Elementary

Vector gira
A.M.Macheath —

Second edition =
This book, written with geometrical
applications in view, includes an intro-
duction to co-ordinate geometry of
planes and lines and to spherical trigo-
nometry. The rules of vector algebra are
introduced geometrically, properties of
ordinary space being assumed without
proof. Co-ordinates and components are
introduced at once, vector and cartesian
forms of expression being studied side by
side. There are illustrations from two as
well as three dimensions, helping the
reader to fit vectors into the framework
of his previous knowledge. The second
edition has answers and hints on solutions
to the exercises.

Price
(IN UK ONLY)
12s 6d net
ELEMENTARY VECTOR ALGEBRA
 Digitized by the Internet Archive
in 2024

https://archive.org/details/elementaryvectorOO0O0amma_i4m9
 ELEMENTARY
VECTOR ALGEBRA
Bx
A. M. MACBEATH
PROFESSOR OF MATHEMATICS
UNIVERSITY OF BIRMINGHAM

SECOND EDITION

OXFORD UNIVERSITY PRESS

1966
Oxford University Press, Amen House, London E.C.4
GLASGOW NEW YORK TORONTO MELBOURNE WELLINGTON
BOMBAY CALCUTTA MADRAS KARACHI LAHORE DACCA
CAPE TOWN SALISBURY NAIROBI IBADAN ACORA
KUALA LUMPUR HONG KONG

© Oxford University Press, 1964

First Edition 1964

Second Edition 1966

Printed in Great Britain by Butler & Tanner Ltd,

Frome and London
 PREFACE
The aim of this little book is not to prepare the reader for any
specific examination, but to provide an introduction to vector
algebra for those who know nothing about it and wish to learn.
The approach is geometrical and non-axiomatic, intuitively
acceptable properties of three-dimensional euclidean space being
assumed without proof and sometimes even without explicit
statement. Apart from this intentional omission, I have tried to
make the treatment fairly thorough.
It is hoped that a good average pupil could master most of the
material without assistance from a teacher or a lecturer. Many of
the exercises are suitable for an average student who has studied
the text carefully, if he is willing to spend a bit of time and effort.
A few more difficult exercises have been included because of their
intrinsic interest.
The book contains no mechanics, because there are plenty of
good textbooks now available which deal with the subject from a
vectorial point of view. However, I have of course tried to present
the material in such a way that readers will be able to apply the
methods both in mechanics and in geometry. Algebraic ideas and
terms (like ‘distributive’ and ‘commutative’) are introduced so as
to prepare the student for a later study of abstract algebra, of
which vector algebra is a part.
The usefulness of the vector method in solid analytical geometry
still seems not to have been fully appreciated. It helps to develop
three-dimensional intuition and to unify algebraic and geometrical
concepts. Lecturing on the subject has led me to introduce the
terms ‘direction vector’ of a line and ‘normal vector’ of a plane,
instead of the direction cosines of other writers. This seems
to be a logical change, and I hope that it will be generally ac-
septed.
A fair amount of emphasis is laid on explicit work with co-
ordinates and components, because this seems to reassure students
who are unhappy with too much abstraction. It is a mistake to
magine that vectors can completely replace cartesian methods,
‘hough they greatly assist understanding, particularly in dealing
vith the plane and the straight line in solid coordinate geometry.
6 PREFACE

T have restricted myself to this easy part of the subject, where the
vector method is most clearly relevant.
The absolute value sign is used to denote the length of a vector
and x is chosen instead of r for a typical position vector. These
notations are not yet standard, but the first is usual in the theory
of higher dimensional space and the second is more consistent
with the use of 2, y, z for coordinates. Generally speaking, the aim
has been to give a firm basis for further work, so that the minimum
of ‘unlearning’ will be needed, whether the reader goes on to study
applied mathematics, abstract algebra or functional analysis.
A first draft of parts of the book was circulated in cyclostyled
form to students at Queen’s College, Dundee. I should like to
thank some of them for helpful comments. I should also like to
thank Dr. Iain Adamson and Mr. Henry Jack for several con-
structive criticisms.
A. M.M.
1963
 CONTENTS
1. MATHEMATICAL NOTATION

2. VECTOR ADDITION
. Introduction
. Vectors
. The sum of two vectors
. Notes on Definition 2. The commutative law
. Use of brackets. Associative law
. The sum of a number of vectors
. Subtraction of vectors
. Geometrical illustration
. Multiplication of a vector by a number
Doar
oman
De
|

. The distributive laws

. Parallel vectors
. Position vector
. Point dividing a segment PQ in a given ratio
ell=.
enBm
alDD
CO
© Centroid of a triangle
Gp Worked examples
Exercises on Chapter 2

3. PROJECTIONS, COMPONENTS, COORDINATES

i Introduction
. Unit vector
. Orthogonal decomposition of a vector. Projections
. Projections of sums and multiples of vectors
. Components of a vector
. Coordinates of a point
b
&
P
Ol. Formulae
“Im for vector operations in terms of components
(i) Addition
(ii) Number-vector multiplication
(iii) Length of a vector
. Application of the formulae
. Components and coordinates in plane geometry
10. The use of vectors in plane geometry
 CONTENTS

LG Change of coordinate system in the plane 45

12. Worked example
Exercises on Chapter 3

4, THE SCALAR PRODUCT

. Introduction 50
. Definition of the scalar product
. Immediate consequences of the definition
. The distributive laws 51
. Formula in terms of components 52
. Consequences of the formulae 52
WON
r
Ano. Geometrical illustrations
. Angle between two vectors. Numerical examples
2) . Trigonometrical application
(oie

10. General remarks

Exercises on Chapter 4

5. THE VECTOR PRODUCT

. Introduction
. Left- and right-handed system of three vectors
. The vector product
. Rotation of vectors in a plane
Vector product with a fixed vector
The distributive laws
e
WW
or
AA . The formula in terms of components
. Worked examples
. Cross-multiplication
. The vector triple product
ee= . The scalar triple product
Oo
©
O

12. The use of cross-multiplication in elimination problems

Exercises on Chapter 5

6. COORDINATE GEOMETRY OF THE PLANE

AND STRAIGHT LINE
Introduction
Locus defined by a single explicit equation
Basic assumptions about the plane. Normal vector
ee Equation of the plane
ee
 CONTENTS
5. Examples on the equation of the plane
6. The angle between two planes
7. Perpendicular distances
8. Loci in space
9. Loci defined by two explicit equations
10. Parametric equations of a curve
11. Direction vector of a line
12. Parametric equations of a line
13. Non-uniqueness of the equations
14. Line as the intersection of two planes
15. Angle between two lines. Angle between line and plane
Exercises 6.1
16. Planes containing a line of intersection
17. Planes containing a line of intersection (theory)
18. Problems
19. The common perpendicular of two skew lines 101
Exercises 6.2 104

7. OTHER APPLICATIONS 107

1. Change of coordinate system 107
(i) Alteration of origin 107
(ii) Change of base 107
2. Products of four or more vectors 109
3. Geometry on the surface of a sphere. Spherical triangles 110
4. The cosine rule 112
5. Worked example 114
6. The sine rule 114
7. The polar spherical triangle 116
8. The area of a spherical triangle 1 ay
9. The tetrahedron 119
10. The cube 123

SOLUTIONS TO THE EXERCISES 126

INDEX 137
 i > Pee

i = —

ed

” . a a.
i! = wy ait Aba 2cece Pal
y ‘te vod? i bbe + LS
' iy A iit» BY pa fein nines
waren ee
aheN | vig, ley te Patyigi, wrt £ i,
| ——
a

NOP featead ad TAR waht He.

j
( aoeak gh IL
Dae! t eeutloryen BT.
mer ah . hit watil | id RTS re ihe = at
oe mm . ) Wo baa Pell) sail « golsietged esgall Ti *
. tA , : apreidar ft a
ut | i ; ea ' Seat) a hed tell (dor) ; imentnas al? aM

a | .
oes
£0 exelent
1 . BROTPATIIGA BE HTC
4 1) or ET <r Sa pinch) a ay
oe
frgity Ts Odeo 7
heat Yo aycmih> (i) “Fei éi
enw Me tad tis fovbort =<=?
) ngcoome aaalh ps net 7
hin il a 7
~ ah iste“ rd aor
e108
es Manele iain
gen pveudips o We ag
—< a ele
. _ - M
 1

MATHEMATICAL NOTATION
To understand this book one must be willing to alter one’s ideas
about the use of symbols in mathematics. Hitherto, for most
readers, symbols have denoted only numbers or points, but now
they will be used to denote other things as well. This extension of
their use is desirable because symbols help to achieve economy
and precision. It is easy to forget how much one gains by their use.
For instance, a simple formula like ‘a?—b? = (a—b)(a+b)’
becomes much more complicated when stated in words: “The dif-
ference of the squares of two numbers is the product of their
difference and their sum.’ The verbal statement is longer and less
precise, for it is not clear whether the difference referred to is
a—b or b—a; and if a more complicated formula had been chosen,
it would very probably have been impossible to put it adequately
into words at all. A wider use of symbols to include other things
besides points and numbers can be expected to lead to further
simplifications.
The algebra of vectors, with which we deal in this book, arose
naturally out of the attempt to apply symbolic methods in certain
parts of geometry and mechanics. A vector is a combined number-
direction, not merely a magnitude but a magnitude associated with
a direction in space. For instance, the velocity of a moving particle
is not specified simply by answering the question how fast the
particle is travelling, but one must also know in what direction it is
moving. Thus a force is required to keep a body moving uni-
formly in a circular path. Though the speed is constant, the
direction of motion does not remain the same, so the velocity is
continually changing. A force, tending to move a body from its
steady uniform motion, can also be represented by a vector. To
understand the action of the force one must know not only how
strongly it is tending to move the body, but one must know too
in what direction it is tending to move it. As well as these applica-
tions in physics, vectors have an application in geometry which
will be explained in the next chapter and which forms the main
topic of this book.
12 MATHEMATICAL NOTATION Ch. 1

In addition to symbols which denote objects, there are symbols,

such as +, —, Xx, ., which are used to express various processes
or operations performed on the objects. Thus a+b denotes the
result of the operation of adding the two numbers a, b. Some opera-
tions may be expressed without using an operation symbol at all,
like the product ab or the line PQ. Brackets are another important -
type of symbol. They do not themselves denote operations, but
indicate the order in which the operations are to be carried out.
Thus (a xb)-+c denotes the result of two operations: ‘Multiply
the number a by the number 6 and add ¢ to the result.’ The ex-
pression a x (b-++c) is the same as the other except for the position
of the brackets, yet it means something quite different: ‘Add
btocand multiply a by the result.’ In the first instance the brackets
would normally be left out because, by convention, multiplication
takes priority over addition if there are no brackets. However, if
there are two or more symbols of operation in the same expression,
brackets are necessary unless some special reason allows them to
be left out. The role of brackets in vector algebra, and indeed in
the whole of mathematics, is essentially the same as in number
algebra, and the rules governing their use are the same.
Ordinary usage assigns a meaning to the symbols +, —, x, .,
when these are associated with numbers. If a, b are vectors, how-
ever, no meaning has been attached to the expressions a+b, a xb,
a.b, and we are free to give them any meaning we please. The
meaning that is always given to the + sign in vector algebra is
connected with the applications in physics. The well-known
‘triangle of forces’ and ‘triangle of velocities’ express the fact that
forces and velocities combine according to the law of vector
addition.
It might seem better to use a completely new symbol for vector
addition and so avoid confusion with the addition of numbers.
However, if different symbols were invented for every possible
operation, the total number of symbols would have to be very
large. One reason for using the + sign for the composition of
vectors is simply the need for economy in the number of symbols.
Another reason will be given in a moment.
We all know that if a and b are numbers, then a+b = b+a. In
the next chapter a similar result will be found for vectors: if a and
b are two vectors, then a+b = b-+a. Why does this require dis-
cussion? Is it not obvious? To understand the answer, one must
Ch. 1 MATHEMATICAL NOTATION 13

remember that there is, a priori, no logical connexion between the

+ of vector algebra and the + of number-algebra. Vector addi-
tion is another use of the symbol +, and the two kinds of addition
are different operations with the same name. If, knowing that
a+b = b-+a for numbers, we conclude that the same is true for
vectors, our logic is hopelessly faulty. It is as if there were two
people called Smith, and we conclude that, because one of them
has fair hair, the other must have fair hair also. To discover the
colour of one person’s hair, we do not look at somebody else’s,
even if their names are the same. Similarly, if we want to know
how vectors behave when they are added, it is no use considering
numbers, with their quite different addition. We have to examine
vector addition separately, using its own geometrical definition.
It turns out that vector addition does indeed satisfy many of
the formal laws of operation satisfied by number addition. This,
as well as economy in the use of symbols, is another reason why
the + sign is used to represent it. Once one has proved the laws,
habits acquired in number-algebra can usefully be transferred to
vector algebra. However, even after the laws have been estab-
lished, one should avoid the casual, almost mechanical attitude
that many people seem to acquire in the study of elementary
algebra. Every identity in vector algebra is equivalent to a geo-
metrical theorem, sometimes quite a complicated one. For in-
stance, the*simple identity a+b = b-+a is equivalent to the
theorem that the opposite sides of a parallelogram are equal.
As well as vector addition, three different kinds of vector multi-
plication will be defined later. One of these combines numbers with
vectors, the other two combine vectors with one ancther. In ordin-
ary algebra there are three ways of writing down the product of
two numbers a and 6, namely ab, a.b, and a xb. These notations,
taken over into vector algebra, denote three different kinds of
multiplication: ka for the product of the number k and the vector
a, a.b for one kind of product of two vectors, and a xb for the
other kind of product of two vectors. These operation-symbols,
interchangeable in number-algebra, have different meanings in
vector algebra, and it is important not to confuse them. In this and
in several other ways the analogy between number and vector
multiplication is not so close as the analogy between the additions.
Other points of difference will be noted in later chapters.
Though we shall refer to some of our discussions as ‘proofs’, no
14 MATHEMATICAL NOTATION Ch. 1

attempt is made to set up a rigorous deductive structure. For us

‘proof’ will mean a deduction from intuitively clear properties of
the space in which we live. Mathematical proof, in the strict sense,
means deduction from a number of basic assumptions, called
axioms, which should be clearly and precisely stated. It is par-
ticularly difficult to give an axiomatic treatment of pure euclidean
geometry of three dimensions, and such a discussion would be out
of keeping with the rest of the material in this book. The reader
who is interested in an axiomatic treatment of solid geometry
should consult, for instance, The Foundations of Geometry, by
(P+Q)+R,

Q
FIG. 1 ‘Addition of points’ in the plane.

G. de B. Robinson (Toronto, 1940) and the references given on

p. 162 of that book.
One way to become convinced of the need for thorough dis-
cussion of the rules in a new kind of algebra is to consider an
algebra in which the rules break down. As an example we define
an ‘addition of points’ in the plane which has some unusual pro-
perties. It is not, of course, a standard definition, but is devised
merely as an illustration. Let capital letters P, Q, R, ... denote
points of the plane. Let P+-Q denote the third vertex of the equi-
lateral triangle on PQ as base which is on the left side of PQ to an
observer looking along PQ from P towards Q. With this definition
the points P+Q, Q+-P will be different, being the vertices of the
two equilateral triangles on opposite sides of the base PQ. Also, if
Ch. 1 MATHEMATICAL NOTATION 15

P, Q, Rare three distinct points, the points P+(Q-+R), (P+Q)+R

are different, as illustrated in Fig. 1. This means that the ex-
pression P+Q-+R is ambiguous, having two possible meanings
which must be distinguished by the use of brackets. Though the
addition just defined does not satisfy the laws satisfied by number
addition, it satisfies other identities which number addition does
not. For instance, the following identities hold.
(ee tee ee (PQ) )) yy) =O).
P+(P+(P+Q) ) = (Q+P)+(P+Q).
IfP+Q=R, then Q+R =P.
(P+Q)+R = (P+(Q+R) )+Q.
Of course there is no limit to the new algebras one might in-
vent, and the one I have just described to illustrate the failure of
the rules is of no importance, mathematical or otherwise. Vector
algebra, on the other hand, though at first sight it may seem arti-
ficial, is important and fundamental. The reader will have to take
this on trust at first because a:.certain amount of groundwork must
be done before really interesting applications are possible. After a
time, however, it will become obvious that vector algebra is the
natural method for tackling problems in solid geometry. This view
is not contradicted by the fact that many of these problems were
done originally without vectors, since mechanics and solid geo-
metry were well developed before vectors were invented. Problems
solved for the first time in mathematics are usually done by a
rather difficult method, the solution then being simplified by later
work. The Greeks and Romans could, with difficulty, add and
multiply numbers without having Arabic numerals, but this is no
argument against using Arabic numerals today.
Once a few chapters have been mastered, the reader should take
an older textbook on solid coordinate geometry and try to re-
write as many as possible of the equations in vector form. He will
soon be convinced of the value of the vector notation, in which one
simple equation can often replace three complicated numerical
equations. It is quite possible to find oneself becoming impatient
with the writer of such a book because he does not use vectors,
though of course many of the equations and formulae were found
long before vectors were invented. It is perhaps best to think of
vectors as a suitable notation for certain mathematical problems,
so suitable that no one who knew about it would dream of tackling
16 MATHEMATICAL NOTATION Ch. 1

the problems in any other way; just as no one who had to multiply
two large numbers together would think of using any notation
except Arabic numerals, though Roman numerals may do very
well on the dial of a clock.
Another unexpected advantage of the vector method is the way
in which it links up with other parts of mathematics. For instance,
the vector product, as we shall see in Chapter 5, has a close con-
nexion with the method of ‘cross-multiplication’ in dealing with
linear homogeneous equations, and the scalar triple product,
written out in terms of components, is nothing but a three by
three determinant. Some parts of plane geometry and trigono-
metry, too, include formulae which are easily recognizable as
special cases of vector equations, for instance Pythagoras’s
Theorem and the formula for the cosine of the difference of two
angles. However, the real power of vectors becomes clear in three
dimensions, as will be seen in Chapters 6 and 7 where we deal with
solid coordinate geometry and spherical trigonometry. Time spent
on the study of vector algebra is thus time well spent, because it
helps to understand material which used to be thought to belong
to several different branches of mathematics.
 2

VECTOR ADDITION
1. Introduction
This chapter begins by explaining what vectors are and des-
cribing the triangle rule for vector addition. Historically, vector
addition was devised in order to deal with the composition and
resolution of forces and velocities. The triangle rule, or, more cor-
rectly, the triangle definition, was dictated by the triangle of
velocities and the triangle of forces in mechanics. Later it became
clear that vector algebra was helpful in many geometrical problems
and, later still, more general vector spaces (outside the scope of
this book) were invented and applied to the theory of linear equa-
tions and other parts of mathematics. Nowadays the concept of
vector addition is a basic part of the equipment of any pure or
applied mathematician.
Most of this first chapter and some parts of the following chap-
ters are necessarily concerned with spade work, and the real value
of the method will not be clear to the reader until later. However,
a few illustrations are included to indicate how vector algebra can
throw fresh light on some known geometrical situations. There are
no applications to mechanics in this book, but the reader can
easily find out about such applications from any modern textbook
on mechanics, since vectors are now a standard method in this
field.

2. Vectors
Definition 1. A non-zero vector is a combination of three things:
(1) a positive number called its magnitude or length,
(2) a direction in space,
(3) a sense (making more precise the idea of direction).
The direction may be described by specifying that the vector is
parallel to a given line, and the sense by specifying one of the two
possible directions of travel along the line. The exceptional zero
vector, which will always be denoted by 0, has zero magnitude, but
its direction and sense are not defined. Symbols in bold type (a,
18 VECTOR ADDITION Ch.2

b, ...) will be used to denote vectors. The magnitude of the vector

a will be denoted by |a |.
Two vectors are defined to be equal if they have the same magni-
tude, direction and sense. It is most important to understand that
two vectors are not equal unless they are equal in all three re-
spects—magnitude, direction and sense. Vectors should never be
called equal if they are equal in magnitude only.
Let P and Q be two points of space. Then the line segment join-
ing P to Q represents a vector whose magnitude is the distance
between P and Q, whose direction is that of the line PQ and whose
sense is the direction of travel from P to Q. The point P will be
called the initial point and the point Q the endpoint of the segment.
The vector which is represented by the segment with initial point
—> —_> —
P and endpoint Q will be denoted by PQ. The vectors PQ and QP
have the same magnitude and direction but opposite sense. Any
‘degenerate’ line-segment PP with the same initial and endpoint
may be taken to represent the zero vector 0.
Since two vectors with the same magnitude, direction and sense
are equal, it follows that the same vector can be represented by
many different line-segments. For example, if ABCD is a parallelo-
—> —
gram, the vectors AB and DC are equal. On the other hand, if
ABC is an equilateral triangle, the vectors BC, CA, AB are all
different, since they differ in direction.

3. The sum of two vectors

The simplest of the vector operations is that of addition.
Because of a formal analogy with the addition of numbers the
same sign + is used to represent it.

Definition 2. Let a, b be two vectors, eer eran we line seg-

O a+b Q ments OP = a, PQ =b, SO
"that the endpoint of the
segment representing a is the
initial point of the segment
representing b. Then the
vector a+b is defined to be
—
FIG. 2 “Vector addition. OQ.
§ 3 VECTOR ADDITION 19

Definition 2 is often called the triangle rule, for if OP, PQ are not
parallel, the vectors a, b, a+b are represented by the three sides
of the triangle OPQ (Fig. 2). However, the definition still has a
meaning if O, P, Q do not form a triangle in the usual sense, if O,
P, Q lie on a line, for instance, or when one of the vectors con-
cerned is the zero vector and two of the points O, P, Q coincide.
These possibilities require study.

Collinearity. Suppose that a and b have the same direction and

sense, so that P lies between
O and Q on the line 0Q a+b
(Fig. 3). Then the vector ; =
——S
OQ =a-+b has the same 0% 2 P b Q
direction and sense as a and Pipe
b, and its magnitude is the sum of their magnitudes;
Jat+b| =|a|+|b]. (1)
This relation holds only if a, b have the same direction and sense.
In general, the sum of the lengths of two sides of a triangle is
greater than the length of the third side and we have the in-
equality
|a+b| <]a|+ |b].
Coincidence. (i) If P and Q coincide in Definition 2, the vector
> =>
b = PQ = PP is the zero vector 0. The triangle rule then gives
a+0 =a. (2)
This fits very nicely into the analogy with number-algebra.
(ii) If the point P coincides with the point O, then a = @, and
we derive
0+b = b.
Finally, if O coincides with Q, we find that
> >
OP+P0 = 0.
Thus the sum of two vectors which are equal in magnitude and
direction but opposite in sense is the zero vector. The analogy
with number-algebra then suggests the following definition.

Definition 3. The vector —a, called the negative of a, is the

vector which has the same magnitude and direction as a, but
opposite sense.
20 VECTOR ADDITION Ch. 2

With this notation, the last result can be written:

a+(—a) = 0. (3)
Note also that —(—a) =a. (4)

4. Notes on Definition 2. The commutative law

=>
The vectors a, b may be represented as sides OP, PQ of many
different triangles OPQ. It might appear, therefore, that Definition
2 does not lead to a unique sum. However, all such triangles OPQ
are congruent and similarly placed; so the different line-segments
0Q will all represent the same vector, and the sum is, in fact,
unique.
The vectors a and b do not enter symmetrically into the defini-
tion, since the initial point of b coincides with the endpoint of a.
To obtain the sum b-+-a, complete the parallelogram OPQR. Then
— —> —
OR = b, RQ = a, so from the triangle ORQ (Fig. 4), b-+-a = OQ.
This proves the identity, known as the commutative law,
a+b =b-a (5)

FIG. 4 a+b=bta

Exercises

1. Show that, if a, b have the same direction but opposite sense and
|a| >| b|, then a+b has the same direction and sense as a. What is
the magnitude of a+b in this case?
2. The proof given of the commutative identity (5) is valid only
when the directions of a and b are different. Consider separately the
cases when
(i) a, b have the same direction and sense,
 §5 VECTOR ADDITION 21
| (ii) a, b have the same direction and opposite senses,
(ili) b is the zero vector.

5. Use of brackets. Associative law

Brackets are used in vector algebra with the same meaning as in
number-algebra: the expression inside a bracket is to be treated
as a single symbol for the purpose of combining with anything
outside the bracket. Thus the expression
(a+b)+e
denotes the result of the following sequence of operations:
(1) evaluate the sum a--b, obtaining a vector d,
(2) evaluate d-+-c.
On the other hand, the expression
a+(b+c)
is evaluated as follows:
(1) evaluate b-+-c to obtain e, say,
(2) evaluate a+e.
It does not seem at all clear that these two sequences of opera-
tions should lead to the same answer; but we shall now prove that
in fact they always do. That is,
a+(b-+c) = (a+b)+ec. (6)
This result is known as the associative law for vector addition.

A ls) 4a C

a+b
B
FIG. 5 a+(b+c) = (a+b)+c

Proof (Fig. oy oethe ee a, b, c be represented by the

/line-em OA, AB, BC peepee yee|Erom Peston 1,

atb = OA+AB = OB, so (a+b)+c = OB -+BC = OC. Again

22 VECTOR ADDITION Ch. 2

by Definition 1,
> Sl es sa
b+c = AB+BC = AC, so a+(b+c) = OA+AC = OC.
Thus the processes indicated by the left and right sides of the
>
equation both lead to the same result, the vector OC.
6. The sum of a number of vectors
The associative law allows us to omit brackets without am-
biguity in a sum of three vectors, the value common to the left
and right sides of equation (6) being written, simply a+-b-+-c. This
result can be extended to a sum of any finite number of vectors.
The expression
Al Ag cn ean
have a value independent of the order in which the additions are
performed.
The sum of » vectors may also be obtained directly by con-
structing a diagram of line-segments in which the endpoint of each
segment is the initial point of the next one. The sum is represented
by the segment joining the first initial point to the last endpoint.
For instance, if m = 5 and the vectors a, b, c, d, e are represented
es
by segments OA, AB, BC, CD, DE, then the sum a+b+c+d+e
——Ss
is represented by the segment OE (Fig. 6).

FIG. 6 The sum of a number of vectors.

7. Subtraction of vectors
Having defined the negative —a of a vector a, we define sub-
traction as adding the negative, that is, b—a = b+(—a).
Subtraction is then the operation inverse to addition (as in
number-algebra) and x = b—a is the unique solution of the
 § 8 VECTOR ADDITION 23

vector equation x-+-a = b; for the triangle OPQ with OP =X,

— —>
PQ = a, OQ = b, represents equally the equation
> —> —> — => —->
x+a = b(OP+PQ = OQ) and x = b+(—a)(OP = 0Q+QP)
(Fig. 7).

O b=xta

HIGa 7
We can therefore take a term from one side of a vector equation
to the other, changing its sign, for this is the same thing as sub-
tracting the term from both sides. Similarly we can change all the
signs in a vector equation. For example, from the equation
a+b =c,
we deduce by adding —c—b~—a to both sides, that
—c = —a—b.
8. Geometrical illustration
It is well known that, if all the sides of a parallelogram have
equal length, that is, if the parallelogram is a rhombus, then its
diagonals are perpendicular. (We cannot give a vector proof of
this result at present, but see
Chapter 4, p. 52.) If ais the
vector represented by one
side of a parallelogram, and
b is represented by an ad-
jacent side, then the diagonals
(with suitable choice of sense)
represent a+b and a—b.
FIG. 8
Thus the perpendicularity of
the diagonals of a rhombus can be indicated, in vector language,
as follows: ‘If |a| =|b|, then a+b is at right angles to a—b
(Fig. 8).
Now let ABC be a triangle and let O be the centre of the circle
, > > >
through A, B and C, so that the vectors OA, OB, OC are all equal
24 VECTOR ADDITION Ch.2

in length. Denote these vectors, for shortness, by a, b and c. Let

ae
H be the point defined by the vector equation OH = a+b-+e.
Then from the triangle COH,
— > —>
CH = CO+0OH = —c+a-+b-+c =a-+bD.
> => —>
Again, BA = BO+0OA = —b-+a = a-—b.
But now, |a| =|b|, so by the result about the rhombus,
CH and BA are at right angles, that is H lies on the altitude
mae =P . . - .

through C. By symmetry, H lies also on the other two altitudes.

This proves again the well-known result that the three altitudes of

FIG. 9

a triangle meet at a point H (the orthocentre). In addition, we

> > Shc
have obtained the vector relation OH = OA+OB-+0C.
The proof just given of the orthocentre property depends only
on the associative law, enabling us to omit brackets in the sums,
the commutative law, concealed in the phrase ‘by symmetry’, and
the fact that the diagonals of a rhombus are perpendicular. The
proof is therefore equivalent to a simple geometrical proof, and
one might be tempted to think that the vector algebra is not
essential. In a sense it is not absolutely essential, but the geo-
metrical construction required if one is to do without vectors is
surprisingly complicated, as one can see from Fig. 9, where all the
parallelograms are drawn in. Once this construction has been
made, the geometrical proof is not difficult, and readers who feel
§ 9 VECTOR ADDITION 25
unsure of the vector algebra (quite a natural feeling at this stage)
may find that it helps to study this figure. They are bound to
agree, though, that the vector proof is better.

9. Multiplication of a vector by a number

Sometimes the same vector may occur twice or more often in
the same sum as in the expression (Fig. 10)
atata+b-+b.
It is natural to write this in the shorter form
3a+2b,
as in number algebra.

FIG. 10

Similarly (—3)a would naturally denote the sum —a—a—a.

We are thus led to the idea of multiplying a vector by a positive
or negative whole number.
Now consider the magnitude, direction and sense of the vectors
3a, 2a, ... , —a, (—2)a, ... Leaving aside the exceptional case
a = 0, we may consider a to be the vector represented by the
SS
segment OP, where O, P are distinct points. Continue the line
OP in both directions, and mark on this line a sequence of points
P,, P;, Py, ... spaced at equal distances | OP | beyond P, and
another sequence of points P_,, P_s, ... spaced at equal distances
Se tee P ;
|OP |beyond O. The segments PP,, P,P3, &c., equal in magni-
—_

tude, direction and sense to OP, all represent the same vector a.
26 VECTOR ADDITION Ch. 2

From the definition of vector addition,

—> > —> aa,
OP,, = OP+PP.+ ... +Pin-1Pim
=a-ta+... +a (m times)
= ma.

a a a

Pi P_, O P=P, P, P,
FIG. 11

Thus if m is a positive whole number, ma is the vector with the

same direction and sense as a, but with m times the magnitude.

Similarly, (—m)a, represented by the segment OP_.,,, is the vector

with the same direction as a, opposite sense and m times the
magnitude. This all suggests the following definition in which the
multipliers need not be whole numbers.

Definition 4a. Let a be a non-zero vector and k a non-zero num-

ber. The vector ka is defined by the following rules, giving, in
order, its magnitude, direction, and sense.
(1) |ka| =|] lal,
(2) Direction of ka = direction of a
(3) Sense of ka and a are the same if k is positive, opposite if
k is negative.
This definition does not allow the zero vector or the number
zero to occur in a product. The obvious value for such products is
the zero vector (for many reasons). Hence

Definition 4b. The product of any number and the zero vector
is the zero vector. The product of the number zero and any vector
is the zero vector. That is, 0a = k0 = 0.
Note. Definition 4 may seem a little cumbersome by comparison
with the simple idea of repeated addition of a vector to itself
mentioned at the beginning of this section. However, the simpler
idea is unsuitable as a definition because it only describes multi-
plication by a positive or negative integer, and it is convenient to
attach a meaning to expressions like 4/2a, or a, which cannot be
defined by successive addition.
 § 10 VECTOR ADDITION 27

10. The distributive laws

In this section we prove the following identities, known as
distributive laws. The last of these laws enables us to omit brackets
in the expression kla, which would otherwise be ambiguous. In
this it resembles the associative law (6).
(k+l)a = ka+la. (7)
k(a+b) = ka-+kb. (8)
k(la) = (kl)a. (9)
It will be noticed that the proof of (7) is split up to cover rather a
large number of particular cases, according to the sign and relative
magnitude of the numbers &k, 1. The situation, geometrically, is
that of three points A, B, C on a line, and there are six possible
orders, reading from left to right, in which they can be arranged
(neglecting the possibility that two of them might coincide).
Though it is rather tiresome to have to consider all these cases in
the proof, it is a great advantage of the distributive law that it
does, in fact, cover all cases, and once the law is established it can
be applied automatically, without any worry about drawing
diagrams of different cases (sometimes without drawing a diagram
at all). A good instance of this occurs in § 13 of this chapter, deal-
ing with the point dividing a segment in a given ratio, the same
formula covering points of subdivision both inside and outside the
segment.
Case (1) in the proof of (7), when & and / are both positive, is, in a
sense, the fundamental one, since the other cases are deduced
from it by taking terms across from one side to the other of equa-
tions. The reader may find it interesting to give direct proofs of
Cases 2 to 6, drawing a diagram for each case.

Proof of (7). Case (1). Suppose that k > 0,1 >0, a 40. The
vectors ka, Ja then have the same direction and sense—that of a—
so by formula (1), § 2, the magnitude of their sum is
|ka] + la] =|] [al + [@}lal = (+) /al.
Now &-+1 is a positive number, so the vector (&-+/)a has the same
direction and sense as a, and its magnitude is (k+1) |a|. Thus
the left- and right-hand expressions in (7) represent vectors with
’ the same magnitude, direction and sense, so they are equal.
Case (2). Suppose that k, J are both negative, say k = —p,
28 VECTOR ADDITION Ch. 2

1 = —q, where p and q are both positive. Then by Case (1),

(p-+q)a = pa+gqa. Changing signs across, we find (k-+-l)a = ka-+la.
Case (3). Suppose that & is positive, J negative, |k| > |1].
Let p = k+l, q = —I, 80 that p and q are positive and by Case (1)
(p+q)a = pa+gqa, or ka = (k+1)a—la. Taking Ja across to the
left side of the equation we obtain Formula (7).
Case (4). k negative, J positive. |k| >|1|. The result follows
from Case (3) by changing signs across.
Cases (5), (6). b <0; LeeOpel ki | 13. or: k > 08
|k| <|J1|. These follow from Cases 3, 4 by interchanging the
numbers k, J and using the commutative law.
Case (7). If k = 0 or 1 = 0, or k+1 = 0, or a = 0, the result
follows from Definition 4b and (2) or (3).
This completes the proof of (7).

Proof of (8). We only prove (8) in the case when k is a positive

number, because the result is clear when k = 0, and the case
k < 0 follows from k > 0 by changing signs across. Suppose that
——
i > 0, then. Let OPQ be a triangle with vector sides OP = a,
— >
PQ = b, so that OQ = a+b (Fig. 12). Let O’P’Q’ be a triangle
with sides parallel to the corresponding sides of OPQ and let

FIG. 12 Similar triangles and the distributive law.

—>
O'P’ = ka. Since the triangles OPQ, O’P’Q’ are similar, we have
P’Q’ = kb and O0’Q’ = kOQ = k(a+b). Hence
/ , f; ,

—> —> —
ka+kb = O’P’+P’Q’ = 0'Q’ = k(a+b).
Strictly speaking, this proof does not cover the case when a, b
have the same direction, and the reader should supply his own
proof in this case.
§ 11 VECTOR ADDITION 29

Proof of (9). If any of the three symbols k, J or a represents a

zero, formula (9) is true, since both sides of the equation represent
the zero vector. We assume, therefore, that none of them is zero.
The left- and right-hand sides then represent vectors with the
same magnitude, namely |k||/||a|, and the same direction,
that of |a | . The sense depends on the sign of the multiplier. Since
two changes of sense return a vector to its original sense and two
negatives multiply to give a positive, it is easy to verify that both
sides of the equation agree in sense too.

11. Parallel vectors

Non-zero vectors which have the same direction are said to be
parallel. It is clear from Definition 4 that if a is a non-zero vector
and k is a non-zero number, then the vector ka is always parallel
to the vector a. Conversely, again from Definition 4, any vector b
parallel to a can be expressed as a numerical multiple ka. The
multiplier k will be + |b|/| a], the plus sign being taken if a, b
have the same sense and the minus sign if they have opposite
senses. Thus we have the important result:
A vector b is parallel to a vector a if and only if there is a
number & ~ 0 such that b = ka.
This means that if b and a are given, and k is regarded as un-
known, the equation b = ka does not always have a solution.
There is no such thing as division of one vector by another. Num-
ber-vector multiplication is only the first of three kinds of multi-
plication to be defined in vector algebra, but they all have this in
common—no division of vectors, in the sense of an operation
inverse to multiplication, is possible.
Though division of one vector by another is not possible, it is
permissible to cancel a non-zero vector from both sides of an equa-
tion, thus: if ka = la, and a ~90, then k = l. This kind of can-
cellation is occasionally useful.
Also, if k is a number different from zero, it is sometimes con-
a 1 ;
venient to write z for the vector 72 It follows from (9) that this
vector is the unique solution x of the vector equation kx= a, so
‘the process may be regarded as a form of division of a vector by a
non-zero number.
30 VECTOR ADDITION Ch. 2

12. Position vector

Let O be a point fixed in space, called the origin. Let P be any
as
point. The vector OP is called the position vector of P, or, in the
case of ambiguity about the origin,
Q the position vector of P relative
q to O. If points P, Q have position
vectors p, q respectively, then
O 2G= oa PO =4q-p.
PQ (10)
-p aa

P For, from the triangle POQ, PQ

FIG. 13 = —-

13. Point dividing a segment PQ in a given ratio

Suppose that two distinct points P, Q are given. Let X be a
third point on the line PQ. The ratio of the lengths PX : XQ does
not determine the point X, for there are two points with the same
value of the ratio, one lying between P and Q, the other lying out-
> >
side the segment PQ. In the first case the vectors PX, XQ have
the same sense, while in the second case their senses are opposite.
It is convenient, therefore, to say that the ratio PX : XQ is posi-
tive if X lies between P and Q, but negative otherwise. With this
convention, one value of the ratio will not give more than one
point of the line. (There is no point X corresponding to the ratio
BX 2 -XQ ===.)
Let X, P, Q be three collinear points, with position vectors x,
P, q respectively. Suppose that PX : XQ = k:1, using the con-
vention of sign just indicated. Then we shall show that
__ kq+lp
GE a9 (11)

Proof. Using Definition 4, we have PX = 7x@, so, by (10),

I(x—p) = k(q—x). Hence (k+1)x = kq-+Ip, so

_ ka+p
[ae ie
as desired.
 § 15 VECTOR ADDITION 31

Corollary. If we take k =1=1, we find that the position

vector of the midpoint of PQ is 1(p-+q).

14, Centroid of a triangle

Let A, B, C be the vertices of a triangle, with position vectors
a, b, c respectively. The midpoints L, M, N of the sides BC, CA,
AB, in order, have position vectors, by (11)
1=3(b+c), m=(c+a), n = 4(a+b).
Consider the point G on AL such that AG: GL = 2: 1. By (11)
its position vector is #(a+2l) = }(a+b-+c). This formula is un-
altered if we change the order of the letters a, b, c, so G is also the
point on BM such that BG : GM = 2: 1 and the point on CN such
that CG: GN = 2:1. We have thus proved, by vector methods,
that the three medians of a triangle meet at a point, known as the
centroid; that this point is a point of trisection of each median;
and that its position vector is one-third of the vector sum of the
position vectors of its vertices.

15. Worked examples

In working out examples, which will often be formulated in
geometrical terms without any mention of vectors, the first step
is to choose an origin and name the position vectors of the various
points. It is usually best to take corresponding letters for a point
and its position vector, so that the point A, for example, will have
position vector a, and the vector AB will be b—a. There must be
some exceptions to this convenient rule, though, since the origin,
even if it is not labelled O, must always have position vector 0.
Our first example concerns the tetrahedron, the solid figure which
is the natural generalization of a triangle. It consists of four non-
coplanar points, A, B, C, D, the four triangles ABC, ABD, ACD,
BCD called its faces, and the six line-segments AB, AC, AD, BC,
BD, CD called its edges. Two edges which do not meet at a vertex
are called opposite edges. There are three pairs of opposite edges:
AB, CD; AC, BD; AD, BC.

Example 1. Show that the three line-segments joining the mid-

points of pairs of opposite edges of a tetrahedron bisect each other.
,

Solution. Let A, B, C, D have position vectors, in order, a, b,

 ADDITION Ch. 2
32 VECTOR
point and position vector
c, d. (This convention—same letter for
mention.) If L, M
_-will be used from now on without repeated
are the midpoints of AB, CD, then
= (a+b), = 3(c+d),
r
and the midpoint P of LM has position vecto
p = Ml-+-m) = Ha+b+c+4). ging the
Now this expression is symmetrical, unaltered by chan
also the midpoi nt of
order of the points A, B, C, D. Therefore P is
that of BD and also
the segment joining the midpoint of AC to

FIG. 14

the midpoint of the segment joining the midpoint of AD to that

of BC, and all three segments bisect each other as required.
The point P obtained above is called the centroid of the tetra-
hedron.

Example 2. ABCD is a parallelogram and E is the midpoint of

AB. Prove by vector methods that DE and AC trisect one another.
cast:
Solution. Choose A as origin, so that b = AB is the position
oes
vector of B and d = AD is the position vector of D. Since ABCD
—
is a parallelogram, c = AC = b-+d. By (11) the position vector
 § 15 VECTOR ADDITION 33

of E is $b. There are two points of trisection of DE and two of AC.

Working out their position vectors from (11), we find
trisecting AC: 4(b-+d), 2(b+d),
trisecting DE: %3d-++%b, 4b+4d.
Example 2 illustrates a method of showing that two given lines
cut in a certain ratio, by using formula (11) to show that the
points of subdivision coincide. To apply it in the form above it is
necessary to know, or to guess in advance, what the ratio of sub-
division is going to be. The calculation of the position vector of
the centroid in § 14 is similar in this respect. However, the next
example shows that this type of method can be modified, by using
unknown ratios which are later adjusted, so as to apply to prob-
lems where the ratio is not known in advance.

Example 3. Coplanar points A, B, A’, B’ have position vectors

a, b, ra, sb, in that order, where a, b are non-zero and not parallel.
Find the position vector of the point of intersection of the lines
AB, A’B’.

Solution. Let X be the desired point. Since X lies on the line

> —
AB, the vectors AX, AB are parallel, that is, there is a number k
> —
such that AX = kAB, or,
x—a = k(b—a). (i)
Similarly X lies on the line A’B’, so there is a number / such that
— >
AX PAB, Or,
x—ra = I(sb—ra). (ii)
Subtracting equation (ii) from equation (i) and simplifying, we
find
—((1—k)—r(1—l) )a = (k—sl)b. (iii)
If both sides of (iii) were non-zero, we should have a vector
parallel to a on the left equal to a vector parallel to b on the right,
,and that is impossible since a is not parallel to b. Thus both sides
of (iii) are zero vectors, and since neither a nor b is zero, the co-
efficients are zero, that is,
{(l1—k)—r(1—l)} = k—sl = 0.
EV A-—B
 ADDITION Ch. 2
34 VECTOR
= (s—r) lL. On substi-
Added together, these equations give (l—r)
get the answer
tuting the value (1—7)/ (s—r) for J in (2), we
. (s—1)ra+(1—r)sb
s—r
by zero is
This formula is meaningless when s = 7, since division
that case, because
impossible; but no answer can be expected in
then the lines are parallel.

Exercises on Chapter 2
ively a, b,
1. The position vectors of points A, B, C, D are respect
ES en
AC, AD, BC,
3a-+-b, —a+2b. Express in terms of a, b the vectors AB,
—> —> segments
BD, CD. Find also the position vectors of the midpoints of the
ACD, BCD.
AB, BC, CD, DA and the centroids of the triangles ABC,
of
2. ABCD is a parallelogram and a, b, c are the position vectors
A, B, C. What is the position vector of D?
t
3, Give an alternative proof that the position vector of the midpoin
of PQ is 3(p-+q) by constructing the parallel ogram POQR, O being
the origin.
4. If L, M, N are the midpoints of BC, CA, AB respectively and
O is any point, show that
ln lin
(i) OA+0B+0C = OL+0M-+ ON,
> OCU
(ii) AL+BM-+CN = 0.
5. ABC, A’B’C’ are two triangles and G, G’ their centroids. Prove
> >.lc —>
that AA’+BB’+CC’ = 3 GG’.
6. Six points A, B, C, D, E, F are given in space. P, Q, R, S are
centroids of the triangles ABC, ABD, DEF, CEF. Show that P, Q, R, 8S
are the vertices of a parallelogram.
7. A,B, C, D are four points not all in the same plane. P, 0, R, Ss,
U are the midpoints of AB, BC, CD, DA, AC, BD respectively. Prove
that PQRS, PTRU, QTSU are all parallelograms.
8. If a, b, c, d are position vectors of the vertices A, B, C, D of a
tetrahedron, show that the lines joining the vertices to the centroids of
the faces opposite them meet at the point G with position vector
Ex. 2 VECTOR ADDITION 35
3(a+b+c-+d). A’, B’, C’ are points on AD, BD, CD such that
AA’/A’D = BB’/B'D = CC’/C'D = 3,
Show that G is the centroid of A’B’C’.
9. If ABCD and A’B'C’'D’ are two parallelograms not necessarily in
the same plane, show that the midpoints of AA’, BB’, CC’, DD’ are also
vertices of a parallelogram.
10. If O, G, H are the circumcentre, centroid and orthocentre of a
triangle ABC, show that O, G, H are collinear and that G trisects OH.
(Use the formulae of § 8, § 14.)
11. Suppose that ABCDEF is a regular hexagon inscribed in a circle
— —>
centre O, radius r. Let AB = p, BC = q. Express in terms of p, q the
SS ee ee > —>
vectors CD, DE, EF, FA and also the vectors OA, OB, etc. Show that
the orthocentres of the triangles ABC, BCD, ... , FAB, also form a
regular hexagon and lie on a circle centre O and radius 2r.
12. Four non-coplanar points A, B, C, D lie at equal distances from
a point O. A point P is defined by the vector equation
> o> Oe arr lc
2 OP = OA+0B+0C+OD.
Show that the line joining P to the midpoint of any edge of the tetra-
hedron ABCD is at right angles to the opposite edge.
13. Suppose that O is the centre of a circle A,A,A, of unit radius in
a plane II. If the point B, is the point defined by the vector relation
—» => > ; :
OB, = OA,+0A,, show that B, is the centre of the other circle of unit
radius in I] which passes through A, and Ag. If, further, two other
circles of unit radius and centres B, and B, are drawn through A3, A;
and A,, A, respectively, prove that the three circles centres B,, B;, B;
a on
meet at C where OC = OA,+0A,+0A3.
 3

PROJECTIONS, COMPONENTS,
COORDINATES
1. Introduction
two per-
The advantages of using coordinates (distances from
points of a plane are well known.
pendicular axes) to label the
in solid geomet ry, but now three
Coordinates are just as useful
are needed instea d of two, and every
mutually perpendicular axes
that the
point has three coordinates. We shall see in this chapter
of space can be labelle d by means
vectors, as well as the points,
ents. The two ideas are related , for
of numbers called compon
point are the compo nents of its positi on
the coordinates of a
vector.
We shall find out in this chapter how to write down the com-
ponents of the sum of two vectors or of a numerical multiple of a
vector, and, in later chapters, when we define different kinds of
vector multiplication, we shall see how they can be expressed in
terms of components too. It will then be possible to translate every
vector equation into either one number equation or three number
equations—three, because each vector has three components.
Though the component form of equation is more cumbersome, it
is often needed in practical calculations, so one must be thoroughly
familiar with it. Historically, the component form is the original
form in which many of these equations first occurred, and in a
sense vector algébra developed as a shorthand method of writing
numerical equations. When the component form of some of these
relations is compared with the short and simple vector form, it
will be clear why vector notation has superseded the other when-
ever possible.
2. Unit vector
A vector u is called a unit vector if its length |u | is equal to 1.
If a is any non-zero vector, there are two unit vectors parallel to
a: a/|a| with the same sense, and —a/|a| with the opposite
sense. If u is a unit vector with a given direction, then any vector
s Vv NE
REAM REMAND, VUIMEUINEINEDS, CUUNRDINALES od

(unit or not) with that direction has the form ku, the number &
being equal, apart from sign, to the length of the vector.

3. Orthogonal decomposition of a vector. Projections.

—
Let u = OL be a unit vector, II the plane through O perpen-
: —
dicular to u and let a = OA be an arbitrary vector (Fig. 15). If P
is the foot of the perpendicular from A on II, then a = OP-+PA,
so any vector a can be expressed as the sum of two vectors, one
—> —
(PA) parallel to u and the other (OP) perpendicular to u. Since
any vector parallel to u is a numerical multiple of u, we have
a=a,u-+p (p, u at right angles). (1)

FIG. 15

When w has been fixed, there is only one such expression for a,
since, if there were two, say
a=a,utp —b,u+q (q,p both at right angles to wv),
we should have, on subtracting,
(a, —6;)u = q—P.
If the left and right expressions of this equation were not both
zero vectors, then one would be paralle! to u and the other would
be perpendicular to u and they could not be equal. Thus both
sides are zero, from which we deduce that a, = 0, and p = q. In
other words the two expressions for a are the same, and the resolu-
tion of a into a vector parallel to u and a vector perpendicular to
u is indeed unique.
The number a, defined by equation (1) is called the projection of
ily a
a on the unit vector u. If v is a non-zero vector, not necessar
unit vector, the projection of a on v is defined to be the same as
 COMPONENTS, COORDINATES Ch. 3
38 PROJECTIONS,
same
the projection of a on the unit vector u = v/| v | with the
V, that is,
direction and sense as V. If 0 is the angle between a and
on and
if 6 is the amount of rotation needed to bring the directi
from the right-
sense of a into line with those of v, then it is seen
ion of a on V is
angled triangle OPA (Fig. 15) that the project
e, since a sense of
|a| cos 6. The angle 0 may be assumed positiv
rotation in the plane OAP has not been defined , and in any case
cos 6 = cos (—8).
The vector p defined by equation (1) is called the projection of a
on the plane Il. Notice that the projection of a vector on a plane
is a vector, whereas the projection of a vector on another vector
is a number.

4, Projections of sums and multiples of vectors

Let u be a unit vector and II a plane at right angles to u. If x
is any vector, let p(x) denote the projection of x on u and let P(x)
denote the projection of x on II. Bold type is used for the pro-
jection on the plane II, because it is a vector, and italic type for
the other projection, which is a number. The relations defining the
projections of two vectors a, b are then
a = p(aju+P(a) (P(a) at right angies to u),
b = p(b)u+P(b) (P(b) at right angles to u).
Added together, these give
a+b = [p(a)-+p(b)]u +[P(a) +P(b)]. (2)
P(a)+P(b), the sum of two vectors parallel to IT, is itself parallel
to II. Thus equation (2) expresses a+b as a sum of two vectors,
one parallel to u and one perpendicular to u. The first member of
the sum, by definition, is the projection of a+b on u, and the
second member is its projection on II. In symbols:
p(a+b) = p(a)+p(b), P(a+b) = P(a)+P(b). (3)
In a similar way, multiplying the first equation by a number &,
we find
p(ka) = kp(a), — P(ka) = kP(a). (4)
5. Components of a vector
Carrying the process of §3 a step further, take three unit
—> — —
vectors, u, = OL, u, = OM, u; = ON, each perpendicular to the
§ 6 PROJECTIONS, COMPONENTS, COORDINATES 39
—>
other two. Let a = OA be an arbitrary vector. As before, drop
AP perpendicular to the plane MON. Within the plane MON drop
PQ perpendicular to the line ON.
a > >
en a = 0Q+QP-+PA. (5)
> > >
The vectors PA, QP, OQ, parallel, in order to u,, Ug, U3, are
multiples @,U,, @.U,, a,U3, and (5) becomes
a = d,U,+a,U,+4
U3. (6)
The expression of a in the form (6) is unique, for in any such

FIG. 16

expression a, must be the projection of a on U,, since a,U,+d3Uz

is at right angles to u. Similarly a, is the projection of a on u, and
dz its projection on U3. The quantities a,, @,, a, are called the
components of a relative to the base U,, Us, Us. When there is no
ambiguity about the base, the vector with components 4, 2, @s
will be denoted by (a, @, 3).

6. Coordinates of a point
Suppose that we are given a fixed point O called the origin and
a base consisting of three mutually perpendicular unit vectors
a co-
U,, Uz, Ug. Such a collection (O, U1, Us, u,)—is said to form
vector
ordinate system. If A is any point and a = OA its position
 COMPONENTS, COORDINATES Ch. 3
40 PROJECTIONS,
a are called the
relative to O, the components (a1, @2, 43) of
coordinates of A (relative to the given coordi nate system).
en points and their posi-
We shall not always distinguish betwe
mean the point with position
tion vectors. Thus ‘the point a’ will
43) may mean, according to
vector a, and the expression (a, @2,
coordi nates or the vector with
context, either the point with these
these components.
s, and the
There are infinitely many possible coordinate system
coordi nates in differe nt systems.
same point P will have different
system can often simpli fy calcula-
A careful choice of coordinate
system will be used throu ghout
tions. Usually only one coordinate
to the ‘coord inates of A’ or ‘the com-
an argument, and one refers
repeat edly specif ying the origin and
ponents of a’, without
base-vectors.

7. Formulae for vector operations in terms of components

(i) Addition
Let a and b be two vectors, with known components
a = (dy, 43) and b = (dj, de, bs).
We wish to find the components of their sum a-++b. From equa-
tion (6),
A = 4,U,+4.U,+43U3, b = bu, +b,u,+,U3.
Hence, by the laws of vector algebra,
atb = (a,+ b,
+(42+02 )¥
)U2+(4 ; s.
3+63)U
Thus the vector a--b has compon ents (a,+b;, @,+02, @3+0s).
Each component of the sum is obtained by adding together the
corresponding components of the two vectors.

(ii) Number-vector multiplication

Let a = (dy, Gg, a3) and let k be a number. We wish to find the
components of the vector ka. We have
ka sad k(a,U,+a.U.+a,U3) — ka,u,+ka.u,+kauy
= (ka, ka2, kas).
Each component of ka is obtained by multiplying the correspond-
ing component of a by &.
Note. Since a,, az, @g are projections of a on the corresponding
unit vectors, the two formulae just obtained could be regarded as
special cases of those of § 4.
 § 8 PROJECTIONS, COMPONENTS, COORDINATES 41
(vit) Length of a vector. In equation (5) of this chapter, the vectors
> > >
~0Q, QP, PA are mutually perpendicular. By Pythagoras’s
Theorem,
cad — => aaa
| O |? = [OQ |?-- | QP |? | PA|?.
Hence |a|? = a,2+4@,?+4,3.
The square of the length of a vector is the swm of the squares of its
components.

8. Application of the formulae

Some numerical examples are now given, illustrating the use of
the formulae of § 7.

Example 1. Tf
at Ay be, and c= Gi— 190);
find
(i) the components of the vector 3a—4b-+7c,
(ii) the length of the vector a+b.
(i) Where two or more vectors are to be added, it is convenient
to write the vectors underneath one another, the numbers in each
column being then added. Using 7 (ii) we have
3a = (—3, 9, 12)
—4b = (—8, 8, 4)
7c = (21, —7, 0):
Adding the numbers in the separate columns we obtain the sum
3a—4b--7c = (10, 10, 16).
(ii) As under (i) we find a+b = (1, 1, 3). By 7 (ili) we have
jatb] = (12+12+3%) = VI.
Example 2. Find the lengths of the sides of the triangle whose
vertices are the points A(—1, —1, —1), B(2, —1, 2), C(2, 1, 0).
A point P divides AB internally in the ratio AP: PB = 2:1.
Find the coordinates of P and show that BPC is a right angle.
' Use p, a, b, c for the position vectors of P, A, B, C. By Chapter
2813,
Pp =
a+2b
= (1, —1, “bY
1);
 COMPONENTS, COORDINATES Ch. 3
42 PROJECTIONS,
r
the coordinates being worked out as in Example 1. From Chapte
=
of the
2, §11, BC = c—b = (0, 2, —2), 80 by 7 (iii) the length
sides CA, AB can be
side BC is 1/8. Similarly the lengths of the
worked out and are found to be 1/14, +/18. We also find

PB = (1,0, 1),PC = (1,2, —1),soPB2+PC? = 2+6 = 8 = BC?

> —>

Thus BPC is a right angle by the converse of Pythagoras’s

are
Theorem. (A better method of testing whether two vectors
perpendicular will be given in the next chapter.)

Example 3. Under what conditions is the vector (a,, G2, 3)

parallel to the vector (0,, bs, bs)?
If the vectors a, b are parallel, then b = ka, where k is a num-
ber. Comparing components, b, = ka,,b, = kas, b3 = kaz. Assum-
ing that a1, dg, a; are all different from zero, we may eliminate &
to obtain
b b b
Pers ee OF aida < Ge gt 0, 1105 | Ue

(Though division by zero is impossible, this condition can be used

(cautiously) when zeros occur if we make the convention that 0/0
can take any value; so that, if any component of a vanishes, the
same component of b must also vanish. When in doubt, one
should always use the basic condition b = ka.)

Example 4. Are the following points collinear?

A(1, 2, 3), B(3, 3, 2), C(7, 5, 0).
> >
The points A, B, C are collinear if the vectors AB, AC have the
same direction. By subtraction
— —>
AB = (2,1, —1) and AC = (6, 3, —8).
The condition for parallelism (Example 3) is satisfied:

Thus the points are collinear.

§9 PROJECTIONS, COMPONENTS, COORDINATES 43
9. Components and coordinates in plane geometry
Suppose that we are considering only vectors in a fixed plane
Let u, be a unit vector in the plane and let u, be the vector ob-
tained by rotating u, through a right angle anticlockwise. An
arbitrary vector a in the plane will clearly be expressible in the
form a,U,-+a,U,, and may be thought of as having only two com-
ponents (a, a,). Just as in space, the components of a sum of two
vectors will be the sum of the correspondiiig components of the
individual vectors.
There is another way of labelling the vectors of the plane. A
vector a is completely specified if we know its magnitude r and the

FIG. 17

angle 0, measured anticlockwise, needed torotate the unit vector u,

to such a position that its direction and sense are the same as those
of a. The quantities 7, 0 are called the polar coordinates of the
plane vector a, and we write a = [r, 6]. Clearly (Fig. 17), the polar
coordinates are related to the components by the equations
Ge = 7 COs v, é, =r sin 0. (7)
the
We shall now use these ideas to give a vector proof of
sines and the sum of
formulae in trigonometry for the sum of two
vectors of equal magni-
two cosines. Suppose we have two plane
— >
tude r, say a = OA = [r, 6], and b = OB = [r, 4].
of the figure
The triangle OAB is isosceles, and the symmetry
is at right angles
shows that if P is the midpoint of AB then OP
OB. Thus the angle
to AB and bisects the angle between OA and
 COMPONENTS, COORDINATES Ch. 3
44 PROJECTIONS,
ey
between OP and u, is }(0-++¢) and the magnitude of OP, from the
right-angled triangle ee is 7 ae Wiss Comparing com-
ponents in the formula OP = 3(OA+OB), we deduce, after
simplification
cos 0-+cos f = 2 cos 3(0—¢) cos 3(9+¢),
sin 9-+-sin 6 = 2 cos $(0—¢) sin 3(6+-9).

FIG. 18

10. The use of vectors in plane geometry

In plane geometry too, as in space, the coordinates of a point
are the components of its position vector, and this is sometimes a
helpful way of looking at problems, although it does not simplify
as dramatically as in solid geometry. As an example, we shall study
in the next section the problem of finding the equations of trans-
formation from one coordinate system in the plane to another.
The method given is not different in principle from the standard
method contained in textbooks of coordinate geometry, but it is
better because the argument clearly works for rotation through
any angle from 0 to 2z, and for points whose coordinates have any
sign. Textbooks on coordinate geometry do not as a rule pay much
attention to these points, and a study of the diagrams in the sec-
tion on rotation of axes will reveal that they really deal only with
§ 11 PROJECTIONS, COMPONENTS, COORDINATES 45

the case of rotation through an acute angle and points whose co-
ordinates are positive in both systems—rather severe limitations!
A large number of separate diagrams and detailed consideration
of signs would be needed to make sure of the formula in its com-
plete generality. When one uses vector algebra, however, all
possibilities of different sign are included, since the distributive
laws apply with multipliers of all signs. The trouble of considering
seven separate cases in the discussion of these laws is well worth
while, because on so many later occasions, like this one, a dis-
cussion of separate cases is avoided.

11. Change of coordinate system in the plane

Let XOY be a cartesian coordinate system in the plane. If u is
the unit vector with the direction and sense of OX and V is the
unit vector with the direction and sense of OY, then it is clear that
the coordinates (x, y) of a point P are given by the vector equation
—>
OP = wzu-+yv. (8)
The reader should check, by drawing pictures to represent
points P in all four quadrants, that the above equation always
gives the correct sign for the coordinates. We shall always assume
that our coordinate systems are right-handed, so that v is ob-
tained by rotation of u anticlockwise through an angle 3z.
Suppose now that X’O’Y’ is another cartesian coordinate
system, u’ the unit vector with direction and sense of O’X’ and
v’ the unit vector with the direction and sense of O’Y’. The same
point P will have coordinates (2’, y’) (usually, of course, different
from x, y) in the new system, where
—— >
O'P = w'u’+y/’'Vv’. (9)
It is obviously useful to be able to change our coordinate system
to simplify calculations in particular problems, but if we do vary
the coordinate system we must know what is happening to the
coordinates of each point when the change is made.
We consider the operation of changing from the XOY-system
to the X’O’Y’-system, so we shall call the XOY-system the old
system and (x, y) the old coordinates of the point P. Similarly,
X’O’Y’ is called the new system, and (w’, y’) the new coordinates
of the point P. We wish to find equations which will give us x, y
when we know 2’, y’, and conversely.
 COORDINATES Ch. 3
46 PROJECTIONS, COMPONENTS,
the component
Throughout the argument, we shall reserve
are J, m in the old
notation (1, m) for the vector whose components
r whose
system: (1, m) = lu--mv. To avoid confusion, the vecto
to be written
components are J, m in the new system will have
out in full, Ju’-+mv’.
ation
Before we can obtain equations, we require some inform
s—the
about the relative position of the two coordinate system
old coordinates of the new origin O’, which we suppose are (h, k),
and the angle 6 of rotation anticlockwise required to bring the old
axes into a position parallel to the new ones. Then the polar co-
ordinates of u’ in the old system are [1, @], so, by (7)
u’ = (cos 8, sin 6).
Since v’ is obtained by a further rotation 42 from u’, we have
v’ = (cos (0+42), sin (0-+43) ) = (—sin 0, cos 6).
— —> : ‘
Also, OO’ = (h, k), and OP = (, y). Hence, using equation (9),
> Sol el
(x, y) = OP = 00’+0'P = 00'+42'u' +y/'Vv’
= (h, k)+<2'(cos 6, sin 0)+y'(—sin 6, cos 6).
Comparing components,
x =h-+z2' cos @—y’ sin 0,
y =k-+<2' sin 6+y’ cos 0.
Two special cases of this formula are particularly important. If
we put 0 = 0, we obtain the formula for change of origin, with
parallel axes:
eo == =Lh; y=y +k. (10)
The formula giving the new coordinates in terms of the old is,
in this case:
x’ =2-h, y =y—k. (11)
If, on the other hand, we put h = k = 0 in the general formula,
we obtain the formula for rotation of axes, with origin fixed:
« =x’ cosO—y'sinf, y=2x'sinf+y’cos6. (12)
Hither by solving these equations for 2’, y’, or, by noticing that
the old axes arise from the new by a rotation —0, we obtain the
reverse equations: |
x’ = “cos 6+y sin 8, y = —xsin6+ycos@. (13)
Any change of coordinates in the plane can be done in two steps,
§ 12 PROJECTIONS, COMPONENTS, COORDINATES 47

a change of origin only, followed by a rotation of axes only, so the

formulae (10) to (138) are all that are needed. The general formula
at the beginning is hardly ever used.

12. Worked example

The reader will find plenty of applications of the formulae of the
last section in any textbook of plane coordinate geometry. Some-
times equations can be simplified by changing the coordinate
system so that certain coefficients vanish in the new equation. It
seems worth while to work one example.

Example. Position in the plane is defined by two systems of co-

ordinates, (x, y) and (x’, y’). The origins are the same, but the
(a’, y’)-axes are obtained by rotation 0 anticlockwise from the
(x, y)-axes. In the (a, y)-system a certain locus has equation
v?4+3ey+(1++/3)y? = 1.
Obtain a value of 6 which will ensure that the (x’, y’)-equation
of this locus has no term in w’y’.

Solution. The new equation is obtained from the old by substi-

tuting for x, y in terms of a’, y’ from equations (12). We find
(a’ cos 6—y’ sin 0)?+
+3(x’ cos @—y’ sin 6)(x’ sin 0+’ cos 6) +
+(1+4+/3)(x’ sin 6+y/’ cos 6)?

The coefficient of x’y’ is

—2 cos 6 sin 6+3(cos? 6 —sin? 6) +2(1+ /3) cos 6 sin 6
= 3 cos 20+ 4/3 sin 20.
Tf this is zero, tan 20 = — 1/3. A solution is 20 = 3a, 0 = 3a.
Then cos 6 = 4, sin@ = 44/3, and the (z’, y’)-equation is
(14$-/3)2'2+(1—}-V/3)y’? = 1.
Since 44/3 is less than 1, both coefficients are positive, and the
equation in its new form is easily seen to represent an ellipse,
although this was by no means obvious in its original form.
 COMPONENTS, COORDINATES Ch. 3
48 PROJECTIONS,

Exercises on Chapter 3
(—3, 1, 0) and
1. If the vectors a, b, c have components (2,1, —1),
vectors:
(0, 1, —2) respectively, find the components of the following
3at+2b—7c, 4a+5b, 3b-+7c, atb—c.
2).
2. Points A, B, C have coordinates (1, 1, —1), (4,1, 2), (—2, 1,
Find the coordinates of the following points:
(i) the point P on AB between A and B such that AB = 3AP;
(ii) the point Q on BA produced such that AB = 3AQ;
(iii) the midpoint of CQ;
(iv) the centroid of the triangle ABC.
3. The points A, B, C have coordinates
A(—1, 4, 3),
B(2, —1, 0), C(5, 2, —8).
A point P lies on BC between B and C so that 2BP = PC. A point Q
is chosen on AP between A and P so that AQ = }AP. Find the co-
ordinates of P and Q and prove that the angle PBQ is a right angle.
4, Ifa = (1, 2, 0), b = (0, 1, 2), c = (1, 0, —1), find numbers 9, q, r
such that pa+qb-+rc = (1, 1, 1).
5. By applying Pythagoras’s theorem to the right-angled triangle
with vector sides a = (@,, Gg, @3), and b = (b,, bg, b3) and a+b, show
that, if the directions of a and b are right angles, then
,b,+a2b2+a43b3 = 0.
(Another proof of this important formula is given in Chapter 4.)
6. Which of the following sets of three points are collinear?

(i) (8,6,1), (9,9,2), (1,8, 2);

(1) ee 2; 5), (2, 4; 4), alk 2, 7)3

(iii) (5, 1, 7), (3, —1,1), (6, 2, 11).

In those cases where the points are collinear (if any), find the ratio in
es the second point divides the segment defined by the first and the
third.
7. Using the condition in Question 5, show that the vectors
wil?
MS (3, 3, —3), Viet (3, =F 3 Te Hig (Gate 3, 3)
are three mutually perpendicular unit vectors. What are the com-
ponents, relative to the base v,, V2, V3 of the vector (3, 5, 7)?
8. The vertices ofa triangle are A(3, 1, 1), B(1, 0, —1), C(4, —3, 2).
M isthe point dividing BC internally so that BM = }MC. Show that
AM is perpendicular to BC.
Ex.3 PROJECTIONS, COMPONENTS, COORDINATES 49

9. Points A, B, C in the plane have coordinates (4, 2), (2, —4),

(—1, —8) respectively. Find the coordinates of the centroid G of the
triangle ABC. A new coordinate system is taken, with the same origin,
but with the new axes obtained from the old by rotation through an
acute angle @ such that cos 0 = 3. What are the coordinates of A, B,
C, G in the new system?
10. In one plane coordinate system two points have coordinates
(16, 9), (14, 6). After rotation of the axes through an acute angle 0, the
points have coordinates (a, c) and (b, —c). Find sin 6, cos @ and the
coordinates of the points in the new system.
11. A new coordinate system (a’, y’) is obtained by rotating an old
coordinate system (x, y) through half a right angle. The origin is kept
fixed. Find the new equations of the loci whose old equations are as
follows.
(i) @+y?=1,
(ii) 2®-+y24+e-+y+1 =0,
(iii) ay? = 3,
(iv) 2?+ay+y?+3r—5y+4 = 0.
12. Find a change of origin, axes fixed in direction so that, in the new
system, the locus with old equation
a*+-xy+y?—3y+5 = 0
should have an equation in which only terms of degree zero and two
occur.
 4
THE SCALAR PRODUCT
1. Introduction
In vector algebra there are three kinds of multiplication:
(1) number-vector multiplication (already defined);
(2) the scalar product ;
(3) the vector product.
These three operations are of quite distinct types. In operation
(1) a number and a vector are multiplied and the result is a vector.
In both operations (2) and (3) two vectors are ‘multiplied’, but
the result of operation (2) is a number and that of operation (3) is
a vector. Since it is obviously essential that no confusion should
arise between these operations, different symbols are used for
them. The types of vector multiplication are listed in the following
table.
Name Type Symbol
Number-vector multiplication Number times vector = vector None
Scalar product Vector times vector = number
Vector product Vector times vector = vector Xx
The scalar product is a useful means of dealing with many
questions involving distance and angle. It is also useful in
mechanics—the work done by a force is an example of a scalar
product—and generalizations of the concept play an important
part in modern algebra and analysis. The vector product, dealt
with in the next chapter, is useful in the solid geometry of planes
and lines. Like the scalar product, it has applications in mechanics,
and its algebra is a stepping-stone to more advanced theories. Its
applications and generalizations are not, perhaps, as fundamental
and universal as those of the scalar product.

2. Definition of the scalar product

Definition 5a. Let a, b be two non-zero vectors, 6 the angle
between them. Then the scalar product a.b is defined to be the
number |a||b| cos 0.
§4 THE SCALAR PRODUCT 51

Definition 5b. If either a or b is the zero vector, the scalar

product a.b is defined to be the number zero, that is, 0.a = a.0
==0;
Note on the measurement of 0. The angle 6 is the amount of
rotation needed to bring the direction and sense of a into line with
those of b. This is the method of measurement used in Chapter 3,
§ 3, where the projection of one vector on another was considered.
Thus if »(b) denotes the projection of b on a, we have
a.b =|a||b|cos? =|a| p(b). (1)
That is: the scalar product of two non-zero vectors is equal to the
length of one vector multiplied by the projection of the other upon it.

3. Immediate consequences of the definition

(i) The condition for two non-zero vectors to be at right angles
is that cos6=0, or a.b=|al||b|cosé=0. (2)
(ii) a.a=|a||a|cos0 =|a|?. (3)
(iii) The components (a;, @:, 43) of a vector a relative to the
base U4, Us, Uz are the projections of a upon Uj, U,, U; respectively.
Since the base-vectors are unit vectors, we have, by (1)
a, =a.U,, A, =A.Ug, As = a.Uz. (4)
(iv) a.b =.b.a. (5)

4, The distributive laws

In this section the following distributive laws are proved.
a.(b+c) =a.b+a.c, (6)
a.(kb) = k(a.b). (7)
The reader should be quite sure that he understands the meaning
of these formulae. In (6), for instance, the sign + is used with two
meanings; on the left of the equation it denotes vector addition,
on the right number addition (the sum of the two numbers
a.b, a.c). In (7) three different kinds of multiplication occur, the
scalar product, denoted by ., number-vector multiplication
(kb), and, on the right, the ordinary multiplication of numbers
(the number & multiplied by the nwmber a.b).
If a = 0, both sides of both equations (6), (7) reduce to zero.
We need only prove the formulae, therefore, on the assumption
that a + 0. The proof depends on formula (1) and the properties
of projections.
 THE SCALAR PRODUCT Ch.4
52
ion of
Proof of (6), (7). As above, let p(a) denote the operat
projection on the vector a. By Chapter 3, § 4,

p(b+c) = p(b)+p(c), —-p(kb) = kp(b).

Multiply by the number | a |, and apply the formula (1). We
deduce
a.(b+c) =a.b+a.c, a.(kb) = k(a.b),
as desired.

5. Formula in terms of components

Let a = (aj, dg, Gy), b = (b,, bg, 63). We wish to find the value
of a.b. By the distributive laws (6), (7) we have
a.b = a.(b,u,+0),.U,+6,U,)
= b,(a.u,)+b,(a.u,) +b,(a.Us),
so, by (4), a.b = a,b,+0a,b,+45)3. (8)
If b = a, this reduces to the formula (using (3) )
| a |? = ay? +42" +03",
already fonnd by another method.

6. Consequences of the formulae

Because of the distributive and commutative laws (5) and (6),
it is possible to expand the scalar product of two sums of vectors
term by term. Thus
(a+b).(c+d) = (a+b).c+(a+b).d
=a.ct+b.cta.d+b.d. (9)
Putting c = a, d = —b, this becomes
(a+b).(a—b) = |a|?—] b|?.
This formula shows that if |a| =| b|, then a+b is at right
angles to a—b, by (2). In other words, the diagonals of a rhombus
are perpendicular—the fact required in Chapter 2, § 8. If, instead,
we put c = a, d = b in (9), we obtain the formula
|a+b |? =|]a|2+ |b|2+2a.b.
. This is the vector form of the ‘cosine rule’ in trigonometry, for
ifa, b represent two sides of a triangle, a+b represents the third
side and a.b is the product of the lengths of the two sides with the
cosine of the exterior angle at the vertex common to the two sides.
It is the exterior angle, because one of the vectors has a sense
§ 7 THE SCALAR PRODUCT 53

directed towards the vertex and the other is directed away from
it. This accounts for the plus sign in the formula. In the trigono-
metrical form an interior angle is normally taken and the sign is
minus.

7. Geometrical illustrations
Example 1. The orthocentre property. The use of the scalar product
gives a very short proof of the orthocentre property of a triangle.
Let ABC be a triangle and let H be the intersection of the altitudes
through A, B. Then HA is perpendicular to BC and HB is perpen-
dicular to CA. We have to prove that the third altitude also passes
through H, or, in other words, that HC is perpendicular to AB.
Take H as origin and let A, B, C have position vectors a, b, c
respectively. Using condition (2) for perpendicularity, we have
—_> >
HA.BC = 0, that is a.(c—b) = 0, so a.b =a.c. Similarly,
since HB is an altitude, b.c=b.a. Hence b.c = a.c, so
(b—a).c = 0, and HC is perpendicular to AB as required.

Example 2. Let G be the centroid of a triangle ABC and let X

be any point. Show that
XA?+XB?2+XC? = 3XG?+GA?+GB?+GC?.
This result is most simply proved by taking G as origin and
letting A, B, C, X have position vectors a, b, c, x. Since the
centroid has position vector }(a+b-+c), it follows, when the
centroid is the origin, that a+b-+c = 0. We have
XKA?+XB?4+ XC?
=|x—al?+|x—b|?+|x—el?
= |x |?—2a.x+|a|?+|x|?—2b.x
+ |b |?+|x|?—2c.x+ |c|?
= 3|x |?—2(a+b+c).x+]a|?+|b|?+]|c|?
Now a+b+c =0, so the second term vanishes, and our
expression is 3 |x |?-+ | a |?+ |b |?+ |c|?, as was to be proved.
Questions involving distance and perpendicularity can often be
solved by using scalar products. The method of attack is straight-
forward. First choose an origin and name the position vectors of
the various points. Next write down as vector equations the con-
ditions given and the result to be proved. Straightforward algebra
usually enables us to deduce the result from the given conditions.
 THE SCALAR PRODUCT Ch. 4
54

A good choice of origin may simplify the algebra. For instance,

in the example above on the orthocentre, H is the obvious choice
of origin, since any other choice would destroy the symmetry
between A, B, C. In the second example, X or G should, for the
same reason, be chosen as origin in preference to A, B, or C. In
some ways, X might appear to be a more natural choice than G.
The reader may find it interesting, as a comparison, to solve the
problem using X as origin.

8. Angle between two vectors. Numerical examples

If neither a nor b is the zero vector, and if 6 is the angle between
a and b, then @ may often be calculated from the formula
a.b
cos G =
Ja||b]
This formula is very useful in coordinate geometry. A few examples
will illustrate its importance.

Example 1. If a = (5, 3, 7), b = (2, —8, 4), find the angle

between a and b. ©
The first step is to evaluate the scalar product a.b using
formula (8). This is conveniently done by writing the vectors one
under the other, multiplying corresponding components, and
adding the products, thus:
a I= (5, 3, 7)
b = (2, —8, 4)
a.b = 10—24428 = 14.
Since, by Pythagoras’s theorem, a = 4/83, b = 1/84 = 24/21,
we deduce from (10) that cos 0 = 7/4/(83.21). (If a numerical
answer is needed, cos 6 can now be calculated to a suitable degree
of accuracy and 6 read off from trigonometrical tables.)

Example 2. Find the lengths of the sides, cosines of the angles,

and the area of the triangle with vertices
A(3, —1, 4), B(2, —2,1), ©(5,1, 3).
— —>
By subtraction, we find AB = (—1, —1 —3),AC = (2, 2, —1),
so AB = 4/11, AC = 3, and by (8) AB.AC = —2—2+3 = <1,
§ 9 THE SCALAR PRODUCT 55

' 1
Using (10), cos A = —3yil* Since sin? A+cos? A = 1, sinA

eee)
sa
14/2 a= AB
ae BAT |. Area AC sinA = 7,1/2/2.

Another method of finding the area of a triangle will be explained

in Chapter 5.

Example 3. Find a unit vector which makes an angle of 45° with

the vector a = (2, 2, —1) and an angle of 60° with b = (0, 1, —1).
Let the unknown vector be u = (a, y, z). We shall require three
equations to determine the three unknowns. These are:
(i) u is a unit vector:
v?+y?+2*% = 1.
(ii) u makes 45° angle with a, so u.a =|u||a_| cos 45°, or

24-+2y—z2 = ae °
4/2
(iii) u makes 60° angle with b, so u.b = |u| |b| cos 60°, or
1
Ue = V2 5

The problem has been reduced to one of pure number-algebra:

three equations in three unknowns. Solve equations (ii) and (iii)
for y and z in terms of 2, finding
(iv) y = —247+ 7/2, 2= —2¢+4v2.
Substitute for y, z in equation (i), obtaining the quadratic in «:
3
9x°—6 2% +5 =3.0,

with roots x = 1/1/2, 1/34/2. Once x is known, y and 2 are

determined by equations (iv). There are, therefore, two solutions
of the problem,
1 A} 1 4 1
w= (S54) = (gana)
9. Trigonometrical application
As in Chapter 3, § 10, we may restrict consideration to vectors in
a fixed plane, so that each vector has two components. By exactly
 THE SCALAR PRODUCT Ch. 4
56

the same arguments as were used earlier in this chapter, the scalar
product of (a, 6) with (¢, d) is ac-+-bd. As an application of this
consider two unit vectors a = [1, 6], b = [1, 4]. (See Fig. 18 in
Chapter 3, § 10.) Clearly the angle between them is @ —d¢, and, since
the vectors are unit vectors,
a.b = cos (9—4).
On the other hand, a = (cos 9, sin 6),
b = (cos 4, sin 4),
so, by the formula,
a.b = cos 6 cos d-+sin 6 sin ¢.
Comparing the two expressions for a.b, we have a new proof of
the trigonometrical formula,
cos (0—¢) = cos 6 cos ¢+sin 6 sin ¢.
This method of proof, besides being simple in itself, gives one an
easy way of remembering the sign in the formula. The angle
between the two vectors is the difference of the two angles, but the
formula for the scalar product involves a plus sign.

10. General remarks

Since the scalar product of two vectors is a number, one cannot
expect the analogy between the scalar product and products of
numbers to be very close. For instance, b.c is itself a number, so
the expression a.(b.c) has no meaning, and there is no question
of any associative law for scalar products. It is also impossible
to associate a division operation with the scalar product, and
cancellation of a vector from both sides of an equation involving
scalar products is not permissible. Thus, if a.b = a.c, the vector
b need not be equal to the vector c. For instance, the equation
a.b = a.c occurs in the proof of the orthocentre property of a
triangle given in § 7, certainly without any need to assume that
the vertices of the triangle coincide. If you prefer a numerical
example, take
a= (1,.0,.0), bi (1,2-1),) cles eee

Exercises on Chapter 4
1. Find two vectors of unit length which make an angle of 45° with
(1, 0, 0) and are perpendicular to the vector CO7O81)
Ex. 4 THE SCALAR PRODUCT 57

2. Find the two unit vectors which make an angle of 60° with both
the vectors (1, —1, 0) and (1, 0, —1). Find also the vectors which make
angles of 45° with both the above vectors.
3. Find the components of the two unit vectors which make an
angle of 45 degrees with (—1, 0, 1) and an angle of 60 degrees with
(—2, 2, 1). Show that these two unit vectors are perpendicular.

4. The position vectors of the vertices of a triangle are 0, a, b.

Show that its area A is given by the formula
4A? — |a [7] b|2—(a.b)?.
5. ABCD is a tetrahedron. Show that if the edge AB is perpendicular
to the edge CD, and if also the edge AC is perpendicular to the edge
BD, then the edge AD is perpendicular to the edge BC.
6. Prove that in any parallelogram the sum of the squares on the
diagonals is equal to twice the sum of the squares on two adjacent sides.
7, The sum of the squares on the six edges of any tetrahedron is
equal to four times the sum of the squares on the three segments joining
pairs of midpoints of opposite edges.

8. If O is the circumcentre of the triangle ABC, and RF is the radius

of the circumcircle, express the scalar products, in pairs, of the vectors
> oOo
OA, OB, OC in terms of R and the trigonometric functions of the angles
A, B, C. Deduce that
bc cos A = R2(1—cos 2B—cos 2C+-cos 2A).

9. If H is the orthocentre of the triangle in Question 8, show that

pa
OH? = 3R2+2R?(cos 24-+cos 2B--cos 2C).
(Use the formula of Chapter 2, § 8.)
10. Three vectors a, b, p are given. Determine the numbers bisin
terms of their scalar products, in such a way that the vector
p—ka—lIb
should be perpendicular to both a and b.
vectors
11. Four points O, A, B, C are given in space, each of the
— > >
OA = a, OB = b, OC = ¢ having unit length. The angles BOC, COA,
the
AOB are denoted by «, B, y. M is the midpoint of BC. ;Express
a ine
if one of the
vectors CA, AB, OM, AM in terms of a, b, c. Show that,
a right angle, so is the other, and that the
angles BAC, OMA is
 SCALAR PRODUCT Ch. 4
58 THE
is
condition for these two angles to be right angles
1-+-cos « = cos B-+cos y.
which
12. Which of the following expressions represent vectors and
represent numbers?
(i) b.cte.a+a.b, (ii) (b.c)a+(c.a)b+(a.b)c, (iii) (b.c)(c.a)a,
(iv) [(b.c)c+(b.a)a].(b+2a).
Evaluate these when a = (1, 0, 1), b = (1, 1, 2), c= (OF 21):

13. Three points A, B, C have coordinates A(1, 6, —2), B(2, 5, 4),

C(4, 6, —3). Find the lengths of the sides, cosines of the angles and area
of the triangle ABC.
14. The angles between the non-zero vectors b and ¢, c and a,
a and b are «, f, y respectively. The vectors u and v are defined as
follows.
u = (a.c)b—(a.b)c, v = (a.c)b—(b.c)a.
Show that if u and v are perpendicular, then
cos? 8 = cos « cos f cos y.
15. All the edges of a tetrahedron OABC are of the same length.
G is the centroid of the triangle ABC, and P is the midpoint of OG.
Show that PA, PB, PC are mutually perpendicular.
16. The four vertices of a tetrahedron have coordinates
(0, 0,0), (8, 2,2), (5, —4, 2), (5,2, —4).
Show that the areas of all four faces are equal.
 5

THE VECTOR PRODUCT

1. Introduction
It is often necessary to find a vector which is perpendicular to
each of two given vectors. There are, of course, many such vectors,
and it is most useful to choose one whose length is the product of
the lengths of the original vectors multiplied by the sine of the
angle between them. This product has its uses in mechanics, for
instance when dealing with moments of forces and with angular
velocity. Regarded from an algebraic viewpoint, it has a close
connexion with the method of cross-multiplication in dealing with
linear equations.
2. Left- and right-handed system of three vectors
Let a, b and c be three non-zero, non-coplanar vectors, repre-
sented by segments OA, OB, OC with a common initial point O.
>
The least angular rotation which moves OA to coincide in direction
aotet
and sense with OB will appear clockwise to an observer on one side
of the plane OAB, but anticlockwise to an observer on the other
side. If an observer at C sees the rotation as anticlockwise, the
system [a, b, c] is called a right-handed system. Otherwise, if the
observer at C sees the rotation as clockwise, the system [a, b, c]
is called a left-handed system. Clearly, if [a, b, c] is right-handed,
then [a, b, —c] is left-handed, and conversely. What is more, the
order in which the vectors are written plays a part. If [a, b, c] is
right-handed, the systems [b, c, a], [c, a, b] obtained from it by
cyclic interchange are also right-handed, but the systems [b, a, c],
[c, b, a], [a, c, b] obtained by interchanging a pair are left-handed.
In Fig. 19, the vectors a, b are shown in the plane of the paper.
Since the idea of right- and left-handed systems is essentially
three-dimensional, it is impossible to draw a plane picture which
illustrates it accurately, so the vector c cannot lie in the plane of
the paper and is not drawn in; but if the vector c is directed away
from the reader, into the plane of the paper, then [a, b, c] is a
right-handed system, whereas if the vector c is pointing at the
 THE VECTOR PRODUCT Ch. 5
60

reader, out of the plane of the paper, then the system is left-
handed. Another way of picturing the situation is to consider an
ordinary right-handed screw. If a and b are vectors in the plane
perpendicular to the axis of the screw and Cc is directed along the
axis of the screw, then [a, b, c] is right-handed if, on rotating the

, au) if
i
b 6
FIG. 19
(i) ¢ into paper, right-handed; c out of (ii) b into paper for right-handed
paper, left-handed. system.

screwdriver from a to b, one is forcing the screw in the sense given

by c.
Until now it has not been necessary to consider whether the
system [u,, U,, U;] of base-vectors is left-handed or right-handed.
All the formulae for addition, number-vector multiplication and
scalar products, are the same in both types of coordinate system.
This is not true of the formula for the vector product. We assume
in what follows that the base-vectors form a right-handed system.

3. The vector product

Definition 6a. Let a, b be two non-zero vectors, not both parallel
and let 6 be the angle between them. The vector product a xb is
a vector defined as follows.
Magnitude: |a| |b || sin @|.
Direction: perpendicular to both a and b.
Sense: so that [a, b, a xb] forms a right-handed system.

Definition 6b. If either a or b is the zero vector or if a is parallel

to b then axb = 0.
It is necessary to give a separate treatment of the cases covered
by Definition 6b because in these cases the second and third parts
of Definition 6a either do not make sense or else do not give a
unique direction and sense ;but Definition 6b is the only possibility
consistent with the first part of Definition 6a, which implies that
§ 4 THE VECTOR PRODUCT 61

the vector product should be zero if either a or b is the zero

vector or if sin 0 is zero.
Note 1. The magnitude of a xb is the area of the parallelogram
with vector sides a and b. Therefore a x b is sometimes referred to
as the vector area of the parallelogram.
Note 2. If [a, b, c] is right-handed, then [b, a, c] is left-handed,
so [b, a, —c] is right-handed again, by § 2. Hence, in particular
[b, a, axb] is a left-handed system, so we have, in contrast to
number-algebra,
axb = —(bxa). (1)
Note 3. If [a, b, c] is right-handed, then [—a, b, c] is left-
handed, so [—a, b, —c] is right-handed. It follows that
(—a) xb = —(axb). (2)
It is clear, directly from the definition, that if k is positive,
(ka) Xb = k(axb).
By (2), this is true for negative values of k as well. More generally
(ka) x (Ub) = kl(a xb). (3)
Note 4. The relation a xb = 0 holds only if a is parallel to b or
if a or b is zero. Thus if a and b are non-zero vectors the following
three statements are all equivalent.
Gi) axb = 0,
(ii) a = kb (for some number k), (Chapter 2, § 11)
(iii) a is parallel to b.

4. Rotation of vectors in a plane

Let II be a plane, looked at from a point on one side of it, so that
it is possible to distinguish between clockwise and anti-clockwise
rotations in II. If a is any vector in II, let a, denote the vector
obtained in II from a by anticlockwise rotation through an angle
6. That is, a, is equal in magnitude to a, and the angle between a
' and a,, measured anticlockwise, is equal to 0.
If a, b are any vectors in I, then
(a+b), = a,+Dp. (4)
> =>
Proof. We omit the trivial cases a = 0 or b = 0. Let OA, OB
62 THE VECTOR PRODUCT Ch.5

represent the vectors a, b. Complete the parallelogram BOAC, so

aa > =>
that c = OC represents the sum a+b. Let OA’, OB’ represent the
vectors a,, b, in order, so that the angles AOA’, BOB’ are both
equal to 6, measured in an anticlockwise sense.
The triangles AOB, A’OB’ are easily seen to be congruent, since
the two sides of one meeting at O are equal to the two sides of the
other meeting at O and the contained angles are equal. It follows
easily that the parallelograms AOBC, A’OB’C’ are congruent,
—> —>
| OC|, | OC’ | are equal, and the angle COC’ is equal to 6. Thus
> => a a es
OC, = OC’, and (a+b), = OC’ = OA’+OB’ = a,+b,.
(eu

A
FIG. 20

5. Vector product with a fixed vector

—>
Let a = OA be a fixed non-zero vector. Let II be the plane
. =>
through O perpendicular to OA and let x = OX be any vector.
The operation of passing from x to a xx can be split into three
operations, as follows (Fig 21).
Step 1. Replace x by its projection x’ = OX’ on II.
Step 2. Rotate x’ through a right angle in II in the sense which
appears anticlockwise from A, to obtain x” = OX".
§ 6 THE VECTOR PRODUCT 63

Step 3. Replace x” by |a |x”.

The vector x’ has magnitude |x||sin6|, the vector x’,
obtained from x’ by rotation, also has magnitude | x | |sin 6],
so |a|x” has magnitude |a] |x]||sin @|. In addition, | a| x”
is at right angles to both the vectors a, x and forms a right-
handed system with them. Thus |a| x”, the result of the three
steps indicated above, is the vector product a xx.
The first step above is a projection, the second a rotation, and
the third, multiplying the vector by a positive number, may be
called an expansion. Thus the operation
omer WO
has been shown to be equivalent to a succession of three operations
P R E
MR
ee a a Ky
where P is a projection, R a rotation, E an expansion.

X
x”

O '

1 x’
'
1
|

EV
FIG. 21
(i) View to an observer at E in (ii). (ii) View of II to an observer at A
in (i).
6. The distributive laws
We now prove the distributive law
ax(b+c) =axb+axce. (5)
, From (5) and (1) follows also the right-handed distributive law
(b+c)xa = bxa+cxa,
which arises from (5) simply by changing signs across.
aS
Proof of (5). Let OA = a and let II be as in § 5. Write b+c = d
 PRODUCT Ch. 5
64 THE VECTOR
s of b, c, d on II, and
and, as in § 5, let b’, c’, d’ be the projection right
each through a
let b’, c”, d” be obtained by rotating b’, c’, d’
of proje ction s (Chapter 3, § 4), we
angle in II. By the properties )
rotations (§ 4 above
see that b’--c’ = d’, and by the property of
” = d”. Hence , by formu la (8), Chapter 2,
it then follows that b”-+c
|a|b’+]alc’ = |a|d’, or, by § 5 above,
axb-axc =axd =ax(b+¢).
still true
This proves (5) when a #9. If a = 0, the equation is
because both sides reduce to zero.

7. The formula in terms of components

The formulae (3), (5) enable us to express the vector product in
terms of components. The unit vectors U,, U2, Us are at right
angles, and form a right-handed system. Thus u, XU, = Us,
u, XU, = —U,, U,XU,; = —Us, UZXU3; = U,, Us XU = Uy,
u,XU, = —U,. This information is set out below as a multi-
plication table, where the left-hand vertical column gives the first
factor and the top horizontal line gives the second factor in each
product. (Tables of this kind have some theoretical importance in
the theory of finite groups and other parts of algebra.)
u, Uy, Uz
u, 0 Us. —Uy,
u, —U; 0 u,
Us Uz —U, 0
If a = (a), da, a3) and b = (),, bg, bs), then
axb = (4,0, +a.U,+a@U3) X (b,U, +5,U, +0304).
Writing this out fully, using the two distributive laws, we have
nine terms
ab,(U, XU,) +445 (U, X Uy) +a,b,(U, X Ug) +
+G9b,(Uy XU) +49b.(U,X Uy) -+a_b,(U, X Ug) +
+A3by (Ug X Uy) +44) (Ug X Ug) +3b3(Ug X Us).
Using the multiplication table above, this reduces to
axb = (42b3—a3b2)U, +(a4b; —a,b)U,
+(a,b, —Agb,)Us.
The components of a xb are therefore
(@2b3—Ggd2, 3b,—a,b3, a,b,.—aQb,). (6)
§ 8 THE VECTOR PRODUCT 65

The following method of obtaining the components is often

useful. Write down the components of the two vectors one below
the other, the components of each vector being written down in
cyclic order, beginning and ending with the second;
ay Ae as ay

i NVeas
ayaaNE 87
The products corresponding toa downward arrow havea plus sign,
the others a minus sign.

8. Worked examples
Example 1. If a = (3, 5,7), b=(1,0,1), ¢ = (2, —1, 3),
verify that a x (b xc) = (a.c)b—(a.b)c.
To evaluate the left side we first obtain b xc, using ‘cross-
multiplication’ as in § 7.
: SGN 1 1
BOS 0

ENON. 4
bxc = (1, —1, —1).
The same process applied to the vector a and the vector b xc
just found yields a x (b xc) = (2, 10, —8). Again, by the formulae
of earlier chapters, a.c = 22, a.b = 10, so the right-hand side is
22b—10c = (2, 10, —8), and the given equation is verified.

Example 2. Find the area of the triangle with vertices

A(1, 3, —2), B(4, 3, 0), C(2, 1, 1).
This problem can of course be solved without the use of vector
pro-
products, as we saw in the last chapter. However, the vector
, since the area AB AC sin A
duct gives the answer very quickly > >
With
is just half the magnitude of the vector product AB x AC.
the numbers given we have
> — > =>
AB (N02) AC Sanh y= 2313) spa XAC = (4; —7, —6).
$+/101.
The magnitude of the last vector is 4/101, so the area is
EVA—C
 THE VECTOR PRODUCT Ch. 5
66

Example 3. Let a, b, c be unit vectors such that axb = axXC,

and b makes an angle of 60 degrees with c. Show that b—c = a.
Since axb = axc, we have ax(b—c) = 0, s0, by § 3, Note 4,
b—c = ka, where k is a number. Thus
|b—c |? = k?|a|? = k?, since a is a unit vector.
k? = |b |?+ |c |?—2b.c
= 2—2 cos 60° = 1.
Thus k= +1, b—c=ka= +a.
The principle used in Example 3, that if p xq vanishes then
p = kq (unless one of p, q is zero), is very important. We illustrate
its use by another example.

Example 4. If bxc =cxa =axb ~0, show that

at+b+c=0.

Proof. From bxc =cxXa = —axc, we deduce (b+a)xc

= 0, so b+a = kc. Multiply (vector product) both sides of the
last equation by a:
bxa+0 = k(c xa).
But it is given that bxa = —(axb) = —(c xa).
Thus k=-—1, bta=-c, and a+b+c=0.
The statement in Example 4 is false if we do not require the
three equal vector products to be different from zero. Thus, if
a=b=c ~0, all the other hypotheses are satisfied but not the
conclusion. At what points in the argument was the assumption
axb ~0 used implicitly?

9. Cross-multiplication
Very often in algebra it is necessary to deal with two homo-
geneous equations in three unknowns, for instance
AX + AgX,+Azt3 = 0, (7)
byt, +bet.+bax, = 0.
If we multiply the first equation by ,, the second by a,, subtract
and simplify, we obtain
Xs ie Xs
a3;b,;—a,b, fis ayb,—ab, j

Combining this with the similar equation obtained by eliminating

§ 10 THE VECTOR PRODUCT 67

x3 between the equations, we derive

La Xs Le
k, (8)
A2b3—Azb. fhAb; —Ayb5 "i A1b2—A ab, iv,
where & is a number.
Now look at the denominators in equations (8). Obviously they
are obtained in the same way as the components of a vector
product; so let us reconsider equations (7) and (8) in terms of
vectors a, b and x whose components, in a certain fixed coordinate
system, are (41, G2, As), (by, be, bs) and (x4, Xg, #3): Equations (7)
say a.x = b.x = 0, in other words x is perpendicular to both a
and b. Equations (8) say that x = k(a xb), in other words x is
parallel to the vector product. Thus a purely algebraic argument
has been found to show that if a vector is perpendicular to each of
two non-parallel vectors, it is parallel to their vector product.
The algebraic form of this is sometimes useful in elimination
problems.
To avoid complication in the above argument, we have not con-
‘sidered the possibility of the denominators in (8) being zero. If all
are zero, then axb = 0, a is parallel to b, and the equations (7)
do not determine the direction of the vector x. If some, but not all,
are zero, the argument has to be modified because division by
zero is never admissible, but equations (8) remain true in the
modified form x = k(axb). (It is also possible to interpret
equations (8) using the convention that 0/0 can take any value, as
indicated at Chapter 3, § 8, Example 3. The reader is not advised
to use this convention in his own working at this stage, but he
must be familiar with it if he is to read standard texts on Euclidean
and projective analytical geometry.)

10. The vector triple product

Another result which is not so obvious geometrically can be
proved by the method of elimination. In this case we begin with
the non-homogeneous equations, writing the component and the
vector form of the equations side by side.
4%, +G2%.+0.%, = k, PW Gehl iy
b,27,1+b,7,+b52, = l, Dies
Let us eliminate x, between these equations, as before, multi-
plying the first equation by ,, the second by ay, and subtracting.
 VECTOR PRODUCT Ch. 5
68 THE

We derive
(aby —40 5) +(Goh1 —Ayby)a_ = kb, —lay.
of the
As in § 9, the coefficients of 73, , on the left remind us
we let
components of the vector product a xb. In fact, if
d —axb, and denote its components by (d,, d2, ds), the above
equation can be rewritten
dot,—dt_ = kb,—la,. (9)
There is no special reason why the component #, should be given
particular consideration. If we eliminate x, between our first two
equations, we derive
d,t,—d,", = kb,—la., (10)
and if we eliminate x3, we derive
d,H_,—d 9%, = kb,—las. (11)

Now look at the equations (9), (10) and (11). Once again, they
remind us of the components of a vector product, this time the
product dx. What about the right-hand sides? A moment's
study will show that the right-hand sides of (9), (10) and (11) are,
in order, the three components of the vector kb—la. Thus these
three equations are the component form of the vector equation
d xx = kb—lIa. (12)
The short and simple form of the equation (12), replacing the
three longer equations (9), (10) and (11), illustrates the brevity of
notation which we have already mentioned. It shows that vector
algebra can save a bit of writing, if it does nothing else!
If, in Equation (12), we remember that d was introduced to
denote a xb, and that k and / are, by the equations from which we
started, the scalar products a.x and b.x, we can transform (12)
into the vector triple product identity
(axb) xx = (a.x)b—(b.x)a, (13)
valid for any three vectors a, b and x. This formula has some
applications in spherical trigonometry, as we shall see in Chapter
lis

11. The scalar triple product

A parallelepiped is the natural generalization to three dimensions
of the parallelogram in the plane. It is the part of solid space cut
off by three pairs of parallel planes—like a cube, but with the
§ 11 THE VECTOR PRODUCT 69

angles not necessarily right angles and the edges not equal. It has
six faces, eight vertices and twelve edges. In Fig. 22 we illustrate
a parallelepiped with vertices O,B,C,D, A,B’,C’,D’. Every edge is
> OS
equal and parallel to one of the three OA, OB, OC, and we say
; > SS
that the parallelepiped has vector edges OA, OB, OC.
It is convenient to pick one of the plane faces and call it a base.
The perpendicular distance between the base and the plane face
C’

FIG. 22 The parallelepiped.

parallel to it is then called the height. Thus if the face OBDC is

chosen as base the perpendicular distance between the plane
OBDC and the plane AB’C’D’ is the height. By arguments
analogous to the plane case, it can be shown that the volume of the
parallelepiped is equal to the area of the base multiplied by the
height. We shall now try to express the volume in terms of the
> —> —>
vectors OA = a, OB = b, OC = c. Let P be the foot of the
perpendicular from O on the plane AB’C’D’ and let 6 be the angle
POA (Fig. 23). The area of
the base is | b xc |and the P A plan of B,C’,D
height is | a |cos 6. Hence
the volume V is
V =|al||bxc] cos 6. (14) 6
—>
Now the vector OP has) o plan of B,C ,D
the same direction as b xc FIG. 23
and it has the same sense
if a, b, c is a right-handed system. Thus, if a, b, c is right-
handed, @ is also the angle between a and b xc, and the formula
 THE VECTOR PRODUCT Ch. 5
70

above for V can be expressed as a scalar product

V =a.(bxc).
If, on the other hand, a, b, c is a left- handed system, OP will
have the direction and sense of —b xc, and then we have
V = —a.(bxc).
Of course the volume is independent of the order in which the
vectors are taken, and bearing in mind the effect of re-ordering on
left- and right-handed systems (§ 2), we have the identities
a.(bxc) = b.(cxa) =c.(axb) =
(15)
—a.(cxb) = —b.(axc) = —c.(b xa).
Therefore, in the expression a.(b Xc), it does not matter if the
position of the dot and cross are interchanged. For this reason the
expression is often written [a, b, c] and called the scalar triple
product.
From the component formulae for scalar and vector products,
we have
a.(bxc) = > a,(bx¢2—bs¢2),
the sum being taken over cyclic interchanges of the suffixes 1, 2
and 3. Some students may recognize this as a determinant
Q A, as
b; bs bs |
Gy, ACar nes |
the rules (15) above corresponding to the rules giving the effect of
interchanging rows in a determinant.

12. The use of cross-multiplication in elimination problems

We end this chapter by a digression indicating how the algebra
of § 9 is often useful in elimination problems. We saw in § 9 that,
if a.x = b.x = 0, then x is of the form k(axb), where k is a
number, provided only that a xb does not vanish.
Algebraically, this means that the equations
ax+by+cz = 0,
a’x+b'ytc’z = 0
imp]oe
= k(be'—b'c), y =k(ca'—c'a), z = k(ab’—a’'b),
§ 12 THE VECTOR PRODUCT 71

where k is some number. This is valid only if at least one of the

expressions in brackets does not vanish. We give three examples
to illustrate the use of this fact.

Example 1. Eliminate 6 between the equations

acos@+6 sin 6é+c = 0,
a’ cos 0+0’ sin #@+c’ = 0.

Solution. If we apply the above formula with x = cos 0,

y = sin6, z = 1, we find
cos 6 = k(bc’—b’c), sin @ = k(ca’—c’'a), 1 = k(ab’—a’b).
We neglect for the moment the possibility that all the expressions
in brackets should vanish. Since cos? 6+sin? 6 = 1, we find, on
squaring and adding the equations and dividing by k?,
(bc’ —b’c)?-+-(ca’ —c’a)? = (ab’—a’b)?. (16)
If the expressions in brackets should vanish, this is still true, so
- (16) is the desired relation.

Example 2. Eliminate x’, y’ between the equations

px qy’
aa’ yy’ a2 yy?

Bah ak ae = 1, aa Et = Me

(Note. These equations express the fact that there is a point

(x’, y’) on the ellipse with standard equation such that (x, y) and
(p, q) both lie on the tangent to the ellipse at this point. Thus the
result of the elimination is the equation of the pair of tangents to
the ellipse from (9, q).)

Solution. We may simplify by writing wu = x'/a?, v = y'/b?, so

that the equations take the form
uxtovy =1, uptrumg=1, a?u?2+by? = 1,
From the first two equations we derive, by ‘cross-multiplication’,
i ac Aa ee Sag
U v —l

Combining this with the last equation, we derive

a?(y—q)?+b?(p—x)* = (qu—py)?.
 THE VECTOR PRODUCT Ch. 5
72

Example 3. Solve the equations, where a, b, c are unequal:

xtytz=0,
ax+by+cz = 0,
aa by +c?z2+xyz = 0.

Solution. From the first two equations, as before, we derive

x =k(b—c), y=k(c—a), 2 =k(a—b).
Substitute in the third equation:
k Y a?(b—c) +k3(b—c)(c—a)(a—b) = 0.

Now > a?(b—c) = —(b—c)(c—a)(a—b).

Hence k? = k, and k = 0, 1, or —1.
There are three solutions:
(i)*=y=z=0,
(ii) e¢=b—c, y=c=—a, 2 =a hi
(ili) « =¢c—b; y= a—c, Zz =u —a .

Exercises on Chapter 5

1. Prove from the definitions that

|axb |? =|a|?|b |2?—(a.b)*
2. If the vertices of a triangle have position vectors a, b, c, show that
the area of the triangle is half the magnitude of the vector
bxc+cxataxb.
Deduce that the three points lie on a line if
bxc+cxataxb = 0.
3. Show that
(ra--sb) x (fa+-ub) = (ru—st)(a xb).
4. (a) Let n be any unit vector, and let P denote a plane perpen-
dicular to n. Show that the projection of x on P is
X—(X.n)n.
(b) A rigid body is rotated through an angle of 90 degrees in a right-
handed sense about an axis passing through the origin. The unit vector
giving the direction and sense of the axis of rotation is n. If the position
vector of a point of the body before rotation is x, show that its position
Ex. 5 THE VECTOR PRODUCT 73

vector after rotation is

nxXx-+(x.n)n.
5. Ifa ~0 and a.b = 9, show that the set of points whose position
vectors x satisfy the equation
axx=b
is a straight line.
6. Let a = (3, 2, —1), b = (1, —1, —2), c = (4, —3, 4). Evaluate
the following expressions, giving the three components if the answer is
a vector.
(i) ax(bxc), (ii) (axb)xe, (iii) (axb).(axc),
(iv) (axb).(bxc), (v) [ax(axb)].c.
Tota (34), 2) b= (002, 1), c = (Lal, 1) and
d = bx(cxXa) +(a.c)a, show that b is perpendicular to d.
8. Let v be a unit vector and a a vector such that a.v = 0. Show,
directly from the definition, that, if b= vxa, then vxb = —a.
Verify this when a = (2, —2, 1), v = (4, 3, %).
9. Let a, b, c be non-zero vectors. Ifaxb = c and bxc = a, show
that a, b, c are mutually perpendicular, that two of a, b, c are of equal
magnitude and that the third is a unit vector. Under what condition
does cx a = b also hold in addition to the above relations?
10. Suppose that a, b, c are distinct unit vectors such that axc
= bxc and the angle between a and c is 45°. Show that a is at right
angles to b.
11. (a) The direction and sense of an axis L through the origin is
given by a unit vector v. If a rotation through an angle 0 in the right-
handed sense is carried out, show that the displaced position of a point
which originally had position vector x, is given by the expression
(1—cos 8)(x.v)v-++cos 0x+sin 6(v X x).
(b) What is the displaced position, after rotation through an angle
45° about the axis given by the vector (2, 2,1) of the point whose
original position was (1, 1, 1)?
12. If c = pa+gqb+7(axb), where a and b are unit vectors at an
rangle of 60°, show that
|c|? = p?+9?+pq-+
dr”.
Obtain similar expressions for c.a, c.b. What values must p, q, 7
have if c is a unit vector making an angle of 60° with a and at right
angles to b?
 6

COORDINATE GEOMETRY
OF THE PLANE AND
STRAIGHT LINE
1. Introduction
This chapter deals with the simplest loci, the plane and the
straight line, and their definition by means of equations. A straight
line requires two equations to define it, a plane only one. Before
considering the details, one must be quite sure what is meant by
the phrases ‘the equation(s) of a locus’ and ‘the locus defined by an
equation’. This is explained in § 2 and in §§ 8 to 10. As one might
expect from plane geometry, the plane and straight line in space
are defined by equations of the first degree.
The algebra of vectors is helpful in deriving the equations of
these loci, and even more helpful in giving a geometrical meaning
to certain constants in the equations. Two sets of constants are
particularly concerned, the direction ratios of a line and the
direction ratios of a plane. Each of these sets of constants is the set
of components of a vector, in the first case a vector parallel to the
given line and in the second case a vector perpendicular (or
normal) to the given plane. The numbers have significance only
through the vector of which they are components—it is the vector
that counts. Though there is always a risk in altering established
terminology, it seems that this is a clear case for change, and we
have introduced the terms direction vector and normal vector. This
makes discussions clearer and helps to keep the geometry of any
situation before us. One would like to think that this terminology
may in time gain acceptance and that the term ‘direction ratio’
may fall into disuse. At present, however, there are many excellent
textbooks in which the standard terminology is used, and the
reader will have to be familiar with both. It will help the reader to
understand the relation between vector algebra and coordinate
geometry if he looks through an older textbook, such as Salmon’s |
Analytical Geometry of Three Dimensions, trying to re-write the
 § 2 COORDINATE GEOMETRY 75

equations in vector form. He will also see that the introduction of

the equation is made much easier by the use of vector notation
and terminology.

2. Locus defined by a single explicit equation

Hitherto we have used a notation (x,, %2, x,) for the components
of a vector x, and also for the coordinates of a typical point. In
this chapter we shall most often use the letters (x, y, z) for the
coordinates of a typical point, though it will occasionally be
convenient to use the earlier notation still.
Consider the ‘equation’
v2+y?+2? = 1. (1)
If we substitute three numerical values for x, y, 2 in (1) the
resulting statement may be true or false. If it is true, we say that
the numerical values satisfy the equation. Thus the values
=th,y =z = 3, satisfy the equation, but the values x = 1,
y = 2,2 = 3 do not.
Now consider some fixed coordinate system in space. Each
point of space has three coordinates in the system. The points
whose coordinates satisfy the equation are said to form the locus
defined by the equation. The equation (1) itself, for instance, defines
a sphere of unit radius, centre at the origin, because %?-+-y?-+z? is
_ the distance from the origin of the point (x, y, 2).
The statement that an equation defines a given locus is really
a combination of two statements:
(i) if a point lies on the locus, then its coordinates satisfy the
equation;
(ii) if the coordinates of a point satisfy the equation, then the
point lies on the locus.
The phrase ‘if and only if’ is often used in mathematics to
combine two statements related in this way:
‘A point lies on the locus if and only if its coordinates satisfy the
equation.’
There are two other ways of expressing the relationship defined
by the phrase ‘if and only if’. One can use the phrase ‘necessary
and sufficient condition’, thus:
‘A necessary and sufficient condition for a point to lie on the locus
is that its coordinates should satisfy the equation.’ The third way
 COORDINATE GEOMETRY Ch. 6
76

is to use the term ‘equivalent’. Two statements are called equiva-

lent if each implies the other, thus:
‘The following two statements are equivalent:
(i) the point P lies on the locus L;
(ii) the coordinates of P satisfy the equation of L.’
For shortness we shall sometimes use expressions such as ‘the
locus x?-++-y?+22 = 1’ to mean ‘the locus defined by the equation
a?+y?+z? = 1’, Similarly we shall often say that ‘the point P
satisfies the equation’, meaning, strictly, that ‘the coordinates of
the point P satisfy the equation’.
It should be noted, finally, that the locus defined by a given
equation depends on the choice of coordinate system. A change of
coordinate system may, and usually will, assign a different locus
to the equation.

3. Basic assumptions about the plane. Normal vector

In obtaining the equation of the plane, as elsewhere in this book,
we assume that the space with which we are dealing has the
intuitively obvious properties of the space in which we live. In
most cases we make these assumptions without stating specifically
what they are, but to fix ideas we now state explicitly the assump-
tions that we need to obtain the equation of a plane.
We assume that, given any plane, there is a unique direction,
said to be perpendicular to the plane, which is at right angles to
the direction of any vector which lies in the plane. Any vector with
direction perpendicular to the
plane is called a normal vector of
the plane. Given a point P and a
non-zero vector a, we assume that
there is a unique plane II con-
taining P with normal vector a.
Finally, we assume that if II is a
P plane, P a fixed point in II, anda
a normal vector of II, then a point
>
X lies in II if and only if PX is at
right angles to a.
FIG. 24 Plane swept out by the
family of lines throughagivenpoint . paly we cimpotics os ¥
NERDOE
P perpendicular toanormalvectorn, iS taken into account in deciding
§ 4 COORDINATE GEOMETRY au

whether it is normal to a given plane. Thus if a is a normal vector

of a certain plane, so is the vector ka, k +0, which has the same
direction as a. On the other hand, the zero vector 0, for which no
direction is defined, cannot be a normal vector.

4, Equation of the plane

Let a <0 be a fixed vector, P(p) a fixed point, X(x) a variable
point. Let II denote the plane through P normal to a. Then, the
following statements are equivalent.
(i) X lies on II.
(ii) a is perpendicular to PX.
(iii) a.(x—p) = 0.
(ivy) a.X =a.p.
Equation (iv) is true if and only if X lies on II. In other words,
equation (iv) is the equation of II. Writing & for the constant
a.p, we find:
The equation of a plane can be written in the form
a.x =k, (2)
where k is a constant and a is a normal vector of the plane. In terms
of coordinates (a, %2, #3), the equation of a plane may be written
with the aid of the formula for scalar products:
04%, 14 :%ota,t, = k. (3)
If we adopt the notation (x, y, z) for coordinates, the equation
of a plane is
ax+by+cz =k. (4)
Thus the equation of a plane is linear in the coordinates 2, y, 2.
We shall now show conversely that any linear equation represents
a plane. As above, such an equation can always be written in the
vector form a.x = k, where a is a vector different from zero. Let
p be any point on this locus (for instance p = ka/ |a |?). Since p
is on the locus, a.p = k, so the equation can be written a.x = a.p,
and it is the equation of the plane through p with normal vector a.
It is important to notice that the equation of the plane gives us
its normal vector immediately, its components being the coeffici-
ents on the left-hand side of the equation. In the above equations
(2), (3), (4), the normal vectors are respectively, a, (d1, @:, @3) and
(a, b, c).
 COORDINATE GEOMETRY Ch. 6
718

Note. In most textbooks, including all older ones, the term

normal vector is not used, and the components of a normal vector
are called direction ratios of a plane. If the normal vector is a unit
vector (there is, of course, a unit vector among the normal vectors
of any plane) its components are called direction cosines of the
plane. We have explained our reasons in § 1 for not using this
terminology, but it is still standard and the reader ought to be
familiar with it. If he has thoroughly mastered the vector opera-
tions and the formulae for them given in the previous chapters,
he should have no difficulty in passing from the one viewpoint to
the other and back again when necessary. Some students have
told me that the new terminology actually helps in understanding
texts which are written in the standard notation.

5. Examples on the equation of the plane

In working examples involving the equation of the plane, the
following three facts are so important that it is worth while to
repeat them.
1. A point lies on the plane if and only if its coordinates satisfy
the equation of the plane.
2. The equation of a plane is linear, and conversely a linear
equation always represents a plane.
3. The vector a is the normal vector of the plane a.x = k.
The third fact is particularly useful. It means that if we know
a normal vector we know the left-hand side of the plane’s equation,
and only the constant k remains to be found. If we wish to find the
equation of a plane, then, one of the first questions we should ask,
is: Can we find its normal vector? As an illustration, consider the
problem of finding the equation of a plane which passes through
three given points. The method is indicated in the following
example.

Example. Find the equation of the plane containing the three

points A, B, C in each of the following two cases.
(i) A(0, 1, =1), Bas 1, 0), C(1, 2, 0).
CAI; (233); B(0;:3; 2), CSF eae
Let us begin with case (i). Consider the vectors
— —>
AB (1,0, 1), . AC (lean
 § 6 COORDINATE GEOMETRY 79

found by subtraction. The normal vector is perpendicular to both

these so it is parallel to their vector product (—1, 0, 1). The left-
hand side of the equation must thus be of the form —1lx+0y+lz,
and the full equation will be —x+z = k. To determine the con-
stant k, consider the point A. Since A lies in the plane, its coordin-
atesz = 0,y = 1,z = —1satisfy the equation, that is —0+(—1)
=k. Thus k = —1, and the equation of the plane ABC is
—“t+z2= —l1.
(To check correctness of working, it is useful to verify that the
points B, C also lie on the plane.)
This example might lead one to think that there is always a
unique plane containing three given points. However, we shall
see that this is not so in case (ii). This time the vectors
> >
AB = (—1,1, —1) and AC = (2, —2, 2)
are parallel. Their vector product is the zero vector 0, which can-
not be a normal vector, since it has no direction. This does not
mean that there is no plane containing the points. The points lie
on a line and there are many different planes containing them.
- The complete solution is left to the reader (Ex. 2 below).

Exercises

1. Which of the following sets of three points are collinear? Find the
equations of the planes containing the non-collinear sets.
(i) A(3, 5, 2), B(7, 4, 1), C(4, 6, 3);
GipA0, 1,.9),/ BLT, =) C6, 7, 20);
(iii) A(4, 5, 6), B(5, 4, 3), C(3, 6, 8).
2. In the worked example (ii) above, show that the distinct planes
x—z+2 = 0, +y—3 = 0 both contain the points A, B and C. Show
also that any plane containing the points A, B and C has an equation
of the form
k(x—z+2)+l(2+y—3) = 0,
where k and / are constants.

6. The angle between two planes

Definition 7. The angle between two planes is defined to be the
angle between their normal vectors.
 GEOMETRY Ch. 6
80 COORDINATE
the
This may seem a strange definition, but a reference to
the inter-
figure will show that it is reasonable. Fig. 25 shows
sections PQ, PR of two planes with the plane of the paper, which
is assumed to be perpendicular to their line of intersection. The
> = —
normal vectors PA, PB of the two planes then lie in the—> plane of
>
the paper. The angle @ between the normal vectors PA, PB is

FIG. 25

equal to the angle between the lines PQ, PR and represents the
amount of rotation necessary about the line of intersection to
bring the plane containing PQ into coincidence with the plane
containing PR. The angle 6 between the planes with normal
vectors a, b can be worked out from the formula given in Chapter
4,
cos6 =a.b/|a||b|.
As an example, if the planes are x+2y+z = 0, anda+y = 0,
the normal vectors are (1, 2, 1) and (1, 1, 0), with scalar product
3, 80 cos 0 = 3/4/6. 4/2 = 34/3, and 0 = 30°.

7. Perpendicular distances
Suppose given a point P(p) and a plane a.x = k. We wish to
find a formula for the perpendicular distance of P from the plane.
If Q is the foot of the perpendicular from P on the plane, then we
know
(i) that Q lies on the plane, and
(ii) that PQ is parallel to the normal vector a (just another way
of saying that PQ is perpendicular to the plane).
§ 8 COORDINATE GEOMETRY 81

Express these two facts in terms of vectors, obtaining

G)a.q=k (Q lies on plane), (5)
(ii) p—q =ta (PQ is parallel to a). (6)
Take scalar products with a on both sides of (6). We deduce,
using (5), that
t|a|?=a.p—a.q =a.p—k. (7)
Now the perpendicular distance we require is the length PQ,
that is, by (6) and (7),
[p—a|=[é||al =|a.p—kl/al. (8)
It will be useful later to have a formula for the perpendicular
distance between two parallel planes, which will, of course, have
a common normal vector a. Let their equations be a.x =k, ©
a.x = |. If pis any point on the second plane, thena.p = 1,so the
perpendicular distance of p from the first plane is, by (8),
|a.p—k|/|a| =|k—I|/al. (9)
The right-hand side is the expression required.

Examples. The distance of the point (1, 2, 1) from the plane

a—y+2z = 5 is |1—242.1—5 |/+/(1? +1242?) = 4/+/6. (Here
ais (1, —1, 2), k = 5 and p is (1, 2, 1).) As an example of (9),
the distance between the parallel planes x+-y+z = 2 and z+y+z
= —] is | 2—(—1) |//3 = V3. (Here k = 2,1 = —1 and ais
(1, 1, 1).)

8. Loci in space
In plane geometry the only kind of loci considered are lines,
straight or curved, defined by a single equation. A ‘locus’ in the
plane defined by two equations would not be of much interest ;two
equations in two unknowns usually only have a finite number of
solutions, and such a ‘locus’ might consist only of a finite number
of points. In space geometry, however, there are three coordinates,
so two equations can be expected to define a locus which is not
trivial. It is easy to convince oneself that, in general, loci defined
by a single equation are surfaces, like the plane and the sphere, for
which we have already found equations; while those defined by
pairs of equations are lines, straight or curved. The definition of a
line by means of a pair of equations (each representing a surface)
 COORDINATE GEOMETRY Ch. 6
82

is an expression of the fact that two surfaces may be expected to

intersect in a line.
It is also possible to define loci by using equations involving
extra variables, or parameters, in addition to the coordinates
themselves. Both the explicit and the parametric type of equation
are needed to deal with the straight line. These two methods of
defining a locus are explained in detail in the next two sections.

9. Loci defined by two explicit equations

Consider the two simultaneous equations
i), e*+y*2+z2? = 1. (10)
Three numerical values for x, y and z may satisfy both equations,
‘or one and not the other, or neither. The points whose coordinates
satisfy both equations are said to form the locus defined by the
equations. Thus the point (0, 2, $) is on the locus, but the point
(4, 2, 2), which satisfies only the second equation, and the point

FIG. 26 Circles as the intersection of a sphere with a plane through its centre.

(3, 2, 1), which satisfies neither, are not on the locus. As in §2,
we can say: ‘A point lies on the locus if and only if it satisfies both
equations.’ The special equations (10) define a circle, because the
points which satisfy both equations lie in the plane 2 = 0 (first
equation) and are at unit distance from the origin (second
equation). Thus they form a circle in the plane x = 0 with unit
radius and centre at the origin. The first equation by itself defines
a plane, the second by itself represents a sphere, and the two
equations combined define the intersection of the sphere with the
§ 10 COORDINATE GEOMETRY 83

plane through its centre, giving an example of two surfaces inter-

secting in a curved line (Fig. 26).

10. Parametric equations of a curve

Consider the equations
Va bl*, sist, 92 = £. (11)
If ¢ is given a numerical value, the numerical values of x, y, and
z are determined, and define a point of space. If tis given a different
value, we obtain another point, and by allowing ¢ to take all
possible real values, we obtain a whole family of points in space
called the locus defined by the equations (11). Equations of this
type, involving one or more extra variables besides the coordinates,
are called parametric equations. Any extra variable, such as ¢ in
equations (11), is called a parameter.
The letter t is very often used for the parameter, the historical
reason being that such equations first arose in connexion with the
path of a moving particle, equations such as (11) giving the co-
ordinates of the point of space occupied by the particle at time t.
It is a great help to intuition to think of parametric equations in
this way.
Though parametric equations have many uses, they do not
give a direct criterion whether a point does or does not lie on the
locus. If we want to know this we have to consider whether there
is any value of ¢ which will, on substitution, give the coordinates
of the point. For example, to find out whether the point (—4, 2, 3)
lies on the locus, we have to decide whether there is a numerical
value of ¢ such that
ft = —}, ?@=4, PP =h.
Now the first equation is satisfied by = —4 only, the second by
t = +4, and the third by ¢ = +4 only. Thus there is no value of ¢
which satisfies all three and the given point does not lie on the
locus. On the other hand, the point (—}, 4, —4) corresponds to the
value ¢ = —3 and so does lie on the locus.
If a locus has been given parametrically and you wish to change
to an explicit definition without parameters, it is often possible
to do this simply by eliminating the parameters using the methods
of ordinary algebra. Thus we could eliminate ¢ from the equations.
 COORDINATE GEOMETRY Ch. 6
84

(11) in two ways to obtain the equations

y=(x—-y)?, 2 = Y¥(%&—Y)s
which are indeed a pair of explicit equations for the locus defined
parametrically by (11).
However, the method of elimination should be used cautiously,
because there is a danger of introducing extra loci in this way. For
example, if we obtain two equations from (11) by eliminating ¢
between the x and y equations and then by eliminating ¢ between
the y and z equations, we derive
y=(z—-y), P= y?.
The locus defined by these two equations does, of course, contain
the curve (11), because the method of obtaining it shows that
every point given by (11) must satisfy it; but it contains other
points too, because the point (—}, 4, 4) lies on it though it doesnot,
as we saw above, lie on the curve (11).

11. Direction vector of a line

The crucial step in obtaining the equation of the plane was
defining a direction associated with it. Similarly to obtain equa-
tions of a line, we consider the direction of the line itself and any
vector with this direction is called a direction vector of the line. If
a is a direction vector, and P is a fixed point on the line, we assume
that a point X lies on the line if and only if PX is parallel to a.
Similarly we assume that, given a point P and a non-zero vector
a, there is a unique line through P with direction vector a.
Everything that was said about the normal vector of a plane
applies, with suitable modifications, to the direction vector of a
line. Thus if a is a direction vector, so is ka if k 0, and the zero °
vector cannot be a direction vector.
In many textbooks, the components of a direction vector of a
line are called direction ratios of the line, or, if it is a unit vector,
they are called its direction cosines. In addition to the reasons
already given for modifying this terminology, it does not seem
satisfactory to use the same terms in referring to planes and lines,
when the relationship is so different. The direction cosines of a
line (or a plane for that matter) are not completely unique. There
are two unit vectors with any given direction, having opposite
senses, so there is a choice of sign; but one cannot change the sign
 § 12 COORDINATE GEOMETRY 85

of one direction cosine without changing the sign of all three.

Continual use of direction cosines (or, in our terminology, unit
direction vectors) seems to lead to unnecessary introduction of
irrational square roots, and we have found it better to use direction
vectors of any length although they are not unique.

12. Parametric equations of a line

Let a ~ 0 be a fixed vector, P(p) a fixed point, X(x) a variable
point. We wish to find an equation for the line Z which passes
through P and has direction vector a. The following four state-
ments are equivalent.
(i) X lies on ZL;
(ii) PX is parallel to a (or is zero);
Sate . °

(iii) x—p = ta, where ¢ is a number (Chapter 2, § 11);

(iv) x = p-+ta, for some number f. (12)
Equation (iv) is thus a parametric equation of the line L, the
parameter ¢ being allowed to take all real values including the
value zero. The value t = 0 corresponds to the point P itself.
In terms of coordinates (x, %2, X3), this single vector equation
splits into three component equations
X= Py +hd,,
Ly, = Pottds, (13)
Ly = Pgttds,
Here (p,, M2, P3) is a point on Land (ay, dz, A) are the components
of its direction vector.
If the (a, y, z) notation is used for coordinates, the parametric
equations of the line through (9, q, r) with direction vector (a, b, c)
are
x = p-+ta,
y = q+tb, (14)
z2=r-te.
If none of the components of the direction vector is zero, one can
eliminate the parameter ¢ to obtain the following explicit forms
' instead of (13), (14) respectively:
MP Lea Pe! ta Ps (15)
ay ae a3
 COORDINATE GEOMETRY Ch. 6
86

mph pgb Umer a6)

a c
If any of the direction ratios is zero, each of the last two forms
of equation (15), (16), interpreted in the ordinary way, is meaning-
less, since division by zero is impossible. However, it is then
possible to attach a meaning to these equations by using the con-
vention indicated in Chapter 2, § 8, Ex. 3, that 0/0 can take any
value, and a denominator can be zero provided the numerator is
also. When in doubt, one should always return to the parametric
form of equation.

Example. Find equations of the line joining the points A(1, 0, 2)

and B(2, 1, 0).
cos
The direction of the line is given by AB = (1, 1, —2), and
A(1, 0, 2) is a point on the line. Using formula (16), with (a, b, c)
= (1, 1, —2) and (p, g, r) = (1, 0, 2), the equations are
Bl “Yrpeiese2
1 1 —2’
or, in parametric form,
e=14+t, y=t, z= 2—26.

13. Non-uniqueness of the equations

The equations (15), (16) of §12 are often referred to as the
standard form of the equations of a line. Yet, even if we are
restricted to this form, the equations of a given line are by no
means unique. For instance it can easily be verified that the lines
x—l Ys ae euee x—2 we z+1
2 4 6’ 1 2 3
are both the same, the first equations being obtained from the
second simply by adding the constant 1 to all three expressions and
then dividing across by 2.
It is easy to find out whether or not two lines are distinct if we
are given their equations in standard form. First we check whether
their direction vectors are parallel, then we take a point on one
of them and check whether it lies on the other. If the vectors are
parallel and if also the lines have a point in common, then the
 § 14 COORDINATE GEOMETRY 87

lines are the same, but if either of these two conditions is not
satisfied, the lines are distinct.

14. Line as the intersection of two planes

When the equations of a line are written in the form (16) it is
seen that the line consists of the points whose coordinates satisfy
two linear equations. Each linear equation taken separately
represents a plane, so the points of the line are the points common
to the two planes, or, the line is the intersection of the two planes.
Of course the linear equations (16) are of a rather special form, one
involving x, y only, the other involving y, z only. It is intuitively
clear, however, that the intersection of any two planes is a line.
We shall now give an algebraic proof of this fact, though some
readers may be willing to take this as a basic assumption and
omit the proof.
Let a.x = k, b.x = 1 be the equations of two planes which are
not parallel, so that the vector b is not parallel to a. Thus if 6 is
the angle between a and b, cos 0 #1land| a |?| b |?—(a.b)? 40
_ We show first:
There is a point p which lies on both planes.
Let p = ja+ub. We shall try to determine values of A, u
which make p lie on both planes. The conditions for this are
a.(Aat+ub) =k, b.(da+pb) = I.
If we solve these simultaneous equations in A, u in the usual
way by elimination, we find
A([a|®|b|2—(a.b)?) = k| b |?—U(a.b),
yw ({al?| b [2—(a.b)) = 2] a |?—k(a.b).
The common coefficient of 4, u is not zero, so there is a solution
and thus we obtain a point p common to the two planes.
It is now easy to show that the intersection is a line. If a point
x lies on both planes we have a.(x—p) = b.(x—p) = 0. The
vector x—p is perpendicular to both a and b, so it is parallel to
their vector product. The locus of x is thus the line through p
with direction vector a xb.
The idea behind this proof may be clearer if the following
numerical example is studied.

Example. Express in standard form the equations of the line of

 COORDINATE GEOMETRY Ch. 6
88
intersection of the planes:
x—2y+3z = 1,
2Qa+y+z = 3.

In order to obtain the equations of the line in standard form, we

require two things: (i) a point on the line, (ii) a direction vector.
(i) We put z = 0 in the above equations, so that they reduce to
two equations in two unknowns. Solving, we find x = y = —1.
The point (—1, —1, 0) thus lies on the line.
(ii) Since the line lies in both planes, its direction is perpen-
dicular to the normal vectors of both planes. Thus the vector
product of the normal vectors is a direction vector for the line of
intersection. The normal vectors of the two planes are, respectively
(1, —2, 3) and (2, 1, 1) with vector product (—5, 5, 5). The equa-
tions of the line in standard form are thus

wel bay y+ _ 2
> 5 5.
or, simplified: —(x+1) =y+l1 =z.

Note. It sometimes happens that, if we put z = 0 in the equa-

tions of the two planes, the resulting equations in x, y have no
solution. In such a case it will be possible to find a point on the line
by putting x = 0 or y = 0, unless, of course, the planes are
parallel.

15. Angle between two lines. Angle between line and plane
There is no difficulty about defining the angle between two lines
in space.
Definition 8. The angle between two lines is the angle between
their direction vectors.
This is an extension of the idea of angle between two lines in
plane geometry, since two lines in space do not as a rule intersect
and thus do not ‘contain an angle’ in Euclid’s sense. However, the
definition is in line with the more modern concept of amount of
rotation, because it is the amount of rotation necessary to bring
one of the lines into a position parallel to the other. The angle
§ 15 COORDINATE GEOMETRY 89

between two lines can again be calculated from the formula

a.b
cos 9 =
[a||b|-
The problem of defining the angle between a line and a plane is
not quite so straightforward. One might be tempted to define it as
the angle between the direction vector of the line and the normal
vector of the plane, but a moment’s thought shows that this is not
at all suitable. For instance, if a line actually lies in the plane, the
angle between the line and the plane ought clearly to be zero,
yet it is at right angles to the normal vector. Again, a line parallel
to the normal vector ought clearly to make an angle of 90° with

normal
vector line

FIG. 27 Angle between line and plane.

the plane. These considerations, together with a study of Fig. 27,

suggest the following definition.
Definition 9. If ¢ is the angle between the direction vector a of
a line and the normal vector b of a plane, the angle 6 between the
line and the plane is defined to be
Tt
6 — oe

The angle 0 can be calculated from the formula

ma ) a.b
sin 6)(=.cos'¢) = =>:
Ja||b|
Two other definitions of the angle between a line and a plane
can be given. (i) 9 is the angle between the line and its projection
on the plane. (ii) 6 is the minimum angle between the line and
any line which lies in the plane. It is easy to verify that these
definitions lead to the same value of the angle as Definition 8.
 COORDINATE GEOMETRY Ch. 6
90
The condition that a line with direction vector a should be
parallel to a plane with normal vector b is that the scalar product
a.b should vanish, in contrast to the usual situation where
vanishing scalar product is associated with perpendicularity.

Example. Find the angle between the plane x+z = 0 and the
line of intersection of the planes x = y, x+-y—z = 1.

Solution. The normal vector a of the first plane is (1, 0, 1). The
direction vector of the line is the vector product of the normal
vectors of the second and third planes:
(1, —1, 0) x(1, 1, —1) =, 1, 2).
The desired angle is given by the relation
sin6 =a.b/|a||b| = 3/+/2./6 = 4v3.
Thus the angle is 60°.

Exercises 6.1

1. Find the equation of the plane containing the three points

(1;:0;.1),. (1s: Die Celik)
2. The vertices of a triangle are (1, 0,1), (0, 1, —1), (—2, 2, —2).
Find the lengths of its sides, the cosines of its angles and its area. What
is the equation of the plane containing it? Find the same information
for the triangle with vertices (1, 1, —1), (2, 0, 1) and (—4, 2, 8).

3. A triangle has vertices (—1, 1, —1), (2, 3, 4), (3, 4, 6). Find its
area and the perpendicular distance from the origin to the plane
containing it.

4. A plane passes through the points (0, 1, 2) and (1, —1, 0). It is
parallel to the direction (1, —1, 1). Find the equation of the plane and
the foot of the perpendicular on it from the point (3, 0, 3).
5. For each of the following triangles, find the lengths of the sides,
cosines of the angles and the coordinates of the foot of the perpendicular
from the origin on the plane of the triangle.
(i) (—2, 0, 1), (—1, l, 2), (—3, 2, 0);
(ile (Lad) .0(2,.0,1), (4/3, 2).
 Ex. 6.1 COORDINATE GEOMETRY 91
6. Three points A, B, C have coordinates
A(1, 2, 3), B(2, 3, 1); C(3, 0, —1),

and O is the origin. Find the cosine of the angle between the planes
ABC, OBC.
7. Show that the equation of the plane which cuts the coordinate
axes at the points (a, 0, 0), (0, b, 0), (0, 0, c), where a, 6, and c are
non-zero, is
asaat
tLZ
Gurr’
Two planes cut the axes in A, B, C and A’, B’, C’ respectively. If O is
the origin and G, G’ are the centroids of the triangles ABC, A’B’C’
respectively, show that OG is at right angles to the plane A’B’C’ if and
only if OG’ is at right angles to the plane ABC.
8. Find the point of intersection of the line with parametric equations
x= 44+, y = 1—t, z = 3¢ and the plane 27+4y+z = 9.

| 9. Show that the intersection of the line x = a--tb (parameter £)

. | and the plane n.x = k has position vector
°
k—n.a
at mas b.

10. Find equations for the line through the point (3, 5, 2) perpen-
dicular to the plane 5e—7y-+4z = —2. Obtain the coordinates of the
foot of the perpendicular on the plane from the point.

11. A plane is parallel to both the lines

Le EY et Se etre Ley tate
eeead at, 2
It also passes through the point (1, 0, —1). Find its equation.

12. Find the cosine of the angle between the following two lines:
(i) e—l =y = 2—I, (ii) e=14+4, y=5t, 2-= —t.
Find also the coordinates of the foot of the perpendicular from the
origin on the second line.
13. Show that the three planes y = 2+-1,z2 = 2+1, x = y+] inter-
sect the plane x+y-+z = 0 in three lines which are sides of an equilateral
" triangle.

14. Find the direction cosines of the lines of intersection of the

 COORDINATE GEOMETRY Ch. 6
92
following three pairs of planes:
(i) ety+2z2=2, *—-y—-z= (50
(ii) 2a—y = 3, a+y+4z2=1;
(iii) e—ytz=4, 2e—2y—3z2 = 5.
Find also equations of these three lines in parametric form, and the
cosines of the angles between them.
15. Find the angle between the line « = y, z = 0 and the plane
yt+z=0.

16. If1 is the line

and m is the line through the point (5, 4, 2) which cuts / at right angles,
find equations for m and the coordinates of the point of intersection
of J and m.

16. Planes containing a line of intersection

If we are given two equations, there are many ways of deriving
new equations from them by the ordinary rules of algebra. Con-
sider, for instance, the two equations of § 14,
x—2y+3z = 1, 2eaty+tz = —3. (17)
One could eliminate z, multiplying the first equation by 2 and
subtracting the second, to obtain
—5y+5z = 5, (18)
or, dividing by 5,
—ytz=1. (19)
Similarly one might eliminate y to obtain
xtz=—l. (20)
Again, one could simply add the two equations together:
3x—y+4z = —2. (21)
The equations (18), (19), (20), (21) are all derived from equations
(17) by the same kind of process; each of the two equations (17) is
multiplied by a constant and the resulting equations are added.
Thus, if equations (17) are denoted by U, V, equation (18) is
2U—V, (19) is 3U—4V, (20) is 3U-+2V and (21) is U+V. All
these are included as particular cases of the general formula
§ 16 COORDINATE GEOMETRY 93
kU-+1V, which, written out in full as an equation, is
(k+2l)e+(—2k4+])y+(8k+Dz = k—3l. (22)
This method of deriving new equations is the same as that used
in the solution of simultaneous equations. Any solution of the
simultaneous equations (17) must satisfy also the equations (18),
(19), (20), (21) or, more generally (22).
What we have done so far is pure algebra, and no reference has
been made to any coordinate system or to any loci in space; but
of course every one of the equations we have written down does,
in fact, define a locus. Since the equations are all of the first
degree, they represent planes. Let us now consider the geometrical
interpretation of these equations.
Any solution of the simultaneous equations (17) gives the
coordinates of a point on the lineL of intersection of the two planes
x—2y+3z = 1, 2e+y+z2 = —3.
Thus any point of L satisfies (18); in other words, equation
(18) defines a locus which contains the line L. Also the locus (18)
has a linear equation, so it is a plane. The equation (18), therefore,
defines a plane containing the line L. The same argument shows that
the other equations (19), (20), (21), (22) also define planes con-
taining L.
Equation (22) is a general type which includes the others; for
instance (18) is obtained from (22) by putting k = 2,1 = —1 and
(21) is obtained by putting k = 1 = 1. We shall see in the next
section that every plane containing L can be obtained from (22) by
giving suitable values to the constants k, 1. First of all we illustrate
the usefulness of the method by working two examples. (These
examples can, of course, be worked out without using the kU+1V
method, by obtaining the equations of the line Z in standard form
and using methods of the kind given earlier in this chapter. The
reader will find that these alternatives are quite a bit longer.)

Example 1. Find the equation of the plane containing the point

(1, 1, 1) and also the line LZ defined by equations (17).

Solution. Equation (22) represents a general plane containing L.

We wish to determine the constants k, 1 so that the point (1, 1, 1)
 COORDINATE GEOMETRY Ch. 6
94

satisfies the equation, that is,

(k-+21).1+(—2k+]).1+(3k+1).1 = k—3l.
Simplifying this, we find k = —7l. A possible solution is
k =1,1 = —1, giving, for the equation of the plane required,
5a—l15y+20z = 10,
or, divided across by 5,
x—3y+4z = 2.

Example 2. Find the equation of the plane containing the line L

defined by equations (17) which is parallel to the vector (2, 1, 0).

Solution. Again we consider equation (22), which represents a

plane with normal vector
(+21, —2k4+1, 3k4+1) (23)
If the vector (2, 1, 0) is parallel to the plane, then it is at right
angles to its normal vector (23) and their scalar product vanishes.
That is,
2(k4+-21) +1(—2k4+-1)+0(3k+41) = 0.
This reduces to / = 0, and the desired plane is
x—2y+32z = 1,
the first equation (17), which does, of course, define a plane
containing L.
The reader may be surprised that in these examples we arrive
at one equation in two unknowns fk, l. This is not really surprising,
since the equation of a plane may be multiplied by a constant
factor without altering the locus. Thus it is the ratio k:1 that
matters. For instance, equations (18), (19) denote the same locus,
though (18) arises from & = 2, 1 = —1 and (19) arises from

17. Planes containing a line of intersection (theory)

Suppose given two planes
a.x—k = 0, b.x—l =0 (24)
with line of intersection Z. We assume, of course, that the planes
are not parallel, so that the vector a is not a multiple of b. If 2,
# are two constants, not both zero, we may multiply the first
 § 18 COORDINATE GEOMETRY 95
equation by A, the second by yu and add, to obtain
(Aa+ub).x—(Ak+ul) = 0. (25)
The vector 4a-+yb cannot be zero, since a is not a multiple of
b, so equation (25) is a linear equation and represents a plane.
We now prove three facts about the plane (25).
I. The plane (25) contains the line L.
Il. If p is any point not on L, the constants 1, u can be so chosen
that equation (25) represents a plane passing through p.
III. Every plane containing L has an equation of the form (25).

Proof of I. If a vector x satisfies both equations (24), then it

satisfies equation (25), so the position vector of any point of L
satisfies (25). Thus every point of L lies on the plane (25) and the
plane contains the whole of L.

Proof of II. The condition that p should lie on the plane (25) is
simply A(a.p—k)+pu(b.p—l) = 0. This is certainly true when
A, u take the special values 1 = b.p—l, uw= —(a.p—k). These are
not both zero, since p does not lie on L.

Proof of III. Let II be any plane containing LZ. Let p be any

point of II which does not lie on L. By JZ, we can find a plane II’
with an equation of the form (25) which passes through p. Then
II and II’ must coincide, for their intersection contains the point
p as well as the line LZ, though the intersection (if any) of two
distinct planes is a line only. Thus II = II’, and II does indeed
have an equation of the form (25).

18. Problems
The theory that has now been given includes the basic informa-
tion needed to solve a fair variety of problems about planes and
straight lines in space. Many different types of problem can be set,
and one cannot expect to prepare in advance for all types. One
must be ready to meet problems of types one has not seen before
and expect to spend some time planning a sequence of steps for the
‘solution before beginning calculations. Problems will vary from
relatively simple ones, like the equation of a plane containing
three given points, to quite intricate ones, such as the problem of
 COORDINATE GEOMETRY Ch. 6
96
lines,
finding the shortest distance between two non-intersecting
in § 19. With these more involv ed
which we shall deal with in detail
problems, it is a good idea to write down a plan on paper before
beginning to calculate. First you should attach names or symbols
to the points, planes, lines and vectors occurring; it 1s a good idea
to include normal vectors and direction vectors, though these will
not always be needed in the solution. Then, write down, in the
form of equations if possible, any geometrical information given
in the question. After that, plan your way through the question
step by step. Do not be in too much of a hurry—once a plan has
been made, the detailed working will require surprisingly little
time. Each problem has to be considered on its merits. There is no
standard method which will solve all problems, nor is there a
single ‘right method’ of solving any particular problem. Any
logically valid method is right, though some may be longer than
others. We shall illustrate this point by giving more than one
method of solution for some of the problems given below.

Problem 1. A plane contains the line

x—-l y-—2 2-1
1 —1 Dy
and is parallel to the line

Find its equation.

First method. Call the first line LZ, the second M and the plane
P. Let the direction vectors of L, M be a, b, and the normal
vector of P be n. Clearly n is at right angles to a, b (§ 15), so we
may take n = a xb. This gives the left side of the equation of P.
To find the constant, we need one point on P—a point on L will do.
In the actual problem as given, a = (1, —1, 2), b = (1, 2, 3), so
that n = axb = (—7, —1, 3), giving for P an equation of the
form —7x—y+3z =k. Now the point (1, 2,1) lies on Z and
therefore on P, so k = —7.1—2+3.1 = —6. Thus the answer is
—Tx—y+3z = —6.

Second method. Use the kU +1V theory (§ 16, Ex. 2). For this the
 § 18 COORDINATE GEOMETRY 97

line Z must be defined as the intersection of two planes, which we

may take to be
ted,
x—l
1
2g.
—2
—1
4 oer
—
—l
_
pe
or, rty =3 and 2y+z2=5.
Multiply the first of these equations by k, the second by J, and
add, to obtain the general plane containing L, that is:
ka +(k+2l)y+le = 3k+51.
The normal vector (k, k+2l, 1) must be at right angles to the
direction (1, 2, 3) of M. Thus the scalar product of these two
vectors vanishes: k+2(k+2l)+3l = 0. This gives k = —7,1 = 3,
and the equation —7z—y+3z = —6, as before.

Problem 2. Find the equations of the projection of the line

—25 =y 2-4 (26)

on the plane x+t+2y+z2 = 3. (27)

Solution. The projection of a line or curve on a plane, of course,

is the locus of the feet of the perpendiculars from its points on the
plane. This is a type of question where the parametric form of
equations is clearly useful. The parametric equations give the
coordinates of a ‘typical point’ on the locus. We can find the
coordinates of the foot of the perpendicular from this typical
point on the plane. This will be a typical point on the projection,
and the expression for it will constitute parametric equations for
the projection. To carry out this plan, let us write equations (26)
in parametric form:
P(x, y, z) = (5+38, t, 4+1)
If Q is the foot of the perpendicular from P on the plane, then
—> —
PQ is parallel to the normal vector (1, 2, 1), that is PQ = (k, 2k, k),
where k is a number. Then the position vector of Q will be
> => =>
,0OQ = OP+PQ, or
(543t+kh, t-+2k, 4+¢+h).
Now Q lies on the plane (27), so its coordinates satisfy the
Vv A—D
 COORDINATE GEOMETRY Ch, 6
98
equation (27) and we have
5+3t+k-+2(t4+2k)+4+t+k = 8,
reducing to k = —t—1. Substituting for &, we find that Q is the
point (4-+2t, —2—t, 3). This is the desired expression for the
coordinates of a typical point on the projection, and the para-
metric equations for the projection are
e= 412i, y= —2-t, 2=3.
This should be an adequate answer to the question as set above,
for if it is not specified whether explicit or parametric equations
are required, one should assume that either will do. If we had been
asked for explicit equations for the projection, however, we should
now eliminate the parameter ¢ to find
w—-4 y+2
5 ay 2 ee AY:

Problem 3. Let O be the origin, and let LZ be the line with

equations

Find a point P on LZ such that OP makes an angle of 45 degrees

with L.

First method. The direction of L is a = (1, 2, 1). If P lies on L,

—>

then, for some value of t, we have OP = (é, 1+2#, 3+). For the
cash
angle between OP and a to be 45°, we must have, by the definition
of scalar product,
—> —>
a.OP =|a||OP|/+72,
which reduces, on using the formulae, to
GE+5 = /6 4/(6f2+10£+10)/
1/2.
Squared and simplified, this reduces to
1812305 = 0,
giving t = 3(—5+ 1/35). The two values of ¢ give two points P
§ 18 COORDINATE GEOMETRY 99
satisfying the conditions:
(—§+3V35, —§$+3V/35, 42+44+/35),
(—$—-6V35, —§3—$/35, 12—3+/35).
Second method. If X is the foot of the perpendicular from O on
L, then XP must be equal in length to OX (Fig. 28). Let us begin

line L
x |?
FIG. 28

by finding the coordinates of X. Since X is on the line L, we must

have, for some value of ¢,
—>
OX ='(@ 14-28, 3-40). (28)
——Se
We want to choose the value of ¢ which makes OX perpendicular
——>
to L, that is, the scalar product of OX with the direction vector
(1, 2, 1) of Z must vanish. The condition for this is found to be
5+6t = 0, t = —5. Substituting for ¢ in (28), we find
cts,
OX = (—4,6? —3,32)
39 and | OX |? = 38.
Now XP is parallel to (1, 2, 1), since it joins two points on L.
cat
Thus there is a value of uw such that XP = (u, 2u, uw). From the
equation |XP |? = | OX |?, we find that vw= +34+/35, so that
ee eo
OP. = OX+ XP = 3(—5, —4, 13)+3+/35(1, 2, 1),
as found before.

Problem 4. A line contains the point (1, 0, 2) and meets each of

the lines
e=y =2+2, (29)
e4+3 = —ly =k. (30)
Find its equations.
 GEOMETRY Ch. 6
100 COORDINATE
wn line NV.
First method. Call the known lines L, M, the unkno
point on V
Let (a, b, c) be a direction vector of N. Then a typical
the point P
will be (1-tat, bt, 2+-ct), where ¢ is the parameter. If
where L meets N corresponds to the value t = k of the parameter,
the coordinates (1-++ak, bk, 2+-ck) of P must satisfy equations (29),
giving
l+ak = bk = 4-+ck. (31)
If the point Q where M meets N corresponds to the value
— lof the parameter, it follows by a similar argument that
4+al = —}bl = 4(2-++l). (32)
All our information has now been written down in the form of
equations (31) and (32), giving only four equations in the five
unknowns a, b, c, &, l. It might thus seem that we have too few
equations, but again, as in § 16, we have a case where only the
ratios of three of the variables matters. The variables a, b, c in
fact represent a direction, and it is merely their ratio that is
important. To complete the solution, eliminate k from equations
(31) and J from equations (32). We find
4a—3b—c = 0, 2a—5b—4c = 0.
From these we find that a:6:¢ = 1:2: —2, and the equations
of the line are

Second method. Let A denote the point (1, 0, 2). The plane
through LZ containing A, which we shall call the plane (A, Z),
must contain JN since it contains two points of it, the point A and
the point P. Similarly, the plane (A, M) contains N; in fact, NV is
the intersection of the planes (A, Z), (A, M@). Our plan then is:
Find the equations of the planes (A, LZ) and (A, M). These two
equations together define the line NV.
The plane (A, Z) contains the point A(1, 0, 2) and also the
point (0, 0, —2) on L. Thus its normal vector is at right angles to
the vector (1, 0, 4) joining these points. Its normal vector is also
at right angles to the direction vector (1, 1, 1) of ZL. A possible
normal vector is thus the vector product (—4, 3, 1). Using the fact
that the point (0,0, —2) is on L, the equation of the plane
§ 19 COORDINATE GEOMETRY 101

(A, Z) is found to be
—4¢4+4+3y+z2 = —2. (33)
An exactly similar method gives us the plane (A, Jf):
24 —5y—4z = —6. (34)
Equations (33), (34), taken together, are a set of equations for
N. If equations in standard form are required, they can be worked
out as in § 14.

19. The common perpendicular of two skew lines

Two lines L, M in space each defined by two equations, represent
a total of four equations in three unknowns. There is not, in
general, a solution, so the two lines do not usually intersect. Two
lines in space which do not intersect and are not parallel, are
called skew. If L, M are skew lines, one may look for two points,
P on L, Q on Y, for which the distance PQ is as small as possible.
It is easy to convince oneself that this minimum value occurs
when PQ is at right angles to both lines. The length PQ is then

(b)
FIG. 29

called the shortest distance between the lines, and the line PQ is
called their common perpendicular.
The problems of finding the points P, Q, the equations of the
common perpendicular, and the shortest distance when the equa-
tions of L, M are given, can be solved in several different ways.
Before considering these, let us study the diagram consisting of the
two lines L, VM and the points P, Q at the ends of the common per-
pendicular (Fig. 29a). To make the diagram more symmetrical, it
seems natural to add to it the line L’ through Q parallel to Z and
the line MU’ through P parallel to M (Fig. 29b). The new diagram
EVA—D®
 COORDINATE GEOMETRY Ch. 6
102

resembles two wedges placed edge to edge. It contains five lines

PQ, L, L’, M, UV’, four planes PQLL’, PQMM’, LM’, L’M and the
two points P, Q. Notice that the planes LM’, L'M are parallel,
that PQ is a normal vector of each of them and that the shortest
distance between LZ and I is the perpendicular distance between
these parallel planes. It is now easy to see many ways of attacking
the problem. We give two methods, taking the following numerical
example.

Example. Find (i) the shortest distance,

(ii) the equations of the common perpendicular,
(iii) the coordinates of the feet of the common
perpendicular,
for the two lines
Lh: x-l=y =2+5, (35)

M: 5= wee2 es (36)

First method. Write down the direction vectors of L, M:

a= (It) 1), = bee 2).
Any point P on L has coordinates

and any point Q on J has coordinates

(8u, —2+2u, —2+4). (38)
We have to choose the values of the parameters t, w so as to
—>
make sure that PQ is at right angles to both a and b, that is, so
—> —>
that the two scalar products PQ.a and PQ.b both vanish. By
subtraction, we find that
eS
PQ = (—1—#+3u, —2—t4+2u, 3—#4+u).
Using the formula for scalar product, we find the two equations
— —>
PQ.a = —3t+6u = 0, PQ.b = —4—6t+14u = 0.
These have the solution ¢ = 4, u = 2, so the desired points
P, Q, obtained by substitution in (37), (38) are P(5, 4, —1),
Ss
Q(6, 2, 0). Thus PQ =(1, —2, 1), and the shortest dis-
tance = |PQ| = 6.
 § 19 COORDINATE GEOMETRY 103

Equations of the line PQ, from its direction PQ and the point P
on it, are:
y—4
—5 = 2 = 2-1.
2Ep,

Second method. We find PQ as the intersection of the plane

PQLL’, which we call II,, and the plane PQMM’, which we call
IT,.
Detailed plan
(i) Work out direction vector of PQ. Call this c.
(ii) Work out equation of IT,.
(iii) Work out equation of II,.
(iv) Obtain Q as the intersection of II, and /.
(v) Obtain P as the intersection of II, and L.
(vi) Shortest distance = |PQ |.
Actual working
(i) PQ is at right angles to a, b, soc = axb = (1, —2, 1).
(ii) II, contains PQ, parallel to c, and it also contains L, parallel
- to a. Normal vector of II, =cxa = (—3, 0, 3). Dividing by
common factor 3, II, has equation —x+z = k. Point (1, 0, —5)
on LZ is on II,,so k = —1—5 = —6.
Equation of II,: —a+z = —6. (39)
(iii) As in (ii), one finds
Equation of II,: —2%+y+4z = —10. (40)
(iv) Q is on YZ, so its coordinates (x, y, z) satisfy the equations
(36) of JZ. Also Q is on II,, so it satisfies equation (39). Solving
equations (36) and (39) we find that Q is (6, 2, 0).
(v) A similar calculation gives P(5, 4, —1).
(vi) As before, |PQ| = -/6.
Notes
(a) In the second method, the equations of the common per-
pendicular are obtained as (39) and (40) simultaneously. There is
no need to go further than Step (iii) if only the equations of the
common perpendicular are needed. On the other hand, if all the
information is required, the first method is rather shorter.
(6) Neither method works out the problems in the order in
 GEOMETRY Ch. 6
104 COORDINATE
ons in
which they are posed. The first method answers the questi
(ii) and the second method answer s them in the
the order (iii), (i),
order (ii), (iii), (i). It is a mistak e to suppos e that the logical order
for answering a number of related questio ns is necessa rily the
order in which they are asked.

Third method (for shortest distance only)

Sometimes only the shortest distance is needed, not the actual
points P, Q or the equations of PQ. In that case the following
method is best.
The shortest distance is the perpendicular distance between
the planes LM’, L'M. Both these planes have normal vector
axb = (1, —2, 1), so their equations are of the form x—2y+z = k,
a—2y+z =k’. The first plane contains the point (1, 0, —5), on L,
so k = —4, and the second plane contains the point (0, —2, —2),
on M, so k' = 2. The distance between the two planes is, by § 7,
[kk |/|c| = 6/6 = 6.

Exercises 6.2

1. Two planes (A), (B) each pass through the origin. The plane (A)
contains the line

and the plane (B) contains the line of intersection of the planes
atytz=1, 2u—y+3z = 2.
Find the cosine of the angle between the planes (A) and (B).
2. Find the feet of the common perpendicular of the two skew lines

x we YE PY 2 a
2 Dee ee ae
3. A line L is the intersection of the two planes
ctyte=1, x—2y+32z = 2.
Find the equation of the plane containing LZ and passing through the
origin. Show that the plane just found makes an angle of 60° with the
plane y+z = 0.
 Ex. 6.2 COORDINATE GEOMETRY 105

4. Find the equations of the common perpendicular of the skew lines

e230 -y 2=6: e+5 z
i So ty 2?
and show that the shortest distance between them is 1/3.
5. Find the angle between the planes
3x—4y+5z = 10, x—4y+3z = 15.
Find also the equation of the plane which passes through their line of
intersection and is parallel to the line x = 2y = 32.
6. Find the equation of the plane which contains the first of the
following two lines and also contains their common perpendicular.
z—1 y
i595) = —— of = —z—l, ctl = no => 2—3.

7. A line L is given by the equations

x z

Find the equation of a plane II parallel to LZsuch that it passes through

the line of intersection of the planes
3a+y+3z = 0, xt+y+2z2 = 0.
Find the perpendicular distance from L to II.
8. A line L is the intersection of the two planes
ety+z = 3, x—2y+3z = 2.
Another line M is given by the equations
e—l = y—2 = 2-3.
Find direction ratios of the line of intersection of the planes joining
L and M to the origin.

9. Four points in space have coordinates

Al. i, 0), B(s;-0-1),-C(1s.0).2),. D1, 1, 3).
Find the equations of two parallel planes, of which one contains A and
B and the other contains C and D. Deduce the shortest distance between
the lines AB, CD.

10. If O is the origin, P the point (1, 2, 3) and L the line of inter-
section of the planes a+y = 2, z+1=0, find the angle between the
~ plane containing O and L and the plane containing P and L.

11, Find the equation of the plane through the intersection of the
 COORDINATE GEOMETRY Ch. 6
106

planes
Qa—y—z = 2, 3a—2z = 5,
which is parallel to the line
are
ae tae 1c

Hence deduce the shortest distance between the line and the line of
intersection of the first two planes.
12. Two lines are given in space by the following equations:
= 4

Find the equations of the following planes:

(i) the plane containing the first line and parallel to the second,
(ii) the plane containing the second line and parallel to the first,
(iii) the plane containing the first line and passing through the
origin,
(iv) the plane containing the first line and the common perpendicular,
(v) the plane containing the second line and the common perpen-
dicular.
13. Find equations of a line with direction vector (1, 2, —3) which
meets both the lines
x—-2 ytl 2-3
“Ss gtSoe ee
14. Find equations of the projection on the plane
6x—3y+2z2 = 1
of the line of intersection of the planes
ety+2z = 3, sa+y+3z = 4.
15. Show that there are two planes which pass through the line
e-51 yl
—l
_ 2433
and make an angle of 60 degrees with the plane y = z. Find their
equations.
16. Find the equation of a line through the origin which meets both
the lines given in Question 12 above.
 7

OTHER APPLICATIONS
1. Change of coordinate system
One often needs to change from one coordinate system to an-
other. A choice of coordinate system involves two choices, a choice
of origin O and a choice of base-vectors u,, U2, Us. The change
from one coordinate system (O, u,, Uz, Uz) to another coordinate
system (O’, u,’, u,’, Us’) can therefore be done in two steps. First
change the origin O to the new origin O’, keeping the same base-
vectors. Second, change the base from u,, U,, U; to the new base
u,’, Uu,’, U;’, keeping O’ fixed as origin. We now consider the
equations governing these two moves.

(1) Alteration of origin ed

Let O be the old origin, O’ the new origin and let OO’ = a. We
wish to find the relation between the position vector x of a point
X relative to O and its position vector x’ relative to O’. From the
a eo
triangle OO’X we have OX = OO’+0’X, that is
x =a+x’, (1)
—
In terms of components (x, y, z), this becomes, if OO’ = (a, b, c),
C= Ce, =O, Zi e-F2: (2)

(2) Change of base

Suppose now that we wish to express our vectors in terms of a
new base u,’, u,’, u,’. Let the components of a vector x referred
to the old base be (x1, #2, v;) and let the components of the same
vector referred to the new base be 2’, x,’, x;’. We distinguish
between vectors given in terms of components relative to the first
base and those given in terms of the second base by the use of
round and square brackets. By the definition of components
(Chapter 3),
X = %,'U;'+2%,'U,’ +2,'U;’. (3)
Now we must also know the components of the new base-vectors in
 APPLICATIONS Ch. 7
108 OTHER
s:
terms of the old base. Suppose these are as follow
uy’ = (1,, mM, %);
Uy’ = (12, Ma, No)s
Us = (13, Mg, Ns).
We can now write down the component form of equations (3),
referred to the old base-vectors.
ay = lyr,’ +109’ +1 sts’,
Ly = MyLy'+Myly' +Mshs\, (4)
Le = N42 + No +NgXq
These are the equations we want, expressing the old coordinates
in terms of the new ones. Of course it is sometimes also necessary
to know the equations expressing the new coordinates in terms of
the old. One way would be to solve the equations (4) for 2’, x2’, x5’.
However, the following method is simpler. Since J, is the u,-com-
ponent of u,’, we have J, =u,’.u,. Similarly, 7, = u,’.u,,
1; = U,'.U,. Thus, in the new system, the components of u, are |
[1,, U2, 13]. This and two similar arguments yield |
u, = [1,, Le, Us],
U2 = [M,, M2, Ms],
Us = [M%, Ne, Ns].
Now, writing x = x,U,+2,U,+2%,U, in components referred to the
new base, we deduce
ay’ = 1,2, +m,%_,+N,Xs, |
Ge! = Let +Mg%e+Ngg, (5)
Ge! = 13%, +Mg%o+Nghy. |
Finally, note that the quantities 1,, ... , n3 are not arbitrary, |
because u,’, u,’, U;’ are mutually orthogonal unit vectors and we |
must have
L?+m?+n? = Lo?+m?-+no? = 1,2+m,?-+n,? = 1,
la +mym,+nn, = Ilgt+m m3+ngns (6)
= 1,1, +m m,+nn, = 0.

Exercises

1. Show that (3, Op —*)s (%, —t, 3), —4, &, a) form a system of
three mutually orthogonal unit vectors. If these are taken as a base for
a new coordinate system, with origin unchanged, what is the equation
§ 2 OTHER APPLICATIONS 109

in the new system, of the locus with old equation

x?taytazte? = 1?

2. Derive the first equation (5) by multiplying the equations (4) by

l,, M4, 2, and adding. (Use relations (6).) Obtain the other two equations
(5) by a similar method.

3. Show that, if equations (6) hold, then also

together with four similar equations.

4. If u,, u,, uz are three mutually orthogonal unit vectors forming

a right-handed system, fill in the missing components in the table below.

2. Products of four or more vectors

The formulae for triple products contained in Chapter 5 enable
one to transform products involving four or more vectors in a
variety of ways and thus to derive vector identities. One example
is worked to illustrate the method.

Example. Transform the product a x(b x (c xd) ) in two ways.

Solution. (i) Let us denote the above product by w and let

p = cxd. Then, using the vector triple product formula,
w =ax(bxp) = (a.p)b—(a.b)p
= [a, c, dJb—(a.b)(c xd).

(ii) Again use the vector triple product formula:

b x(c xd) = (b.d)c—(b.c)d.
Take vector product with a on both sides:
w = (b.d)(axc)—(b.c)(a xd).
1 Tf we equate the two expressions obtained for w and rearrange
the resulting equation in a more symmetrical form (remembering
that axd = —d xa), we derive
[a, c, dJb = (a.b)(c xd) +(b.d)(axc)+(b.c)(d xa). (7)
BV A—E
 OTHER APPLICATIONS Ch. 7
110

Exercises

= qxr,
1. If p= bxc, q = cxa, r = axb, show that [a, b, cla[a, b, c]?.
and two similar formulae. Show also that [p, q,r] =
Deduce that if p,q, r are known, then a, b,c are determined apart
from sign if [p,q,r] 49.
=—0 and |p| =|[a|=|]r]=]|s],
2. If p+q+r+s show that
|pxa|=|rxs|.
3. If a, b, c satisfy the conditions
|bxc| =|cxa|=|axb|=|(a—b)x(a—c) |,
[a, b, c] #0,
deduce from Questions 1, 2 that |a| = |b—c}.
Hence show that if the faces of a tetrahedron are all equal in area,
then the length of any edge is equal to the length of the opposite edge.
4. Show that (axb).(cxd) = (a.c)(b.d)—(a.d)(b.c).
5. Show that
(i) (axb)x(cxd a, b, dJc—[a, b, c]d,
(ii) (ax b)x(cxd a, c, djb—[b, c, dja.
Deduce that
[b, c, dJa—[a, c, d]b+-[a, b, d]c—[a, b, c]d = 0
Find an alternative proof of the last formula, using (7).

3. Geometry on the surface of a sphere. Spherical triangles

Let S denote the surface of a sphere centre O and radius r. Thus
Sis the set of points whose distance from O is equal to the constant
r. The geometry on the surface S has great practical importance,
because the surface of the earth is almost spherical. If I is any
plane containing O, the intersection of II with S will be a circle
centre O and radius r. Such a circle on S lying in a plane through
O is called a great circle, since there cannot be a circle of larger
radius lying on S. If A, B are two points on S not diametrically
opposite, then there is just one great circle containing A and B,
given by the intersection of § with the plane OAB. The minor arc
of this great circle is the shortest path in S from A to B. Thus the
great circles play the same role in geometry on the surface of the
sphere as the straight lines do in plane geometry.
A spherical triangle is the figure formed on S by three points
 §3 OTHER APPLICATIONS lll

A, B, C and the three minor arcs of the great circles BC, CA, AB
joining them in pairs. It is assumed that no two of the points A,
B, C are diametrically opposite. The angle between the arcs AB,
AC at A (that is, the angle between the tangents to these arcs at A)
is called the angle A and the angles B, C are similarly defined. In
Fig. 30, B’, C’ are the intersections of the great circles AB, AC
(produced, if necessary) with the plane B’OC’ through O perpen-
dicular to OA; and AT, AU are the tangents at A to the great
circles AB, AC. Since the tangent to a circle is at right angles to
the radius it follows that the angle A of the spherical triangle is
equal to the angle B’OC’, which in turn is equal to the angle be-

FIG. 30

tween the planes OAB, OAC (see Chapter 6, § 6). Notice however
that the angle A is defined with more precision than the angle be-
tween two planes, because a normal vector of a plane may be given
with a choice of two possible senses (it may point ‘above’ or ‘below’
the plane), giving two possible values of the angle which add up to
180°, while the tangents AT, AU will of course be given the sense
which is consistent with the direction of travel along the ares AB,
AC in the direction away from A. The angles are always taken to
be positive and less than 180°, as in plane trigonometry.
The sides a, b, c of the spherical triangle are defined to be the
' angles BOC, COA, AOB subtended at the centre of the sphere. If
the actual length of one of the ares BC, CA, AB is needed, it can
always be calculated by multiplying the appropriate angle,
measured in radians, by the radius r of S. Really, in spherical
112 OTHER APPLICATIONS Ch. 7

trigonometry, the radius of the sphere is taken as the unit of

length. The reader may find this measure of length more natural
if he remembers that a similar angular measure is used at sea, the
nautical mile being one minute of arc.
4, The cosine rule
The basic problem of spherical trigonometry is, as in plane trigo-
nometry, the solution of triangles. Given three of the quantities a,
b, c, A, B, C, to determine the other three. The main formula, from
which all the others can be deduced, is the following analogue of
the cosine rule:
cos a = cos b cosc-+sin b sinc cos A. (8)
Because of the importance of this formula, two proofs are given
(and one further proof is suggested later as an exercise). In the
> SO
first proof, let a, b, c be the position vectors OA, OB, OC of the
vertices of the triangle relative to the centre of the sphere. To
avoid unnecessary factors, we take the radius of the sphere to have
unit length, so that a, b, c are all unit vectors. The vectors a xb,
a xc are normal to the planes OAB, OAC, and it is easy to see that
the angle between them is the angle A, taking the sense of the
vectors into account and measuring the angle A as indicated in
§ 3. Moreover the magnitudes of these vector products are sinc,
sin 6 respectively. Thus, by the definition of scalar product,
(axb).(axc) = sinb sinc cos A. (9)
We can modify the expression on the left by using the scalar
triple product formula p.(axc) = a.(c xp). Putting p=axb
and using also the vector triple product formula, we find
(axb).(axc) = a.(cx(axb)) =a.((c.b)a—(c.a)b) =
= |a|*(c.b)—(a.b)(a.c) = cos a—cos b cos ¢ (10)
Comparing the two expressions (9), (10), we derive (8).
The second proof of this formula requires the introduction of a
coordinate system, and a study of it may help the reader to under-
stand how to choose a coordinate system to suit particular prob-
lems. It is helpful to think of the vertex A as the ‘North Pole’ and
the great circle AB as the ‘Greenwich Meridian’. The latitudes of
the points B, C will be 3 —c, $a —b respectively, and the longitude
of C will be the angle A. If the Greenwich Meridian cuts the equator
at X and the 90° meridian cuts the equator at Y, we can use
 § 4 OTHER APPLICATIONS 113

OX, OY, OZ = OA as axes of a rectangular cartesian coordinate

system. If we can work out the coordinates of the points B, C in
this system, we shall then be able to find the cosine of the angle
BOC from the formula for the cosine of the angle between two
vectors. (See Chapter 4.)
Let the great circle AC meet the equator OXY at W. Let BM,

FIG. 31

CN be perpendicular to the plane XOY so that M lies on OX,

N on OW. Then, assuming unit radius BM = cosc, OM = sinc
and OB is the vector (sinc, 0, cosc). Similarly, CN = cos 6,
re . °

ON = sin}. Also the angle XON is the angle A (between the

planes XOA, WOA). Thus the w- and y-coordinates of N, which are
the same as the x- and y-coordinates of C, are ON cos A, ON sin A,
, and the coordinates of C are

OG = (sin b cos A, sin b sin A, cos b).

Using the formula for scalar product in terms of components,
 114 OTHER APPLICATIONS Ch. 7

we have
> —->
cos a = OB.OC = sin b sin c cos A+cos b cos ¢, (8)
as desired.

5. Worked example
The usefulness of the cosine rule will be illustrated and the
principle behind the second proof made clearer by a study of the
following example.

Assuming the earth to be a sphere of radius 3963 miles, find

the distance from New York (40° 40’ N., 73° 50’ W.) to Moscow
(55° 45’ N., 37° 42’ E.).
Take the spherical triangle ABC with A at the North Pole, B at
New York, C at Moscow. The arc from the North Pole to any
point in the Northern Hemisphere subtends an angle 90°—I,
where J is the latitude. Thus, in the given spherical triangle,
b = 34° 15’, ¢.==49°:20%
Since New York is West of Greenwich and Moscow is East, the
angle A is the sum of their longitudes, so A = 111° 32’. For the
side a, representing the distance we want, the cosine rule gives
cos @ = cos 34° 15’ cos 49° 20’—sin 34° 15’ sin 49° 20’ sin 21° 32’.
(To come within the range of the tables, which run from 0 to 90°,
use the relation cos A = cos (90°+21° 32’) = —sin 21° 32’.)
We calculate each of the products with logarithms, and subtract
to find cos a. The details are below.
log cos 34° 15’ = 19172 log sin 34° 15’ = 17504 0:5386
log cos 49° 20’ = 18140 log sin 49° 20’ = 18799 0:1567
1-7312 log sin 21° 32’ = 1-5647 06-3819
product = 0:5386 1-1950
product = 0-1567
Hence cosa = 03819 and a = 67°33’. In radian measure,
Snr 4
= 67x ren and the distance required is
Geil

6723 x 33-1416 x 3963

= 4 672 miles.
180

6. The sine rule

Another useful formula which can be established with the aid of
 §6 OTHER APPLICATIONS 115

vectors is the sine rule:

sna sinb_ sine
= = ; 11
snA4 sinB sin@ ay
As in § 5, let a, b, c denote the position vectors, relative to
the centre of the sphere, of the three vertices of the spherical tri-
angle. The magnitude of the vector product (axb) x(axc)
is sin 6 sin c sin A, since the magnitudes of axb, axc are
sin c, sin b, and A is the angle between them. Now, by the
vector triple product formula,
(axb) x(axc) = ((axb).c)a—((axb).a)c = [a, b, cla.
Since a is a unit vector the magnitude of this expression is
| [a, b, c] | , and, by symmetry, we deduce
| [a, b, c] | =sindsincsin A = sincesinasinB
= sin asin b sin C,
from which (11) follows.
An alternative geometrical proof for (11) may be found interest-
ing (Fig. 32). Drop a perpendicular AL on the plane BOC and drop

FIG. 32

LP, LQ perpendicular to OB, OC respectively. Then the plane

' APL is at right angles to OB, because the lines AL, LP are. Simi-
larly the plane AQL is at right angles to OC. Since AP, PL are
both at right angles to OB, the angle APL is the angle B of the
spherical triangle and similarly the angle AQL is the angle C.
116 OTHER APPLICATIONS Ch. 7

From the right-angled triangles OPA, ALP,

AL = AP sin B = OAsincsin B.
By a similar argument, AL = OA sin b sin C, and (11) follows.

7. The polar spherical triangle

If a great circle is given on the sphere, the two points where the
line through the centre perpendicular to the plane of the great
circle cuts the sphere are called the poles of the great circle. If, for
example, the sphere is the earth and the great circle is the equator,
then its poles are the North and South Geographical Poles. If
ABC is a spherical triangle, let P be that pole of the great circle
BC which lies on the same side of this great circle as A lies. The
angle POA is then an acute angle. Let the poles Q, R of CA, AB
be similarly defined. The spherical triangle PQR is said to be the
polar spherical triangle of ABC. If p, q, r are the position vectors
of P, Q, R and a, b, c those of A, B, C, then we have the relations
p.a>0oO p.b=0 psci=10;
q.a=0 q.b>0 qQic 0,
ra 0 tb =, Lice 70;
It follows from the symmetry of these defining relations that,
if PQR is the polar triangle of ABC, then ABC is the polar triangle
of PQR.

FIG. 33
 § 8 OTHER APPLICATIONS 117

The angle A is the angle between the planes OAB, OAC, and
OQ, OR are normal vectors of these planes. The angle A is there-
fore equal either to the angle QOR, or to z—QOR. Consideration
of Fig. 33 will convince the reader that the second is the correct
expression. Now the angle QOR is, by definition, a side of the
polar triangle. Thus the three sides of the polar triangle are
a—A,ax—B,x—C. Because ABC is also the polar triangle of PQR,
we deduce in the same way that the angles of PQR are z—a,
2z—b, x—c. These facts enable us to derive a new formula from
any known formula in spherical trigonometry by putting 7—A,
az—B,ax—C,x—a,x—b,x—c instead of a, b,c, A, B, C respectively.
For example, if we apply the cosine rule (8) to the polar spherical
triangle we obtain
—cos A = cos Bcos C—sin B sin C cos a.

8. The area of a spherical triangle

It is known that the area of the surface of a sphere of unit radius
is 4%. Now consider the area of a segment of the sphere bounded by
two great semicircles which meet at an angle 0. It is clear that this

FIG. 34 Area of a segment is proportional to the angle 0.

, area is proportional to 0, and when 6 = 2, it is half the area of the

sphere. Thus the area of such a segment is 20.
Now consider a spherical triangle ABC. Let A’, B’, C’ be the
points diametrically opposite to A, B, C on the surface of the
118 OTHER APPLICATIONS Ch. 7

sphere. The three great circles ABA’B’, BCB’C’, CAC’A’ sub-

divide the surface of the sphere into the eight triangular regions
listed below (Fig. 35).
ABC, - ABC) “ABC. Face,
A’B'C’,, ABC’, “BCE LAB.
Each triangle is diametrically opposite to the one listed below it,
and therefore has the same area. Let x, y, z, w denote the four
areas in order. The first and second triangles in the top row taken

FIG. 35 Subdivision of the surface of the sphere into

; eight triangular portions by three great circles.

together form a segment with angle A. Hence, by the formula just

found for the area of a segment, x+y = 2A. Similarly x+z = 2B,
x--w = 2C, Again, since the sum of all the eight areas is the whole
surface of the sphere, x+y+z+w = 2x. If we add together the
first three equations just found and subtract the last one we
obtain «= A+B4+4+0—xz.
The area of a spherical triangle is the excess of the sum of its angles
over 7.

Exercises on Spherical Trigonometry

1. In Fig. 30 let b,c, b’, c’, a be the position vectors relative to
the centre of the sphere of the points B, 0, B’, C’, A. Show that
 § 9 OTHER APPLICATIONS 119

b = cosca+sincb’, c=cosba+tsindc’,
and that b.c = cosa, b’.c’= cos A.
Deduce the cosine rule.
(This proof was given in a terminal examination by a first-year
student of engineering at Queen’s College, Dundee.)
2. Deduce the sine rule from the cosine rule by using (8) and the
identity sin? A = 1—cos? A.
3. In an equilateral spherical triangle whose sides are all equal to a
and whose angles are all equal to A, show that
cos a = cos A+-cos a cos A,
4 cos? 4a sin? 44 = 1,
4. Show that if s = $(a+6-+c) then
sin s sin (s—a)
COs? 4A ea ee
: sinbsince ’
ary sin (s—b) sin (s—c)
sin? 4A = : : :
sin 6 sin c
sin (s—b) sin (s—c)
tan? 4A =
sin s sin (s—a)
5. If C = 3a, prove the formulae
cos c = cos
a cos b,
cos A tan c = tan
b,
sin A sin c = sin
a,
tan a = tan
A sin J,
cos A = cosa sin B.
6. If «, B, y are the angles between the line OA and the plane BOC,
between the line OB and the plane COA, and between the line OC and
the plane AOB respectively, show that
sin @ sin « = sin b sin f = sinc sin y = sin b sinc sin A,
sin? a cos? « = cos? b-++-cos? c—2 cos a cos 6 cos c.

9. The tetrahedron
If ABCD is a tetrahedron, one may wish to know the values of
various quantities associated with the figure, for instance:
(i) the six lengths BC, CA, AB, AD, BD, CD;
,
(ii) the twelve angles of the four plane triangular faces;
(iii) the six angles between pairs of plane faces;
(iv) the three shortest distances between pairs of opposite edges ;
(v) the angle between a face and an edge not contained in it;
120 OTHER APPLICATIONS Ch. 7

(vi) the volume;

(vii) the area of the plane faces.
These quantities are by no means independent, since, for instance,
if we know the lengths of the sides of a plane triangular face we
can deduce the angles and area by the methods of ordinary trigo-
nometry. In fact, there are only six independent constants needed
to define a tetrahedron, so if we take any seven of the quantities
above, there must be a relation between them. It is useful to take
one of the vertices, say, D, as origin, and work in terms of the
> —> — ;
vectors DA =a, DB =b, DC =c. A knowledge of the six
quantities
(a2) Dl], el? es ca, aD (12)
will enable us to calculate all the other quantities listed above.
Notice that the quantities (12) can be obtained at once from
equations of the type
2b.c = | b |?+ |c |?— |b—c|?
if we know the lengths of the six sides.
As an instance, let us show how to express the volume V in
terms of the quantities (12). The volume of a tetrahedron is known
to be one-third of the area of the base multiplied by the height, so
6V =a.(b xc). Now
| a.(b xc) |? =| al?| bx<ej2?— | ax(b xc) |”.
Simplifying this with the help of the vector triple product
formula,
36V2=|al?|b|?\e p> | a |?(b.c)?+2(b.c)(c.a)(a.b) (13)

Here the >» sign denotes summation over all the expressions
derived from the given one by cyclic interchange of the letters
a, Dec:
As another example, let us take the shortest distance between
the two opposite edges DA, BC. The line DA is given para-
metrically by the equation
x ta, (14)
and the line BC is given parametrically by
y = b+u(b—c). (15)
(By this we mean, as in Chapter 6, that any point on DA has a
§ 9 OTHER APPLICATIONS 121

position vector x given by (14), with a suitable choice of the real

number ¢, and any point on BC has a position vector y given by
(15), with a suitable choice of the real number w. Notice that we
have to use different letters for the coordinate vectors x, y,
because, of course, the points are not the same, and similarly we
need different letters for the two parameters ¢, wu.) Suppose now
that t, u are given values in (14), (15) so that x, y are the feet of
the common perpendicular, which will happen if x—y is parallel
to ax(b—c), the vector product of the direction vectors of the
two lines. We thus have
x—y = k(ax(b—c) )
or ta—b—u(b—c) = k(ax(b—c) ). (16)
In (16), take scalar product with a x(b—c) on both sides. We
deduce that
k|ax(b—c) |? = —b.(ax(b—c) )= b.(axc) = +607.

Thus |x-—y| =|k||ax(b—c)| = as : (17)

We have already seen how to express V in terms of the quanti-

ties (12) and the denominator in (17) can be expressed in terms of
them too by the use of the formula |p xq |? = |p |? |q |?—(p.q)’.
In geometrical terms, the formula (17) states:
The shortest distance between two opposite edges of a tetrahedron ws
equal to six times the volume divided by the product of the lengths of
the edges concerned multiplied by the sine of the angle between them.
Tf one wishes to work out angles between faces of the tetra-
hedron, the formulae of spherical trigonometry are helpful. For
let us consider the three plane faces that meet at the vertex A,
and imagine them produced indefinitely beyond the points B, C,
D. The three planes will cut out a spherical triangle on the sphere
with unit radius centre A, and the three sides of this spherical
triangle will be the angles BAC, CAD, DAB while its three angles
will be the angles between the faces of the tetrahedron which meet
at A.
To illustrate all these points, we evaluate some of these quan-
tities when ABCD is a regular tetrahedron, that is one with all its
faces equilateral triangles. If J is the length of one of its edges,
then, since all the angles between intersecting edges are $7, and
 122 OTHER APPLICATIONS Ch. 7

we have
faft=[b[?=[cl/? =P, = Hl?
b.c=c.a=a.b
For the volume V we find, from (13), 6V = 1?/4/2. To derive the
shortest distance between a pair of opposite edges, we note the
fact that, in this case a.b = a.c and a is at right angles to b—c.
Thus | ax(b—c)| =|a]||b—c| =/?. Substituting in (17) we
see that the shortest distance is 1/4/2. Finally, to find the angle «
between two faces we use the formula
cos a = cos b cosc-+sin b sin c cos «.
In this case we have a spherical triangle with a = b =c = Ty
Ol

so
i
mace
ee ye)
sree COS «, and cosa = 3,

By symmetry, the centroid (Chapter 2, § 15) $6= }(a+b-+c) of

a regular tetrahedron is equidistant from the four vertices. This
can be verified by evaluating the lengths:

ls[?= (> ja ?+2 > b.c) = 3)2,

ee byes al b-+c|?—6(b+c).a+9|a ?)= 312

For applications in chemistry, it is useful to know the angle 6
subtended at the centroid by the edge DA. We have
312 cos 6 = $.(6—a) = —H?.
Thus cos 9 = —i, = 7%—cos—! 3,

Exercises on the tetrahedron

1. Show that the points (0, 0, 0), (0, a, a), (a, 0, a), (a, a, 0) are the
vertices of a regular tetrahedron of side a/2. Find, by the methods of
coordinate geometry, its volume, the angles between its faces and the
shortest distance between pairs of opposite edges, thus verifying the
results obtained at the end of the last paragraph.
2. The lengths of the edges of a tetrahedron are BC = 3, CA = 3,
AB = 5, AD=BD=CD = 4, Find its volume and the shortest
distances between pairs of opposite edges.
§ 10 OTHER APPLICATIONS 123
3. Let 1, 1’; m, m’; n, n’ be the lengths of the three pairs of opposite
edges of a tetrahedron. Let 0 be the angle between the edges of length
1, 1’. Show that
2Ul’ cos 6 = m?2+-m’2—(n?+n’2).
4. If 0, a, b, c are the vertices of a tetrahedron, show that the
position vector x of the centre of the circumscribing sphere is given by
the equation
[a, b, c]x = $| a |*(bxc)+3| b|%(cxa)+3]| c|%(axb).
10. The cube
The solid figure known as a cube, which has six square faces, is
more important than the other regular solids because cubes can be
stacked so as to fill space. A glance at the diagram shows that the
cube has
(i) six square faces falling into three pairs of parallel opposite
faces,
(ii) eight vertices,
(iii) twelve edges, falling into three sets of four parallel edges.
IB

FIG. 36
Each vertex lies on three edges and on three faces. Each edge
lies on two faces and contains two vertices. Each face has four
 124 OTHER APPLICATIONS Chesi

edges and four vertices. Two vertices of a cube are called opposite
if they do not belong to any common face. If V is a vertex the
opposite vertex V’ is the intersection of the three faces opposite
to the three which meet at V. In Fig. 36, A and G, B and H, C and
E, D and F are pairs of opposite vertices.
The line joining two vertices, if it is not an edge, is called a
diagonal. If the two vertices belong to one face, the diagonal is
called a face diagonal. A diagonal joining two opposite vertices
is called a body diagonal.
Since the three edge-directions are mutually perpendicular, it is
natural to choose a coordinate system in which these are the
directions of the base-vectors. For symmetry, we chose the origin
of coordinates to be at the centre of the cube—the point of inter-
section of three planes each parallel to a pair of faces and lying
midway between them. If the length of an edge of the cube is 2a,
the equations of the plane faces are
C=, C= —t, Y= 4, Y= —G, 2=—6, 2 = —o.
The coordinates of the vertices are (a, ta, +a),2x2x2=8
choices of sign giving eight vertices in all. The equations of the
twelve edges are
+x =4+y =a,
+y = +2 =4,
ae = Shy = Oh
The equations of the other points, lines and planes associated
with the cube can be worked out from these by using the methods
of coordinate geometry. Some possibilities are suggested in the
following exercises, but the reader may find others which we have
missed. It may be found helpful to have a wooden or cardboard
model of a cube to help in tuition.

Exercises on the cube

(Unless otherwise indicated, the cube referred to in these exercises

will have side 2a and the coordinate system will be the one described in
the last paragraph.)
1. The equation of the three face diagonals through (a, a, a) are
t=Y, 2=4; Y¥=2, =a; 22>, Y=a4.

2. The equations of the body diagonals are x = ay = +2.

Ex. 7 OTHER APPLICATIONS 125

3. How many face diagonals are there? How many body diagonals?
4. If d is a given body diagonal, how many face diagonals meet it?
How many face diagonals are skew to it? Show that the shortest distance
between a face diagonal and a body diagonal skew to it is ay/2/4/3.
5. All the body diagonals bisect one another at the origin.
6. How would you define a pair of opposite edges of a cube?
7. If f is a given face diagonal, how many body diagonals meet f?
How many other face diagonals meet f? How many face diagonals are
parallel to f? How many are skew to f?
8. Show that a face diagonal is at right angles to any body diagonal
which is skew to it. Show also that the face diagonals skew to a given
body diagonal form the sides of two equilateral triangles.
9. If two face diagonals are skew to one another, show that either
(i) they lie in opposite faces and are at right angles to one another or
(ii) they lie in adjacent faces and the angle between them is 60°. Find
the shortest distance between them in each case.
10. A cube has its centre at the origin and two of its adjacent
vertices are (5, —1, —1) and (3, 3, 3). Find all its vertices.
11. The eight vertices of a cube fall into two sets of four, each set
being the vertices of a regular tetrahedron. No two vertices in the same
set are adjacent vertices of the cube. Each edge of one of these two
tetrahedra is a face diagonal of the cube. If one of the two inscribed
tetrahedra is known, the cube is uniquely determined.
 SOLUTIONS TO THE
EXERCISES
Tn all solutions, unless otherwise stated, points A, X, &c., will have
position vectors denoted by the corresponding small letters a, x, &c.
Whenever a point is labelled O in the question as set, this point will be
taken as origin unless the contrary is explicitly stated.

Exercises page 20
1.|a|—|b|.
2. (i) The vectors a+b, b-+a then have the same direction and
sense (that of a and b) and their magnitudes |a|-+|b|,|b|-+|a|
are equal, by number-algebra. (ii) Assume, interchanging a and b if
necessary, that |b| <|a|. As in Q. 1, it can be seen that a+b and
b-ta are both vectors with the same direction and sense as a and
magnitude |a|— |b]. (iii) follows from the equations a+-0 =a =
0-La (p. 19).

Exercises on Chapter 2 page 34

1. —a+b, 2a+b, —2a+2b, 3a, —atb, —4a-+b. Midpoints
3(a+b), $a+b, a+b, b. Centroids 4(4a+2b), a+b, 4(2a+4b).
2. at+c—b.
3. The midpoint of OR is also the midpoint of PQ since POQR is a
— >=
parallelogram. OM = 4OR = }(p+q).
4. 1+m-+n = }(b+c)+h(c+a)+f(a+b) = atb-+e.
>lc Oo lhl>
AL+BM-+-CN = I—a+m—b-+n—c = 0.

5. 3 GG’ = 3(8'—§) = a’+b’+c’—(a+b-+c) = AA'4 BB’ LCC.

6. The midpoint of PR has position vector
Mp+r) = ia+b+e+d+e+f)
= 3(q+s), the position vector of the midpoint of
QS. Thus PQRS is
a parallelogram.
Ve As in Ex. 1, p.
30, it can be shown that PR, QS, TU all have
the
same midpoint. Thus PQRS, PTRU, QTSU are all
parallelograms.
 SOLUTIONS TO THE EXERCISES 127

8. Let P,Q, R,S (position vectors p,q,r,s) be the centroids of

BCD, CDA, DAB, ABC, in order. The position vector of the point
dividing AP in the ratio 3: 1 is }(a+3p) = }(a+b+c-+d). This is the
point G, G lies on AP and similarly G lies on BQ, CR, DS, so these lines
meet at G. By (11), p. 30, a’ = }(3a+d), b’ = }(3b+d),c’ = 1(3c+d),
so the centroid of A’B’C’ has position vector
}(a'+b’+c') = datb+c+d) = g.
9. Since ABCD, A’B’C’D’ are parallelograms, we have b—a = c—d,
b’—a’ = c’—d’. Adding these two equations and dividing by 2 we find
$(b+b’)—#(a+a’) = (c+c’)—4(d+d’), ie., the four midpoints form
a parallelogram.
> >
10. Take O as origin. By § 8, § 14, OH = a+b+c = 3 OG.
—> —> — —>
11. CD = —p+q, DE = —p, EF = —q, FA = p—q. OA = —q,
OB = p—q, OC = p. Since O is the circumcentre of the triangle ABC,
its orthocentre has position vector (§ 8) a+b-+c = 2b. Similarly the
orthocentre of BCD has position vector 2c, &c. Thus the orthocentres
lie on a circle centre O and radius 2r.
12. Let M be the midpoint of CD. Then
—>
MP = 3(a-+b+ce+d)—4(c+d) = a+b).
—
Since |a| = |b |, this is at right angles to b—a = AB.
13. Since |A,B, |= |a;| = 1, and |A,B, |=| a,| =1, the point
B, is centre of a unit circle through A,, A. Similarly b, = a,+a,,
—> > —>
b; = a,+a,. Then B,C = a,, B,C = a,, B,C = aj. Since these are unit
vectors, the three circles of unit radius with centres at B,, B,, B, inter-
sect at C.

Exercises on Chapter 3 page 48

1. (0, —2, 11), (—7, 9, —4), (—9, 10,,—14), (—1, I, 1).
2. (i) (2, 1, 0); (ii) (0, 1, —2); (ai) (—1, 1, 0); Gv) (1, 1, 1).
—> —
3. P(3, 0, —1), Q(0, 3, 2), QP = (3, —3, —3), QB = (2, —4, —2),
BP = (1, 1, —1). Hence QP? = 27 = QB?+ BP?, and BPQ is a right
angle.
4.9=0,¢=r=1.
128 SOLUTIONS TO THE EXERCISES

5. If a and b are at right angles, then by Pythagoras’s Theorem,

since a+b represents the hypotenuse, we have
Ja|?+]b|?
=| a+b |?.
Written out in components, this simplifies to the result given.
6. (i) Collinear, —4; (ii) collinear, —?; (iii) not collinear.
7.,(8; 0; 4)-
— =>
8. Mis (2, 1,0). Hence AM = (—1, —2, —1), length 1/6, BM = (1,1, 1),
length 4/3, AB = (—2, —1, —2), length 3. AB? = AM?+ MB?.
9. Centroid is (2, —3). In the new system A (4, —2), B (—2, —4).

10. cos 0 = 2/4/5, sin 6 = 1/4/5; (41/4+/5, 2/4/5), (34/4/5, —2/4/5),

11. (i) w’2+y'? = 1.| (ii) v2? +-y'2+ 4/2a’+1 = 0. (iii) 2a’y'+3 = 0.
(iv) 3”’2+-y'?4 4/2(8x'+2y’)+8 = 0.
12. Take new origin at (—1, 2).

Exercises on Chapter 4 page 56

Lis ome 1 1
1. |=, —, 0) and: |—— = 0)
& 2 V2 (y2 v2
2: (0mS
Sie bs ~ Js)
2779
and (34/2).34/2"
ga a 34/2)”
te
(1,0,0) and ( mae -5).
ee es

3. (ee V2 3420/2
6 = § a> 8G
me (= 2/2 _v2 ao).
6 6 6
4. By the sine formula for the area, 4A2—J/al2|b/2b|? sin?
<p 6
=| a|?| b |2(1—cos? 6) = | a|?| b |2—(a.b)2. |a|?|

5. Take A as origin. Then b.(c—d) = c.(d—b) = 0, so

b.c=c.d =d.bandd.(b—c) = 0,
i.e., AD is perpendicular to BC.
6. Let a,b be the vector sides of the parallelogram. Then the vectors
given by the diagonals are a+-b. The result follows from the identit
|a—b |?+ | a+b] ? = 2(|a|?+ |b|?), ay,
 SOLUTIONS TO THE EXERCISES 129

7. Let O, a, b, c be the position vectors of the vertices. The sum of

the squares on the six edges is
x |a|?+2
| b—c |? = 32 | a |?—2db.c,
while the sum of the squares on the segments joining the midpoints
of opposite edges is 2 |a—b—c |* = 1/32 |a|2?—2Db.c).

8. The angle between OB and OC is 2A and their lengths are R.

>= >>> >>>
Hence OB.OC = R? cos 24, OC.OA = R? cos 2B, OA.OB = R2 cs 20.
—>
Now (denoting OA by a &c.) we have be cos A = (b—a).(c—a) =
|a|?—a.b—a.c+b.c = R? (1—cos 2B—cos 2C'+ cos 24).

9. Using the notation of Chapter 2, § 8 and Q. 8 above,

OH? =|a+b+c |? = X|a|?+2Xb.c = 3R?242R?2 . cos 24.

10.
_ (p.a)| b|?—(p.b)(a.b) 7 _ (p-b) | a |?— 2 b) .

|a|?|b|2—(a.b)? ” | a| 7] b|?—(a.b)?
> “> >
11. CA = a—c, AB = b—a, OM = }(b-+c), MA = a—i(b+c).
> >
BAC is a right angle if O = CA.AB = a.b+a.c—b.c— |a|?,
OMA is a right angle if
>
O = OM.MA = }(2a.(b+c)-+ |b|?+ |c|?—2b.c).
If we substitute 1 for |a|?, |b |?| ¢|?, cos « for b.c &c., both these
conditions reduce to the desired form.

12. (i) Number 8; (ii) vector (5, i, 9); (iii) vector (4, 0, 4);
(iv) number 45.

13. AB = 1/38, BC = 34/6, CA = 1/10;

3 =
41 C=
13 . Area = 44/371.
cos A =
Deon eer ee bv
14. If u and v are perpendicular, then
O = u.v = (a.c) |b|?—(b.c)(c.a)(a.b) = cos* B—cos « cosB cos y.

15. If 1 denotes the common length of all the edges, we have

2=|b|?=| c |? = | b—c |? = | b [?—2b.c+ |c |?, so 2b.c
= 22,
 130 SOLUTIONS TO THE EXERCISES

> -—>
PA.PB = [sarbte—a| i [Fatb+e—P|

= sq(—5 |a[2—5 |b [2+ |c|2—4b.c—4c.a+26a.b)

ee= it ee
eee +13)
J—2—2-+-13) = 0.

16. All four faces have area 91/6.

Exercises on Chapter 5 page 72

2. The area is half the magnitude of the following vector product
(Note 1, p. 61):
(b—a) xX(c—a) = bxc—bxa—axc+axa
= bxc+cxa+axb (Note 2, p. 61).
3. Since axa = bXb = 0 and bxa = —(axb),
(ra+sb)x (tat+ub) = rt0—st(axb)+ru(axb)+su0.
4, (a) If v is the projection of x on P then x = v-+kn, where k£ is a
number and v.n = 0. Hence x.n = v.n+k|n|? = 0-4, since n is
a unit vector. Then v = x—kn = x—(x.n)n.
(b) Since n is a unit vector, the expansion E mentioned in § 5 will not
occur when one takes a vector product with n. Thus nx will be the
result of rotating the projection v through an angle of 90 degrees in the
plane P. The vector (x.n)n, parallel to the axis of rotation, will be
rotated into itself. Hence the vector x = v+(x.n)n will be rotated
into nx x-+(x.n)n.
5. Since the vector product ax x must be perpendicular to a, there
will be no vector x satisfying the equation unless a.b = 0. Ifa.b = 0,
there will certainly be a vector p in the plane perpendicular to b with
the property that the area of the parallelogram (a, p) is equal to |b].
Then axp will be +b, since it agrees in magnitude and direction,
though it may differ in sense. On changing p to —p,if necessary, we may
take it thata x p = b. Ifnowaxx = b, it follows that ax (x—p) = 0,
so that x—p is parallel to a. Thus the locus of all such x is the line
through p parallel to a.
6. (i) (—10, 7, —16); (ii) (—5, 0, 5); (iii) —20; (iv) —15; (v) 20.
8. Since v is a unit vector and a is at right angles to v, it follows that
b = vxXa is obtained by rotating a through a right angle about v.
Then vxXb is the result of rotating a through two right angles, i.e.
vxb=—a. '
 SOLUTIONS TO THE EXERCISES 131
9. From aXb = c it follows that c is at right angles to a and b and
from b xc = a it follows that a is at right angles to b and c.
10. (a—b)xc = 0, so a—b = kc. Hence |a|?—a.b = ka.c, ice.,
l—a.b = k/4/2, and |a |?—2a.b-+ |b |? = k?| c|?, 2—-2a.b = k?,
Thus k? = ky/2, and since a and b are distinct, k 40, sok = 4/2,
and a.b = 0.
11. (a) Let p be the projection of x on the plane perpendicular to
v, let x’ be the vector in its rotated position and p’ the projection of x’.
As in Question 4b we have
x’ = p’+(v.x)v
xX = p+(v.xv).
Also v x x lies in the plane perpendicular to v and is obtained from p by
rotation through a right angle. From the diagram

VXxX

fie

we derive the equation

p’ = cos 6 p+sin 0(v xx).
Substituting for p and p’ yields the required formula.
(b) The unit vector v with the same direction and sense as (2, 2, 1)
is v = (2, 3, 4). Taking this vector as v and 0 as 45° in formula (a) we
find x’ = (10+ 4/2, 10—24/2, 5424/2).

12. c.a = p+}q, C.b=3p+9,p =3,9= —5, 7 = t+.

Exercises page 79
1. (i) Non-collinear, y—z = 3; (ii) collinear; (iii) non-collinear,
ety = 9.
2. If a plane contains the points A, B, C its normal vector (p, q, r)
must be perpendicular to the vector AB =(—I,1,—1), so
—p+q—r = 0. The most general solution is r= k, g = 1, p = k—1.
The equation of the plane takes the form
(k—lat+ly—kz = t.
The condition that B should lie on the plane yields ¢ = 3/—2k.
132 SOLUTIONS TO THE EXERCISES

Exercises 6.1 page 90

124-2 — 13:
2. Sides 1/6, 1/6, 1/22; cosines of angles septic St nile ;aread,/11;
24/33’ 6 24/33
plane x+3y+z= 2. Sides 1/6, 21/11, +/42; cosines of angles
1 2 20
; area 4/62; plane 37+-7y+2z = 8.
34/7 1/66’ 1/462
3. Area $4/3; perpendicular distance 1/1/3.
Aa of — 7 = 1s (a a a)
ae ; Lesy2
5. (i) Sides 3, 1/6,4/6, V/ 1/3; cosines of angles
g 0, V3 V3 ; foot
—, ~— oot of of perpen =

dicular (—}, 0, $). (ii) Sides 1/2, 1/14, 4/14; cosines of angles
1 ee 8 :
D7? IT ve foot of perpendicular (—3, —4, 3).

6. 3/4/455.
a The given equation represents a plane and it is satisfied by the
coordinates of the points given. Let (a’, 0, 0) &c. be the coordinates of
the points A’, B’, C’. Then G’ is (4a’, 4b’, 4c’) and the normal vector of
e/a
the plane ABC is & 5° ‘):For these to be parallel, we must have

aca’ = bb' = cc’. The symmetry of this condition proves the result.
8. (1, 4, —9).

dO hos so = . (a ee e2)

11. 54+6y—Tz = 12.

12. 3. (2%, 5,
2Q7IA\V7? st).
V7

13. Follows from symmetry. The sides of the triangle are

permuted
by cyclic interchange of x, y, z, so they have the same length.

Wa) es, 2); ae ee ope lea

 SOLUTIONS TO THE EXERCISES 133

Exercises 6.2 page 104

(eee
24/105"
2e(1; 4, rand (2,.4...0),
3. at4y—z =

y Nana
—l
ao setae
i
eal1
5. cos! aE a—4y+32 = 15

6. y—z=1

7. x—y—z = 0; |

8. (S11):

9. aty—z = 2; 2+y—2 = —1; V3.

10. cos“? ay
11
11. ety—z = 3; V3.

12. (i) 5u—3y—z = —3; (ii) 5u—3y—z = —7; (ili) x 22 =0;

(iv) y—32 = 1; (v) 5¢+1ly—8z = 49.
yt2 242
13. 2+ Y —
5 =
3
—_

62-2. y—l1. 22-1

Maga A eo tas 2S
15. <+y = 6 and x+4y+z = 6.
134 SOLUTIONS TO THE EXERCISES

Exercises page 108

1. 7x'2+10y’2+2'2+112'y’—2'a'+
8a'y' = 9.
3. In equations (5) put z,=1, *,—2,=0. We find 2,=4h,
2, = 1,, #3 =1,. Substitute these values back into equations (4) to
obtain the result.
4, u, = (3, —7, —#), Us = (—7, 7 —7)-

Exercises page 110

1. qXr = (cXa)xXr = (c.rja—(a.r)c = [a, b,/ejaysince a.ri== 0:
Thus p.(q Xr) = [a, b, c](a.p) = [a, b, c]?. If p, q, r are known, then
a, b,c is now determined apart from sign, and a is given by the
formula a = qxr/(a, b, c].
2. From p+q-+r-+s = 0 follows
|p|?+|a[?+2p.a=|p+a|?=|—(r+s)|?=|r [2+ |s[2+2r.s,
Hence p.q =r.s,| pxq |? =| p|?|q|?—(p.q)?
= |r|?|s|?—(r.s)? =|rxs 2
3. As in Question 1, let p = bxc &c., and put s = (a—b) x (c—a).
Then p+q+r-+s = 0, so by Q.2|pxq|=|rxs|,ie.,
|[a, b, c] a] =| [a, b, c](b—c) }.
In the tetrahedron OABC, p, q, r, s are twice the vector areas of the
faces, and a, b—c define opposite edges.
4, Put y=axb. Then (axb).(cxd) = y.(cxd) = (yxc).d.
Now yXc = (axb) xc = (a.c)b—(a.b)c.
Hence (yXc).d = (a.c)(b.d)—(a.b)(c.d).
5. (i) is the vector triple product formula for r x (cx d) withr =aXxb :
(li) is the same for (axb)xs with s = cxd. The second part follows
either by equating these expressions, or by expanding each term
using (7), obtaining twelve terms which cancel in pairs because of the
anticommutativity axb = —bxa.

Exercises on spherical trigonometry page 118

2. sin? b sin? c sin® A = sin? 6 sin? c—sin? b sin? ¢ cos? A
= sin 6 sin c—(cos a—cos b cos c)?, by the cosine rule,
= {cos (b—c)—cos a} {cos a—cos (b+c)}
= 4 sin s sin (s—a) sin (s—b) sin (s—c), where 2s = a+b-+te
.
The sine rule follows from the symmetry of the above express
ion.
3. Put 6b= c = a in the cosine rule to obtain
cos a = cos? asin? a cos A, so cos a (l—co
s a) = (l—cos? a) cos A.
 SOLUTIONS TO THE EXERCISES 135

Dividing across by 1—cos a, which is not zero because a ~ 0, the result

follows. The second equation follows on putting cos a = 2 cos? ja—1
cos A = ]—2 sin? $A.
4,
sin 6 sin c+-cos a—cos b cos ¢
cos? 44 = 4(1+-cos A) =
2 sin 6 sin ec
__ cos a—cos (b-+c) __ 2 sin s sin (s—a)
2 sin 6 sinc 2 sin 6 sinc
The formula for sin? 4A follows by a similar argument, and that for
tan? 4A is obtained by division.
5. (i) Put cos C = 0 in the cosine rule:
cos c = cos a cos b+ sin a sin b cos C.
(iii) Put sin C = 1 in the sine rule: sin C sina = sin A sin c. (v) Put
cos C = 0,sinC@ = 1 inthe cosine rule for the polar triangle:
cos A = —cos B cos C-+sin B sin C cos a.
sin 6 cos ¢
(ii) By (i), (iii), cos A = sin B cos a = (iv) Multiply the
sin c cos b
left side of (i) and (ii) by the right side of (iii), equate to the right side
of (i) and (ii) multiplied with the left side of (iii) and simplify.
6. The vector b xc has magnitude sin A and direction perpendicular

to the plane BOC. The angle between a and bxc is therefore =e

We have
sin a sin « = [a, b, c] = sin b sinc sin A (p. 115).
Also sin a cos « is the magnitude of the vector
ax(bxc) = (a.c)b—(a.b)c.
Hence sin? a cos? « = (a.c)?| b | 2+(a.b)?| c| ?—2(b.c)(c.a)(a.b).
= cos? b-- cos? c—2 cos a cos b cos ¢.

Exercises on the tetrahedron page 122

1. By the distance formula, every edge has length a4/2, so the
tetrahedron is regular. Volume = scalar triple product of vector
edges = 3a%, shortest distance = a. These results reduce to those in
the text on replacing a by 1/4/2.
2. Volume 53;+/95; shortest distances: AB to CD 34/95, AC to BD
54/19, AD to BC §1/19.
136 SOLUTIONS TO THE EXERCISES

3. If 0, a, b, c are the position vectors of the vertices, then | = |al,

=| b—c|;
2Ul’ cos 8 = 2a.(b—c)
(tb)]*2 [aac |e ea Oe
m2+-m'2—(n®?+n'?).
4, Since |x|? =|x—al|?=|x|?+|a|?—2x.a, so we have
x.a = }]|a|? and two similar equations. Now apply formula (’7),
p- 109 with b replaced by x, c and d replaced by b, c.

Exercises on the cube page 124

3. 12 (two for each of six faces) ;4 (joining the eight vertices in pairs).
4, 6 (three meet it at each of its two endpoints); 6 ( = 12—6).
6. An edge joins two vertices, and the opposite edge joins the two
opposite vertices. Opposite edges are parallel.
To 2,0; 4508

9. (i) 2a, (ii) ma

10. If P Q are the adjacent vertices, the other vertices are obtained
by rotating through +90° and 180° about an axis parallel to PQ
through the origin. Use the formula of Ex. 4b, p. 73 to obtain
(5, a, =i (5, ue ih); (3, 3, 3), (—3, —3, —3)
(1, Ls = 0), (aoe, ts 5), (1, =o; 1), (=I, 5, ie
 INDEX
Addition of vectors, 18 Negative of a vector, 19
Angle between vectors, 54 Normal vector, 74, 76
— between planes, 79 Notation, mathematical, 11-16
— between lines, 88
— between line and plane, 89 Orthocentre, 24, 53
Area, 61
— of spherical triangle, 117 Parallelepiped, 68
Associative law, 21 Parallel vectors, 29, 42, 61
Axioms, 14 Parameter, 83
Parametric equations, 83
Base, 39 Polar coordinates, 43
— of a parallelepiped, 69 — spherical triangle, 116
Brackets, 12, 21 Position vector, 30
Projection, 37, 38
Centroid, 31, 32 Proof, 14
Commutative law, 20
Components, 38 Queen’s College, Dundee, 6, 119
Coordinates, 39, 40, 107
Right-handed system, 59
Cosine rule, 52
— —, spherical, 112 Scalar product, 50
Cross-multiplication, 65, 70 — triple product, 70
Cube, 123 Sense of vector, 17
Sine rule, 114
Determinant, 70 Skew lines, 101
Direction cosines, 78, 84 Spherical triangle, 110
— ratios, 74, 78, 84 Standard form of equation of line,
— vector, 74, 84 86
Distributive laws, 27, 51, 63 Subtraction, 22
Sum of several vectors, 22
Endpoint, 18 — of two vectors, 18
Equivalent statements, 76
Explicit equations, 75, 82 Tetrahedron, 31, 119
Triangle rule, 12, 19
Initial point, 18 Triple products, 67, 70

Left-handed system, 59 Vector product, 60

— triple product, 67-68
Multiplication of vectors, three
kinds, 13, 50 Zero vector, 17
 >

;
LUECVIT q
Weas e
?

; Tigo ‘S a : . :
BY moron by oalttlé.
fenuualt So i ga toh
bt — SEtenalwtapa
owetal
eae=
lane yt A actig Sie tal ar tndl: :
. ‘ Uke
D Meg cet ls 1 ,aytende? [acarsods wey
wi 1 y) oF 7 : : i wel sai ney ;
16 4 a2 4 ots
CN atic) yo Gay wie ' =. x
. d eo b a i)
yi ' 7 r
tal brs op loll cone @
{ yes? 4 » by | fe) : ;

. ‘ oy yr 7

oe 8 ' ds

o em ail }
i a, A2 rep iis 4
Dil
a i
G) fable = “4 iE eet a)
F
Varese
‘ Vets 4 se : a4 tants oan <
“ar { laligty a ff gepownits
; Late he | 7

BET tupantls ; ;
\ 4 xeeill ean Sat gy
CHL apo! lectus i «
18 )* eth
ug fo 0 ij gees!
Trea ce
(ie ]
ad ee
So metres BAA ES aaehy eta
Sh wing? lai yes P| ad

Kl Sweepers owe fr

- “4 Fh VG inva bak

Ai; " Ql.) Sihtoipeee

aT PO Atha wae —
CH) pair tt Hout
MBAS) pou berny why eh

ti ehaay ae Se ‘a oe
ry 7 ; i. F
 OXFORD BOOKS
MATHEMATIGse EO
SGIENICE
By W. L. FERRAR
340 pages
This book is written expressly for scientists in the sixth forms
of schools and for classes at Technical Colleges and Universities.
It selects those mathematical techniques that they are certain
to use at a later stage and introduces them in a direct way with-
out much of the detail that would be proper in a book written
for mathematical specialists. Although proofs are given, it indi-
cates firmly when they are essential to the first reading and
when they can be left until later.
There are two main sections: non-calculus, which covers
trigonometric and hyperbolic functions, complex numbers,
inequalities, analytical geometry, and vectors; and calculus,
which covers differential calculus, Maclaurin and Taylor ex-
pansions, the indefinite integral, the definite integral and its
applications, differential equations, integration by parts, and
partial differentiation.

DISCOVERING MODERN
ALGEBRA
By K. L, GARDNER
240 pages
This book is about groups and matrices. It is hoped that the
reader will gain a working knowledge of these topics which,
although not over sophisticated, will well fit him to tackle
more formal texts. As such it may be suitable as a first course
for training college students, school teachers, sixth formers or
first year undergraduates. The material collected itself as a
result of five years’ worth of lectures to members of the first
two sets.
The other mathematical topics touched on are linear pro-
gramming, numerical analysis, motion geometry and vector
analysis. The treatment is light and the reader is encouraged
to think ahead for himself, thus making a considerable part of
the text irrelevant!

OXFORD UNIVERSITY PRESS

[832523/11/65]