Aristotle on Mathematical Objects

Edward Hussey

1. Introduction
2. The nature of mathematical objects: a first look at Metaphysics M 3
2a. 1077bl7-1078a5
2b.1078a5-17
2c.1078al7-28
2d.1078a28-31
3. Metaphysics M 3: foundations of an interpretation
3a. 'Considering as'
3b. 'Separation'
4. Representative objects
5. Aristotle's mathematical objects as 'representative objects':
Metaphysics M 3 revisited
5a. 'Considering qua...', abstraction, representative objects, and
the threat of incoherence
5b. The potential existence of mathematical objects
5c. Separation
5c(i)· Mathematical practice
5c(ii). The need for separation
5d. Are mathematical objects necessary to mathematics?
6. Further considerations
7. Conclusion

1. Introduction

'Aristotle's philosophy of mathematics' is sometimes understood nar-

rowly as including only the thoughts on mathematical objects set out
hastily in Metaphysics M1-3. But there is a great deal more to Aristotle's
thinking about the nature of mathematics, evenBrought
if it is tonot
youexpounded
by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
106 Edward HusSey

in a connected and explicit way but has to be searched for in scattered

passages of the Physics, de Caelo and elsewhere.1 A full account of
Aristotle's philosophy of mathematics would have to cover a much
wider territory and provide a map of the connections between different
areas. This paper confines itself to the problem of the objects of mathe-
matics and is necessarily centered on the third chapter of Metaphysics
M.2

2. The nature of mathematical objects: a first look

at Metaphysics M

Aristotle's primary purpose in Metaphysics M 1-3 is a negative one. He

is concerned to show that the existence of mathematics as an unques-
tioned science and source of truths, having (apparently) objects of a
special kind, does not entail the truth of any theory of a Platonic kind
about mathematical objects, and hence that no argument can be drawn
from mathematics to the existence of non-sensible substances. He ar-
gues in the first two chapters that mathematical objects cannot be
substances of any kind, whether 'separated' in Platonic fashion or 'in
sensibles' according to some other theory.3 This leaves the question:
what then are mathematical objects, if not substances? They must exist
'in some other wayO076a35, cf. 1077bll). In M 3, down to 1078a31,
Aristotle provides the sketch of an answer. As so often, he is presum-
ably in many places recalling by compressed allusions thoughts famil-
iar to himself and his intended audience. We do not get a full-dress
exposition, and some essential points are left unexplained. In this
section and the next, a first look at M 3 (-with a flashback to the first two
chapters) will concentrate on establishing, not an interpretation, but
certain constraints on any interpretation worth considering.

1 Much but not all of the relevant material is collected in Heath (1949).
2 This paper builds on, but also substantially revises and extends the account given in
sections 5-8 of Additional Note I ('Aristotle's Philosophy of Mathematics') of Hussey
(1982), 179-184. Other works to which this paper is much indebted are: Mueller
(1971); Annas (1976), 26-11, and the notes on M 1-3; Lear (1982); Annas (1987).
3 On the nature and purpose of Metaph M1-3, see especially Annas
Brought (1976),
to you by | 26-28.
ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 107

Aristotle takes it for granted throughout M 1-3 that there are such
things as 'mathematical objects' (in particular, numbers and geometri-
cal figures), and that they are not straightforwardly identical with any
entities of a more ordinary kind. By 1077bll Aristotle takes it as
established that mathematical objects (A) do not exist 'apart from'
sensible objects; (B) are prior to sensible objects in definition, but (C)
posterior to them in being/substance. From (B) and (C) it is clear that
mathematical objects are not just identical with sensible objects.
Points (A) and (C) hang together. Mathematical objects must exist in
some way that is derivative from the existence of sensible objects.
Otherwise, there is a pointless (and potentially unending) multiplica-
tion of entities which corresponds to and is supported by nothing in
the actual practice of mathematics or its application. Nor can we see
how these entities could be the sort of things to be capable of existence
apart. But (B) mathematical objects are prior in definition to sensible
objects (and therefore not identical with them). The definition of 'sen-
sible body' involves the definition of 'body' as three-dimensional mag-
nitude; and this in turn presupposes the definitions of one- and
two-dimensional magnitudes.4 Because mathematical objects are prior
in definition, they may be called 'abstracts'; the 'abstraction' here
envisaged seems to be or to correspond to the logical splitting up of a
definition into its component parts.
At 1077bl2 there follows the transition to the passage of principal
interest, 1077bl7-1078a31, on the way in which mathematical objects
exist, a way which must be 'not straightforward' (1077bl6), i.e. not the
way of substantial existence. It will be convenient to divide the discus-
sion into subsections.5

4 This point is important in relation to the 'exactness' of mathematics and the hierarchy
of branches of mathematics and mathematical physics, topics not discussed in this
paper. For mathematical objects' priority to and distinctness from sensible sub-
stances, see also Metaph Γ 2,997b32-998a6 and 6,1002bl2-16; Γ 5, 1002a4-8; El.
5 The usual punctuation of the Greek text (e.g., that of Ross and Jaeger) obscures the
structure of the passage. We need full stops after μόνον (1077b30), ωσαύτως (1078al7)
and ύπάρχειν (1078a28); the translation of Annas (1976)Brought
gets thistopoint right.
you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
108 Edward Hussey

2a. W77bl7-1078a5

In 1077bl7-1078a5 Aristotle draws three parallels, each introduced by

the word hösper. Each is meant to help us understand the relationship
between mathematical objects and the actual world. The first parallel
(1077bl7-22) invokes the relationship between more general and more
special branches of mathematics. The more general study of 'pure
quantity'6 is not concerned with objects that exist in separation from
numbers and magnitudes. It is concerned with numbers and magni-
tudes, since all actual quantities are one or the other; but it is concerned
with them not specifically as numbers or magnitudes respectively, but,
we may supply, as quantities. Applying the analogy to mathematics
and sensible objects, Aristotle says that mathematics is concerned with
sensible objects, but concerned with them not as sensible, but as having
mathematical properties.7
The second parallel (1077b22-30) is now brought in to support or
explain the first one. It invokes the relationship between the objects
studied by other sciences, such as physics, and the actual world. Phys-
ics can study and be concerned with changing sensible things, consid-
ered as changing and disregarding other aspects of them. This does not
commit it to introducing changing substances which are apart from the
actual world or embedded as different substances within it. So too with
mathematics, when (with applications to the changing world) it studies
solids, planes, lines and points.8 It too, we are apparently meant to
conclude, can be said to be concerned with changing sensible things
considered in a certain way.
In these two parallels, then, the link with the actual world is empha-
sized. But from the fact that mathematics and other sciences are con-
cerned with sensible substances considered in certain ways, it does not

6 On this see Lear (1982), 165-7.

7 I use the cumbersome 'concerned with' for peri (likewise, I would use it later for epi
or for the simple genitive in the same sense), rather than 'about' in order to minimize
ambiguity. 'Mathematics is about X' might be taken to mean that X is the (real, only,
primary) subject matter of mathematics. We must not assume that Aristotle's aim
here is to identify the primary subject matter.
8 Numbers are not mentioned, but may be included by means of the phrase 'qua
indivisible only" (b30). Aristotle is concentrating on geometry here; for a brief
discussion of arithmetic, see §6 below. Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 109

follow that the proper objects of these sciences, what they are most
precisely 'about', are just sensible substances. In fact, as already re-
marked, Aristotle takes it that that is not so. Nonetheless the implica-
tion of these two parallels is that there is at least a close logical tie
between sensible substances and those proper objects. The proper
objects of mathematics must be such that mathematics is 'about' them
just because, and just to the extent that, it is concerned with (actual or
possible) sensible substances considered in their mathematical aspects.
The conclusion drawn (1077b31-4) from these two parallels is that,
just as it is 'straightforwardly true to say' that the changing things with
which physics is concerned exist,9 so it is straightforwardly true to say
that mathematical objects do. This conclusion is, of course, closely
linked to what precedes. To preserve the parallel, we must not think of
'changing things' here as particular changing sensible substances, but
they must be, once again, things which are closely tied, logically, to
changing sensible substances. In particular, it must be 'straightfor-
wardly true to say' that 'changing things' exist, just because, and just
in so far as, it is straightforwardly true to say that (actual or possible)
changing sensible substances exist.
The third parallel (1077b34-1078a5), like the second, invokes the
other sciences. But here the proper objects of those sciences are explic-
itly invoked. Each science has its proper object, which can be distin-
guished from irrelevant properties that may happen to be associated
with it. The science of medicine has as its object 'the healthy7; but not
'the white', even if it should be the case that every particular thing that
is healthy is also white. So too with geometry. It may be the case that
every particular thing that has geometrical properties is a sensible
object. It does not follow that the proper objects of geometry are
sensible, or sensible objects.10 On the other hand, neither does it follow
that these proper objects exist in separation from sensible objects.
It is clear enough in a general way what Aristotle is trying to do, and
what he is not trying to do in these parallels. He is trying to give models
for the relationship between mathematics and the actual physical

9 Hap/os must not be taken with einai; three words separate them, and Aristotle has
already said (1077bl6) that these objects do not exist haptös. I take haptos to qualify
atethes, which it immediately precedes, rather than apein as Annas (1976).
Brought
10 The syntax of 1078a2-5 is tortuous, but the sense required is nottoinyou by | ULB Bonn
doubt.
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
110 Edward Hussey

world; models that will help us to see how it can be that mathematics
(a) is 'about' the physical world and not about some detached realm;
yet (b) has proper objects that are not physical bodies (nor, even, limits
of physical bodies), though not detached from them, but logically
closely tied to them. He is not trying to explain just what the proper
objects of the sciences are or just how they exist.
Why, then, are his explanations themselves so obscure? They hardly
help, because without antecedent knowledge we cannot be expected to
understand how Aristotle sees the parallel phenomena either. But the
whole section does at least serve to stake out the extent of the territory
Aristotle wishes to occupy. He wishes, for one thing, to show that
mathematics is basically in the same position as other sciences,11 and,
for another, to insist that all the sciences are concerned with the ordi-
nary world of sensibles. The objects of the sciences cannot themselves
be substances. Therefore, insofar as the sciences are about anything real
(and they are) they must be concerned with ordinary sensible sub-
stances. And this makes sense, since ordinary sensible substances do
embody, in a straightforward and non-controversial way, actual par-
ticular 'exemplifications' of the proper objects of study of the sciences:
they contain actual particular points, lines, changes, examples of health,
diseases, etc. So we must distinguish mathematical objects and their
actual particular 'exemplifications'12 which verify the theorems of ge-
ometry in the actual world. But so long as these proper objects are not
themselves substances, no argument from the sciences to non-sensible
substances can be derived by Platonists or others.
The point of claiming that the existence of these objects can be
'straightforwardly' affirmed is to emphasize, against Platonic doubts,
the closeness of the link between ordinary particulars and the objects of
the sciences. That closeness is affirmed over again, in different words,
by the claim that the sciences — whether mathematics, kinematics or
medicine — are 'concerned with sensible substances'; and that they
'consider [sensible substances] qua three-dimensional/moving/

11 Thus, the point of the second parallel appears to be, not to enlighten us about how
mathematical objects exist, but to emphasize the fact that mathematics is in just the
same position as any other science so far as the question of existence of its objects
goes.
12 This term is intended to be non-question-begging. Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 111

healthy, etc.', i.e., regard them as matter for theoretical treatment when
considered in a certain special aspect. Again, of course, something vital
is left unexplained, namely, just how considering actual sensible sub-
stances under these aspects is related to considering the proper objects
of the sciences. It is obvious that, e.g., physics applies its truths to the
understanding of actual situations in the real world considered under
the aspect of 'changing bodies of such a kind'. It is not obvious how
these truths are arrived at or in terms of what sort of bodies they are
formulated.
Such is Ans totle' s line of thought in this section. As already remarked,
it looks incomplete to us, in that he does not specify just what the objects
of the sciences are and in just what way they exist.13 That is not his concern
here. He concentrates on the parallel with other sciences and on the link
with actual sensible substances. But that must not mislead us into iden-
tifying them as the proper objects of study for the sciences. That is ruled
out by explicit statements made here and elsewhere. On this point
Aristotle shares common ground both with his Platonist opponents, and
indeed with ordinary common sense about what geometers do.

2b. 1078a5-17
This section continues the thought of the last, but develops it in order
to reach an understanding of the internal structure of mathematics. The
first point (1078a5-9) is that the objects of mathematics, whatever they
are and though they are not sensible particulars, have properties which
are studied by mathematics and which they 'inherif from the sensible
particulars from which they are abstracted. This accounts for the fact
that the subject matter of mathematics is just what it is. It is concerned
with just those properties of sensible particulars that 'hold true of them
in their own righf in so far as they are considered mathematically. In
this respect mathematics is just like any other science, e.g., biology.
From this point Aristotle is then able to go on (1078a9-17) to account
for the hierarchical relationships existing within 'pure' mathematics
and between 'pure' and 'applied' mathematics. The objects of the
various branches of mathematics may be arranged in order of

13 We are told that the existence of the objects of any science can be 'straightforwardly'
affirmed. Not till later are we are given a due about how this can be, and what sort
of existence is involved. Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
112 Edward Hussey

'simplicity', by which Aristotle explains that he means priority in

definition. This order is also an order of 'exactness'; the simpler
branches are the more exact. Although Aristotle does not say so explic-
itly, there is perhaps here also the explanation of why mathematics is
distinguished from other sciences by its particularly 'exact', 'self-con-
tained', and 'detached' nature. The proper objects of mathematics are
prior in definition to sensible objects, as stated in M 2,1077a36-bll. The
consequence which is relevant to the context and follows from this is
that mathematics, unlike other sciences, is particularly 'self-contained'.
In spite of its close links with the natural world, it involves no necessary
reference to that world at all. This is why it may appear to verify Platonist
claims by being not merely timeless but easily 'detachable'. By contrast,
in the case of any branch of 'mathematical physics', the proper objects
must be changing things, which involves a complication, compared with
pure mathematics (1078al2-3).

2c. Ί078αΊ7-28
In this section we are told more about what the mathematician does.
He 'posits' his objects 'in separation from accidental properties', and
considers them as such (1078al8-9). Aristotle is here principally con-
cerned to insist that no falsehood will result from this procedure and
that it is a good way of studying because it removes what is irrelevant.
What he is talking about is clear in a general way. It is a familiar fact
about mathematics that as a theoretical science it can operate as though
in a complete vacuum, making no essential reference to anything
whatever outside itself. But his problem is to explain how this can be so
if a Platonist view is rejected. And it must be said that what Aristotle
says here is again far from self-explanatory. The obscurity (to us) in the
formulation of his account seems entirely due to the fact that he has not
told us what sort of things mathematical objects are. We know that, e.g.,
in geometry they are points, lines, plane figures, etc. of some sort. We
also know that they do not exist in the way that sensible substances do
and are not substances at all. We know that they are 'from abstraction',
i.e., they are 'arrived at' in some sense14 and by some kind of logical
operation from sensible substances. Once a decision is arrived at on

14 It is at this point unclear whether they (a) exist antecedently and are found or grasped
by thought, or (b) are brought into being by construction. See below
Brought to you§3b.
by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 113

these questions, this section too may become clearer; but it would be
ill-advised to attempt to interpret this section in advance of any answer
to them.

2d. 1078a28-31
This short final section does not remove the fundamental obscurity,
although it adds something very important. The objects of mathematics
do really exist; for 'to exist can be taken in two ways: in actuality, or as
matter does'. It is natural to understand this as equivalent to 'poten-
tially in the same sense as matter' elsewhere;15 but this phrase in turn
needs elucidation.

3. Metaphysics M 3: foundations of an interpretation

The first and most fundamental requirement for any interpretation of

Aristotle's account of mathematical objects is that it should recognize
the radical incompleteness of the account offered by Metaphysics M 3. Any
interpretation that assumes that all necessary information is provided
within the chapter is bound to run into trouble.16 Nor is there any
guarantee that the missing evidence will be supplied elsewhere.

3α. 'Considering as'

Take, to begin with, the operation of 'considering' (or 'studying') sensi-
ble objects 'as...'. Is this operation all that we need to know about in order
to grasp the nature of mathematical objects? That cannot be. We know,
in the first place, that (a) mathematical objects cannot just be ordinary
sensible objects considered under some description. For (apart from the
evidence of other passages already cited) which sensibles are these? And
what if there exist no actual sensibles of the kind we need (e.g., regular
icosahedral solids)? Nor (b) can mathematical objects plausibly be taken

15 Aristotle says that the infinite exists 'potentially as matter' at Phys ΠΙ6,206bl4-15; cf.
206al8-b3,207a21-32,207b35-208a3.
16 The sketchiness of M 3 is nghtly emphasized by Annas (1976), but her comment that
Brought
'it is straightforward and the sense requires little elucidation' to is
(148) you by | ULB Bonn
over-optimistic.
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
114 Edward Hussey

to be objects somehow brought into existence, or made present to

thought, by the mere act of 'considering as...', applied to particular sensi-
bles. For once again the unanswerable problem of existence rears its
head; what if there are no sensibles of the required kind? Nor (c) can we
improve the last suggestion by saying that it is not particular sensible
objects, but actual sensible objects generally, that are to be 'considered
as...'. For then one no longer sees what might be meant by the operation.
And (d) if we seek to circumvent the problem of existence by extending
'considering as...' to possible sensible objects, then the mathematician
will have to know in advance which are possible and which are not,
which (at least in geometry) implies that he already knows a consider-
able amount of mathematics. Even if this last point is disregarded, there
is actually no warrant in the passage for saying that 'considering' actual
or possible sensible objects in a certain way leads to the creation of, or is
identical with the consideration of, mathematical objects; or that the
existence of mathematical objects means only that sensible objects can be
considered in a mathematical way.17
In fact, as already suggested, the primary point of the talk about
'considering as...' is to emphasize the mathematician's links with the
actual world, not to explore the ways in which he turns partly away
from it.18 The mathematician considers actual particulars under mathe-
matical descriptions in order to apply his theorems, or possibly in order
to discover them, but not in order to state and to prove them.
It is true, of course, that Aristotle introduces the talk about 'consid-
ering as...' as part of an effort to explain something about mathematical
objects. However, what Aristotle wants to do is not to explain just what

17 Lear (1982), 170 claims that such a warrant is supplied by 1077b31-4: 'So since it is
true to say without qualification not only that separable things exist but also that
non-separable things exist (e.g., that moving things exist), it is also true to say without
qualification that mathematical objects exist and are as they are said to be.' (Transla-
tion of Annas (1976).) The words 'so since' do indeed link this sentence logically with
what immediately precedes; but they do not prove that it expresses the conclusion
of an argument based solely on what precedes
18 As seen by Mueller (1971), 158. Aristotle also emphasizes the link with the real world
at Phys Π 2, 193b22-25, which however need not be read as saying that it is (only)
physical points, lines, etc., with which the mathematician is concerned. Lear's render-
ing ((1982), 162) of 193b25 'and these are the subject-matter of mathematics' is
potentially question-begging; better 'and these are things that the mathematician is
concerned to investigate'. See note 7 above. Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 115

they are, nor how they are arrived at, nor in what sense they exist.
Rather he wants to point to the fact that they and what can be said about
them are closely and necessarily linked to the actual world and what
can be said about its actual and possible contents. At a minimum this
requires that the considering and studying of mathematical objects
should somehow involve the considering of sensible substances in a
certain way, and that any theorem which is proved in mathematics is,
at least potentially and approximately, applicable to the actual world, just
in virtue of being a theorem of mathematics.19 Any interpretation of
'considering as...' has to show that mathematical objects, as it describes
them, are related in this very close way to the actual world. To consider
sensibles 'as mathematical' must involve considering mathematical
objects. And, at the very least, the existence of mathematical objects
must be bound up with the (possible or actual) existence of sensible
objects with mathematical properties.

3b. 'Separation'
The constraints on the understanding of 'separation' formulated at
1078al7-28 can be expressed by a series of interwoven questions to
which any interpretation must give defensible answers. Physics II 2,
193b22-194al2 and de Anima III 7, 431M2-19 must also be taken into
account.

19 According to Lear (1982), 170, One can say truly that separable objects and mathe-
matical objects exist, but all this statement amounts to — when properly analyzed
— is that mathematical properties are truly instantiated in physical objects and, by
applying a predicate filter, we can consider these objects as solely instantiating the
appropriate properties.' (There is a similar view in (pseudo-)Alexander's commen-
tary on this passage at 734.1-739.18, and in some other recent treatments.) Apart from
its lack of any positive support in the text, this interpretation does not explain why
Aristotle should speak of existence of objects at all in such a context, let alone why
he should speak of potential existence. But there is a more fundamental objection to
Lear's position. Thysical objects' must here include possible physical objects as well
as actual ones (otherwise most of the required objects will not 'exisf even in the
tenuous sense). But to determine which physical objects are possible, it is already
necessary to have recourse to geometry and to show first by geometrical construction
the possibility of the corresponding geometrical objects. This shows that there is no
hope of reducing the existence of geometrical objects entirely to that of physical ones
in the way Lear seeks to do. (Syrianus makes similar objections from a Platonist
standpoint [in Metaph 94.29-100.13]) Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
116 Edward Hussey

(1) The process of separation is apparently a purely mental

operation (Phys 193b34, de An 431 bl 6). Does it reveal new types
of objects (or new aspects of objects, or new truths about them)?
Or does it create, or purport to create, them? Or does it assert
or assume their existence?
(2) Does the process of separation (i) start with the objects of
mathematics already in existence, and then separate (or pur-
port to separate) them somehow from sensibles or from sensi-
ble properties? Or (ii) does it start from sensibles only, and
somehow create or reveal mathematical objects by the process
of separating out them or something else?
(3) If (i) is correct, does separation create, reveal, or purport to
create or reveal yet a third type of object ('separated mathe-
matical object')? Or does it simply involve treating mathemati-
cal objects in a new way?
(4) It seems to be said that separation involves some untruth or
fiction. Is this really said or implied? And, if so, is separation
supposed to introduce fictitious objects? Does it involve the
making and/or use of some false assumptions about fictitious
or non-fictitious objects? How is the 'positing' of 1078al7, 22,
and 24 related to separation?
(5) Is separation necessary to the mathematician, or is it an
optional, convenient extra?
(6) Why does the untruth produced by separation not infect the
science concerned? The remark that the falsehood 'is not in the
premises' seems to show that it enters, at worst, only into the
way the mathematician thinks about the objects, not in what he
asserts about them as a mathematician. Does he 'treat them as
separate' when doing pure mathematics because the link to the
actual world can be disregarded?
As a start to unwinding this tangle, it seems that, at least on the
fundamental question (1) some reasonable certainty is to be had. The
threefold repetition of forms of the verb tithenai in the sense of 'posit',
'assume the existence of, seems to show sufficiently (at least for a
working hypothesis) that what happens in separation is this: one as-
sumes the separated existence of something that does not in fact exist
in separation. The account of 1078al7-28 shows that the arithmetician's
Brought to you by | ULB Bonn
separation consists of assuming the existenceAuthenticated
in separation of, for
| 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 117

example, an indivisible unit which is not a man or any other sensible

substance but/usf a unit. Now we know that this involves untruth: such
things do exist, but not in separation. The untruth is in something that
is assumed, not necessarily in something that is asserted. But it still has
to be explained how the untruth in the assumption does not affect for
the worse the assertions which the assumption helps to enable.
We now have, then, the outlines of an account that is based closely
and economically on 1078al7-28 and can accommodate Phys 193b32-
194al as well. 'Separation' involves making a false assumption: the
assumption that mathematical objects, which do exist but not in sepa-
ration, exist in separation. It starts then with mathematical objects
already given; it does not create them or assume or reveal their exist-
ence. It purports to create or reveal a new type of object, the 'separated
mathematical object', but this is in fact a fiction. This story accounts for
the texts mentioned satisfactorily with a minimum of fuss and extra
hypothesis.20 It must therefore be considered a strong candidate to be
the basis of an eventual interpretation. It answers questions (1) to (4)
above, but leaves (5) and (6) open. And, of course, it still does not
explain the exact nature of mathematical objects and their existence.

4. Representative objects

The position that we have reached is in summary as follows. Antece-

dently to M 3, we know that mathematical objects are some kind of
numbers, points, lines, surfaces, solids, etcv which are not to be identified
either with sensible particulars or with any substances. From M 3 (taken
together with Phys 193a23-194al2) we gather rather little extra, and that

20 This account (already in outline in Hussey (1982), 180-181) agrees with that of Mueller
(1971), 159-161 and Annas (1987) in making separation a further step which takes us
beyond proper mathematical objects. It differs from Mueller and Annas, but agrees
with Lear (1982), 170-3 in taking separation to be something more than 'consider[ing]
the result of abstraction in isolation', and in taking it to involve a fiction or false
assumption. Minor problems remain. (1) On arithmetic here, and 'the man qua
indivisible' see §6 and note 56 below. (2) Why does Aristotle say at 1078a24-25 that
after the separation the arithmetician 'considers whatever is incidental to the man
qua indivisible'? This need not imply that the arithmetician is still thinking about the
man; the reference back to the man may be just a way of reminding us that the objects
of arithmetic can be connected back to the actual world. (3) What is the point (and
the correct text) of 1078a26-28? Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
118 Edward Hussey

mostly negative. In particular, 'considering sensible substances as...'

does not of itself create or reveal mathematical objects. But it may help to
uncover tru ths about them and to show what their properties are; and in
general the possibility of 'considering as...'. must be closely connected
with the existence of mathematical objects. As for 'separation', it does not
create or reveal mathematical objects; on the contrary, it starts with
mathematical objects and separates them from sensibles. The only extra
positive piece of information from M 3 is that mathematical objects exist
potentially in some sense, 'in a matterlike way'.
All these ideas do not by any means add up to a complete under-
standing, and a new idea is needed. In the interpretation to be here
proposed, the objects of mathematics according to Aristotle are taken
to be 'representative objects'.21 These are objects of which the logic has
been examined in recent work of Kit Fine, who calls them 'arbitrary
objects'. Fine has shown that (contrary to what has been suggested by
Frege and others) such items can be painlessly and even elegantly
accommodated within ordinary predicate logic.22 As Fine remarks,
arbitrary objects are also highly natural objects of philosophical study,
as they enable us to mirror more closely within logic the actual reason-
ings of mathematicians and others.
It is characteristic of representative objects that it is true of them by
definition that they possess just those properties which (i) are shared
by all (actual or possible) individual members of the class they repre-
sent, and (ii) are representative properties, i.e., belong to the individuals
qua members of that class.23 For example, it will be true of a repre-
sentative 'case of fever' that it has all and only the properties that people

21 The interpretation of Aristotle's mathematical objects as representative objects is

close to that of Mueller (1971), 164, who takes geometrical objects as 'denuded objects'
which are like ordinary objects, but with many properties 'removed' by abstraction,
and 'with a special mallei'.
22 All references to Fine are to his (1985), which may be consulted for refinements and
further details not mentioned here.
23 Fine's term for these properties is 'generic'. See Fine (1985), 13-15 on the problem of
marking off generic predicates from others and of the possible ambiguity of some
predicates as between a generic and a non-generic reading. The formulation I have
given is not sufficient to overcome or avoid these problems, nor is it intended to be.
They cannot be considered here, but on parallel difficulties in Aristotle, see §5a. For
the reasons for making the formulation metalinguisticBrought
see Fineto(1985),
you by9-10.
| ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Anstotle on Mathematical Objects 119

suffering from fever have qua suffering from fever. If this kind of object
is possible at all, it can be used in mathematics too. In the case of
geometry we start from actual and possible individual physical bodies.
Abstracting, we reach 'the representative sphere', 'the representative
isosceles triangle', etc. A higher degree of abstraction, subsuming both
arithmetic and geometry, is given by the 'pure theory of quantity',
where we consider 'representative quantities' in their relationship to
one another. In arithmetic we start from actual numbered totalities;
then, when we have distinguished units, couples, triples, etc., we may
go on to consider 'the representative unif, 'the representative couple',
'the representative triple', etc. These are objects having all the proper-
ties which belong to every actual particular unit, couple, triple, etc., of
actual individuals, but no properties that do not. They thus have the
required properties to be the numbers 1,2,3.... There is a satisfactorily
close link between these ideas and what Aristotle says about numbers
and counting in Physics IV 12-14.24
This kind of object has sometimes been thought to be open to
devastating logical objections. Fine shows how these may be circum-
vented, without any sacrifice of the original motivation for the idea or
of its intrinsic naturalness. There is a price to be paid, of course. We
have to abandon, for these purposes, not the theorems of classical
prepositional logic, but certain parts of its traditional semantics, nota-
bly the principle of bivalence.25 For the purposes of interpreting Aris-
totle, it is not necessary to explore all the difficulties that might be raised
about representative objects. It is enough initially to make it plausible
that Aristotle could have thought that he had ways of making them
logically watertight against the most immediate difficulties. The three
principal types of objection that Fine (1985), ch. 1 distinguishes are:
(1) ontological objections: do such things really exist,
and if so how?
(2) objections of logical incoherence.
(3) objections of indeterminacy.
To these may be added the further questions:

24 On numbers and arithmetic see further §6 below.

25 How high a price this really is is a matter on which opinions may
Brought differ.
to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
120 Edward Hussey

(4) What is the theoretical appeal of such objects?

Are they really necessary?
If the interpretation of Aristotle's mathematical objects as 'repre-
sentative objects' is correct, we would expect that Aristotle would be
aware of the need to meet at least objections (1) (2) and (4), particularly
as they are closely related to objections that he himself and others raised
against Platonic Forms.26 It turns out, in fact, that we find Aristotle's
(admittedly sketchy) answers to these problems, just where we might
expect, in M 3, as the next section will try to show.

5. Aristotle's mathematical objects as 'representative

objects': Metaphysics M 3 revisited

On the interpretation I am suggesting M 1-3 can be understood as a

clarification of certain doubtful points about mathematical objects,
these being understood from the outset as 'representative objects' in the
sense explained. The obscurity, to us, of M 1-3 in particular is due to
the fact that neither here nor elsewhere does Aristotle state expressly
that he takes mathematical objects to be representative objects.27 More
likely he is just assuming (in the way he often does) that this cardinal
point, in which after all he is more or less in agreement with Plato,28
would be familiar to his audience.
M 3 is devoted to clarifying the following points:
(A) How are representative mathematical objects related to
actual sensibles? In particular, how can they be related to
sensibles and yet not have sense perceptible properties? These
questions correspond to Fine's Objection from incoherence'.
Their answers are given in the discussion of 'considering as...'
(1077bl7-1078al7).

26 I have not found anything to suggest that Aristotle was aware of objections of
indeterminacy.
27 Metaph Z 10-11 can perhaps be read as carrying the implication that there are
representative objects in mathematics (individuals without sensible matter which are
not universale: 1036a2-12,1036b32-1037a5), but it would be unwise to build anything
on these difficult passages.
28 On Forms as representative objects, see §6. Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 121

(B) In what sense can these objects be said to exist at all (since
M l and 2 have shown that they are not substances)? This
corresponds to Fine's Ontological objection'. The answer is
given at 1078a29-31.
(C) how can these objects do what the mathematician requires
and yet escape the objections to Platonic Forms? The answer is
given in the discussion of separation (1078al7-28).
In connection with the answers to these questions, it will also be neces-
sary to tackle another question which is in the background in M 3:
(D) Are mathematical objects, whether in the original or in the
separated state, necessary to mathematics?
I shall treat each of these issues in a separate section.

5a. 'Consideringqua...', abstraction, representative objects, and

the threat of incoherence
Most of M 3 is concerned with the link of mathematical objects to the
real world. As Fine brings out in his chapter 1, representative objects
exist, if any exist, for a wide range of classes of actual things. But
mathematical objects seem to be in a special position in that they seem
to be peculiarly remote from the ordinary world, being (apparently)
unchanging and without sense-perceptible properties.29 Aristotle has
the task of explaining how this can be.
Aristotle's answers depend upon his use of the notion of 'considering
qua...'. We have already seen that considering an actual particular qua,
say, three-dimensional does not of itself bring one to consideration of the
relevant representative object, say, 'solid body'. But, given that there
exists the relevant representative object, then considering actual bodies
qua three-dimensional will help us to see what the properties of the
representative object must be. For 'considering qua...' is just the same
thing as (in Fine's terminology) marking off the generic predicates (in this
example, those that a particular individual has qua three-dimensional)

29 Cf. de An I 1, 403a29-bl6: the mathematician unlike the physicist or craftsman is

concerned with things not separate but with them not as 'affections of matter', but
'in abstraction'; and Metaph N 3 1090a29-30- if mathematical objects were separate,
we wouldn't find mathematical properties m actual bodies.
Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
122 Edward Hussey

from the others (those Fine calls 'classical'). So a knowledge of the

properties of the representative object (though not the object itself) may
be arrived at by the process of 'considering qua...'. Since the problem of
determining which are the 'generic' predicates is central to the whole
enterprise, it is no surprise that Aristotle gives such prominence to the
manoeuvre.
This operation also explains why Aristotle says that representative
objects are reached 'by abstraction'. The operation is one of logical
stripping. Take any particular individual (actual or possible) of the
required kind and strip away every property except those that it has qua
(say) three-dimensional. You will be left with jus t those properties which
the representative three-dimensional object has. Once again, the 'qua-
manoeuvre' is being used, not to construct representative objects, but to
define the scope and the consequences of their representativeness.30
Neither Aristotle's nor Fine's formulations remove all difficulty
from the distinction between the 'generic' or 'representative' properties
and others. It is not always clear just which properties qualify as
'representative'. For example: all physical solids have matter and ex-
tension, weight (or lightness), mobility, color (or transparency), heat or
coldness. Which of these do they have simply qua three-dimensional?
Aristotle's answer seems to be that qua three-dimensional they have
matter and extension but none of the others; hence it is only matter and
extension that are carried over to representative three-dimensional
mathematical objects.31 Aristotle's silence on how we tell 'what has

30 This account of the way in which mathematical objects are related to 'considering
qua ..' is close to that of Mueller (1971). Other evidence for 'considering qua...' or
'lack of attention' as related to abstraction is found at Metaph K 3 and 4, especially
1061a28-b6/ where abstraction is most fully described, and Mem 1,450al-7; see also
Sextus Empiricus M III 57.
It is easy to see how the procedure of arriving at the representative object (by
exclusion of irrelevant properties) came to be called 'abstraction'. In various places
(APo 15, Top 115,107a36-b5, VI3,140a33-bl5, VII2,152blO-16, and VIII2,157bl5-33;
Metaph Z15,1040a29-33) Aristotle describes the modification of a proposed definition
or theorem by 'abstraction', i.e., by substitution of more general predicates for less,
in particular by the omission of one conjunct from a conjunction. (At Top VIII11-30
it seems that premisses are being added/subtracted.) On 'abstraction' in Aristotle
see Cleary (1985), whose conclusions seem generally compatible with the view taken
here.
31 On matter in mathematical objects, see §6. Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 123

what qua what' is paralleled by Fine's admission (Fine (1985), 13-15)

that there may be no general procedure for distinguishing generic
predicates from others.32
A further point is that 'abstraction' may be carried on in two or more
stages. Representative objects may be more or less 'abstract', according
to the generality of the class they represent. Thus, 'magnitude' and
'number' are less general terms than 'quantity7; correspondingly, the
representative objects are less abstract, and the sciences of magnitudes
and of numbers are less general than the all-embracing science of
quantity. There is a hierarchy of abstraction,33 which corresponds to the
hierarchical ordering of various branches of mathematics.34

5b. The potential existence of mathematical objects

On existence, Aristotle affirms robustly in passing (1077b32) that
mathematical objects do exist. Only at the end of the passage (1078a29-
31) does he explain that the Other way' in which they exist is a
matterlike way (hulikös), not 'in actuality' (entelecheiai).35 To say that
something exists 'potentially as matter' is to say at least that it does not

32 At Metaph Z 11,1036a26-b7 Aristotle shows awareness of a related difficulty: how

do we tell 'what are parts of the form and what not?'
33 For the hierarchy, see the references in note 4. The problems and those of 'applied'
or 'mixed' mathematics in Aristotle cannot be tackled here. It has been pointed out
to me that a rigidly ordered hierarchy would imply a fixed order of abstraction. There
is no sign that Aristotle envisaged anything like this. "Mixed' mathematics must in
any case be distinguished from physics though it is used by the physicist. PA I 1,
641bl 1-13 says that physics does not have abstracts among its objects of study. And,
according to Phys II2,193b36-194al, the objects of physical sciences are 'less sepa-
rable' than those of mathematics, where in view of what follows (194al-12), the point
seems to be that in the case of physics or medicine even a fictitiously separated object
would necessarily have properties which refer us in an essential way to the natural
world. Cf §2b above.
34 This account does not absolutely require that the physical world obey the truths of
mathematics exactly. But everything is much simpler if this is so, and in fact there is
no reason why it should not be so. In this regard I agree with Lear (1982), 175-181, as
against, e.g., Mueller (1971) and Annas (1982). But there is more to be said on the way
that this view complements the view of mathematical objects as representative.
35 See note 15. This way of existing is a less obvious kind, hence, the existence of
numbers is less obvious than that of hot and cold (APo Brought
110,76bl7-18).
to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
124 Edward Hussey

have the kind of potentiality which can ever be realized by its actual
existence. Consider the matter of a statue. Even when the statue actu-
ally exists, the matter still exists only potentially as matter. And even
before the statue exists the matter still exists in exactly the same way as
it does later. Whether the statue exists or not makes no difference to the
way in which the matter of the statue exists.
This indifference to what substances with what properties actually
exist is suitable to mathematical objects. The mathematician needs to
be able to use them in just the same way regardless of just what there
is or is not in the actual world at the moment. Whether or not there
actually exists a totality with 23,745 members or a regular icosahedron
must not make any difference to the fact that the number 23,745 and the
regular icosahedron exist. So all geometrical figures and all of the
natural numbers 1,2,3... will always exist. There is no danger that they
will constitute an actual infinite totality and so infringe the require-
ments of Aristotelian finitism. For their usual way of existing is poten-
tial. Hence they do not constitute an actual infinite totality, any more
than the points on a line do.36 So there is just as much, and much the
same, reason to say that representative mathematical objects exist as to
say that matter or the infinite exists. Just as matter never appears 'as
itself but always transiently as the matter of some actual individual,
different from time to time, so a representative object may be 'embod-
ied' in various different sensible individuals at different times, but is
never itself a sensible individual. In both cases to reach the matterlike
aspect requires a process of logical 'stripping' of an actual individual.37
The fact that mathematical objects exist only potentially could also
be used by Aristotle to explain the failure of the principle of bivalence.

36 A line, which can be divided 'anywhere', potentially contains an infinite collection

of points. But this is a potentiality that can be realized only progressively, not all at
once. Corresponding to the actualization of a point as a division of a line, there is the
actualization of a geometrical figure in the shape of an actual physical body (or a
surface or line bounding one), and the actualization of a number in the form of a
totality of actual physical objects. It is dear that for Aristotle only finitely many
figures or numbers can be actualized at any one time (even if, as may possibly be the
case, thinking about them is sufficient to actualize them). On the potentiality of
mathematical objects see Metaph Γ 5,1002al5-bll, which puts forward dialectically
a preliminary case on the question; see also de An III 4-7.
37 On 'shipping' to arrive at matter see Metaph Z 3,1029alO-19.
Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 125

As Fine (1985), 9-11 shows, this failure is a natural consequence of using

representative objects. Aristotle must have been aware, for example,
that it cannot be true to say of a representative triangle either that it is
isosceles or that it is not isosceles (though it must be true to say of it
that it either is or is not isosceles). But failure of bivalence is charac-
teristic of what exists only potentially. These bricks are potentially a
house; they are, then, both potentially a house in the Georgian style,
and potentially a house in some non-Georgian style. But it therefore
cannot be true that the potential house, which is constituted by these
bricks, is Georgian, nor that it is non-Georgian, though it must be true
that it is either Georgian or non-Georgian.38
That Aristotle's mathematical objects must in some ways be 'matter-
like' has long been seen.39 It may nevertheless seem paradoxical. In a
bronze circle it is natural to take the bronze as the matter and the circle
as the form. To say that the representative circle is present in the bronze
circle in a matterlike way is then a confusing complication. But, in the
first place, it is evident that a representative object, or a mathematical
object, can never be the form of anything.40 Nor, secondly, is the shape
of anything a form in the sense of 'substance-like, naturally self-propa-
gating form'.41

38 Compare the apparent failure of bivalence for future contingents discussed in

de Int 9. There might seem to be a difficulty about 'proof by cases', if, e.g., we
successively suppose that a representative triangle is and is not isosceles. On the view
here advocated neither of these hypotheses can be true (or false); yet by combining
the consequences they have in common, we reach a true result. But this is just how
the logic of representative objects works. If we like, we may say that we are really
reducing the problem to two smaller ones: one about a representative isosceles
triangle, one about a representative non-isosceles one.
39 See particularly Mueller (1971), 159-168, to which these remarks are indebted.
40 APo I 13, 79a6-10, which says that 'mathematics is concerned with forms', is an
isolated and puzzling passage.
41 Matter is needed for the matter/form analysis of individuals, an analysis which sheds
light on the metaphysics of existence and predication, and on the logic and the
physics of change. Can we find a correspondingly vital, if less all-pervading, matter-
like role for representative mathematical objects? Evidently the place to look is in the
question of the relation of mathematics to the physical world. The hypothesis which
this interpretation suggests is that the roots of Aristotle's theory of mathematical
objects lie in the need to analyze and explain this relation. See §6toand
Brought younote
by |59 below.
ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
126 Edward Hussey

5c. Separation

The provisional account given of separation in §3 leaves some ques-

tions open. The most urgent need is to connect Aristotle's account of
separation with both: (i) the actual practice of mathematicians; and (ii)
his view (as interpreted above) that the proper objects of mathematics
are non-separated representative objects.
5c(i). Mathematical practice. In the first place, what is it in the actual
practice of mathematicians that Aristotle is describing as a purported
separation of their objects? Aristotle describes the purported sepa-
rables as 'separable/separate in thought', and sometimes he says that
the mathematician 'thinks them as/as if separable/separate'.42 This
must be something more than just that mathematicians take them as
objects of thought. It must mean that they reason about each of their
objects, for example, the representative triangle, in some way as though
it were an actually existent particular triangle. Now such a procedure
corresponds to the practice of 'universal generalization', familiar from
mathematical reasoning. To prove that every triangle has property F, a
mathematician will say 'let ABC be a triangle'. He then demonstrates
(using, of course, only theorems that apply to all triangles) that ABC
has property F. He can then conclude that, since ABC is an arbitrary
triangle, all triangles have property F. Here 'the triangle ABC' purports
to designate some particular triangle, which may even be 'represented'
(inside or outside the mathematician's head) by a representation of an
actual triangle. Particular parts or aspects of it, e.g., 'the angle ABC',
may be referred to in the course of the proof. But there are serious
objections to supposing that ABC is indeed some particular triangle.43

42 'Separable/separate in thought': Phys II2,193b34; 'thinks as (η) separate': Metaph Μ

3,1078al8; 'thinks as if (ως) separable/separate': de An III 7,431b 15-7. The ambiguity
in choristos ('separable' or 'separate') does not seem to have troubled Aristotle. I
assume that he felt no need to keep his practice uniform (which is understandable in
cases where both 'separability' and 'separatedness' are equally fictitious). In case of
need, he could use kech rismenos, which is unambiguously 'separated'. See Annas
(1987), 138-142 and references in her note 23 to the discussions of 'separation' in
Aristotle by Morrison and G. Fine.
43 For this point and on 'universal generalization' generally in relation to 'arbitrary
objects', see Fine (1985), especially 76-77 and ch. 12. So far as I know, Euclid and other
Greek mathematicians never make the procedure explicit. But it is obvious that they
talk about the triangle that is the subject of their proofBrought
as though it were
to you an actual
by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Anstotle on Mathematical Objects 127

Aristotle therefore interprets it as a fiction. Since it has only the prop-

erties of the representative triangle, he understands it as constructed
from the supposition that the representative triangle exists in separa-
tion.44 Notwithstanding the use of this fiction, Aristotle claims that
'nothing false results' or that '[the mathematician] asserts no false-
hood'. This cannot be just because the existence of ABC is not formally
asserted; for the fictitious object is certainly used in the proof. The
existence of ABC must be like a false assumption which (provided the
reasoning obeys certain rules) is eventually 'discharged', as the logi-
cians say. Aristotle may be unable to say with sufficient clarity to satisfy
a modern logician why or how this is so, but he sees dearly enough that
it is so.45
5c(ii). The need for separation. If this account is correct, then it is
natural to ask whether, in Aristotle's view, mathematics needs the
artifice of 'separation'. Aristotle's ideal scheme for a science in the
Posterior Analytics does not leave any place for 'universal generaliza-
tion' in the march of the syllogisms. But it is plausible that the Posterior
Analytics is meant only as an ideal scheme for setting out the results
of a science, not for discovering or proving them.4* If we consider only
the practical business of discovery and proof, it is overwhelmingly
plausible that theorems of any complexity, involving more than, say,
one or two bound variables, simply could not be handled without the
use of such fictions. We might want to ask whether the 'could nof,
here, is a matter of psychology or of logic. But we have already outrun
the available evidence. Aristotle says little in all about 'separation',
and does not positively claim anything more than that the fiction is a

individual; they do so even more obviously when they talk about mathematical
objects as 'given' for the purposes of tasks: On a given straight line to erect a
perpendicular through a given point', etc. See the paper by Taisbak in this volume.
44 Is it something like this that Aristotle has in mind at de An III 8, 432a3-10 when he
says that a phantasma is involved in thinking about abstract objects? He seems to be
thinking of such fictions at Phys IV 1, 208b22-5 when he says that mathematical
objects, though not in place, have a conventional left and right.
45 Compare in general APr I 41, 49b33-50a4 on the logical manoeuvre of ektithesthai,
which seems to involve a fiction in the same way.
46 Cf. Barnes (1975). Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
128 Edward Hussey

convenient and harmless device. On the other hand, he gives no hint

that mathematicians could do without it.47

5d. Are mathematical objects necessary to mathematics?

A further and perhaps deeper question is this: are even unseparated

representative objects, which according to this interpretation are the
proper objects of mathematics, necessary to mathematics? There ap-
pears to be an argument to the contrary. After all, we might say, for the
mathematician to be able to reason about representative objects, what
is needed is just that he can specify what properties necessarily would
belong to any actual object of the relevant kind. The knowledge he
initially needs, then, is of necessary propositions which apply to all
actual or possible objects of the kind in question. Indeed, according to
the Posterior Analytics, such necessary propositions are the staple con-
stituents of a science. But this suggests that mathematical objects are
not necessary to mathematics, and more generally that representative
objects are not necessary to any science. For to know the relevant
necessary propositions is already to have knowledge of the relevant
science. Therefore, it seems, the relevant science can and indeed must
exist before representative objects are introduced. The science therefore
can and must be able to dispense with them.
To see what is correct and what is wrong about this argument, we
must first distinguish between the statement of the results of a science
and the study of a science, whether study involves discovery or just
going over what is known. Aristotle might have believed that the
results of mathematics could ultimately be presented in a form which
involves no introduction of mathematical objects. On this view geome-
try could be presented as the science of the possible spatial structures
of physical bodies, and arithmetic as the science of numerical properties
of sets of physical objects. It would be immensely cumbersome, but
modern logical techniques could perhaps do it.48 But for mathemati-

47 Unless 'the best way of studying anything' (M 3, 1078a21) merely means the most
convenient way.
48 For example in place of '2+2 = 4' there would be some such monstrosity as the
following: for any sets A, B, if there exist distinct individuals x, y, and if A has χ and
y as its only members, and if there exist distinct individuals z, to
Brought w,you
and by
if Β has 2Bonn
| ULB and
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 129

cians to study mathematics, it is mistaken to claim that they need to

know in advance general propositions about mathematical properties
of actual bodies. On the contrary, it is characteristic of mathematics that
it requires next to no experience of the actual world.49 That is precisely
because it has proper objects of which the structure is completely
determined by a handful of simple postulates. These are indeed based
on and discovered from properties of objects in the physical world. For
arithmetic, for example, we need to postulate the existence of a unit,
and the possibility of constructing larger totalities from totalities by the
addition of one extra member; and these truths are originally derived
from the physical world.50 But the claim that the study of mathematics
must be able to dispense with mathematical objects has not been made
out. On Aristotle's view 'representative objects' are the proper objects
of mathematics, which explains why its formulations are so much
clearer and more convenient when put in terms of them. For the same
reason, these formulations, together with separation, exhibit the way
in which mathematicians actually think.51

6. Further considerations

In considering Aristotle's theory of mathematical objects this essay has

hardly looked beyond Metaphysics M 1-3, and the related passage in

w as its only members, and if A and B are disjoint, then x, y, z, w are pairwise distinct,
and there exists a set C which is the set-theoretic union of A and B and which has x,
y, z, w as its only members. (For the sake of argument I here leave out of account the
serious problems about how Aristotle could be supposed to have known the tech-
niques for making translations of this kind and whether and how they could be recast
to preserve syllogistic form.)
49 Cf. NE VI 8, 1142all-20 where Aristotle says that young people can be good
mathematicians because mathematics requires no experience of the world.
50 According to Aristotle, the potential infinity of number is derived from the fact that
(physical) magnitudes are infinitely divisible (Phys III 7,207bl).
51 Compare Fine (1985), ch. 12, 'Accord with ordinary reasoning'. Note that my discus-
sion, conducted on the basis of M 3 alone, gives no real answer to the questions urged
at N 2,1090a2-15: what are mathematical objects needed for, and what are they causes
of? For a suggestion of a possibly more satisfactory Aristotelian response, see below,
§6 on mathematical objects as material causes in the physical
Broughtworld.
to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
130 Edward Hussey

Physics II 2. There is other evidence, most notably (a) those places in

Metaphysics Z and H which mention the 'intelligible matter' of mathe-
matical objects, and (b) those dealing with numbers, counting, and
measuring in Physics IV and Metaphysics K. In default of a fuller account
something at least should be said to suggest that the interpretation can
meet these further tests.52
(a) On the account which has been offered, mathematical objects,
being representative objects, ought arguably to have some kind of
matter. For every particular object that exemplifies, say, 'circle' has
some matter; otherwise, how could it be circular? And since the matter
is essential to its being circular or having any shape at all, it seems that
it has that matter qua circular, not just qua actual physical substance.53
So the representative circle must also, by the principle of generic
abstraction, have some 'representative' matter. Likewise, the repre-
sentative triad, i.e., the number 3, must have some 'representative
matter'. Every actual triad is a triad of distinct actual individuals. And,
again, that property seems to be essential to its being a triad. So then
also must the representative triad be a triad of distinct 'representative'
individuals, which are its 'matter'. A complete account of this question
would have, among other things, to look more closely at the whole
discussion of definition in Metaphysics Z 10 and 11, which treats ques-
tions which are manifestly closely related to the notion of 'repre-
sentative' objects. But perhaps enough has been said to indicate a way
of accounting for the statements about 'intelligible matter'.
(b) It is sometimes suggested that there is for Aristotle a potentially
embarrassing difference between geometry and arithmetic.54 It is true

52 Other passages on mathematical objects and abstraction which would need consid-
eration in a fuller treatment are Phys IV 2,209b6-ll and Metaph Z 3,1029alO-19 on
the parallel process of 'stripping' objects to get down to the matter; and various
discussions of knowing and thinking about mathematical objects, e.g., APo I 18,
81b2-5 (on abstracts and induction), de An III 8,432a3-10 (the need for a phantasma in
thinking about abstracbs), and Phys ΠΙ 4, 203b22-25 (the claim that mathematical
objects 'do not give out in thoughf; see also III 7,207blO and 8,208al4-22)
53 This point seems to be supported by de An III 4,429bl8-19 on the need for 'continuous
stuff in geometrical abstracts.
54 For example, by Lear (1982), 183-184, who holds (consistently with his view that M 3
is intended as a complete account) that Aristotle was 'not especially
Brought to yousensitive'
by | ULBtoBonn
the
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 131

that geometrical objects correspond to features of single physical objects,

whereas numbers do not; but the resulting difference between arithme-
ticand geometry need not be all that profound for Aristotle. The business
of 'abstraction' and 'separation' can still proceed analogously. In arith-
metic, as Aristotle seems to have realized (at least in Physics IV11 -14), the
numbers correspond, not to features of single physical objects but to
features of collections of objects.55 It is true that the point is not made in M
3, but then this chapter hardly mentions arithmetic at all — under-
standably, since in a brief outline it is clearly much simpler to take
examples from geometry only. The correspondence involved is no more
difficult or less natural in arithmetic than in geometry. As we should
expect, the remarks on these matters in Physics IV show the repre-
sentative objects, i.e., the 'numbers by which we counf, as exactly exem-
plified in the numbers of different actual totalities, 'the numbers
counted'. Again, there is no sign that the representative objects can
somehow be arrived at simply by 'considering as...'. or by abstraction.
For in order to count a totality, we already have to have available the
'number by which we counf, i.e., the representative object. This priority,
it is true, need not hold for the smallest numbers, which could be, and
perhaps are, abstracted from actual collections. But for the number
system as a whole it is indispensable.56

difference between his philosophies of arithmetic and of geometry, and that he was
unable to give a successful account of arithmetic.
55 Cf. Hussey (1982), 160-164.
56 There is certainly a difficulty with the unit. M 3, 1078a21-25 says that the unit
corresponds to the arithmetician's 'considering [some physical object, say, a man] as
indivisible'. Obviously the ordinary sense of 'indivisible' is not in play here, since a
man is not in the ordinary sense indivisible. The required sense is just that which
appears at Metaph 11,1052a29-bl as 'indivisible in number'. (Phys ΠΙ7,207b5-10 and
IV12,220b20-22 do not help here) Another attempt, at Metaph Δ 6,1016b3-6, suggests
that 'indivisible' is to be filled out as 'indivisible qua man'. This suggestion is
preferable since it at least gestures at relevant truths, namely that if it is men you are
counting, you cannot count one man as more than one man; and if you cut a man in
two you do not get two men. But as it stands it is still obscure. We have to admit, I
think, that Aristotle here fails to follow through the insights of Phys IV. What he ought
to have said is that the man is considered as a collection of human beings with just
one member. It is the collection, not the man, which is indivisible in the required
sense.
There is no space here to deal fully with the problems of Aristotle's
Brought to you arithmetic. A
by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
132 Edward Hussey

A more general question about this interpretation is whether any-

thing like 'representative objects' appear elsewhere in Aristotle. It is
natural to think first of all of Platonic Forms. These too are 'repre-
sentative'. However, although Aristotle is very critical of the Forms, he
does not criticize the features which they share with representative
objects. His objections are mostly concentrated on the 'separatedness'
of Forms and on their role as predicates, precisely the features which
differentiate them from mathematical objects. This fact may support the
view that Aristotle did not see any fatal objections to representative
objects in themselves.57
A less obvious parallel exists with Aristotle's treatment of time.
Aristotle's tendency to mathematize and spatialize temporal duration
is evident in Physics IV 11-14. His program for dealing with the prop-
erties of time is parallel to his emphasis on the close connection of
mathematical properties with the actual world. But, as with mathemat-
ics, there is a matterlike residue of structure which, though exemplified
only in the actual world, cannot be wholly reduced to it either ontologi-
cally or epistemologically. To this structure belong that interesting
entity, the 'permanent now', which is analogous to the 'ubiquitous
point' in geometry, and the relation of simultaneity.58 This parallel can
perhaps be extended to other 'mathematically amenable' features of the
physical world, in particular to the continuity of physical quantities,
and the proportionalities holding between them.59

starting-point is given by the very full and useful discussion of Mignucci (1987); see
also Barnes (1985).
57 On Aristotle's criticism of Forms and 'separation', see particularly G. Fine (1982) and
(1984). Aristotle also criticizes Forms as not fitting into the Aristotelian scheme of
'causes' (see G. Fine (1987)), but representative objects are like a material cause (see
below).
58 The parallel with the treatment of time (also suggested by Metaph B 5,1002b5-8) is
noticed by Annas (1976), 28, but her interpretation of it differs from mine. On the
'ubiquitous point' and the 'permanent now' see Hussey (1982), xlii-xlv and 152-157.
59 This suggestion obviously needs much elaboration to be even worth consideration.
For a study of the way in which mathematical considerations dominate some parts
of Aristotle's conception of the physical world, see Hussey (1991).
Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Aristotle on Mathematical Objects 133

7. Conclusion

The interpretation offered here aims to show how Aristotle proposed

to resolve his problem about mathematical objects, a problem which
Mueller (1971), 115 has described as a conflict between a Platonist
epistemology of mathematics and a non-Platonist ontology. Aristotle
accepts that the mathematician treats (in Mueller's words) Objects
which are different from all sensible things, perfectly fulfill given
conditions, and are apprehensible by pure thought7, but 'is not willing
to construe mathematical objects as merely mental constructions de-
pendent on human thought for their existence'.60 Representative objects
serve as a kind of 'matter7, analogous to the more familiar kind, in the
physical world. Mathematics may be seen as an exploration of the
physical possibilities inherent in that matter.61
Aristotle on mathematics is altogether more sensible, more down-
to-earth, and less liable to be carried away on logical hobby horses, than
Plato or, indeed, some of his own subsequent interpreters. And when
he writes about what mathematicians do, he writes about what they
really do do, not what he thinks they ought to do.

60 In Hussey (1982), 182-184, moved by a supposed (but non-existent; see §5) threat to
Aristotle's finitism, I mistakenly suggested that unseparated mathematical objects
must be dependent on thought for their existence. Even Phys IV 14,223a21-29 need
not be read as supporting this view since it is concerned primarily with the count-
ability of actual totalities; see Mignucci (1987), 182-187.
61 It is true that at Metaph N 5,1092b23-25 Aristotle says that number is not a cause as
matter is, but this seems to be part of a criticism of a specifically
Brought toPythagorean view.
you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM
 Brought to you by | ULB Bonn
Authenticated | 131.220.250.11
Download Date | 9/22/13 9:16 AM