object and only by mathematisation on a high
level can be fitted to the pattern of the linear function. I also mention the exposition and composition states (Section 6.4), which are proportionally compared
with each other with no intervention of linear functions. If the linear function

FUNCTIONS

537

is experienced as a connection between variable objects as is natural the

appeal to linear functions for norming and renorming (Sections 6.911) requires
a mental switch to measuring numbers instead of objects, which is only possible
in an advanced stage. The visualisations (Section 6.13) of position and composition states differ widely from the usual visualisations of the linear function.
Criteria for linearity (Section 6.16) will more often be derived from corresponding ones for preservation of ratio than conversely.
Closer connections between proportionality and linearity are found in the
structure states (Section 6.5), particularly in those of geometrical origin. If
internal length ratios in a piece of reality are compared with those in their
images, or internal length ratios in two images with each other, the dominatingly
experienced function is the more global geometric mapping; there is little need
for a separate experience of the length increase or reduction as a function,
unless it is instigated or in an advanced stage it is used for explicit calculations with scales. On geographical maps the centimeter on the paper is the
kilometer or so in reality with no intervention of the function that transforms
centimeters into kilometers. Experiencing such a function could even cause
damage to the operationality of the correspondence of distances. More strongly
than the action of comparing does that of constructive increasing or reducing
work in favour of experiencing functions.
17.48. The the , Proportionality and Function

In Section 17.44 we pointed out the conditions which the structure involved
by a pattern the the should fulfill in order to witness the experience
of the function. Even then this experience might remain qualitative as long as
the pattern the the is not sufficiently specified. Such specification
can tend towards the additivesubtractive or the multiplicative-inverse. Mutual
dependence of ages in the course of time is expressed by an additive constant
a function that becomes familiar to children at an early age. As stressed
before, this does not at all mean that the multiplicative interpretation of the
pattern the the must wait for the acquisition of the arithmetic multiplication or that it is necessarily supported by it. The natural development
seems rather to take place along proportionality understood as equality or
at least qualitative agreement of internal ratio. I recall the example of Bastiaan
(see Section 6.7) who reproduced the ratio altitude of two kinds of clouds by
distances of his hands from the ground the strong need for ratio models
leads to a first step towards the linear function. Recognising and reproducing
topographic data on scale, comparing well or less well arranged mixtures (red
and blue beads in a chain or some other pattern), distribution problems according
to given or wanted ratios require no multiplicative techniques, though at a
later stage they can be formulated and solved multiplicatively, by means of
linear functions. Here too the specification of the pattern the the
by means of the linear function is preceded by that of proportionality.

538

CHAPTER 17

17 .49. Aiming and Hitting as Source of the Function

The link between aiming and hitting is in the experience of the function perhaps
the earliest. Monotonicity plays an important part in this, somewhat refined by
more or less rough quantification. Hitting too high elicits aiming lower, too
far to the right elicits a correction towards the left, and conversely, and this

can be modulated by adverbs like much, a bit, a little bit, a bit more, a bit
less. Marksmanship in general is conditioned by a wonderful feedback system
in the brain but this does not exclude opportunities to consciously steer the aim
variable and observe the hit variable. As a matter of fact aiming and hitting is
not restricted to a spatial context. Times, colours, sound pitches and volumes
can be an aim, which is to be hit.
Even if there is no quantification, successful hitting supposes as a precondition: knowledge about the kind of monotonicity in the relation between aiming
and hitting. By this one gets enabled to steer the right way. The frequency of
qualitative errors in this field even among adults is rather high, even if
there are no misleading cues as with the behaviour before the mirror, looking
through an astronomic telescope, backward driving with a trailer. If feedback is
lacking or weak the reaction often goes the wrong way. A well-known example:
if a lever is to be balanced, young children react by charging the heavy side even
more, in particular if the extra charge does not change the unstable state
of the lever. If older ones perform better it is not a symptom of progressive
development as some psychologists claim but empirical experience. All the
same serious psychological research into the reasons for the wrong way reaction
would be useful.
Repeatedly* I have recommended the aim-hit-relation as a paradigm for
continuity: little causes yield little consequences. Continuity is not being discussed now; we are far away of any problematic regarding continuity. What
counts here, is the dependence itself between aiming and hitting, which is
used to score a bulls eye.
Aiming and hitting cuts in two ways as a source of experiencing the function:
globally and locally.

Globally means if space is concerned the mapping of the

cone of aim directions on the target,
thus a mapping in two dimensions, which can serve as a useful paradigm for

mappings. The local way is operationalised in

scoring a bulls eye,
* For instance Les niveaux de 1apprentissage des concepts de limite et de continuit. Atti

del Convegno Internazionale sul tema Storia, Pedagogia e Filosofia della Scienza; Pisa,
Bologna, Roma 712 Oct. 1971. Accad. dei Lincei 1973, Quaderno 184.

FUNCTIONS

539

that is choosing the aim variable appropriately to get a wanted value for the hit

variable. Here this activity has been placed in the context of the the .
Indeed what happens if one repeatedly aims at a goal is answering deviations in
a perceived sense by corrections in the opposite one.
Both global and local function experience are integrated in aiming at and
hitting of a moving target. Here we come again across a phenomenon we have
met before without paying explicitly attention to it: compensation.
17 .50. Compensation in the Pattern the the as a Source of the
Function
Most often one independent variable is concerned if a dependent variable is to
be controlled. Control of a dependent variable may mean keeping it constant:

hitting a fixed target, balancing a lever or some other contraption, and so on.
The variable z I want to stabilise depends on the variable x I can perceive and
the variable y I can directly control, and the dependences are monotonic at
least in the local domain I am concerned with. The direction of monotonicity
must be known in order control to be feasible and even somehow specified
quantitatively in order the influence to be more effective.
In order to stabilise z I have to compensate for the perceived or anticipated
influence of x on z by a change of y. This activity can be

preprogrammed or exercised

to such a high degree that the compensation is

practically perfect

or perfection can be more or less

approximated
by

aiming and hitting.

Yet the compensation can also

theoretically be steered
by experiencing certain lawlike connections such as

compensating for the length in the width in order to keep the area of

a rectangle constant,
compensation of lever variables to keep equilibrium.

In all cases the influence of x on y that stabilises the variable z can be experienced
as a function a function where the composition via z of
may or may not be an explicit component of the experience.

540

CHAPTER 17

17 .50a. Inter- and Extrapolation as Sources of the Function

Meteorological data read thrice a day,
the precipitation of the last 24 hours registered at 8 AM,
the morning and evening temperature of a patient,
the number of traffic deaths each month in The Netherlands,
the gold treasure of the National Bank at the end of each month,
the cost of living at the first of each month,
the electricity consumption of the Netherlands in a calendar year,
the (estimated) world population in subsequent years,
grades on successive reports of a pupil,
the number of storks breeding in The Netherlands each summer
these are primarily discrete data, often presented by line graphs, and nominated
to be functions of time. They are rightly so, if intermediate values can be ascribed
to them though they have neither been measured nor estimated as such;
hypothetically some of these functions can be extended beyond the limits
of measuring and estimating, to the past or future. Between the measuring points
it is done by linear or somehow smooth interpolation; outside the measuring
interval by continuing an apparent or patent trend.
In some of these cases intermediate states are out of the question. The
births, traffic deaths, the electricity consumption have already been accumulated
monthly; the grades of a pupil express his performance over a period; the storks
leave in the winter. This does not prevent one from constructing functions.
The weight of a more or less coarse matter as a function of the quantity, its
price as a function of the weight are primarily functions of a discrete variable,
continuously interpolated. Sometimes the meshes of the independent variable
are indeed so narrow that interpolation is natural and admissible, even if the
matter consists of indivisible units. This continuisation of the discrete extends
even to mathematics, for instance to the number of primes below
called
if it is estimated by

17 .51. Piaget
The function has also been one of Piagets themes. The work where this is
exposed* is utterly disappointing. A quarter of the booklet, by Piaget & Grize,
is a kind of epistemology: generating functions by what is alternately called
structuring functions, constitutive functions, coordinators, with in

particular by Grize the display of mathematical terminology that is out of

order, and formulae that look cabalistical rather than mathematical. If rather
than five times one would try and read this part fifty times, one would perhaps
* J. Piaget, J.-B. Grize, Alina Szeminska, and Vinh Bang, Epistmologie et psychologie de
la fonction (Etudes dpistmologie gntique XXIII), Paris, 1968. Translation: Epistemology
and Psychology of Functions, Dordrecht, 1977.

FUNCTIONS

541

find out what the authors intended to convey with this theory, but it is a comfort
to notice that the collaborators, responsible for the experimental part (three

quarters of the book) did not understand it either, or at least did not show
that they cared to. The terminology of the epistemological part is only used,
but hardly elucidated, in Chapter 3 and then in research where the relation with

functions is farfetched if existent at all.

The relation with functions is more often lacking. In some cases a phenomenological appearance of the function is belaboured in a manner that

everything that has to do with functions is extracted: the subject is asked to

perform artificial, aimless and useless, operations, the sediment as it were of
what was left after the extraction. (For instance Chapter 1 on permutations.)
Another objection can fairly be illustrated by the experiments of Chapter
2. They are what in Section 17 .40 we called the jigsaw puzzle type, albeit with
a certain flexibility. What is tested, however, is whether children can manage
the puzzle rather than whether the activity of the more or less enforced assignment is experienced as a function. The same objection can be raised more or
less as criticism of the other test series. For instance, in Chapter 5 the subtractive connection between the two pieces in which a string is divided by a
roll over which it is conducted; in Chapter 6 the connection between the two
parts in which a quantity is divided; in Chapter 7 the connection between the
levels of the same quantity of liquid in glasses of different shape; in Chapter
8 the two sides of a rectangle with a given circumference; in Chapters 912
with proportions. It is always the same: about something that is viewed by
the experimenter as a function, asking the subject to predict for a given value
of one variable the corresponding one of the other. What is not done or not
in a sufficiently structured way is: finding out whether the connection is
experienced by the subject as a function. Of course, somebody who experiences
the connection as a function, has a better chance to score high on such tests.
But conversely, a wrong answer can be caused by a function experience, which
happens to be wrong. Intentional experiments could show which is the one
and which the other. Unfortunately the experiments are often artificially
complicated to such a degree that the experimenters themselves were unable to
describe them decently. Protocols of interviews are often lacking or restricted
to examples, which do not allow one to check how the experimenters had
instructed the subjects who were expected to perform the complicated tasks.
The investigation on proportions is also too little tuned on functions; the
tasks are needlessly complex.
The lack of phenomenology there is not compensated by what the authors
call epistemology.

17.5257. The Didactics of the Function Experience

17.52. Experiencing

In this phenomenology the present chapter is the first where I used the term

542

CHAPTER 17

experiencing. I should say I am not too happy with this term. In German I

would say erleben, which is just what I mean. In English I tried sometimes
lived through or felt instead of experienced, but this does not work
too well except in rare cases. In French I would choose prouv though the
undertone of suffered can be misleading. The advantage in German is that
there are two words erleben und erfahren, which in English are covered by
one. What is the difference? A shade? Or isnt it more? Translating is a good
opportunity to analyse what one knows by intuition no, by use. German
Aha-Erlebnis is translated by aha-experience, there is no Aha-Erfahrung; erleben
is more emotional than erfahren. Older people are wiser by Erfahrung and
sadder by Erlebnis is it this way? I think I had better stay close to mathematics. Time, space, infinity if erlebt can be empty, if erfahren they are likely
to be full. It is easier to tell people about erfahren than erleben; erfahren is
easier to be verbalised than erleben.

I would like to understand experiencing in this less pregnant way though I

do not know how to do it. One remembers that from the start onwards I pushed
concept formation far away. Constitution of mental objects was the first that
matters. Small cardinals are early constituted as mental objects, yet one number
is not the other (while any point can be any other); the constitution of big
natural numbers happens within that of the natural number, which moveover
can be differentiated according to different aspects. The point, the straight line,
the circle, the sphere, however, are mutually constituted in all their generality
without any previous constitution of specialisations (except that in the case of
the straight line horizontal and vertical ones can initially play a special part).
In the case of magnitudes, real numbers, ratio, proportion I did not feel the
need for an explicitly indicated period of experiencing preceding that of constituting the mental object, though it would not have been farfetched to do
so.
The mental constitution of the function, in an advanced stage, can hardly
be distinguished from concept formation. I stressed earlier that so much and
so many things are comprised by the function concept that all this hardly
corresponds to one mental object. Moreover beside mental objects of vastly
diverging origin and character, they are linguistic terminological structures,
lumped together conceptually by means of an explicit definition. Investigations up to the highest grades of secondary education* have pointed out that
this conceptual definition does not work, that is, that in instructional practice
it has neither replaced nor overarched the various mental objects.
Various aspects of the functions can grow to become mental objects, though
in an early stage and not without directed didactical support. As a mental object
the function is more complex than number, than geometrical objects, even
than ratio, though one aspect may be more easily accessible than the other.
* See, e.g., R. Barra and J. J. Pensec, Un dfaut de notre enseignement propos de fonctions, Bulletin APMEP 57 (1978), pp. 357362.

FUNCTIONS

543

Therefore a broader margin should be left for experiencing, allied to using and
provoking, of functions, before they are made conscious, accepted as an object
and named. This of course does not exclude didactical attempts at level raising.
This leads us to the problem of how to justify having and letting functions
be experienced. There is not the slightest need for functions in order to base
the cardinal number on them, or to introduce little finite groups. One could,

however, suppose and even maintain that the function, made explicit or not,
can fulfill an important strategic task. Repeatedly I pointed out that mathematicians avail themselves of bundles of strategies, which for a long time could
have been operational before somebody hit on the idea to condense them
in a comprehensive theory. Even at the highest levels of mathematical education
it can be didactically recommendable in this matter to pursue somehow the
historical course, that is, maintaining and teaching as incidental strategies things
that will later be overarched by a theory.

17.53. Geometrical Mappings

According to the title this could be a repetition of Chapter 12. But there geometrical mappings ranged in the context of geometry whereas here they are
meant in that of functions. Nevertheless much could be repeated. I will try
to avoid this.
It is in the visual realisation that the function is the easiest accessible. How
strongly can it be experienced in this approach? Reproduction of a piece of
(genuine or imagined) visual reality is often compository in character: objects
from the piece of reality are reproduced by partial pictures, united in a composition to one picture of that reality, and then according to structure rules that
are arbitrary or difficult to find out a structural relation, which can perhaps

be interpreted but hardly experienced as a function.

Perspective reproduction on paper of a three-dimensional piece of reality is,
at least in its initial stage, as we pointed out earlier, rather a kind of composition; perspectivity as mapping by means of vision lines is a rather advanced
stage.
If we speak of geometrical mappings, we mean a pointwise rather than an
objectwise assignment, which can most strongly be experienced in the relation
between an objective situation and a picture or model of it or between two
pictures or models of the same objective situation. This then is at least in

principle the pairwise mapping which can be

given in order to be stated, or
asked to be made or completed.

A special case is

given or asked for symmetries

in a picture or model.

544

CHAPTER 17

It is not required that such a mapping be a similarity; it can be a subject of

search or discussion to find out the rules governing the mutual ratios (if there are

any). What matters is

experiencing the act of assigning as a pointwise one
and moreover

experiencing the composition and inversion of mappings,

as soon as more pictures of the same objective situation are at stake. In a natural
way this happens if

the same drawing is copied from the blackboard by a number of pupils,

or if

pupils are given copies of the same demonstration picture.

The composition of the mappings
copy of pupil A
copy of pupil B

copy of pupil B
copy of pupil C

and their inversions can here be experienced. In the case of similarities

scale factors and their behaviour under composition and inversion

can be stated or observed. But even with mappings of other sorts

characteristics and their behaviour under composition and inversion

are worth being taken into account.
I considered here pictures and models of an objective situation because they
lend themselves excellently to suggesting pointwise assignment. Geometrical

figures as objects have been dealt with in Chapter 13 with this intention. The
assignment there is in fact restricted to a finite number of points (for instance
the vertices of a tetrahedron) in order to be extended according to geometric
rules a procedure at a higher level than is viewed here. Even if the geometric
figure is strongly prestructured, for instance by means of a lattice or like the
number line, the experience of the pointwise assignment is more demanding
than in the case of naturally structured pictures and models of an objective
situation. In fact it requires, as pointed out above, a knowledge of geometric
relations to perform exact geometric constructions or a kind of aimhit procedure by which supposed assignments are systematically corrected.
17 .54. Classifying by Parametrisation

Classifying is a popular subject in developmental psychology. Two trends of

research can be distinguished: research whether subjects can perform a certain
classification or a certain kind of classifications for instance according to

FUNCTIONS

545

categories like living and non-living, solidliquidgaseous, size, weight, age

or investigating whether subjects master classifying as a formal activity, that is,
using formal criteria which can be imposed on classifications. Both kind of
abilities get sometimes entangled with each other in the course of such experiments, with the consequence that one does not know for sure what is actually
being tested.
Functions have to do with classifications because they can be used in classifying. A function from A to B classifies in A by partitioning A into classes of
originals of the various elements of B, where moreover a possible structure in
B can be transferred to the system of classes. For instance the function that
assigns to everybody ones age, divides the population into age classes, which
can be ordered according to the ages they represent.
Classifying as understood and investigated by psychologists is a more primitive
and rough conception than that which is performed as such in mathematics.
This is particularly true with respect to the operations acting on and between
classes and the connection with functions (as far as ever considered by psychologists). As a matter of fact we will also view the connection between
classifications and functions in a special direction. Rather than in a posteriori
classification by means of an a priori given function we are interested in the
construction of a function for the benefit of, and as a tool for, classification.
Classifying can be done by comprehension, that is, by arbitrarily combining
into classes or by discrimination, that is by distinguishing characteristics that
are interesting in themselves. The second manner is by far the most important.
A characteristic of classification, as intended, can be colour. The function
generating this classification is colour of, that is, a function that to everything
assigns its colour. In a didactical context, aiming at experiencing functions,
it is a wrong example. As far as the colour can or must serve as a distinguishing
characteristic, one can be satisfied with this use rather than dragging in a function, which is a needless duplication of the distinguishing characteristic.
The title of the present section sounds a bit strange, perhaps by its brevity,
but even with more words I could not have made the intention clearer. So I will
start the exposition with an example, which is rich enough to be paradigmatic,
and brief enough to be clear.
I am considering a rhombus, realised by the combination of four equally long
rods which are hinging in joints (Figure 224). The rhombus can take different
shapes: a square, more or less stretched or pressed, up to coincidence of the

546

CHAPTER 17

sides. It is a whole system of geometric figures though all of them rhombuses

produced by this instrument. The term classification in the title aims at
the need for mutual distinction of all these rhombuses. Distinguishing can be
done by characteristics such as mentioned before: square, more or less stretched.
Yet more or less can be made more precise. The term parametrisation
in the title means the choice of a parameter that can be used to distinguish
various rhombuses. If such a parameter is found, then the
shape of the rhombus as a function of the parameter
is a natural experience of a function.

In fact such a parameter is itself a function, to wit of the rhombic shape it

describes. The choice is among a number of possible parameters,

the angle of two rods,

the height of the rhombus,
the width of the rhombus,
the distance between opposite vertices (diagonal),
the area,
a system of variables that can also be viewed in their mutual dependence
qualitatively or more or less specified.

Not only
shape as function of a parameter
but also

parameters as functions of other parameters

can be experienced, as well as the mutual relatedness by means of

composition and inversion.

As a matter of fact the change of perspective
from shape to a required parameter
towards

shape determined by a parameter value

is worth being experienced.

I could have started with a simpler example, which, however, would have
lacked the rich didactical-phenomenological context. For instance, with
Cuisinaire rods colour and length are the primary criteria of classification. I
already pointed out that colour rather irrelevant in the present case can
hardly be an incentive for function experience. Length, measured with the
shortest unit, could be different, were it not that the system of rods is too soon
identified with the parameter: the rod of length 3 is 3. This quasi identity is

FUNCTIONS

547

an impediment in the way of experiencing a non-trivial function. The number

line, right or curved, is not much different: it is just the intention that the points

are identified with the numbers at the points, which trivialises the assignment
required for the function. Number lines on different scales is another case,
which, however, belongs rather to the geometrical mappings dealt with in
Section 17 .53.
Squares, cubes, circles, spheres classified up to congruence is a better example,
though less rich than the rhombuses we dealt with formerly. Parameters, which
introduce themselves, are diameter, side or diagonal, and perhaps area and
volume. They are appropriate as examples for the function experience, but
to begin with, a richer example such as that proposed, looks more appropriate.
A rich example is also that of rectangles classified up to similarity. Parameters
that present themselves are
the angle of the diagonals,
the angle between diagonal and side,

the ratio of sides,

the ratio of side and diagonal,
where the occurrence of ratio in the last example requires broader previous
knowledge than was supposed in the former examples.
A similar, perhaps even richer, example is triangles classified up to congruence
or similarity.

A more elementary case is rectangles with a given circumference, classified

up to congruence, with classifying parameters
length of a side,
length of a diagonal,
angle of side and diagonal,
angle of the diagonals,
area.
A sophistication we did not pay attention to so far: are the considered
rectangles supposed to possess well-distinguished (numbered) vertices, or are the
lying and standing rectangle, if congruent, to be consiedered as equivalent?
According to this decision certain parameters shall be more precisely specified;
certain function can become two-to-one rather than one-to-one.
The preceding examples do not exhaust the possibilities. Rather than with
respect to congruence or similarity one can classify up to translations and, for
instance, choose a

system of circles with the centre on a fixed line and the radius constant
or depending on the centre,
or the

system of equilateral triangles inscribed to a fixed triangle

548

CHAPTER 17

or the

system of squares inscribed to a fixed square,

systems which can nicely be realised by a kinematic contraption (Figure 225):
A closed string as the periphery of the fixed figure with three or four nooses,
respectively, attached at points that divide the circumference of the string in
equal parts, and through the nooses an elastic string.

There are a great many such mechanisms. To terminate this section with an
example of a string construction we mention that of the ellipse: the point of
the ellipse as a function of one of the radii vectores or of one of the related
angles or all of them as functions of each other; and the shape of the ellipse
as a function of focal distance and long axis.

17.55. Time as the Independent Variable

In Sections 17.5354 geometry was dominating. There was no need for visualis-

ing since the data and their mutual relation themselves were visual. An effective
visualisation of the function experience is function graphs. Of course a number
of functions we came across in Section 17 .54 can also graphically be visualised
and for a better understanding this can be useful in some cases. Various approaches towards graphical visualisation of functions will be reviewed later on.
One of them, and perhaps the didactically most effective, is the one where the
horizontal axis is interpreted as a time axis (or time log).
The development of time experience is perhaps even more mysterious then
that of space experience. Though little children have trouble with the distinction
between yesterday and tomorrow, one may ask whether this fact really
proves a disability in distinguishing past and future, or is it rather a Linguistic
disability comparable to the confusion of colour names while colours as such
and the difference of colours is mastered at a quite early age. Whatever the

FUNCTIONS

549

reason may be, in the development of time experience, periodicity of the happenings in time must play an important role, the change of light and dark,
waking and sleeping, school times and Sundays celebrated in the church or
not , vacations, birthdays, Santa Claus. By the events in time, and certainly
by the periodic events, time is structured. The days of the week, the hours
of the day determine an order structure, the unwinding of a primordially
cyclic structure, and even more: a mental rough measurement of time.
Length and time measurement differ greatly. Lengths can be compared by
direct transfer, with an increasing precision. In order to compare and measure
time intervals as reliably one needs more or less complicated instruments. Or
arent we in this respect spoiled by the possession of such instruments?
It seems that in his experiments on the inclined plane Galileo measured time intervals, by
singing a melody. So it is perhaps not that difficult to keep a fixed rhythm to use it for
time measurement as a matter of fact the duration of a pater noster or ave Maria was
from olden times a not unusual time measure.
At a kind of bowling (in rails) where the number of tumbling skittles depends on the
initial ball velocity only, I succeeded after a few observations in predicting exactly the result
which shows that velocities can be observed with a reasonable precision.

Time is a magnitude as is length. But, other than length, time is not primordially a magnitude. Lengths can be compared and displaced within the plane
and space. Time is first of all experienced, that is, the thing experienced is the
distinction between moments and the order structure. Duration is secondary.
Time as a linearly ordered system precedes time measurement and time
interval. With length it is just the other way round. Straight lines can be structured or unstructured. One line is not the other, but to acquire the mental
object of straight line, identifications are required. Time as such is unique
and structured by living in it. As a matter of fact the microstructure of time is
more easily accessible than its macrostructure. Globally time is structured by
history. Elswhere* I pointed out that for 89-year olds the past may still be
unstructured though accessible to structuring.
Structuring of time the order structure as well as the metric one can
intentionally be supported by visualising: the spatial translation of time into
the time axis. It is meaningful to use this model of time from the beginning
onwards on various scales with the structural units of
century, year, month, week, day, hour, minute, second,

to accentuate the structuring by

history, life history, seasons events, school events, day division, clock
division,
* Weeding and Sowing, p. 290. Zentralblatt Didaktik d.Mathematik 10 (1978), No. 2,
pp. 7677.

550

CHAPTER 17

and illustrate the last example by

unwinding the dial upon the line,

a geometrically active illustration.
Not surprisingly does the introduction of the time axis historically coincide
with that of function graphs (Oresmus in the 14th century), that is, graphs of
simple functions. Even now we feel obliged to introduce time as a function
variable by certain contemplations on time as a magnitude and on the time

axis.
Didactically it can be useful to have dependence on time experienced as a
function, and the most natural objects for this are

growth processes.

First of all a somewhat improper but nevertheless effective example, the growth
of age:

age of a person as a function of calendar time

or

age of one person as a function of that of another,

in both cases graphically represented (if scales are equal) by
a line, inclined 45,

provided age is interpreted as a continuous variable. If age is rounded in the

usual way, that is downwards, then the
age of a person as a function of calendar time
is a step function.
Age graphs are a means to answer questions like

when was A double the age of B?

when will A be double the age of B?
at a glance.
The condition A double the age of B can also be visualised by a function graph; problems
that involve this condition can be formulated as a pair of equations in two variables and

solved by intersecting graphs. The function A double the age of B, however, is of an

entirely other character than that which connects the ages of A and B. The first is as it were
of logical, the second of physical origin. I realised this important fact when I noticed with
higher grade students the confusion caused by the difficulty of grasping and appreciating
this difference: they tended to reject the function A double the age of B because it is not

preserved in the course of time. It looks a strange attitude which, however, would be worth
investigating more closely.

FUNCTIONS

551

Age as a function of time is an improper growth process. Growth processes

rather regard such variables as

length, weight, product, number, energy consumption

of
persons, animals, nations, population, industry
as a function of time. I used the word growth which, though not in the
vernacular but by convention, also means
loss as negative growth,

a new example of the habit, since the adoptation of plus and minus, to
mathematise polarities by positivenegative (see Section 14.9). A certain object
becomes in time more or less according to some numerical property because
it gains or looses something.
Another kind is
air temperature, air pressure, prices

as a function of time.
Again of another character, though no growth processes:
path, velocity, acceleration
as functions of time.
The algorithmic operational force of most of these function examples, as
understood in Section 17.37, is not great. Most of them derive their practical
and didactical value from the visualisation by graphs, which demonstrate not
only

growth and loss,

but also
the strength of growth and loss.

Experiencing growth functions certainly includes the recognition of such features. It is quite another thing to experience functions and classes of functions
by

laws of growth.
Certainly experiencing and even verbalising of a
linear law of growth
is possible early. An
exponential law of growth

552

CHAPTER 17

however strongly experienced by trained persons requires much more.

One would guess that handheld calculators should make an early, that is prealgorithmic, experience of the exponential law of growth possible, but I do not
know of any experiments or observations that would confirm this hypothesis.
Special didactic attention what regards experiencing functions, should be
paid to
timedistance graphs.

Such a graph, which indeed describes a motion, easily conveys the misleading
suggestion of picturing the path covered by the moving object.
Pupils who in grades 5 to 6 are for the first time taught graphs and well the time
distance type, for instance graphs of the course of trains along a certain traject, or of a
skating competition, are inclined to interpret intersections of this kind of graphs as intersections of paths.

It seems didactically commendable not to introduce timedistance graphs, and

for this reason also distance as a function of time, unless the pupil has beoome

familiar with graphs of functions in other contexts.

On the other hand richer opportunities are offered by timedistance graphs
and functions than by any other kind, thanks to the special meaning of
intersections of such graphs in one picture
as well as to the particular character of related measures of growth, such as
velocity and acceleration,
witnessed by proper names, which can provoke the experiencing of

new functions and graphs.

It is one of the approaches to

derivatives
which may be possible and useful at an early stage.

Closely connected to the timedistance graphs are what I would call

timepath graphs.

A closed circuit is as it were cut open and mapped on the vertical axis, which
causes unavoidable discontinuities deriving not from the function but from the
cut. An example: the graph of the motion of a clock-hand on the dial (Figure
226) or the (apparent) motion of the Sun or the Moon around the Earth, or one
of many laps of the cycling track. This kind of graph becomes operational in
problems of overtaking and meeting such as:
When do hours and minutes hands cover each other?
Why does an eastbound or westbound globe traveller win or lose a day,
respectively (that is, sees one sunrise less or more)?

FUNCTIONS

553

Distance is the most striking and first experienced motion parameter, but it is
not the only one. We already mentioned velocity. As regards closed circuits it is

elucidating to link the image of an irregular circuit and the velocity graph,
where minima and maxima are determined by bends (Figure 227).*

Another important motion parameter as a function of time is

projection on an axis.
The uniform motion of a point on a circular orbit, projected on the vertical

axis (Figure 228) is a harmonic vibration, or in other words, a trigonometric

* See p. 572 footnote (Janvier).

554

CHAPTER 17

function. Sine and cosine functions can be experienced early in this graphical
way in order to be applied non-trivially even derivatives can be read from the

figure, and this in a stage of still unalgorithmised experience of the function.

Earlier I demonstrated hesitations with regard to exponential functions.
Whereas trigonometric functions can be experienced in a geometric context,

the natural context for exponential functions would be the law of exponential
growth a context which in spite of a good deal of exploration is still not
easily accessible.

17 .56. Height as a Function

If I say height I of course include depth as negative height again an
example of the habit, since the adoptation of plus and minus, to mathematise polarities by positivenegative (see Section 4.19). Height and depth as
understood here are forcefully concretised by the surface of a mountain or

a sea-floor above a part of the terrestrial surface or below the sea level, respectively. Originally I chose functions of two variables as the title of this section.

It would have been a wrong choice because a pair of variables need not be the
means to experience height as a function of the underlying relevant variable.

The ground plane needs no previous coordinatisation that splits the one (point)
variable into two (coordinate) variables.
A visualisation of this kind of function which tries to imitate graphs, is the
concrete model, an unmanageable device. A much more efficient method is

contour lines
that is, lines of equal height a system from which one can read

peaks, valleys, passes, ascent along paths, lines of steepest descent.

By change of perspective the contour model can be used to have functions
other than height experienced, in particular on the terrestrial surface (Figure
229),
temperature, air pressure, density of population

FUNCTIONS

555

as functions of the place, with the corresponding

isolines,

though the terrestrial surface as function domain can also be replaced with a
plane as such,
unstructured or structured (by a pole or as product of two lines).

17.57. Elementary Operations as Functions

By elementary operations I mean here adding, subtracting, multiplying, and

dividing. They are binary operations, thus involving three variables: an improper
one that is fixed in order to characterise the operator, while the two others
are considered as independent and dependent, thus the kind of functions which
are algebraically denoted by

and perhaps more complicated ones, composed from these.

There are many contexts in which to introduce

function tables
of these functions. First of all I mention a not so trivial one under its IOWO
didactics term,

the double decker,

that is: a fixed number of persons is divided over the upper and lower deck
which yields the function table
above
below

of the function
thought of:

0
40

1
39

2
38

...
...

A great many variables as convincing as this can be

pages of a book: read and still left,

distribution of a fixed number of marbles between two persons,

distance covered and still left,

and so on.
There are many more opportunities to draw up function tables I will leave
it at that one example, which otherwise would perhaps escape the attention.

556

CHAPTER 17

A modern linguistic device, which is intended to stress the functional character

of elementary operations is

They can also be used to formulate the traditional dot problems more forcefully: in

two values are being given and the third is asked. On the other hand the use of
the arrow creates an opportunity to relieve the equality sign from improper
use and to stress its static character.
A more concrete, that is, geometric feature is added to the arrows language
by placing
arrows on the number line.
More involved operation patterns can be built with arrows (Figure 230).

A recent though rather verbal concretisation of the function character of

operations is

machines
with an input and output and labels characterising them, which moreover can
be composed in various ways such as (Figure 231)

or even more involved, for instance with several inputs and outputs. This
linguistic device can also replace the traditional dot problems, if again two
values are being given and the third is asked. A further schematisation leads to

flow diagrams.

FUNCTIONS

557

The didactician will not stick to one device. He will rather have the same
function experienced

in a story, as a table, in arrow language, on the number line, as machine,

as flow diagram
where stress will be laid on the
isomorphism of the various function experiences.

17 .5863. The Algebraisation of the Function

17 .58. Approaches to Algebra

Algebraised functions were already discussed in Chapters 15 and 16 (in particular
Sections 15.912, 16.21). In the past it was a popular controversy which
approach to algebra is to be preferred

via identities, that is, where letters indicate indeterminates

or

via equations, that is, where letters indicate unknowns.

At present one would add a third,
via functions, that is, where letters indicate variables.

The didactic goal in the latter case is: letters as variables, and algebraic expressions in the broadest sense as functions of one variable or several variables,
corresponding to the letters. It is quite natural to start a didactical sequence with
the expedients of function experience I exposed in Section 17 .57, thus with a

translation of these expedients of function experience into algebra.

17 .59. From the Elementary Operations to Functions

I will illustrate this with a few examples:
(a)

(b)
(c)

(d)
(e)

Think of a number, add 5, multiply by 2, subtract 5. What is the

result? ... Then you had ... in mind.

Think of a number, multiply by 2, add 10, divide by 2. What is the
result? ... Then you had ... in mind.
Think of a number, add 5, square it, subtract 20. What is the result?
... Then you had ... in mind.
Think of a number, multiply by 2, add 10, divide by 2, subtract the
number you first thought of. The result is 5, isnt it?
Think of a number, add 5, square it, subtract 25, divide by the
number you first thought of. What is the result? . . . Then you had
... in mind.

558

CHAPTER 17

The riddles (a), (b), (c) are of the type that directly springs from the use
of machines. In (d) and (e) the number in mind is fed once more into the
process. This means that in (a), (b), (c) the original number can be retrieved by
inverting the process as a whole:

In (d) and (e) this does not work. Retrieving the original number seems to lead
into a vicious circle, such as a millenia old prototype of this kind of queries:

A stone weighs one pound more than half its weight, what is its weight?
It seems that in order to calculate the weight of the stone one must first know
half its weight, which would be true, indeed, in
A stone weighs the double of half its weight. What is its weight?
What is the difference between these problems? More profound insight leads
to an answer.
Solving by starting at the end, as with type (a), (b), (c) is to my view the
indispensable predecessor for learning to prove algebraic identities and solving
equations: the new aspect of the circular looking problems becomes thus clearer
and the search for new methods gets more strongly motivated.
The search can first mean trying:

The next step would be introducing a variable at the place of the number in
mind:

FUNCTIONS

559

Simplifications underway would lead to a final

which in the cases (a), (b), (c), (e) would be compared with the result in order
to figure out
in the difficult case (c) the way back would be shown by the
way from the number in mind to the result.

17 .6062. The Algebraisation of the Inversion and Composition of Functions

17 .60. The sequence of algebraisation of the function has been started with the
function such as experienced. The algebraisation of the function, however,
includes as well

the algebraisation of inverting and composing

of functions, which in the didactical sequence should finally get entirely
algorithmised. This is a well-known thorny aspect of the algebraisation of the
function. It is quite natural to require that the algebraisation of inverting and
composing continues the

experience of inverting and composing

of function, in machine or arrow language, such as sketched in Section 17.57
and elaborated in Section 17 .59.
The final notation* of the kind of functions we came across is

* The reader will notice that in the sequel I will not mention the
to supersede the rather deficient arrow notation.

I proposed

560

CHAPTER 17

or

The difference, compared with the former arrow language, is not only that the
arrows have been straightened out and the specification of the operation above
the arrows has been dropped. In the definitive notation the original experience
of inversion and composition is lost. The gulf between these two notations is
broad and not easily bridged, and certainly not within the frame of the usual
didactics today, which after the rise of New Math has got addicted to the one
symbol for the variable (or and if there are two of them).
Even well-trained pupils are exposed to confusion by the multiple appearance
of the letter if functions like the above f, g, h, are to be inverted and
composed. This is a serious handicap to the algebraisation and finally algorithmisation of these activities.
In the notation

one can clearly recognise

as respective inverses. Moreover composition of operations in this or that order is

clear in

The need for brackets might be experienced in this context and their use might
be exercised. Beware of the didactical problems that arise for inverting and
composing functions f, g, h, given in their definitive notations!

17 .61. To face these difficulties one has to break the monopoly of the variable
The transition from

to

should be accompanied by that to

FUNCTIONS

561

and so on and to

and so on, and it should be understood that all of these are notations for the
the same function. Not until this has been achieved, can the definitive notation
be tried and then not only in the form

but as well in the form

and so on, and in the form

and so on. Not until this has been attained, can one count on algebraic inversion
and composition of functions in a way which in the long run would be accessible
for algorithmisation. In order to compose

in this order, one makes implicitly present variables explicit by putting

and similarly in order to invert

one writes

for the same function, in order to solve

for x,

and to get the function

which finally can also be written

562

CHAPTER 17

I think it is the way adult mathematicians do it. It is unbelievable that by

restricting the independent variable to the one symbol x, one creates didactical problems for the learner which an adult mathematician would never risk
incurring.
17 .62. Earlier we pointed out that in particular in applications from physics
certain letters are stuck to certain magnitudes. This should not conflict with
the need for free exchange between letters that indicate variables. This need is
felt not only in

composing and inverting

functions but also in
interpreting functions as inverted or composed ones,
whether such functions are
specified or general.

I mean expressions like

log sin 2x

which must be interpreted as functions, in order to be tabulated or graphically

represented, differentiated or integrated. It is the classical procedure to split
it into

It requires routine, for instance in differentiations, to perform this splitting

purely mentally, that is, without introducing new variables. Devices which are

familiar in Calculus the introduction of new variables should be not only

allowed but even promoted at lower levels.
17 .63. Once More: the Function Experience

Composing functions is in a primitive way experienced by coupling machines

in a row. A more sophisticated form is the arrow language. It is algebraised in
the formal substitution. But even at the level of algebraisation, there are aspects
of the functions which remain a matter of experiencing.
I again start with two examples:
Given the graph of
what that of

what does that of

look like,

It is of course the intention to have the second and third function to be

interpreted as composed from the sine function and
and
respectively.

FUNCTIONS

563

More generally: Given the graph of f (Figure 232), what does that of fg look
like, where g is a nice function, linear or at least monotonic.

What happens with the graph of f if the x in f(x) is replaced with

with ax? If f has a maximum or a minimum or takes the value c at
do
and f(ax) do the same?
Answer: at
and
respectively.
In composing with

or
where

the graph of f is shifted horizontally by a,

shrunk horizontally by the factor a,
respectively.

More general: if g is monotonic, the graph of fg arises from that of f by

subjecting the horizontal axis to the transformation
A wave movement in time to the right with the profile f is described by
a uniform swelling by

in other words,
f(x t) as a function of time t

represents a wave progressing in the positive sense,

f(tx) as a function of time t

a decompression.
To this kind of experiencing of the composition of functions more didactical

attention should be paid.

17 .6465. Composing Implicitly Given Functions
17.64. After this side-leap let us return to the theme of algebraisation of the
function. So far our view was directed to
explicitly given functions,

564

CHAPTER 17

that is, algebraic expressions (in the broadest sense) with a variable,

interpreted as a function of that variable.

The algebraisation extends farther. As early as inverting a function f one requires,
at least in principle,
solving

with respect to x.

More generally one is confronted with this problem if an algebraic expression

involving more variables is given,

one of which is considered (implicitly) as a function of the other,
for instance in

the y as a function of x. This leads to the interpretation problems we sketched

in the historical context. In general

yields a relation between x and y: the solution set

which can possibly be split into function graphs, in the case

those of

If univalency of functions is to be maintained, the implicit definition

is to be refined by domain restricting conditions

in the case of
for instance by

Then the solution set appears to be split into two implicitly defined function
graphs, to wit

FUNCTIONS

565

On the other hand it might be possible to describe the solution set of

by
parametrisation,
in the case

by
from which one can return by
elimination of t
to the previous, parameter free, equation. In other words the connections
between x and y described by

and by

are equivalent.

The
elimination

is the pattern that also serves in composing implicitly given functions. Let by
functions

by given. The composed function

can be obtained by eliminating y from

which may be a purely formal action as far as algebraic functions are concerned.
As an example take the elimination of y from

by means of

566

CHAPTER 17

This formal acting springs from experiencing the implicitly given functions:
from the equations

an equation, free from y,

is derived, which thus determines the connection between x and z. In fact,

one has to prove the equivalence of both of them:
as counterpart of the elimination of y from

the parametrisation by means of y

of

In general this will require the choice of appropriate domain restricting conditions

In the previous example, for instance

As a matter of fact, elimination is reduced to substitution if an explicitly

given function

and an implicit one

with

are composed, that is, to

with
17 .65. What is the meaning of this exposition in a didactical context? My
answer is: to give a fresh chance to the natural operations with implicitly given

functions in opposition to the dogmatism of formalist hobbyists.

Implicitly given functions are at present didactically banished. Things like
with
and
with

are nowadays rejected as function definitions. It must be

567

FUNCTIONS

which then may be composed to get

and this is imposed and required, contrary to ones better knowledge, that is,
even if explicit expressions for and are not available while elimination of
y readily yields the wanted connection between x and z. Users of mathematics,
in particular physicists who do not understand these formalist hobbyism, rightly
stick to their own methods that bear witness to insight into what variables and
functions mean, compared with a formalism detached from its context. I considered it useful to stress here the content aspect of working with variables and
functions.

17 .6668. The Algebraisation of the Pattern the the

As we know, the pattern the the is a vigorous source of the function.
It is mathematised in various ways additively, subtractively, by ratio, or more
general, by monotonicity monotonic increasing or decreasing.
Additive and subtractive the the are algebraised, starting from the
function table, the geometric picture or the graph by
and

or implicitly written
with

and

respectively.

The paradigm of the ratio the the is the uniform motion or growth:
in equal times equal distances,
in equal times the same increase.
Or, more general, with two variables,
the growth of y is independent of the initial state determined by that
of x.

In a high level formula:

for all

This supposition, implies that

to the double, threefold, and so on, increase of x corresponds the double,
threefold, and so on, increase of y.

(See Figure 233.) In formulae, with

568

CHAPTER 17

Thus, the growth function

satisfies

Likewise
to the half, the third, and so on, of the increase of x corresponds the half,
the third, and so on, of the increase of y.

(See Figure 234.) In formulae

thus
and similarly
and so on.
If
to the threefold increase of x corresponds the threefold increase of y

is combined with

to half the increase of x corresponds half the increase of y

569

FUNCTIONS

one gets
to the

fold increase of x corresponds the

fold increase of y,

and in general for any fraction

to the

fold increase of x corresponds the

fold increase of y.

In a formula
for (positive) rational r.
This leads to an explicit form for as soon as the increase of x, say for
is supposed to be known (Figure 235 and 236): for

which yields for all rational values of r.

The pattern the the requires monotonicity of the growth function

Each real number can arbitrarily closely be confined between two rationals

570

CHAPTER 17

Since is monotonic increasing or decreasing,

respectively

from which by having

approach

(and each other) one gets

Uniform growth means that the increase of y is a fixed multiple of that of x.

The fixed factor means, depending on the circumstances, a velocity of motion,
of growth, and so on. Uniform growth is algebraically described by a linear
function

called linear because its graph is a straight line.

17.67. Uniformity means that growth is viewed as an additive phenomenon:
in equal times the same is added, which implies a growth proportional to the
time elapsed.
Another view of growth is introduced by the concept of growth factor; of
the two values of a variable at two moments the quotient is taken to tell us how

much the variable has increased in between. It is a most natural idea as soon
as it is posited that the growth from some moment onwards depends on the
state at that moment. Perhaps the first example historically is compound interest:
interest not withdrawn helps to breed again interest, the part of the additive
is taken over by the growth factor
. In radioactive decay the number
of atoms decaying in the next time interval is proportional to the quantity that
is left.
The traditional expression for this is:

to an arithmetical sequence corresponds a geometrical one.

The converse idea

to a geometrical sequence corresponds an arithmetical one

once led to the invention of logarithmic tables as a means to replace time-wasting

multiplications by time-saving additions. The slide-rule is an offshoot of the
same idea: thanks to the logarithmic scale multiplications can be performed by
the additive operation of adding line-segments.

FUNCTIONS

571

To the arithmetical sequence, say,

corresponds a geometrical one, say,

thus the assignment, for

is

However, the growth, observed discretely, takes place continuously even

compound interest could be bred continuously. Though built stepwise the table
of logarithms pretends to continuity.
Corresponding to the arithmetical sequence there is a geometric one. In

arithmetic sequences one can interpolate, halving the step d, or dividing in ten
parts, or even smaller ones, say, N parts, where N may be as large as you wish.
The corresponding geometric sequence is well-determined. But how?
One needs a multiplicative counterpart of halving, dividing into ten parts, into
N parts. Euclids terminology was continual proportion: switching, say
terms between two magnitude values A and B:

Additively dividing into N parts is the inverse of multiplying by N. Multiplicatively dividing into N parts is the inverse of raising to the Nth power.
is a directly visible multiplicative halving of 9 and similarly
a multiplicative dividing into 10 parts of 1024.
In this course root-extraction offers itself as the inverse of power-raising:
asked for the r such that

the Nth root of q, denoted by

because

extends a law for whole exponents, as does the definition

This then is the way to interpolate the assignment

of a geometrical to an arithmetical sequence to

572

CHAPTER 17

for rational r and repeating the monotonicity arguments of Section 17.62

to
for real

perhaps even negative ones; or somewhat simpler:

with C and Q, which can easily be calculated. This then is an exponential function, describing what is nowadays called exponential growth (or decay).

17.68. Questions as to whether

power raising is invertible,

is defined for rational

is meaningful,
are not put here in their didactic context, since I did so earlier* quite circumstantially. One aspect, which was then still of minor importance, should be

added: the use of handheld calculators in what I would call

imparting meaning to existence queries by operationalisation:

the
power, root, exponential, logarithmic functions

being effectuated by computing them.

Properly I should have amplified my arguments at many points by an appeal
to handheld calculators. However, I could not collect sufficient experience
with calculators either by working didactically with them or by consulting
literature. If I ever catch up with this lag, didactical phenomenology of the
handheld calculator ought to be a separate chapter.

17.6976. Graphs
17.69. Graphs have been mentioned several times in the present chapter (for
instance Section 17.50a, 17.55, 17.63); I even promised to deal with graphs
more systematically. Without entering into details, I would like to draw the
attention to research by C. Janvier,** which unites phenomenological and
experimental aspects.
I restrict myself here to line graphs while disregarding histograms and similar
* Mathematics as an Educational Task, Chapter XIV.
** C. Janvier, The interpretation of complex cartesian graphs representing situations
studies and teaching experiments. Thesis, Nottingham, 1978 Accompanying brochure
including appendices. Shell Center for Mathematical Education and Universit de Qubec
Montral.

FUNCTIONS

573

means of visualisation, which have been discussed earlier (e.g., Section 16.13).
With regard to the functions I stick to numerical ones of one variable (mappings
out of R into R). Taking these restrictions into account I state a phenomenological reciprocity between functions and graphs:

functions, visualised by graphs,

graphs interpreted as functions.

17 .70. Janvier put the graphs into a larger frame of devices to describe connections between variables. The media of description are put into a double entry
table with the means of translation from one medium to the other in the compartments even those in the diagonal can be meaningfully filled out.

17 .71. I recall from the beginning of the present chapter the phenomenology of
variables from the
physical, social, mental, mathematical
world with related dependences which as functions can be

stated, postulated, produced, reproduced.

This can indeed happen in a verbal description. Even in the world of mathematics
this is not usual (the chord as a function of the angle, the area (of a circle)
as a function of the radius). The temperature (at a certain place) as a function

of the time can have been put into a table where it can be read, or a thermograph
can have recorded it in the form of a graph.
A formula pretends to have more precision than a table or a graph. The table

explains the values of the dependent variable only on a discrete set of the independent variable, though this can be done with any precision that is relevant to

the problem; moreover the mesh of this discrete set can be chosen as is required
by relevance. The precision of graphs, however, is limited by the unavoidable
thickness of lines. As regards precision no graph of the logarithmic function can
compete with the simplest logarithmic table, but in many phenomena described
by functions one can be satisfied with the precision of the graph.

574

CHAPTER 17

The often decisive feature by which the graph can compete with table and
fomula is its visualising power. The computer, which produces long lists of
numbers, is assisted by a plotter, which lends perspicuity to this chaotic material.
On the other hand we will come across the graph as a device to visualise functions that have only been subjected to qualitative requirements.
17 .71. Data for a graph construction can be:

(a)

an explicit or implicit equation that expresses a dependent by

(b)
(c)

empirical numerical data,

general instructions on the behaviour of the function.

(a)
(b)
(c)

the function is monotonic increasing, decreasing, convex,

an independent variable,

Examples
the temperatures at de Bilt during 1978.
possesses a certain number of maxima with minima in between.
Explanation of (c):
Graphs of this latter, less determined type are used in econometry. One
speaks about demand and supply as functions of the price of a utility. These,

however, are useful fictions. What happens in the economy depends on

too many and too closely entangled variables to be accessible to a successful
isolation of dependences between a pair of variables; fixing a moment the other
variables is a sometimes useful illusion. Demand and supply functions

are quite simplistic models. Nevertheless there is something like a trend: according to the increase of the price of a utility, the demand may decrease and the

supply increase the degree to which demand and supply react on the change
of the price, called elasticity by the economist, can be visualised by the slope
of a graph; for extremely low or extremely high prices the curve can behave
deviantly so cheap that no consumer cares, so expensive that nobody dares

to produce. The behaviour in general or in special cases can be sketched by a

graph. Such a picture can for instance show the equilibrium of demand and
supply at the intersection of the demand and the supply graph (Figure 237).

FUNCTIONS

575

Another example (by Janvier, see Section 1755) is the speed of a race-car
(Figure 238) as a function of the path it is clear that the function must have
a steep minimum and two less steep minima with horizontal pieces in between
only the distances between the minima can quantitatively be more precisely
specified (Figure 239).

17.72. In order to put data of the type a and b (see Section 17.71) into a graph,
pointwise plotting is the appropriate method. The pointwise given graph is
completed by interpolation to yield a linear zigzag function if the in between
is meaningless or irrelevant, or a so-called smooth curve. If the character of
the function that is to be constructed, is established (or accepted, or postulated)
a priori up to a number of parameters, then a corresponding number of graph
dots suffices to determine the graph if it is a linear function, two points are
needed, which are joined by means of a ruler; if it is a polynomial of degree n
-tuple suffices. But apart from such theoretical condition, there is
a certain feeling for what is a smooth curve, which in the practice of the conscientious draughtsman is backed up by the use of models with a multifariously
varying curvature. It is impossible to say, and it does not seem to have been
investigated, what is the origin of this feeling for adapting a curve to graphical
data. The need and feelings for smoothness of a graph constructed from empirical
data can lead to rejecting misfits. If it may be assumed that the empirical data
are affected by observational errors, a cloud of measuring points is replaced with
a smooth line either guided by subjective feelings or according to an objective
method of adaptation to a functional type depending on a number of parameters.
17 .73. Plotting a graph is preceded by the choice of a scale or scales for the
variables. The graph must fit the dimensions of the paper if it is the graph of
a function given by a formula, at least so far that certain characteristics can be
visualised or a proposed problem becomes graphically solvable (for instance

intersections of graphs become accessible). The scales of independent and

dependent variables can differ from each other. For instance this divergence of
scales is badly needed for drawing the frequency curve of the normal distribution if its characteristics are clearly to be displayed; an exaggeration of the
ordinate in the ratio 10 : 1 is not unusual. In a graph of the vibrating string
this exaggeration might even be 100 : 1.

576

CHAPTER 17

Beside linear scales one knows logarithmic and double-logarithmic ones;

exponential growth processes can slip away if plotted on a linear scale. By a
logarithmic scale for the independent variable exponential growth processes
are apparently featured by linear ones in order to be visualised and compared.
17 .74. Reading graphs is done according to the needs
pointwise,
locally,
globally,

whether
a single graph
is concerned or several graphs are

compared separately or in the same picture.

Pointwise means that for a given value of the independent variable that
of the dependent one is looked up or conversely for a given value of the dependent variable those of the independent one. This is usually done by scanning
the graph

vertically horizontally,
horizontally vertically,
while interpolating, if need be, on the axes.
Locally this means comparative investigations of the graph behaviour

near a point:
positive, negative,
increasing, decreasing, constant, minimal, maximal,
steeply or less steeply increasing or decreasing, up to a discontinuous jump
or fall,
rectilinear or bent in the convex or concave sense,

more or less bent in one sense or the other,

sharply or less sharply bent at a minimum or maximum,
changing from one type of bending to the other (turning point).
This qualitative behaviour can quantitatively be specified by considering

a measure of increase or decrease,

the height of minimum or maximum,
a measure of curvature (in general or at an extreme value),

FUNCTIONS

577

in particular by plotting graphically

the derivative, or second derivative

of the function a didactically useful activity, which in Calculus instruction is
not paid the attention it deserves.
Globally this can mean
searching for
and comparing

local characteristics in order to find

intervals of positivity, negativity, confinement between certain bounds,

intervals of increase, decrease, constancy,
minima, maxima,
points of steeper, less steep, steepest, least steep behaviour,
jumps,
intervals with certain curvature,
turning points,
points of sharper, less sharp, sharpest, least sharp bending.
Reading graphs globally beyond the search for the comparison of global
characteristics takes place if one states
monotonicity,
oscillating behaviour,

asymptotic behaviour of various kind,

periodicity,
damped, excited, slightly or more seriously disturbed periodicity,
linearity, quadratic or otherwise algebraic character,
exponential character,
sinusoidal character.
Reading globally a function given in a certain restricted interval can lead to
extrapolation based on certain characteristics mentioned earlier.
If the graphs of two or more functions are confronted with each other, it
is in particular
function values,
slope measures,
places of minima and so on

578

CHAPTER 17

that are compared with each other. If families of functions depending on parameters are discussed

the diverging asymptotic behaviour

for different parameter values can play a part.

17.75. Properties and peculiarities such as those summarised here, can more
easily be recognised in the graphical picture than in formulae, and this is one
of the reasons why graphs are so important.
Even the restriction to one variable is not as absolute as it seems. Graphs
are pictures of function-relations between an independent and a dependent
numerical variable. Functions of more variables can be mastered by declaring
some of the variables as parameters, and plotting functions corresponding
to various parameter values besides or below each other as if they were moving

pictures. Another graphical means to picture functions of more, in particular

two, variables, has been shown in Section 17.56: level lines with the function
value as level.

LIST OF THE A U T H O R S P U B L I C A T I O N S

ON M A T H E M A T I C S E D U C A T I O N

1.

De algebraische en de analytische visie op het getalbegrip in de elementaire

wiskunde, Euclides 24 (1948), 106121.

2. Kan het wiskundeonderwijs tot de opvoeding van het denkvermogen
bijdragen. Discussie tussen T. Ehrenfest-Afanassjewa en H. Freudenthal
(Publicatie Wiskunde Werkgroep van de V.W.O.), Purmerend 1951.
2a. Erziehung des Denkvermgens (Diskussionsbeitrag), Archimedes Heft
6 (1954), 8789. (This is a translated extract from 2.)
3.

De begrippen axioma en axiomatiek in de Wis- en Natuurkunde, Simon

Stevin 39 (1955), 156175.

3a. Axiom und Axiomatik, Mathem. Phys. Semesterberichte 5 (1956),

419.
4. Initiation into Geometry, The Mathematics Student 24 (1956), 83
97.
5. Relations entre lenseignement secondaire et lenseignement universitaire
en Hollande, Enseignement mathmatique 2 (1956), 238249.
6. De Leraarsopleiding, Vernieuwing 133 (1956), 173180.
7. Traditie en Opvoeding, Rekenschap 4 (1957), 95103.
8. Report on Methods of Initiation into Geometry, ed. H. Freudenthal,
(Publ. Nederl. Onderwijscommissie voor Wiskunde), Groningen, 1958.
9. Einige Zge aus der Entwicklung des mathematischen Formalismus, I,
Nieuw Archief v. Wiskunde 3 (1959), 119.
10. Report on a Comparative Study of Methods of Initiation into Geometry,
Euclides 34 (1959), 289306.
l0a. A Comparative Study of Methods of Initiation into Geometry, Enseignement mathmatique 2, 5 (1959), 119139.
11.

Logica als Methode en als Onderwerp, Euclides 35 (1960), 241255.

11a. Logik als Gegenstand und als Methode, Der Mathematikunterricht 13

(1967), 722.
12. Trends in Modern Mathematics, ICSU Review 4 (1962), 5461.
12a. Tendenzen in der modernen Mathematik, Der math. und naturw.
Unterricht 16 (1963), 301306.
13. Report on the Relations between Arithmetic and Algebra, ed. H. Freudenthal (Publ. Nederl. Onderwijscommissie voor Wiskunde), Groningen,
1962.
14. Enseignement des mathmatiques modernes ou enseignement moderne
des mathmatiques? Enseignement Mathmatique 2 (1963), 2844.
15. Was ist Axiomatik, und welchen Bildungswert kann sie haben? Der
Mathematikunterricht 9, 4 (1963), 519.
579

580

16.

16a.
17.
18.
19.

20.
21.
22.

23.
24.

25.
26.

27.

28.

29.
30.

31.
32.
33.

LIST OF THE AUTHORS PUBLICATIONS

The Role of Geometrical Intuition in Modern Mathematics, ICSU Review

6 (1964), 206209.
Die Geometric in der modernen Mathematik, Physikalische Bltter 20
(1964), 352356.
Bemerkungen zur axiomatischen Methode im Unterricht, Der Mathematikunterricht 12, 3 (1966), 6165.
Functies en functie-notaties, Euclides 41 (1966), 299304.
Why to Teach Mathematics so as to Be Useful? Educational Studies in
Mathematics 1 (1968), 38.
Paneldiscussion, Educational Studies in Mathematics 1 (1968), 6193.
Lintgration aprs coup ou la source, Educational Studies in Mathematics 1 (19681969), 327337.
The Concept of Integration at the Varna Congress, Educational Studies
in Mathematics 1 (19681969), 338339.
Braces and Venn Diagrams, Educational Studies in Mathematics 1 (1968
1969), 408414.
Further Training of Mathematics Teachers in the Netherlands, Educational Studies in Mathematics 1 (19681969), 484492.
A Teachers Course Colloquium on Sets and Logic, Educational Studies
in Mathematics 2 (19691970), 3258.
ICMI Report on Mathematical Contests in Secondary Education (Olympiads), ed. H. Freudenthal, Educational Studies in Mathematics 2 (1969
1970), 80114.
Allocution au Premier Congrs International de lEnseignement Mathmatique, Lyon 2431 aot 1969, Educational Studies in Mathematics
2 (19691970), 135138.
Les tendances nouvelles de 1enseignement mathmatique, Revue de
lenseignement suprieur 4647 (1969), 2329.
Verzamelingen in het onderwijs, Euclides 45 (1970), 321326.
The Aims of Teaching Probability, in L. Rde (ed.), The Teaching of
Probability & Statistics, Almqvist & Wiksell, Stockholm, 1970, pp. 151
167.

Introduction, New Trends in Mathematics Teaching, Vol. II, Unesco,

1970.
Un cours de gomtrie, New Trends in Mathematics Teaching, Vol. II,
Unesco, 1970, pp. 309314.
Le langage mathmatique. Premier Sm. Intern. E. Galion, Royaumont
1320 aot 1970, OCDL, Paris, 1971.

34.
35.
36.
37.

Geometry between the Devil and the Deep Sea, Educational Studies in
Mathematics 3 (1971), 413435.
Kanttekeningen bij de nomenclatuur, Euclides 47 (1971), 138140.
Nog eens nomenclatuur, Euclides 47 (1972), 181192.
Strategie der Unterrichtserneuerung in der Mathematik, Beitrge z.
Mathematikunterricht, Schroedel, 1972, 4145.

LIST OF THE A U T H O R S P U B L I C A T I O N S

38.

581

The Empirical Law of Large Numbers, or the Stability of Frequencies,

Educational Studies in Mathematics 4 (1972), 484490.

39. What Groups Mean in Mathematics and What They Should Mean in
Mathematical Education, in Developments in Mathematical Education,
Proceedings of the Second International Congress on Mathematical Education 1973, pp. 101114.
40. Mathematics as an Educational Task, Reidel, Dordrecht, 1973.
40a. Mathematik als pdagogische Aufgabe, Band 1, 2, Klett, Stuttgart, 1973.
41. Mathematik in der Grundschule, Didaktik der Mathematik 1 (1973),
211.
42. Nomenclatuur en geen einde, Euclides 49 (1973), 5358.
43. Les niveaux de lapprentissage des concepts de limite et de continuit,
Accademia Nazionale dei Lincei, 1973, Quaderno N. 184, 109115.
44.
45.
46.

47.
48.
49.

50.
51.
52.
53.

DeMiddenschool, Rekenschap 20 (1973), 157165.

Waarschijnlijkheid en Statistiek op school, Euclides 49 (1974), 245246.
Die Stufen im Lernprozess und die heterogene Lerngruppe im Hinblick

auf die Middenschool, Neue Sammlung 14 (1974), 161172.

The Crux of Course Design in Probability, Educational Studies in Mathematics 5 (1974), 261277.
Mammoetonderwijsonderzoek wekt wantrouwen, University Newspaper
U, State University of Utrecht, June 1974.
Mathematische Erziehung oder Mathematik im Dienste der Erziehung,

Address 21 June 1974, University Week, Innsbruck.

Kennst Du Deinen Vater? Der Mathematikunterricht 5 (1974), 718.
Lernzielfindung im Mathematikunterricht, Zeitschrift f. Pdagogik 20
(1974), 719738; Der Mathematikunterricht 23 (1977), 2645.
Sinn und Bedeutung der Didaktik der Mathematik, Zentralblatt fr
Didaktik der Mathematik 74, 3 (1974), 122124.
Soviet Research on Teaching Algebra at the Lower Grades of the Ele-

mentary School, Educational Studies in Mathematics 5 (1974), 391412.

54. Ein internationaal vergelijkend onderzoek over wiskundige studieprestaties, Pedagogische Studin 52 (1975), 4355.
55. Wat is meetkunde? Euclides 50 (19741975), 151160.
56. Een internationaal vergelijkend onderzoek over tekstbegrip van scholieren,
Levende Talen, deel 311 (1975), 117130.
57. Der Wahrscheinlichkeitsbegriff als angewandte Mathematik, Les applications nouvelles des mathmatiques et lenseignement secondaire, C.I.E.M.
Conference, Echternach, June 1973 (1975), 1527.
58. Wandelingen met Bastiaan, Pedomorfose 25 (1975), 5164.
59. Compte rendu du dbat du samedi 12 avril 1975 entre Mme Krygowska
et M. Freudenthal, Chantiers de pd. math., June 1975, Issue 33 (Bulletin
bimestriel de la Rgionale Parisienne), 1227.
60. Pupils Achievements Internationally Compared the I.E.A. Educational
Studies in Mathematics 6 (1975), 127186.

582

LIST OF THE AUTHORS P U B L I C A T I O N S

60a. Schlerleistungen im internationalen Vergleich, Zeitschrift fr Pdagogie

21 (1975), 889910. (This is a translated extract from 60.)
61. Leerlingenprestaties in de natuurwetenschappen internationaal vergeleken,
Faraday 45 (1975), 5863.
62. Des problmes didactiques lis au langage, pp. 13;Lorigine de la
topologie moderne daprs des papiers indits de L.E.J. Brouwer, pp.
916. Lectures delivered at the University, Paris VII, in April 1975
(offset). (With Krygowska).
63. Variabelen (opmerkingen bij het stuk van T. S. de Groot, Euclides 51,
154155), Euclides 51 (19751976), 349350.
64.
Bastiaans Lab, Pedomorfose 30 (1976), 3554.
65. De wereld van de toetsen, Rekenschap 23 (1976), 6072.
66. De C.M.L-Wiskunde, interview, Euclides 52 (19761977), 100107.
67. Valsheid in geschrift of in gecijfer? Rekenschap 23 (1976), 141143.
68. Studieprestaties Hoe worden ze door school en leerkracht beinvloed?
Enkele kritische kanttekeningen n.a.v. het Colemanreport, Pedagogische
Studin 53 (1976), 465468.
69. Rejoinder, Educational Studies in Mathematics 7 (1976), 529533.
70. Wiskunde-Onderwijs anno 2000. Afscheidsrede IOWO 14 Augustus 1976,
Euclides 52 (19761977), 290295.
71. Annotaties bij annotaties, Vragen bij vragen, Onderwijs in Natuurwetenschap 2 (1977), 2122.
72. Creativity, Educational Studies in Mathematics 8 (1977) 1.
73. Bastiaans Experiment on Archimedes Principle, Educational Studies
in Mathematics 8 (1977), 316. (This is a translated extract from 64.)
74. Fragmente, Die Mathematikunterricht 23 (1977), 525.
75. Didaktische Phnomenologie, Lnge, Der Mathematikunterricht 23
(1977), 4673.
76. (Review, with J. van Bruggen) The Psychology of Mathematical Abilities
in Schoolchildren by V. A.
Proceedings of the National
Academy of Education 4 (1977), 235277.
77. (Review, with J. van Bruggen) Soviet Studies in the Psychology of Learning
and Teaching Mathematics (6 volumes), ed. by J. Kilpatrick and I. Wirszup,
Proceedings of the National Academy of Education 4 (1977), 201234.
78. Mathematik als pdagogische Aufgabe I, Klett, Stuttgart, 2. Aufl. 1977.
(Cf. 40a).
79. Die Crux im Lehrgangentwurf zur Wahrscheinlichkeitsrechnung, Didaktik
der Mathematik, 436459 (ed. Steiner) Wiss. Buchgeselschaft Darmstadt,
1977. (Translated from 47.)
80.
Brokjes Semantiek, Pedagogische Studin 54 (1977), 461464.
81. Cognitieve ontwikkeling kinderen geobserveerd, Prov. Utrechts
Genootschap, Jaarverslag 1977 (1978), 818.
82. Teacher Training An Experimental Philosophy, Educational Studies in
Mathematics 8 (1977), 369376.

LIST OF THE AUTHORS P U B L I C A T I O N S

583

83. La smantique du terme Modle, La Smantique dans les Sciences, Colloque de lAcadmic Intern. de Philos. des Sciences 1974, Archives de
lInstitut Internat. des Sci. Thor. 21 (1978).
84. Bastiaan meet zijn wereld, Pedomorfose 37 (1978), 6268.
85. Address to the First Conference of I.G.P.M.E., 29 Aug. 1977, Educational

Studies in Mathematics 9 (1978), 15.

86. Modern wiskunde-onderwijs? Goed wiskunde-onderwijs, Intermediate
28 April 1978.
87. Weeding and Sowing A Preface to a Science of Mathematics Education,
Reidel, Dordrecht, 1978.
88. Change in Mathematics Education since the late 1950s Ideas and
Realisation, an ICMI report, Educational Studies in Mathematics 9 (1978),
7578.
89. Soll der Mathematiklehrer etwas von der Geschichte der Mathematik
wissen? Zentralblatt Didaktik Mathematik (1978), 7578.
90. Change in Mathematics Education since the late 1950s The Nether-

lands, Educational Studies in Mathematics 9 (1978), 261270.

Vorrede zu einer Wissenschaft vom Mathematikunterricht (Original of 87),
Oldenbourg, Mnchen, 1978.
92. Nacherfindung unter Fhrung, Kritische Stichwrter, ed. D. Volk, Fink
Verlag, 1979, pp. 185194.
91.

93. Rings and String, Educational Studies in Mathematics 10 (1979), 6770.

94. Lessen van Sovjetrekenkunde, Pedag. Studin 56 (1979), 5760.
95. Onderwijs voor de kleuterschool cognitief, wiskunde, De Wereld van

het Jonge Kind 1979, 143147, 168172.

96. Introductory Talk, Congresso Internationale 715 gennuio 1976,
Accad. N. Lincei 326 (1979), 1532.
97. Konstruieren, Reflektieren, Beweisen in phnomenologischer Sicht,
Schriftenreihe Didaktik der Mathematik, Klagenfurt 2 (1979), 183200.
98. How Does Reflective Thinking Develop? Proceedings Conference IGPME
Warwick 1979.

99. Ways to Report on Empirical Research in Education, Educational Studies

in Mathematics 10 (1979), 175303.
100. Mathematik als pdagogische Aufgabe II, 2. Aufl., 1979.
10la. New Math or New Education? Prospects Unesco 9 (1979).
101b. Mathematiques nouvelles ou ducation nouvelle? Perspectives Unesco 9
(1979), 339350.
101c.? Matemticas nuevas o nueva educacin? Perspectivas Unesco 9 (1979),
337348.
102. De waarde van resumerende en tweede hands informatie, Pedag. Studin
56 (1979), 323326.
103. Unexperienza di insegnamento dello lege di Archimede, Instituto della
Enciclopedia Italiana n. 1, 245246. Roma (translation of 96; see 73).
104. Weeding and Sowing, 1980 (= 87).

584

LIST OF THE AUTHORS PUBLICATIONS

105. Invullen Vervullen, De Pedagogische Academie 10 (1979), 197200.

106. Invullen Vervullen, Euclides 55 (1979/80), 6165 (= 105 abridged).

107. Onderzoek in onderwijs, Pedomorfose 12 (1980), no. 47, 5766.
108. Wiskunde en onderwijskunde in de leraarsopleiding, Resonans 12 (1980),
2021.
109. Lernprozesse Beobachten, Neue Sammlung 20 (1980), 328340.
110. Examenpakket en wiskundige kiespijn, Weekblad NGL 13 (1980), 438
439.
111. Four Cube House, For the Learning of Mathematics 1 (1980), 1213.
112. IOWO Mathematik fr alle und jedermann, Neue Sammlung 20 (1980),
633654.
113. Kinder und Mathematik, Didaktik des Entdeckens und Nacherfindens,
Wiskunde op de basisschool, Grundschule 13 (1981), 100104.
114. Flcheninhalt in phnomenologischer Sicht, Mathematiklehrer 19811,
510.
115. Major Problems of Mathematical Education, Educational Studies in
Mathematics 12(1981), 133150.
116. Verslaggeving over empirisch onderwijskundig onderzoek ten onzent,
Pedag. Studin 58 (1981), 141143.
117. Hovedproblemen for matematikkundervisning, Normat 29 (1981), 49
66. (Transl. from 115.)
118. Should a Mathematics Teacher Know Something about the History of
Mathematics? For the Learning of Mathematics 2 (1981), 3033. (Transl.
from 89.)
119. Roltrappen, W. Bartjens 1 (1981/2), no. 1, 34.
120. Binnen is t minder, N. Wiskrant 1 (1981), no. 2, 1820.
121. Mathematik, die uns angeht, IOWO, Mathematiklehrer 2 (1981), 34,
44.
122. Wat is er met het aftrekken aan de hand? W. Bartjens 1 (1981/2), no. 2,
45.
123. Fiftyfifty, N. Wiskrant 1 (1982), no. 3, 78.
124. Winkel messen und berechnen, Mathematiklehrer 1 (1982), 45.
125. Taalfetisjisme, Euclides 57 (1981/2), 297291.
126. Mathematik eine Geisteshaltung, Grundschule 14 (1982), 140142.
127. Kinder und Mathematik, in Grundschule 14 (1982), Arbeitskreis Grundschule, Beitrge 50 (= 113).
128. Ik was moeder en ik doe boodschappen, W. Bartjens 1 (1981/2), 131
133.
129. Differentialen ja of neen en zo ja, hoe? N. Wiskrant 1, No. 4, 1518;
2, No. 1, 1618(1982).
130. Studentenhaver, W. Bartjens 2 (1982), 214215.
131. Mathematika kak
1982. (Translation
of 78.)
132. = 127 = 113.Beltz Praxis 1982.

LIST OF THE AUTHORS PUBLICATIONS

585

133. Fiabilit, validit et pertinence, Educational Studies in Mathematics 13

(1982), 395408.
134. Een visie op onderwijskundig bezig zijn, Utrechtse Pedagogische Verhandelingen 5 (1982), No. 1, 311.
135. De kortste weg, N. Wiskrant 2, No. 2 (1982), 1921.
136. Inderdaad een oud probleem, Euclides 58 (1982/3), 6568.
137. Die Entwicklung des mathematischen Denkens der 1014 Jhrigen,
Mathematiklehrer 1982, No. 3, 513.
138. Ganzeborden andersom, W. Bartjens 2 (1982/3), No. 1, 1819.
139. Kortste Wegen op een krom oppervlak, N. Wiskrant 2 (1983), No. 3,
911.
140. Wo fngt die Geometric an? Mathematiklehrer 1983, 1, 24.
141. Pourquoi de la gomtrie dans lenseignement primaire?, Colloque international sur lenseignement de la gomtrie, Mons, 1982.

INDEX

abacus 31, 72, 91, 118ff., 173

, pictorial 120
Abel, N. H. 526
abridged multiplication 126ff.
abstracting 296ff.
adding lengths 2ff.
addition 69, 74, 77, 81, 96ff., 103, 170,
445
additive structure; see IN
affine, affinity 12, 213, 219, 230ff., 246,
333, 339ff., 349ff., 375ff.
age as a function 550
Alembert, d 516, 525
algebra 473ff.

argument (variable) 497

Aristarchos 224
Aristotelism 330

arithmetical language 463ff.

arrow(s) 441ff.
arrow language 481ff., 556ff.

article; see binding

associativity 104, 110, 113, 216

of adding lengths 52ff.

assortment 52
asymptotic; see reading graphs
atlas 284
attitude 459, 477
Augustine 281
automatism(s) 209, 457ff., 469ff.

in geometry 450ff.
algebraic permanence principle 434ff., 488
algebraisation of functions 557ff.
geometry 415, 433, 450
inverting and composing 560ff.
algebraisation, stage of algorithmisation 118

algorithmic(s), algorithmisation

automaton 496ff.
automorphism(s) 217ff., 231, 344ff.

axiomatisation 349

back to basics 38
Bang, V. 540 footnote
Barra, R. 542 footnote
barrels 331, 335

117, 200,

470, 476, 479ff.

algorithmics in IN 117ff., 172

algorithmisation of the function 561ff.

bathrooms 242
beer can 247
Bernoulli 516, 520

analogy 490
analytic geometry 436
analytical mechanics 411
analysis 411; see function
angle 236ff., 322ff., 357ff., 518
transferred by sliding 364ff.
Antonie-Alexander 254

binding (variables)
universal, existential, article, set forming
function forming, interrogative, demonstrative 474ff.

anontologisation 81

bisector 316ff.
blue-prints 224, 282

Apollonius 410, 519

bodies 264

appellative 473ff.
application 43

; see regular, rigid

Bolzano 254ff.

of areas 519ff.

border 256, 262, 267ff.

approximation 10, 16, 381ff., 401

Borel systems 41, 378ff.

bottles of beer 161

arc 327
arc length 16, 317, 370
Archimedean axiom 8
Archimedes 290, 377, 399, 518, 520

Archimedes principle 102

area 35, 140, 152, 236, 266ff., 401ff.

box(es) 228ff.

boundary 256, 262, 267ff., 380

Bourbaki 31, 90
brackets 470ff.

braiding 277ff.

of circle sector 370

of polygons 385
of rectangle 110, 140, 380

break-make transformations 18ff., 84ff.,

140, 152, 257, 374, 383ff., 394,
401

586

587

INDEX

broken line 10, 15

Brobdingnag 402
Brouwer, L. E. J. 256
Bruner 17, 20, 30, 31
Bryant, P. 12, 47 footnote
calculators 131,463, 572
calculus 327, 403ff.
Cantor 76ff., 526
Cantors discontinuum 256, 379ff.

cardinal (number), cardinality 12, 34, 38,

54ff., 74, 76ff., 83ff., 97, 103,
198, 532
Cartan, E. 220

cartesian product 265, 411, 497

measures 9
compositions 185ff., 196, 198, 536ff.
compository (reproduction) 234, 543

compound interest 570

concave 576
concept(s) 31, 295
concept attainment 31, 225ff., 228, 330,
388, 542
concretisation, false 32, 39ff., 54, 81, 154,
360
Concorde 327
cone 227ff., 232, 322, 402ff.
configurations 296

congruent, congruence, congruence mappings 12ff., 17, 152, 188, 190,

Cavalieris principle 376ff., 388

cavity 270ff.

211, 218, 227ff., 231ff., 245ff.,

330ff., 337ff., 357ff., 367, 375ff.,
378ff., 394ff.
conics (history) 519ff.
connection 257ff.
, multiple 272ff.

celestial motions 516ff.

connections 44

change of perspective 417ff., 461ff., 490,

510, 520, 546, 554ff.
change of standpoint 84
chimney 243
chord 368
circle 227, 238, 246, 314ff., 356ff., 402,
547

conservation 12, 21ff., 55, 83ff., 140, 180,

234ff., 395ff.
constitution (mental) of

Caspar Hauser 399

castles 425
category 34

Cauchy 254ff.

circulation 273ff., 279

cardinal number 54ff.

addition 96
division 114

mappings 338ff.
mathematical concepts 29ff.

circumcircle 317

mathematical structures 29ff.

classification 57
by comprehension 544ff.

mental objects 31ff., 54ff., 233ff., 412ff.

number 56

by discrimination 544ff.
by parametrisation 544ff.
closed paths 272ff.

objects 290
plane 412

clover leaf knot 227ff., 426, 430

space 413
construction box 245

coexistence 218ff., 233

constructs 185ff., 199, 200

collinearity 218ff., 233

combinatoric(s) 210ff., 221, 243, 246, 277,

contact transformations 235

content 392ff.
context 20, 42, 43, 49, 204, 223ff., 228ff.,

279, 294ff., 299, 333, 335, 350,

422ff., 428ff.

232, 244, 250ff., 281ff., 331,

372, 401, 462

column arithmetic 117ff.

commutativity 104,109,113
of adding lengths 4
comparatives 202
comparision of lengths 10
compasses 317, 361ff., 367ff.
complement 39ff.
complex numbers 433ff.
composing
areas 374ff.
long objects 3ff., 29

continued fractions 25
continuity 250ff., 258ff., 349ff., 517ff., 538
contour lines 554ff.
convex(ity) 12, 313ff., 576

cordinate (s) 284

coordinate systems 408ff., 436ff., 446ff.,
527

cartesian 410ff.
celestial 409
cylindric 415

588

INDEX

lefthand, righthand 418ff.

distributivity 109, 113, 117, 224

polar 413

divisibility 36, 131ff.

division 25, 114ff., 168ff., 184
, long 130ff.
with remainder 479ff.
divisor 39

terrestrial 409
counters 439
counting 74ff., 81, 83, 86, 99ff., 531ff.
cube 13, 218, 242ff., 246, 300, 334

Cuisinaire (rods) 101, 546

curvature 316, 320ff., 353ff., 576
curves 10, 15, 16, 250, 261ff., 506ff.
Cusanus 223
cut 270ff.
cyclic; see order
cyclic processes 136, 138, 165ff.

, greatest common 132

double decker 555
doubling 202ff.
drawer principle 38, 67ff., 94, 532
Dubrovnik 90
duodecimal 367

cycloid 520

econometry 574

cylinder 227ff., 232, 245, 322

elasticity 574

Davydov 102

elementary structures 44

dead reckoning 410

elimination 565ff.

elementary operations as functions 555ff.

decimal development 174

ellipse 548

fractions 25, 124, 174

embodiment 32

system 114ff., 470

enactive 17, 20, 30ff.

endomorphisms 344ff.

decimate 139

enlargement 345, 400ff.

epistemology 540ff.

decreasing 576
Dedekind 79ff.
Dedekind cut 9

equality 13, 49ff.

deformations 506ff.
demand/supply 574

equality sign 477ff., 556

equations 450, 557

demonstrative; see binding

denominator 25ff.

equilibrium 574
equipartition 62ff., 65ff., 111ff.
Eratosthenes 224

density 94ff., 198, 206

dependence 494ff., 520ff., 523ff., 528ff.

dependent/independent variable 496ff.

derivative 552, 554, 577
Descartes 224, 410, 433, 450, 520
descriptive geometry 244, 311, 334
diagonal 248, 379
diagrams 284

diamond 245
die 246, 331, 334
differentiable 250ff., 379
differential geometry 302, 320
differential-topological 265ff.
digital 470

Erlanger program 220ff., 226, 228, 231ff.

erleben/erfahren 542
essence/appearance 292
estimate 94, 103
Euclid 31, 75ff., 132, 218, 221, 223ff.,
282, 323, 341, 342, 366, 571

Euclidean algorithm 132

euclidean plane, space, geometry 218, 221,

dihedra 322

224ff., 238, 359

Euclidicity 359
Eudoxos 282
Euler 256, 516
Euler polyhedron formula 256
exhausting with units 374

dihedral group 217

Dijksterhuis, E. J. 520

experiencing 541ff.
exponential 572, 576

dilatation 230ff., 443

dimension 255, 261ff.
directed graph 211

expositions 185ff., 198, 536ff.

extension 39

direction 301ff., 407ff.

Dirichlet 94, 525
distance
1, 19ff., 188, 211, 213, 218,
352ff., 407ff., 552

extensive parameters 203ff.

factorising 113
fair sharing 375
Fermat 520

589

INDEX

figures 296ff., 379ff., 450

filter 37
Fischbein, E. 33 footnote
Fladt, K. 236
flatness 227ff., 297ff.
flexible, flexibility 14ff., 17, 20, 234, 243,
341
flexions 14ff., 17, 152, 212, 217, 221, 246,
374, 394

additive/ subtractive 535ff,

algebraisation of 567
in proportion 536ff.
multiplicative/inverse 536ff.

univalent/polyvalent 525ff.
function theory plane 270
functional 526, 550ff.
functions

flow diagram 466, 485

, composing 522ff., 526, 529, 533, 544,

546, 559ff.

flow distribution 154, 170

, demand/supply 574

football 322
formal languages 463ff., 477

, exponential 572
former; see binding

formalising 68ff., 82ff., 468ff.

formulae 202
four colour problem 274
four cubes houses 332
Fourier series 525
fraction(s) 25, 33, 133ff., 144ff., 201ff.,
433
as a comparer 145ff.
as a fracturer 139ff., 147
as an operator 150ff.

implicitly given 563

aspects 147ff.
as transformer 148ff.
terminology 134ff.

, and ordinal number 139

Frege 31, 75 footnote, 76ff.

Frbel 101
front of a car 247
function(s) 154, 199, 202, 404, 411, 450ff.,
489, 491ff.
function,
: aiming and hitting 538ff.

arbitrary 525ff.
algebraisation of the 557
behaviour
qualitative/quantitative 576ff.
pointwise, locally, globally 576ff.
classifying by a 544ff.
: compensation 539ff.
experience 501ff., 562ff.
: genetic view 528ff.
graph 498ff.
, Leibniz-Bernoulli terminology 516,
520ff.

: historical view 515ff.

: inter- and extrapolating 540
notation 560ff.
of time 548ff.

inverting 523ff., 526, 529, 544, 546,

559ff.
linear 518ff., 536ff.
, logarithmic 572
of n variables 506ff., 523ff., 527

, polynomial 575
, power 572
, quadratic 518ff.
, root 572
, sorts of 506ff.
, trigonometric 517
zig-zag 516ff.

functor 496
fundamental group 273
Galileo 391, 549

Galois 526
Galton board 67

Gattegno 439
gauging 218

Gauss curvature 320ff.

genus p 271
geodesic(s) 302, 321
geographic 326

geometric- algebraical permanence principle

435ff., 450ff., 479
geometry 224ff., 351ff.
gestalt 245
giant 207
glide reflections 13, 425
globe 284, 409

globe traveller 552ff.

graph(s) (combinatorial)

51, 210ff., 256,

284, 350
(functional) 198, 202, 412, 450ff., 455,
525, 540, 550ff., 572ff.

table 498ff., 555

graph construction 574ff.

the ( ) of ( ) 502
the the 533ff.

grids 382, 389, 401

Grize, J. B. 540

590

INDEX

ground plans 284

group(s) 34, 37, 215ff., 216, 217, 219ff.,
230ff., 526ff.
, Moebius 235
of automorphisms 217ff.
of four 215

theory 220
growth factor 570
growth processes 550ff.
linear 551
exponential 551ff.
gun bullet 247

Habermas 28 footnote
Hankel 435 footnote
Hegel 28 footnote
Heidegger 28 footnote
height (as a function) 554ff.
hexagon, regular 370
Hilbert 75, 224, 386

Janvier, C. 553, 572, 575

jigsaw puzzle 245, 324, 531, 541
Jordan, C. 251
Jordan arc 251ff., 257ff., 349
curve 251ff., 260ff., 265, 349, 418ff.,
422ff.
disc 253ff.
sphere 253ff.

Jordans theorem 251ff., 349

JordanBrouwer theorem 254

k-homogeneous set 95
Kant 226
Keitel, Chr. 178
Keplers laws 184
kinaesthetic 419ff., 425, 431
Klein, F. 75, 220
Knotting 277ff.
Knigsberg bridges 279
Kuratowski, C. 527

histogram 152, 198

hole 270ff.
homological/homotopical 272ff.
horizontal 238, 431
Husserl 28 footnote

ladder surface 271

Lagrange 526
language 460, 461ff., 470, 477, 481ff.,
556
learning processes 29ff.

ideal 36
identities 557

least common multiple 36

Lebesgue measure 378

identity (versus congruence) 13

idiom 486
ikonic 17, 20, 30ff., 238ff., 258

left turn 387

legends 284
Leibniz 281ff., 516, 520, 522
length 1ff., 152, 266ff.
expressed by adjectives l0ff.

immersion transformation 395

inclusion 89
increasing 576
indeterminates 477ff., 557
inequality of the triangle 353
infinitestimal 520ff.
infinity 75
Inhelder, B. 223, 225, 228, 361 footnote,
412 footnote
IOWO x, 207, 307, 555

invariance 17
measure 9ff.

measuring 15, 24, 26

of an arc 16
ratio 13
letters 473ff., 477
lightrays 310
Lilliput 402
limit 256

insight 209, 457ff., 469ff.

intensive parameters 203ff.

line segment 13, 267

interrogative; see binding

intersection 36, 39ff.

linearity 184

linear order 258

intuitions 33 footnote, 226

lines 261ff.

invariance

linguistic transformation 461ff.

linked, linking 260, 277
localisation 407ff.
logarithmic 570

12ff., 22ff., 84ff., 92ff., 184,

235ff., 349ff., 378ff.
inversion, didactical 305
isoperimetric 391
isomorphic, isomorphism 89, 93, 162ff.,
200, 210ff., 216, 220, 344ff.

logic blocks 39, 53, 239

logical operations 39

591

INDEX

long object 2ff., 152

ludo board 50ff.

morphisms 496ff.

machine (as function) 154, 496ff., 556

multiplication 66, 74, 77, 81, 108ff., 267,

401
multivalued symbol 524ff.

magnitude

1ff., 101ff., 116, 135, 137,

139ff., 144, 146ff., 160, 181ff.,

266ff., 372ff., 390, 433, 549
, directed 432ff., 445ff.
make and break 18
manipulation 256ff.
mapping models 155
mappings 231, 336ff., 506ff., 510, 516,
523ff., 531

affine (see affine)

motions 342ff., 425

movements 13, 506ff.

IN 57, 73ff.
, additive structure of 104ff., 113,
152ff., 160, 546ff.
, decimal structure of 90
, multiplicative structure of 112ff.
name(s) 473ff.

, proper 470, 473ff., 491ff., 495

at one blow 343ff.

congruence (see congruent)
geometric 330ff., 526ff., 533ff., 537ff.,
543
motions 342, 425
slide reflections 13, 425ff.

, generic 473ff., 495

, of functions 504ff.
, polyvalent 474, 491ff., 506ff.
natural number(s) 73ff., 103, 470ff.
negative number(s) 432ff.

topological 251ff., 349ff., 526

Netherlands Railways 240

maps 284
meaning 471ff.
measure 35, 266ff., 377ff.
of length 9, 359
measuring 351ff.
lengths 15, 24, 26, 317, 351ff., 401
angles 359ff.
areas 372ff., 400ff.
volumes 372ff., 392ff., 400ff.

measuring number 150, 537

mechanics 342, 376
Menger, K. 265
mental arithmetic 120ff.
mental object 33, 53, 54, 223, 225ff.,
245ff., 265, 295, 324, 329, 330,
346ff., 366, 380ff., 388ff., 394ff.,
400ff., 414ff., 536ff., 542
metric space 216
structure 218

system 172ff., 392ff.

miniature 533
minimum/maximum 576ff.
model 51, 284, 416, 436ff., 517

neighbourhood 214
networks 244, 299, 311
Newton 223, 281, 415, 520
n-fold connected 270ff.
nomograms 202, 453ff., 456ff.
non-euclidean 220, 224
non-linearity 205ff., 267, 368
non-standard analysis 522
nooumenon 28

norming 195ff., 537

number concept 31
line 101ff., 152, 160, 546ff.
names 133ff.
numerals 470ff.
numerosity 54ff., 81, 89, 99
Nuremberg scissors 234

obstructions 293ff.
occupation 296ff.
of, 137ff.

a fraction (2/3 of )
one-to-one-ness 258
operation 496ff.
operator 149, 496ff., 526

, wire, plaster, cardboard 241

Moebius geometry 231

order 45ff., 74, 218, 259, 534

group 235
strip 256, 300, 324
monopoly of the variable x 560
mono tonic, monotonicity 456ff., 533ff.,
538ff., 567ff.
Montessori 92

Oresmus 550

of lengths 5, 352ff.
linear 417ff., 456ff.
cyclic 47, 414ff.
oriented, orientation 387, 416ff.
on surfaces 423ff.

orthogonality 245, 317

592

INDEX

pack and stow effect 393ff.

Papert, S. 316

parallel, parallelism

12, 213, 219, 227ff.,

234ff., 245, 303, 368ff.

parallelogram 303, 354ff., 445ff.
parameter 500, 507, 508ff., 546ff.

parameter,
intensive/extensive 203ff.
parametrisation 509ff., 544ff., 565ff.
Pascal 80ff.
Pasch 224
passage 270ff.
path 181, 412, 506ff.

: right-left 286, 288ff.

: right turn-left turn 287, 418ff.
, temporal 416
polygon(s) 16, 217, 313ff.
polyhedra 299ff.

positional systems 118ff., 132, 177

pour and mould transformations 394ff.
precision 247ff.
predicate 39ff.
pretty little girls schools 471
pretzel 271
prick 270ff.

paving 369

primes 540
prism 299ff.

Peano 73, 250, 257

Peloponnesos 391

probability field 41ff.

process 496ff.

Pensec, J. J. 542 footnote

permutation(s) 496ff., 514ff., 526

production/reproduction 296ff.

of composing parts 19
perspective 225, 233, 242ff., 311, 543
phainomenon 28
phenomenology 10, 25ff., 279, 541
, didactical 10, 28ff., 178, 279
, genetic 10, 28ff.
, historical 28
photogrammetry 282
Piaget 21, 22, 47 footnote, 77, 83ff., 90,
190, 221, 223, 225, 228, 232ff.,
238, 256, 279, 288, 297, 305,
325, 361, 396ff., 412ff., 533, 540
pictograms 463

progressive formalisation 470

projection 155, 244, 342ff.

projective, projectivity

214, 219, 225,

231ff., 342, 349

projective plane 263, 300
proportion, proportionality 134, 178ff.,
205, 412, 518, 536ff., 541

continual 571
proportionality tables 202
propositional logic 39ff.
protractor 328, 365ff., 367
Ptolemaeus 518
punctuation 470ff.
pyramid 232, 334, 402

place holder 493

Planck 402
planes 297ff.
plank model 110
Plato 391, 402ff.
: ideas 247
playing ground 291

plotting 575ff.
Poincar 265
points 265, 268
polarities 284ff., 415ff.

: above-below 284
, circular, cyclic 416ff.
: front-back 286
: head-tail 285
: here-there 288
: inside-outside 286
, linear 416ff.
of direction 285

of side 285
: plus-minus 416, 551, 554
: positive-negative 416, 551, 554

quantity 135, 395ff.

quantifiers, quantors 256, 474ff.
race car 575

rack and pinion 328

rain clouds 190
ratio 134, 148, 150, 152ff., 178ff., 212,
282, 337, 359, 433, 518ff., 536ff.,
547
: internal/external 183ff., 537
ratio preserving 189ff., 203ff., 368
, criteria for 202
rational multiple of lengths 7
rational number 133, 149, 157ff.
reading graphs 576ff.

ready knowledge 467

real multiple of lengths 8
reciprocity of change 292ff., 309ff.
rectangle 228, 245, 337, 354ff., 547

rectangle model 60ff., 81, 110

for algorithmic multiplication 129ff.

593

INDEX

rectilinearity 213ff., 227ff., 245, 264ff.,

298ff., 302, 304, 310, 349, 352ff.,
412, 451ff., 455
red currants 56

short memory 121 footnote

shortest line 302, 321, 352ff.

similar, similarity 13, 14, 188ff., 212, 219,

227ff., 232ff., 245ff., 337ff.,
358ff., 370ff., 401ff., 544, 547

reduction 400ff., 481ff.

reference set 39 ff.
reflection(s) 344ff., 424ff.
regular bodies 322
relations 497ff.
relatively 193ff., 209, 536
retention of insight 121

reversibility 12, 21ff.

reproduction 238ff., 543

revolution surfaces 322

rhombus 245, 545ff.
rich didactical sequence 161
Reimann surfaces 525
right angle 227

rigid, rigidity 13, 17, 22, 211, 227ff., 401

rigid congruently and similarly reproducible
bodies 227ff., 245ff.
ring 36

roads model 63
root symbols 524
rotation(s) 230, 343, 347ff.

rotation axis 304, 348

roulettes 152
Royaumont 90
rule of three 203ff.

Ruffini 526
Russell-Whitehead 76ff.
scale(s)

200, 350ff., 359, 363, 544, 549,

575ff.
Schubert, H. H. C. 75ff.
screws 425ff.
sections of trees 247

sector diagrams 152, 196, 198

sequence 496ff., 507ff., 527
arithmetical/geometrical 570ff.

set, sets 34ff.

algebra 39ff.
, power 35, 39
, product 34
, solution 36
theory 39
sexagesimal 91, 367
sextant 361
shadow 198
shake transformation 84
shape 13
shearing 376ff., 394

sheaves 221

sine 518, 554

sinus 518
size 13
skew lines pair 305, 311, 371
slide rule 202, 570
slit 270ff.
smooth graph 575

Snellius 350
Socrates 32, 224
solid geometry 312

solving 489, 524ff.

sources of insight 209
space 223ff., 231, 281ff.
, affine 219, 225, 233ff.
, euclidean 218, 221, 224ff.
, homogeneous 220
, metric 34, 216, 218

, n-dimensional cartesian 411, 446

, projective 219, 221, 231ff.
, topological 217, 219, 221, 233, 267
, vector 445ff.
speed 181
sphere(s) 13, 227ff., 319ff., 402
spinners 152
spirals 252ff.
spring balance 190
square 228, 238, 245
stamp collection 52

standing plane 417

standpoints 290ff., 309, 407ff., 412ff.
Stevin 172
stigma 334ff.
stochastic variable 41
straight ahead lines 302, 321, 352ff.
straight line 233, 238, 264ff., 303ff.

strategies 487ff., 543

structures 34, 53ff., 58, 210ff.

, combinatoric 211
, elementary 44

, elimination of 48ff.
, equipartition 62ff.
, geometric 210ff.
, order 45, 215
, power set 66ff.
product 60ff.
rectangular 60ff., 81

topological 219
substitution 482ff., 488, 522ff., 566

594

INDEX

substratum set 34, 53ff.

subtraction 102, 106, 170
surface(s) 13, 262ff., 506ff.
surveying 282, 408
symbolic 17, 20, 30, 33, 238ff.
symmetric group 514ff.
symmetry 227ff., 245, 298, 341ff., 355ff.,
490, 543
symmetry axis 304, 316
syntactical 471ff.
system constraint 318
Szeminska, A. 225 footnote, 361 footnote,
412 footnote, 540 footnote
tables 411, 416, 516, 555
tactics 487ff.
tangent (tan), 518
teaching tables 122
territories 268ff., 293

by graphs 573
by tables 573
verbal 573
transport 296
tree model 64
triangle 245ff., 314, 402ff., 518, 547
trigonometric, trigonometry 358ff., 518,
553
truth value 482
tunnel 270ff.
turtle 316

ultrafilter 37
uniform motion 181ff., 412, 553, 567ff.
union 39ff., 59, 96
unknowns 476ff., 557
Urysohn, P. 265
Utrecht 232

tesselations 256

van Hiele, P. M. 445, 448

tetrahedron 218, 268, 314, 334, 428

Thales 356
Theaitetos 224 footnote
Theodoros 224 footnote
tiles 248, 268
time (axis) 181ff., 281ff., 412, 517, 548
time-distance graphs 552
time-path graphs 552ff.
times 136ff.
with fractions 136ff.
in cyclic processes 137
toothed bar 328
topographical, topography 244, 250, 281ff.,
299, 325, 361, 371, 407ff., 415ff.,
427, 430, 537
topological, topology 35, 190, 214ff., 217,
219, 232ff., 238, 246, 250ff.,
258, 265ff., 349ff., 526, 527
topology
algebraic/set theory 267
torsi 298
torus 263, 271, 322
tower 207
traffic signs 463
transformations
496ff.; see also breakmake, contact, flexions, functions, immersion, mappings, pour
and mould, shake
transitivity 46, 55, 59, 200, 355ff.
translation (geometry) 230ff., 342ff., 347ff.
translation (algebraic) 489
by formulae 573

variability 528ff.
, qualitative 529ff.
variable(s) 43ff., 473ff., 491ff., 506ff.,
522ff., 528, 557ff.

see also binding

variable objects 492ff.
variation calculus 526
vector 303
vector addition 104
vector space 445
velocity 552ff.
Venn diagrams 39ff., 48ff., 54, 59, 70, 93,
99ff., 239, 318ff.
Venus 291
vernacular 472ff.
vertical 238, 401
vicious circle 558
Vieta 487, 520
viewpoints 309
vision line 233, 304ff., 352ff., 409, 420,
543
visualising 128ff., 151ff., 160, 192, 196,
198, 204ff., 345, 411, 441ff.,
537, 543, 548ff., 551ff., 554,
573ff.
volume 35, 266ff., 372ff., 392ff., 400ff.

Wallace, J. G. 399 footnote

weak force 431
weaving 277
whole/part 141ff.
Winter, H. 479 footnote

INDEX

Wiskobas curriculum 129

word problems 464
world 43

of boxes 228ff.
writing/drawing 239ff.

Wursten 237

yardstick 24, 25

Zorns lemma 34, 37

595