Journal of Philosophy, Inc.

Aristotle on Mixtures
Author(s): Richard Sharvy
Reviewed work(s):
Source: The Journal of Philosophy, Vol. 80, No. 8 (Aug., 1983), pp. 439-457
Published by: Journal of Philosophy, Inc.
Stable URL: http://www.jstor.org/stable/2026162 .
Accessed: 03/12/2012 19:14

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal
of Philosophy.

http://www.jstor.org

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 4-

THE JOURNAL OF PHILOSOPHY

VOLUME LXXX, NO. 8, AUGUST 1983

-
_~~~~~~~4 e

ARISTOTLE ON MIXTURES*
QU UESTIONS about mixture and combination were among
the most central topics discussed by the earliest philo-
sophers. Indeed, Anaxagoras, Anaximines, Empedocles,
and others might be described as philosophers and chemists. Aris-
totle, in De Generatione et Corruptione (Bk. I, Ch. 10), attempted
to solve certain puzzles and to explain mixing. There was some
medieval discussion of the topic of mixture, but it has been virtu-
ally ignored since then.' The topic is worth reviving, both for its

* The present paper comes out of research on mixtures, matter, and mass terms
which got its real start during a visit to the University of Auckland in 1973. A sec-
ond visit for twenty months in 1979 and 1980 provided the opportunity to give a
number of lectures on these topics and to finish a great deal of this work.
Versions of much of this material have been presented at California State Univer-
sity at Northridge (1974), to the American Philosophical Association, Pacific Divi-
sion (1976), to the Australasian Philosophical Association, New Zealand Division
(1979), and at an NEH Aristotle Workshop at Florida State University (1983). Also,
various drafts have been circulating in typescript for some time. Thus, I have been
able to take advantage of quite a bit of good advice from a number of people.
I thank Helen Cartwright, M. J. Cresswell, Lucy Carol Davis, Michael Ferejohn,
Jon Gavrin, Dan Hunter, Dale Jamieson, Cristelle Leaf, Mark Nixon, F. Jeffry Pel-
letier, W. V. Quine, Reed Richter, Denis Robinson, Benjamin Sharvy, James Tom-
berlin, Martin Tweedale, Julian Young, and many others for their comments and
discussions of these matters.
'For commentaries on Aristotle, and historical discussion of the topic, see Alex-
ander of Aphrodisias, PernKraseoiskai Auxese-os ("On Fusion and Augmentation"),
in Ivo Bruns ed., Supplementum Aristotelicum, vol. 2, part 2, Scripta Minora (Ber-
lin: Reimer, 1892) 213-238; Cyril Bailey, The Greek Atomists and Epicurus (Oxford,
1928); Ida Freund, The Study of Chemical Composition: An Account of its Method
and Historical Development (New York: Cambridge, 1904; reprinted New York:
Dover, 1968); David J. Furley, Two Studies in the Greek Atomists, Study I (Prince-
ton, N.J.: University Press, 1967); H. H. Joachim, Aristotle "On Coming-to-be and
Passing Away," (Oxford, 1922); G. S. Kirk and J. E. Raven, The Presocratic Philo-
sophers (New York: Cambridge, 1957) chap. xv; J. R. Partington, A History of
Chemistry, vol. I (London: Macmillan, 1970); Rosamond Kent Sprague, "An Anony-
mous Argument against Mixture," Mnemosyne, 26 (1973) 230-233; and Gregory
-Vlastos, "The Physical Theory of Anaxagoras," Philosophical Review, 59 (1950)
31-57.

0022-302X/83/8008/0439$01.80? The Journal of Philosophy, Inc.

439

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
440 THE JOURNAL OF PHILOSOPHY

own sake as a philosophical problem, and also for its connection to

various questions about matter and about mass terms.
1. The Problem. Aristotle begins his discussion by asking whether
mixture (mixis) is even possible, for "as some say": (i) if the ingre-
dients continue to exist in the supposed mixture and are not al-
tered, then they are not really mixed; (ii) if one ingredient is de-
stroyed, then the ingredients have not been mixed, but one
ingredient has being and the other does not, whereas what has been
mixed should still be what it was before; (iii) if both ingredients are
destroyed, then they are not mixed since they do not even still exist
(327a29-b7).
It is not clear at all who the "some" are who actually advanced
this puzzle, but we can let this question pass and tackle the puzzle
itself and Aristotle's solution. First, he gives some examples. (i) In
a combination of unmashed beans and rice, the ingredients con-
tinue to exist; they are not altered; they are merely physically jux-
taposed; we have a mere combination (synthesis, 327b35) and not a
true mixture.2 (ii) In burning, wood does not mix with fire, but the
fire comes-to-be and the wood passes-away. Properties and states do
not mix with things, for we see them persisting unaltered (a white
man is not a mixture of white and man). (iii) Properties and states
cannot be mixed with each other (e.g., knowledge and white), for
they cannot themselves exist independently (cf. Categories 2, 5).
Aristotle then seems to have certain conditions that a true mix-
ture must satisfy.
Axiom 1. Each of the ingredients in a mixture must have originally
existed separately (327b23).
Axiom 2. The ingredients of a mixture can be separated again (327b29).
These two axioms do not tell us what mixtures are, however, for
a mere combination like the beans and rice will satisfy them, and
Aristotle does not wish to count that as a true mixture. How then
could these two axioms hold if it is also required that the ingre-
dients do not persist unaltered? Aristotle's answer is that the ingre-
dients neither persist actually nor are they destroyed (327b23-32).

But compare his remark that "mixture (mixis) is not always fusion (krasis), for
the mixture of dry substances is not fusion" (Topics IV.2 122b31-32). Here he did
call this mixis, so there seems to be an inconsistency in his use of the term mixis.
Perhaps he was adopting a special technical usage in De Gen. et Cor., for at 328a9
he seems to equate kraszs and mixis. Technical precision may have seemed called
for. Plato had used several diffeient terms for mixture fairly loosely in just a few
lines of the Sophist (252e-253b), and in the Laws he had used symmixis to mean
sexual intercourse (VIII 839a).

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 ARISTOTLE ON MIXTURES 441

I find this idea of a quantity of matter existing potentially-but-

not-actually very mysterious and in conflict with Aristotle's general
views on matter and potentiality. I will discuss this in the last sec-
tion of this paper. In any case, Aristotle seems to believe that an
ingredient's persisting-but-not-actually is logically related to its
not being-preserved-in-small-particles. For he demands
Axiom 3. A inixture must be homeomerous, that is, made up of like
parts which are like the whole, just as any part (meros) of
water is water (328a1-12).

2. Homeomerism.3 What is it to be homeomerous? Leonard and

Goodman4 define the related property of being dissective as a prop-
erty of properties or predicates: a predicate P is dissective if and
only if, if x is part of something that satisfies P, then x will satisfy
P:
P is dissective iff x e (iy *Py) D Px5
For example, the property of being stuff in my pocket is dissective.
However, being homeomerous is not the same thing as being dis-
sective, for it is properties that are dissective or not, whereas Aristo-
tle seems to call kinds of stuff and quantities of stuff homeomer-
ous; his own example was water. Being stuff in my pocket may be a
dissective property, but I don't think that the stuff in my pocket
should be called homeomerous.
Aristotle's own illustration refers to the kind of stuff water itself:
he says ". . . just as any part of water is water." And in Ch. 1 he
had characterized the homeomeries of Anaxagoras as things syn-
onymous with (named the same as) their parts (314al9-21). So let
us try
DI. A kind K* is homeomerous iff x < K* D Kx.

3Parts of sections 2-5 are technical, and parts are digressions. If the reader insists
on skipping them or reading them last, he should at least read the bit on "Zeno's
Blender" at the end of this section. That shows the basis of my view that two quan-
tities are homeomerously mixed if and only if they occupy the same space at the
same time.
4"The Calculus of Individuals and Its Uses," Journal of Symbolic Logic, 5 (1940)
45-55, at 55.
'Remark on the notation: I will use '<' for the being part of relation, and the
schwa 'a' to form an indefinite description 'a/sm P'. Helen Cartwright, in "Hera-
clitus and the Bath Water," Philosophical Review, 74 (1965) 466-485, at 469-472, has
offered the exciting suggestion that 'some' with its pronunciation "sm" is the Eng-
lish indefinite article for mass and plural terms. I will occasionally spell 'some' this
way to call attention to that usage.
'S(ax Px)' is analyzed as '(3x)(Px Sx)'. I also assume that variables are univer-
sally quantified unless otherwise indicated.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
442 THE JOURNAL OF PHILOSOPHY

(The '*' is to remind us that this is 'K' as a variable ranging over

kinds, not a predicate as in 'Kx'.) There are some difficulties in
making sense of this, however. First, to be part of a kind is not
what it might seem. The water now in my glass is not a part of
water; the shrimp salad now on my plate is not a part of shrimp
salad. What are some parts of shrimp salad then? Well, one part of
shrimp salad is shrimp.
Only kinds are parts of kinds. A performance of Beethoven's Sev-
enth Symphony is not a part of it; the viola part, itself an abstract
entity, is a part of it. "Thus the semicircles will be parts not of the
universal circle but of the particular circles" (Metaph., VII.ll,
1037a3-4; see also 1035b). The defining clause in DI would be sat-
isfied by any elementary kind, since an elementary kind would
have no parts (in this sense) but itself. But no elementary kind is a
mixture, and we were looking for an analysis of being homeomer-
ous that might apply to mixtures. Aristotle clearly faults Anaxago-
ras for this error (De Caelo 111.4, 302b12-20), i.e. for the error of
confusing being an element (stoikheon) with being homeomerous.
Our difficulty was with the 'x < K*' clause in DI. 'K*' is pre-
sumed here to have mass terms as substituends. Now 'K*' occurs in
a noun position, e.g., as a variable, in that clause. It is useful here
to apply Quine's view that a mass term in a substantive position is
the proper name of a single concrete object (usually large and scat-
tered): the totality of all that satisfies K.6 So if we read Aristotle's
'any part of water' as 'any part of the world's water', DI becomes
D2. A kind K* is homeomerousiff x ? the world's K D Kx.
Unfortunately, on this definition it will turn out that for any
mass kind K*, that kind K* will be homeomerous if and only if the
property of being sm K is dissective. That is, D2 just collapses
being homeomerous and being dissective. To show this, we need to
show that
(1) x C (ay Ky)
and
(2) x < the world's K
imply each other when K is a mass predicate.
An analysis of definite descriptions built from mass predicates is
needed here, since 'the world's K' is such a definite description. I

6
Word and Object (Cambridge, Mass.: M.I.T. Press) p. 98.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 ARISTOTLE ON MIXTURES 443

have shown elsewhere that the Russellian analysis will not do for a
definite mass description such as 'the coffee in my cup', since there
is not a unique object satisfying the predicate 'is sm coffee in my
cup'.' The coffee in the northern half of the cup satisfies it, and so
does the coffee in the southern half. What the definite mass descrip-
tion 'the coffee in my cup' designates is the total sum or "fusion"
of the various quantities of coffee in my cup-the total quantity of
coffee in my cup. If a mass predicate has any instances, the definite
mass description formed from it must designate something, and
what it designates must itself satisfy that predicate: it must be the
case that the coffee in my cup is sm coffee in my cup. This is sup-
ported by Quine's observation that mass terms are cumulative:
"any sum of parts which are water is water" (p. 91).
My notation for 'the G' is '(0x *Gx)'. (Sharvy, 1980, used a mu; I
now prefer a theta.) My analysis of this is '(ax)(Gx *(y )(Gy D y < x))'
-ssm G that all G is part of. 'F(the G)' is then written '(0x *Gx)Fx',
and the analysis of this is
(3x)(Gx(y)(Gy D y < x) Fx)

This differs from the Russellian analysis only in having the part of
relation where Russell had identity. Since the world's K is itself sm
K, (2) implies (1), and so any dissective mass predicate would be
homeomerous on definition D2. Conversely, (1) implies (2) for
every x also. For the mass predicate K must be cumulative (Quine's
condition), and if (ay *Ky) exists and K is cumulative, then (Gy*Ky)
exists also. (That is, if sm K exists and K is cumulative, then the
world's K exists also.) Furthermore, c is transitive, and Kx D x <
(Oy- Ky). (That is, any K must be part of the world's K.) So if (1),
that is, if x is part of sm K, say of a, then since Ka, a is part of the
K, and then so is x. Therefore (1) implies (2) for every x.
So D2 does not analyze Aristotle's notion of homeomerism. On
D2, being homeomerous just amounts to being dissective. Recall
why being dissective did not work: the property of being stuff in
my pocket was dissective, regardless of what was in my pocket. So
on D2, the mass kind stuff in my pocket would be a homeomerous
kind. On the other hand, perhaps whisky-and-water is the sort of
thing that should be called homeomerous, yet it fails to satisfy the
defining clause in D2: the water in my whisky-and-water is part of
the world's whisky-and-water, but it is not whisky-and-water.

"A More General Theory of Definite Descriptions," Philosophical Review, 89

(1980) 607-624.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
444 THE JOURNAL OF PHILOSOPHY

Now it might seem that a definition of homeomerism could be

given along the above lines by restricting the application of 'part'
to spatially determined parts. Being whisky-and-water is not a dis-
sective property, for the reasons just given. But what seems to be
important is that every spatially determined part of a quantity of
well-mixed whisky-and-water is sm whisky-and-water. A spatially
determined part is a part which is the complete contents of a quan-
tity of space, and if the whisky and the water are well-mixed, it
would seem that the contents of any region within the stuff would
contain both whisky and water. However, this will not work as a
definition of homeomerous, because, again, the property of being
stuff in my pocket would turn out to be homeomerous; and that is
quite counter-intuitive.
A restriction to so-called "natural kinds" is of no help here
either. Water is a natural kind, unlike whisky-and-water which is
an artificial kind. But like whisky-and-water, water is a combina-
tion of two more elementary kinds. Water should turn out to be
homeomerous, just like whisky-and-water, but the property of
being water is not dissective: the hydrogen in the world's water is
not water.8
Aristotle has more to say about his example of wheat and barley
particles which fail to be a true mixture. Such a combination fails
to be a true mixture because the ingredients continue to exist unal-
tered; this definitely does involve certain spatial notions. Aristotle
notes that we might divide the particles into smaller particles,
which might yield an apparent mixture, depending on the ob-
server's sharpness of vision. But this would still not be a true mix-
ture "to the eyes of Lynkeus" (328al5-16).
Aristotle also observes that it is impossible to divide the wheat
and barley particles and arrange them so that every quantity of
wheat is next to a quantity of barley (328al3-18). So long as there
are particles of wheat, these will have spatially determined proper
parts which are also wheat; thus, there will be quantitites of wheat
completely internal to each wheat particle, which will therefore not
be next to any quantity of barley (328al-5). We can imagine a ma-
chine called Zeno's Blender that splits particles in two and then

8For more on this, see my papers "The Indeterminacy of Mass Predication," in

F. J. Pelletier, ed., Mass Terms: Some Philosophical Problems (Dordrecht: D. Rei-
del, Synthese Language Library, 1979) 47-54; "Reponse a Monsieur Pelletier"
(forthcoming in Logique et Analyse); and "Mixtures," Philosophy and Phenomeno-
logical Research, 43 (1982-83) at press.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 ARISTOTLE ON MIXTUJRES 445

shakes them up and re-arranges them. Suppose that we begin with

an arrangement like this:

B W

Then, we run it through Zeno's Blender a thousand times. The re-

sult will still contain the same figure, and within each wheat parti-
cle there is stuff that is only wheat, and not a mixture of wheat and
barley.
We need to approach the analysis of homeomerism with notions
of spatial division. Homeomerism is a topological notion, not a
purely mereological one. Although reading meros (part) in Axiom
3 as 'spatially determined part' did not quite work, it did give us a
hint in the right direction; it did work for whisky-and-water. What
we need is a definition that applies to a quantity thought of as a
sum of discrete parts. And we need a spatial or topological compo-
nent in our definition.
3. Some Mereology and Topology. A useful notion is a partition of
a quantity of stuff:
D3. A set S of subquantities of a quantity Q is a partition of Q if and
only if (i) no two membersof S overlap,and (ii) the sum of S is
Q.
Overlapping here is the strict notion from the calculus of individu-
als.9 Two quantities overlap just if they have a common part. (I
generally prefer to use being part of as the primary relation, with
overlapping as defined.)
It is vital here not to assume (as Goodman does tacitly) that over-
lapping is the same as spatial overlapping. For example, the light
in my room does not overlap the air in my room, although the spa-
ces they occupy do overlap. It is easiest to see this point if we think
of overlapping defined as having a common part: the air and the
light are discrete. An example of a partition would be Ithe whisky,
the water}, which partitions the whisky-and-water. (Please just

9See Nelson Goodman, The Structure of Appearance, 3rd. ed., Boston Studies in
the Philosophy of Science, vol. 53 (Dordrecht: D. Reidel, Synthese Library, 1977).

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
446 THE JOURNAL OF PHILOSOPHY

forget that whisky itself contains water, or make up an example of

your own.)
Call the quantity of space occupied by something its receptacle. I
take this term from Richard Cartwright's "Scattered Objects"10 al-
though I have changed the sense slightly. (Specifically, Cart-
wright's receptacles are sets of points of space, and I do not wish to
presuppose that a quantity of space is a set of points.) Then we can
approach Aristotle's discussion of the wheat and barley problem by
first defining relative homogeneity:
D4. A partitionS of a quantity Q is d-homogenousif and only if every
sphericalregion of space within the receptacler(Q) of Q having
diametergreaterthan d overlapsthe receptacleof each member
of S.

For example, {the beans, the rice} is a one centimeter homogen-

ous partition of my beans and rice, but it is not a one millimeter
homogenous partition of it. Or consider a white and black chess-
board with two-centimeter squares. {the white, the black} is a two
centimeter homogenous partition of the color, but not a 1.9 cen-
timeter homogenous partition of the color. (Treat 'sphere' appro-
priately, i.e., here as a two-dimensional sphere, i.e., a circle.) Im-
agine bisecting each square horizontally and vertically and
recoloring the board. Then {the white, the black} will be one cen-
timeter homogenous, but not less." Then Aristotle's remarks about
the wheat and the barley can be put this way: no matter how many
times the wheat and barley particles are put through Zeno's
Blender, it will never happen that {the wheat, the barley} is a zero-
homogenous partition of the wheat and barley.
This suggests a definition of homeomerous:
D5. A partition S of a quantity Q is homeomerousif and only if S is a
zero-homogenouspartition of Q.
At least this seems to be just what it is not possible to reach by re-
peatedly running the wheat and barley through Zeno's blender.
However, it is the limit thus approached.
Notice that being solid rather than liquid is not a determining
factor. Olive oil does not mix with vinegar. If you put some of each
I?ln Keith Lehrer, ed., Analysis and Metaphysics (Dordrecht: D. Reidel, 1975)
153-171, at 153.
" There are some problems with this definition that I first noticed during a course
of lectures I gave on mixtures in Auckland in 1980. We were not able to solve them.
Exercise: figure out what the problems are and then solve them.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 ARISTOTLE ON MIXTURES 447

in a bottle, even after some shaking, the oil and the vinegar are
unmixed. They might form a one millimeter homogenous parti-
tion of the total, but the oil is preserved in "particles," i.e., in re-
ceptacles containing only oil.
I have avoided any requirement that the partition of a quantity
be one that divides it up into "natural kinds." This would be an
obvious condition to add, but it would leave us with a definition as
soft as the notion of a natural kind itself. If I have a glass of water
and you have one too, if we pour them into a single container, the
partition {your water, my water} will fail to be homeomerous for
the first few minutes, until the stuff is thoroughly mixed."2 That
partition will continue to partition the sum quantity of water even
after they are mixed, although it is certainly not a partition into
natural kinds. We can easily define a notion of "natural" homeo-
merism in the manner of D5, simply by adding the requirement
that the partition S partition the quantity Q into natural kinds.
That notion is then just as clear or unclear as the notion of natural
kind, but the softeness is not due to any problem in the underlying
definition D5.
Notice that if, in the above example, your water was at 10?Cand
mine was at 90?C, the sum of your water and my water does not
have a defined temperature until they are mixed and reach thermo-
dynamic equilibrium; it is just false that temperature is defined as
average kinetic energy of molecules for just any old quantity, such
as your water + my water before they are mixed.
But if a homeomerous mixture, as defined in D5, cannot be
created by splitting particles, how is any homeomerous partition
possible? If the water in my whisky-and-water does not exist as tiny
particles, how does it exist? Aristotle attempts to answer these
questions beginning at 328a18. But before turning to his answers, I
want to discuss some further aspects of homeomerism.
What is the receptacle occupied by the whisky in my whisky-and-
water? If {a, b} were a homeomerous partition of a quantity Q
what would be the receptacles of a, b, and Q? The answer is that all
would have the same receptacle. If a is homeomerously a proper
part of Q, still, the content c(r(a)) of the receptacle of a is identical

12 Perhaps Aristotle would not count this as a mixture, since the quantities would

not seem to be separable again. Questions: can the notion of a natural kind be con-
nected with the requirement of separability? Do quantities belong to two natural
kinds just if they can be separated if mixed? What sort of possibility is intended in
the requirement that they "can" be separated again? See also Denis Robinson, "Re-
Identifying Matter," Philosophical Review, 91 (1982) 317-341.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
448 THE JOURNAL OF PHILOSOPHY

with the quantity Q itself. I shall suggest two models on which

something might be homeomerous.
4. The Density Model. Let the points within my glass which are a
rational number of centimeters from its center be called whisky
points, and the rest water points. Then {the whisky points, the
water points} is a partition of the set of points within the glass, in
the usual set-theoretic sense of 'partition'. And every sphere within
the glass contains both whisky points and water points. Further-
more, the mixture of points is not merely relative to sharpness of
vision-it is logically impossible to have eyes sharp enough to dis-
criminate between the whisky points and the water points. The
whisky points are not next to any water points, for no point is next
to any point at all (Phys., VI.1, 231b7-10). However, each whisky
point is arbitrarily close to some water point.
But this density model is not very satisfactory. First of all, it is
just absurd to think that actual quantities of whisky and water
could be mixed this way. (De Gen. et Cor., I.5, 320b2-4: matter
cannot occupy just a point; b16: matter has points and lines as
limits.) More crucially, it used a partition of a set of points rather
than of a quantity of stuff. This conflicts with Aristotle's theory of
space.
Aristotle holds that quantities of space are not collections of
points (Phys., VI.1-2). Nor is any point part of any quantity of
space; rather, it is quantities of space that are parts of quantities of
space; it is lines that are parts of lines (cp. Phys., IV. I1, 220a20-23).
In Aristotle's topology, quantities of space are the most primary
objects, and then planes and then lines and then points are defined
in terms of them, as limits. Specifically, a limit of something is the
first point beyond which no part of that thing can be found, and
the first point within which every part is contained (Metaph.,
V.17). This "least upper bound" definition nicely leaves it open
whether a limit of something is contained in it or not. (It is inter-
esting to compare mathematical notions of limit that came twenty-
two centuries later.)
Aristotle's topology is thus methodologically like Tarski's 1927
geometrical model of mereology, in which spheres and the inclu-
sion relation are primitive, and points are then constructions from
these.13 Tarski notes that a geometric model of mereology (the so-
called calculus of individuals) occurs by interpreting "the relation
13 See his 1927 paper, "Foundations of the Geometry of Solids," in his Logic, Se-

mantics, Metamathematics (Oxford: Clarendon Press, 1957) 24-29.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 ARISTOTLE ON MIXTURES 449

of a part to a whole as the [set] inclusion relation restricted to non-

empty regular open sets" (p. 29).
Richard Cartwright employs a number of topological notions
and suggests that a set of points of space is a receptacle (i.e., one
with which a material object could coincide) if and only if it is a
non-empty regular open set, i.e., a non-empty set of points of space
identical with the interior of its closure, i.e., identical with the re-
sult of removing the limit points from the result of adding all the
limit points.14
In addition to such structures as the regular open algebra of a
topological space, there are also such things as regular closed alge-
bras of such spaces. Each exhibits the same structure; intuitively,
the difference amounts to whether a thing's border is in it or not.
This strikes me as a real "don't care"-if the Oregon-Washington
border is part of each state, then those states have points in com-
mon, which seems odd; if it is part of neither, then a road crossing
from one to the other has points that are in neither, which also
seems odd; and it would be odd for that border to belong to one of
these states rather than the other. But if we simply think of the
mereology of receptacles as primitive, and stick with Aristotle's
view of points as defined as limits, these little puzzles do not arise.
Now Aristotle's claim that no combination of quantities pre-
served in particles is homeomerous implies this:
Theorem: If S partitions Q, and if the set of receptaclesof membersof
S partitions the receptacler(Q) of Q, then S is not a
homeomerouspartition of Q.
This is so regardless of whether quantities of space are thought of
as non-empty regular open sets, non-empty regular closed sets, or
as primary undefined objects.
Aristotle saw a contradiction which seems to escape writers of
chemistry textbooks today. On its first page, a chemistry textbook
will tell the reader about the atomic theory of matter: all matter
consists of tiny things called atoms, etc. Then the reader is treated
to distinctions between compounds, solutions, mixtures, suspen-
sions, colloids, alloys, etc. (with definitions that differ from one
text to the next). But the typical definition of a solution of x re-
quires that x be uniformly distributed (!) in a continuous medium
(!). This is just what Aristotle saw was impossible if matter comes

'4Cartwright, op. cit., pp. 155-157. For more on regular open algebras, see Paul
Halmos, Lectures on Boolean Algebras, ?4 (Princeton: Van Nostrand, 1963).

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
450 THE JOURNAL OF PHILOSOPHY

in particles. But he was not an atomist, and so he could go on and

still discuss mixtures.
Now suppose that la, b} is a homeomerous partition of a quan-
tity of stuff Q. Then every sphere within the receptacle r(Q) of Q
overlaps both r(a) and r(b), by the definition of being homeomer-
ous, since every sphere has diameter greater than zero. Now every
quantity of space can be thought of as a sum of spheres (they are a
topological basis), so r(a) and r(b) are overlapped by every quantity
of space within r(Q). Hence, r(a) = r(b) = r(Q). That is,
Theorem: S is a homeomerous partition of Q if and only if every mem-
ber of S occupies the same space, which is also the space
occupied by Q.

So, we have homeomerous quantities just if we have quantities that

have no common part, yet which occupy the same space.
This fits in with the density model. If we think of the space oc-
cupied by a set of points as its topological closure (or the interior
thereof), then we can have completely disjoint sets of points which
occupy the same space; this will happen if the two sets are dense in
each other in the manner of the whisky-points and the water-points
discussed above. {The whisky, the water} is a homeomerous parti-
tion of my whisky-and-water just if the whisky itself occupies the
same space as the water and as the whisky-and-water.
It is not clear to me that Aristotle saw this implication of his no-
tion of homeomerism, although he does say in the Physics that if a
quantity is both continuous and homeomerous, its parts have only
potential receptacles (places); but when the parts are divided but in
contact, as in a heap (beans and rice?), then they are actually in
places (IV.5, 212b3-6). And Thomas comments "if the elements in
a mixture are in distinct positions we only have an apparent
mixture. " 15
Aristotle does discuss the idea of two bodies (somata) being in
the same place, and definitely implies that this is impossible
(Phys., IV.1, 209a6, 6, 213bl8-20, 7, 214b7, De Gen. et Cor., 1.5,
321a8-10). But this is always stated in terms of bodies. Could two
quantities of matter be in the same place? At 320b8-10 he says that
if air comes to be from water, it is because its matter is in the water
as in a container, and that nothing prevents an infinite number of
matters being contained in the water. And what is needed for a
mixture is not two bodies in the same place, but two quantities of
stuff.

"5Thomas Aquinas, Summa Theologica, Part I, Bk. VI, Q. 76, Reply Obj. 4.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 ARISTOTLE ON MIXTURES 451

But I see no reason to worry about this. Since Schr6dinger and

De Broglie, we have been encouraged to think of gold, iron and
wood as not so different from light, heat and sound."6 A quantity of
sound can occupy the same space as a quantity of air. Finally, notice
that chemistry itself routinely speaks of the volume of the nitrogen
gas in my room as being equal to the volume of the oxygen and
equal to the volume of the sum of these two. Otherwise, the gas
laws relating pressure and volume make no sense, especially for
partial pressures. And a wine bottle's claim to contain 12%alcohol
by volume is nonsense-the receptacle of the alcohol is that of the
wine itself. Probably what is meant is a conditional: if the alcohol
were separated, then its volume would be 12% of the total. But re-
call that mixing 10cc. of water with 10cc. of alcohol yields about
19cc. of mixture. While it is still mixed, the volume of the alcohol
is 100% of the volume of the wine."7
If a geometric model is wanted which doesn't have the faults of
the density model, we can use

5. A Projection Model. On this, a quantity of matter is four-di-

mensional, and the space it occupies is its three-dimensional spa-
tial projection, like a shadow. Notice that his allows discrete quan-
tities to have the same receptacle. In fact the only formal
requirement on a receptacle function is this:

for any set S of quantities, the receptacle r(Fu(S)) of the fusion of S is

the fusion Fu{x: x = r(ay *y e S)} of the receptacles of the members of S.

This holds both in the density model (on which the receptacle of a
set of points would be the interior of its closure) and in the projec-
tion model. From this, other equalities and inequalities will fol-
low, such as
x c y D r(x) ' r(y)
r(x n y) < r(x) n r(y).

1 See E. Schr6dinger, Science and Humanism, (New York: Cambridge, 1951), and
"Conceptual Models in Physics," in his Science and Human Temperament (New
York: Norton, 1935); Louis De Broglie, Matter and Light (New York: Dover, 1937);
and N. R. Hanson, "The Dematerialization of Matter," in E. McMullin, ed., The
Concept of Matter (Notre Dame, Ind.: University Press, 1963). (This book is out of
print. Many of the articles in it appeared later in two other collections of alrticles on
the history of the concept of matter edited by McMullin, but the Hanson article was
not among them.)
17Discussions of matter and density can found in Bernard Bolzano, Paradoxes of
the Infinite, ??50-69 (London: Routledge & Kegan Paul, 1950); David Sanford, "Vol-
ume and Solidity," Australasian Journal of Philosophy, 45 (1967) 329-340; and E.
Schr6dinger, op. cit., esp. pp. 11-47. See also Robinson.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
452 THE JOURNAL OF PHILOSOPHY

In fact mathematicians call any such function r on a distributive

lattice a prolection.18 Notice that the function m(x), the matter of
x, is itself a projection. If a = the set of rabbits, and b = the set of
integral proper rabbit parts, then a n b is empty, and so is m(a n b).
But m(a) = m(b) = the world's rabbit stuff. This projection model
was suggested by the literal notion of dimension, but has wider
application. The notion of topological closure used in the preced-
ing section to picture the receptacles of sets of points also satisfies
the above requirements.
Questions about the dimension of something can become very
complicated for exotic mathematical structures. The intuitive no-
tion was nicely described by Poincare in 1912:
. . . lines, which can be divided by cuts which are not continua, will
be continua of one dimension.... If to divide a continuum C, cuts
which form one or severalcontinua of one dimension suffice, we shall
say that C is a continuum of two dimensions; if cuts which form one
or several continua of at most two dimensions suffice, we shall say
that C is a continuum of threedimensions;and so on.19
Roughly, the idea is that the dimensional number assigned to
something is one higher than the number assigned to the sorts of
objects that form its boundaries, the things which separate it from
other objects. I will leave for another place the investigation of the
relations between homeomerism and dimension.
The projection model was based on the literal idea of geometri-
cal dimension, but it tells us nothing about how it can happen that
two quantities of matter could occupy the same three-space, yet
differ at a fourth co-ordinate and so be disjoint. However, it is one
way in which we can have a homeomerous mixture of disjoint
quantities of stuff.
Recall Plato's famous remarks on the projection model (Repub-
lic VII, 515b-c) and the underdetermination of theories by observa-
tions. His natives are watching shadows in a cave, projections, and
carrying on quite sensible conversations, at least when they stick to
observation sentences. But they also think that they are making ref-
erences. According to Plato, it is just here that their talk should be
viewed as theoretical rather than purely observational; it is here,
says Plato, that the problems arise. Quine's solution is that

1 Proofs: x c y iff x U y = y, so r(x U y) = r(x) U r( y) = r( y), so r(x) < r( y).

Now x n y < x, so r(x n y) < r(x), and similarly r(x Q)v) < r( y); theni r(x n y) <
r(x) n r( y). See also Halmos, ?7.
'9Quoted in W. Hurewicz and H. Wallman, Dimension Theory (New York: Ox-
ford, 1941, and Princeton, N.J.: University Press, 1948), p. 3.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 ARISTOTLE ON MIXTURES 453

nothing can be said; Plato thinks that something can be said-that

these natives are just mistaken. "If they could discuss things with
each other, don't you think they would maintain that they were
referring to the things that they see passing before them?" (515b).
Plato, that is, thinks that we can, from outside the story, determine
what they are "really" referring to. In Plato's description, the real
things were simply objects of greater spatial dimension. (Notice
that shadows, like ingredients in a mixture, can pass through each
other.)
In any case, the projection model is purely geometric, and al-
though it gives us a mathematical representation of a homeomer-
ous mixture, it does not really explain how such things might
really work. Art is representation; science is representation that ex-
plains. The projection model falls short here. It is very, very sug-
gestive, and brings together a number of philosophical questions
under a single picture. We have two projections which are very
similar: the matter of a thing, and the receptacle of sm matter. Just
as two things (such as I and my body) or two sets of things (such as
the set of rabbits and the set of integral rabbit parts) might have the
same matter, two quantities of stuff, such as the whisky and the
water, might have the same receptacle. Although this picture of
things falls short of being a scientific explanation, it is a suggestive
picture with many applications.
6. Potentiality and Matter. Aristotle turns to a discussion of action
and receipt of action. Agents which do not have the same matter as
their patients act but are not acted upon, and thus cannot mix: the
art of healing does not mix with the bodies of patients (328a21-23).
This suggests another axiom about mixtures:
Axiom 4. Two things can mix only if they can act reciprocally on each
other, and this requires that their matter be similar.

Not every such combination will be a mixture, because the in-

gredients must be within an appropriate range of proportion, or
one may be destroyed, in violation of a previous principle. A drop
of wine added to ten thousand measures of water does not mix; the
form of wine is lost; the wine becomes part of (!) the resulting total
quantity of water (328a27-28). So
Axiom 5. There must be a balance between the active powers of the
various ingredients.

Now Aristotle had already briefly indicated his explanation of

how ingredients could persist in a mixture they persist potentially.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
454 THE JOURNAL OF PHILOSOPHY

Suppose that I drop some ice cubes into a glass of good strong
tea. I do not immediately have any more tea, but as the ice melts, it
becomes part of what then has become tea. Here, it seems that ho-
meomerism is involved in the difference. Some quantities which
were merely in something become part of it when they merge ho-
meomerously with it. In the film The Invisible Man, the title char-
acter puts on a dressing gown when he has a visitor just after eat-
ing dinner. The point was to spare his visitor the sight of chewed
food floating about the room at stomach height. The invisible man
explains that food he eats becomes invisible only when it is di-
gested and thus becomes part of him. (Other possible questions are
tastefully ignored.) Similarly, water preserved in particles, as in a
cloud, is not part of the air, but is merely in the air like a bird or an
airplane. But if it evaporates, it ceases to exist in particles, and be
comes part of the air. Aristotle frequently speaks of water coming
to be from air, and of air coming to be from water (e.g., Phys., IV.7,
214b3, VII.4, 255bl8-19, De Gen. et Cor., 1.2, 317a28-30); he also
speaks of water being potentially air, and air potentially water
(Phys., IV.5, 213a4-11).
Question: do we have in examples of homeomerism examples of
a quantity of matter existing potentially? I believe that this is not
coherent. The notion of potential existence has application to an
individual substance: a particular bust of Hermes may have existed
for only 20 years, although its matter, the wood in it, has existed
for much longer (cp. Metaph., IX.6, 1048a31-33), and the Hermes
had only a potential existence in the original block of wood. But
this explanation requires that an individual substance be some-
thing other than its matter. It requires that being a bust of Hermes
is not a property of that matter. Rather, it is constituting a bust of
Hermes that is a property of the matter. For if being a bust of
Hermes were a property of that matter, then the singular term 'the
bust of Hermes' would in fact designate the wood rather than the
bust-or would ambiguously designate both.20
In that passage in the Metaphysics, we have the Hermes des-
cribed as existing potentially inside the block of wood. This re-
quires that the Hermes be distinct from its matter-otherwise, it
would come to exist in a de dicto sense only. That de dicto sense is
the one in which if my coffee becomes cold, then some cold coffee

20This is an instance of a general principle that is often missed. Here are some
other instances of the principle: if being air were a property of the space in this
room, then 'the air' would designate the space; if 'is a skillet' were true of the iron in
my skillet, and 'is iron' were true of the space it occupied, then it would be true that
the skillet = the iron = the space.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 ARISTOTLE ON MIXTURES 455

has come into existence. But the sugar in the coffee cannot be like
the Hermes in the wood. Whereas the Hermes comes to exist de re
when the sculptor carves the wood, the sugar is matter, and must
exist throughout the process of dissolving and separation.
The notions of form and matter are used to explain the becom-
ing and destruction of individual substances. But the matter in-
volved is something which itself is not subject to becoming or des-
truction. So what sense does it make to say e.g. that the beans in
my beans and rice exist actually, but that the sugar (dissolved) in
my coffee exists only potentially? This is a mistake, since the sugar
is itself sm matter, and cannot have any "further" matter which
persists when it, that sugar, comes to be or passes away (if it does;
but it doesn't). Aristotle says this himself: "For if it came to exist,
there must have existed a primary substratum from which it came.
... For matter is what I call the primary substratum of each thing,
from which it comes to exist . . ." (Phys., 1.9, 192a28-34).
It would be unfair to ignore another, looser, sense of matter that
Aristotle uses, and which might help here. In this looser sense, the
matter of something is, in the plainest sense, that of which it con-
sists, is composed, or is made. In this sense, the matter of a deck of
cards is the cards, the matter of a syllable is the letters, and the mat-
ter of a forest is the trees (Phys., 11.3, 195al5-19). On this account,
the trees in turn have matter-the wood of which they are com-
posed, and the wood in turn has matter-prime matter, perhaps.
Indeed, this chain yields one definition of prime matter: it is the
limit of the matter of relation.
On this account then, some matter can be destroyed. The trees
are the matter of the forest, and they can be destroyed if their matter
loses its form as trees, perhaps in a forest fire. Is it only prime mat-
ter then, that would be indestructible on this account? Notice that
the sugar is matter in a sense in which the trees are not. It is matter
in the unqualified sense, i.e., the non-relative sense. Although the
trees are matter only in that they are the matter of something, the
sugar is matter even if it is not the matter of anything. It need not
exist as a lump; it can just be a formless heap (cf. Metaph.,
VIII.16-17, VIII.6).
The sugar that I added to my coffee may be homeomerously in
the sugar + the coffee, but it certainly exists. It might not still be
sugar-but it, that stuff, cannot pass away. It may also be poten-
tially visible sugar, and actually tasteable potential sugar-it can
be separated again by evaporation of the water. But it, whether ac-
tually sugar or not, still exists. Matter is ultimately the substratum
of coming to be and passing away. That is, no matter itself is ever

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
456 THE JOURNAL OF PHILOSOPHY

created or destroyed. Aristotle's remark at Metaph., IX.8, 1050al5-17,

that matter exists potentially just because it may attain form, just
seems crazy. For what is the it? What is the it, unless something
that actually exists?? Where is the snow of yesteryear? All around
us, and in us.
But Aristotle has called our attention to something very
interesting:
There are some kinds whose parts are necessarily homeomerous.

One example is the water in tea. That is exactly the situation we

had with the ice cubes melting into a glass of iced tea. The ice
cubes are H20 not homeomerously contained, and so are merely in
some tea, surrounded by it; when they melt, that H20 becomes part
of some tea. The water droplets in a cloud are merely in the air, but
when they evaporate, they become part of the air. So air is a kind
for which water is a necessarily homeomerous ingredient. Aristotle
seemed to hesitate at saying that there is actually water in air when
it is a homeomerous part; I would not hesitate at all, but nothing
turns on this difference. What is important is to see that Aristotle
should have said that whether or not what was water in the cloud
is still water when it has evaporated, still, it exists as part of the air.
Chemical knowledge may have now changed the actual circum-
stances under which we learn mass terms. The child learns 'iron'
accompanied by mama's skillet, but also in sentences like 'eat your
spinach; it's got iron in it'. In any case, what is in the spinach may
be potentially iron-in-the-primnary-sense, and actually iron-in-a-
secondary-sense. So there does not exist in the spinach any ironi.
Similarly, when sugar dissolves in water, the sugari ceases to exist;
yet nothing has actually ceased to exist. Rather, some stuff has
changed qualitatively from being sugar, to being sugar2. This is no
more ceasing to exist than the de dicto sense in which a cultured
man ceases to exist when Ronald, a cultured man, becomes uncul-
tured. (See De Gen. et Cor., 1.4, 319b6-31.) We have alteration
(rather than generation or destruction) if and only if we have a sin-
gle matter (1.1, 314b29-315a3).
To the extent that Aristotle's explanation of mixtures depends
upon a quantity of matter itself existing potentially at one time
and actually at another time, his explanation fails to be consistent
with the rest of his theory of matter. The idea of potential existence
properly belongs to the form+matter theory of individual substan-
ces. But the theory of mixtures is wholly within the pure theory of
matter.
On the other hand, he has called our attention to certain kinds of

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions
 FITNESS 457

mixtures, and to a relation between the being part of relation and

homeomerism. Specifically, there are certain kinds of stuff A,
quantities of which contain quantities of other kinds of stuff B as
part if and only if the B exists homeomerously in the A. An exam-
ple is the water in the air, which is part of the air if and only if it is
in it homeomerously, but otherwise that water is merely in the air
like a bird, without being part of the air.
This I find quite exciting.
RICHARD SHARVY
The Avondale Institute

FITNESS*
EBATESabout the cognitive statusof the Darwiniantheory
of natural selection should have ended long ago. Their per-
sistence reflects the steady failure of biologists and philoso-
phers of science to treat the notion of fitness as the quite ordinary
theoretical term which in fact it is. Even the rare expositions of fit-
ness and its role in evolutionary theory that have been correct have
failed to put the methodological controversy over this theory to
rest.' In this paper I shall show that 'fitness' differs from an ordi-
nary theoretical term, like temperature, not in kind, but only in
degree; that this difference sets limits on the measurement of fit-
ness; that these limits give the theory of evolution its undeserved
reputation for vacuousness. I then apply these conclusions about
fitness to laboratory experiments in evolution, with a surprising re-

*The author must thank Jonathan Bennett, Daniel Hausman, Richard Burian,
for detailed comments on an earlier draft, and Peter van Inwagen and Mark Brown
for specific improvements of the current version. Research supported by a John
Simon Guggenheim Memorial Foundation fellowship.
IFor example, D. Hull, The Philosophy of Biological Science, and M. Ruse, The
Philosophy of Biology, both address the allegation that the theory of natural selec-
tion is vacuous but they fail to explain why it persistently attracts this false charge.
Other recent attempts to refute the charge, such as Mary Williams, "Falsifiable Pre-
dictions of Evolutionary Theory," Philosophy of Science, 40 (1973): 518-537, or
S. K. Mills and J. H. Beatty, "The Propensity Interpretation of Fitness," Philosophy
of Science, 46 (1979): 263-288, neither correctly diagnose the source of this error, nor
provide effective remedies for it, and they generate some further obstacles to the dis-
solving of the mistake. For details of these defects see, A. Rosenberg, "The Super-
venience of Biological Concepts," Philosophy of Science, 45 (1978): 368-386, and
"On the Propensity Definition of Fitness," Philosophy of Science, 49 (1982):
268-273.

0022-302X/83/8008/0457$01.70? The Journal of Philosophy, Inc.

This content downloaded by the authorized user from 192.168.72.223 on Mon, 3 Dec 2012 19:14:33 PM
All use subject to JSTOR Terms and Conditions