Sie sind auf Seite 1von 21

On the "Intermedktes"

by Julia Annas (St. Hugh's College, Oxford)


Is Aristotle right when he says that Plato believed in a class of
entities which are "intermediate" between physical objects and
Forms? It may seem unrewarding to ask this question again.
Scholars divide into two groups over the answer, and it may well
seem impossible by now for either side to make any point that
will convince the other. If the issues are well understood, and the
lines of the controversy drawn already, can it be worthwhile to
reconsider the problem ?
In this paper I shall try to reconsider this issue from a somewhat
different angle. It may be that, while the traditional question cannot
be conclusively answered s so stated, the problem can nevertheless
be fruitfully investigated by first considering a related question,
the answer to which has implications for the way in which the traditional question should be answered.
The outlines of the traditional problem are clear enough. Aristotle
often says plainly that Plato believed in a third class of entities,
which are identical neither with Forms nor with physical objects.
But although there are passages in the dialogues where similar
ideas seem to be indicated, Plato nowhere accepts this important
and strnge doctrine in a straightforward way. There are essentially
two ways in which this problem has l^een resolved. Some scholars
maintain that on the basis of Aristotle's evidence we should accept
that the passages in the dialogues show that Plato did accept intermediates1. Others hold not only that these passages do not explicitly
contain the idea, but that they can be better interpreted without
1

A selective list includes Zeller, Die Philosophie der Griechen, 2, l, 790ff.;


Robin, La thoorie platonicienne des idees et des nombres, p. 199221, 260
266; Adam, Appendix l to book VII of bis edition of the Republic; Hardie,
A Study in Plato, eh. VI; G. Martin, "Platons Lehre von der Zahl und ihre
Darstellung durch Aristoteles", Zeitschrift fr philosophische Forschung,
1953; K. Gaiser, Platons ungeschriebene Lehre, p. 89106; Crombie, An
Examination of Plato's Doctrines, vol. ii ch 3, VII A, especially p. 4437.
The most explicit and authoritative Statement of this view is to be found in
A. Wedberg, Plato's Philosophy of Mathematics, chs. 4 and 5 and Appendices
AD.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

147

it. So Aristotle's testimony nmst be wrong; it must be either misleading or confused2.


There is an obvious compromise view, namely that Aristotle
is reporting a subsequent development of ideas expressed in the
dialogues3. On close inspection, however, this does not remove
the difficulty. What Aristotle is taken to be reporting is not a real
development of doctrine, but merely the recognition of a view
beneath the surface of the dialogues. According to the compromise
view, Plato did not change his mind; he merely recognised (somewhat belatedly) a commitment implicitly contained in what he
explicitly held in the dialogues. So the compromise resolves into
the first alternative: Aristotle is reporting a theory of Plato's
which can be found in implicit form in the dialogues, even if it was
not given recognition until later.
These views are clearly addressed to the issue of Plato's belief
in intermediates. I propose to concentrate on the issue of Plato's
grounds for this belief. If we look carefully at Aristotle's references
to intermediates we find that his descriptions of them, and his
frequent criticisms of them, assume that Plato is arguing for them
in a certain way. Aristotle consistently treats the intermediates
s the outcome of a certain line of thought, When we turn to the
dialogues, however, we find that Plato nowhere argues for intermediates in this way. App'arent references to intermediates turn up
in a variety of different contexts, but Plato nowhere explicitly uses
the argument for them which Aristotle treats s Standard. This
suggests that Aristotle is not talking about the intermediates s
these are argued for in the dialogues. So the Aristotelian evidence
need not, and should not, be expected either to support or to conflict with what is said about intermediates in the dialogues. It is
concerned with a different argument. for intermediates. This has
2

Cook Wilson, 'On the Platonist Doctrine of the ",


Classical Review 1904 4 and 9; Shorey, "Ideas and Numbers Again",
Classical Ph ology 1927; Murphy, The Interpretation of Plato's Republic,
p. 1668; Cherniss, The Riddle of the Early Academy, p. 758. Cherniss
rejects s nnreliable all of Aristotle's testimony about Platonic doctrines that
cannot be found in the dialogues.
3
See ROSS* note on Meiaphysics 987 b 14, and p. liiiIvii of his Introduction
to the edition. People who hold the first view often waver towards this view:
Crombie says on p. 446 that "even if Plato feit compelled to postulate
entities to be denoted by number substantives, he did not make much of
them, and would perhaps have been glad to dispense with them if he
could".

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

14-8

Julia A n n a s

obvious imp cations for thc traditional problem and the ways it
has bccn answcrcd.
Aristotle's Statements. What are the intermedi res P
Twice4 Aristotle attributes to Plato by name the view that there
is a third class of entities besides Forms and physical objects. The
first of these passages occurs in Aristotle's account of Plato's
philosophy in Metaphysics A 6, and in the second Plato is contrasted with Speusippus. In both passages the intermediates are
treated s a characteristic part of Plato's doctrine: So it is reasonable
to assume that Aristotle has Plato in mind in his numerous other
references also. Unfortunately many of these are, while suggestive,
not very helpful. Aristotle mentions Plato's "Separation" of the
intermediates from Forms5, and presents the distinction between
Forms and intermediates s one of two kinds of substance6. Aristotle's own terminology and ways of describing Plato's doctrine do
not help us to recover Plato's rationale for the idea. They teil us
that Aristotle took Forms and intermediates to be two distinct
kinds of entity, but this is obvious and not very helpful.
Aristotle frequently7 refers to these entities s "the intermediates"
( ), but he also calls them "the mathematicals" (
). At 992 b 16 he says that the intermediates are mathematicals8 and at 997 b 2 that "they" (the Platonists) say that the
mathematical sciences are concerned with the intermediates. But
while Aristotle always assumes that for Plato the intermediates
are objects of mathematics, this does not yet provide anything
like an argument to show why mathematical sciences, and not
others, should have objects that are neither Forms nor physical
objects.
* 987 b 14^18; 1028 b 1821.
1086 a 12.
6
1028 b 1821; 1069 a 346; 1076 a 1921.
7
987 b 16; b 29; 991 b 19; 992 b 16; 995 b 17; 997 b 1; b 13; 998 a 7; 1002 b 13;
b 21; 1059 b 6; 1077 a 11; 1090 b 35.
8
ROSS in the Oxford translation renders this s "the intermediates (for those
are the objects of mathematics)". This implies that the two sorts of entity
are to be identified, but the Greek does not demand this Interpretation. What
the passage stresses is that the intermediates are special in being objects of the
mathematical sciences, not in that the latter study only intermediates. There
are many passages in the dialogues that are at least compatible with mathematicians studying Forms.
5

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

149

The answer is provided indirectly by three passages which throw


a diiferent light on the intermediates by calling them "many but
eternaT'9. \Vhat this means comes out in the most explicit of these
passages, 987 b 1418:
Further, besides sensible things and Forms he [Plato] says that there are
the objects of mathematics, which occupy an intermediate positibn, differing
from sensible things in being eternal and unchangeable, from Forms in that
there are many alike, while the Form itself is in each case unique10.
5s
, , '
* , .

Intermediates are eternal and unchangeable, like Forms, but they


are unlike Forms in not being unique. While each Form is unique
of its kind, there can be many instances of each type of intermediate, just s there can be many instances of each type of physical
object. So intermediates share some of their features with Forms,
and others with physical objects. This is the idea encapsulated in
the metaphor of betweenness.
These passages give a sense to the idea of "intermediates".
But why should such intermediates be objects of mathematics?
Aristotle does not teil us, and we have to fill in this step in the
argument for ourselves.
A very simple line of thought suggests itself. It is parallel for
arithmetic and geometry. I shall present it for arithmetic first.
Plato begins from the conviction that there are two kinds of
number. When we count, we are counting either numbers of things
or pure numbers11. Each pure number is unique of its kind, This
is partly because it is easy to think of the natural numbers s a
series of unique objects12, partly because for Plato numbers are
0 987 b 1418; 991 a 4; 1028 b 1821.
I use the Oxford translation.
11
The passages in the dialogues where this distinction is made will be discussed
below. Aristotle says at 1090 b 356 that the Platonists put mathematical
number between "Form number and sensible number" (
Kai ), Some Plato's ideas about number may be influenced by
this way of talking s though a number of perceptible objects were a
"perceptible number". Martin discusses the concept and relates it to uses of
"angewandte Zahl" in German,
12
This is not just a pre-philospphical Intuition; it can form the basis of
sophisticated platonism, Cf. Frege, The Foundalions of Arithmetic (tr. J. L.
Austin), p. 49e: "When we speak of 'the number one', we indicate by means
of the definite article a definitc and unique object of scientific study. There
are not divers numbers one, but only one/' It is true that this way of
10

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

150

Julia Annas

Forms, and each Form is unique of its kind. Aristotle says twice13
that for Plato numbers were Forms; Plato never says so explicitly,
but it is clear from Phaedo 101 and 103105 tha he treats numberterms s though they had the same logic s other general terms
which he takes to stand for Forms14.
There is, however, a special problem with numbers which does
not arise in the case of the other Forms. In everyday life we talk
of adding numbers, and of performing other repeatable operations
on them. This is clear from the simplest arithmetical Statement,
like "24-2=4'', or indeed s soon s we begin to talk about "twice
two". What we say seems to have a clear sense and a satisfact iy
use. But what are we talking about when we use such a Statement ?
Not groups of physical objects, for the truth of such Statements
does not depend on the observable behaviour of physical groups.
But surely not the Form number Two either, for this is unique,
and it can make no sense to talk of adding it to itself. Similarly
it makes no sense to talk of Twoness being part of Threeness, or
of getting Twoness by adding two Onenesses. Surely, however,
"2+2=4" must be about numbers? So it appears that it must be
about a third kind of number, distinct from numbered groups and
also distinct from Forms. This will be mathematical number, the
number we do mathematics with.
Exactly the same argument goes through for geometry. Theorems
talk about two circles intersecting, Again, we cannot be talking
about actual drawn circles, for we do not prove the theorem by
measuring them. But we cannot be talking about the Form of Circle
either; this is unique and connot sensibly be said to intersect another
circle. (Which other circle ?) So again it appears that the theorems
must be about intermediate circles. T*he same obviously applies
to geometrical objects generally.
thinking of numbers is not suggested by the way, e. g. Euclid treats arithmetic
in books VIIIX. But Euclid does not treat of specific numbers; see Heath's
edition, vol ii, p. 297.
18
1090 a 46; an argument from the (in Alexander's commentary on
Metaphysics 990 b 17, p. 87. 911), where Aristotle says.
?
.
14
In the Phaedo the true "cause" of a pair of things' being two is "participating
in " where $ seems to be both "twoness" and "the number two".
This is exactly analogous to the way in which a single thing is, e. g. beautiful
by participating in the Form of Beauty. This passage teils against the claim,
based on Hippias Major 301 d302 b, that Plato was aware that numerical
properties belong not to things but to sets of things.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

151

Intermediates are thus required for the mathematical sciences


in order to preserve the apparent .sense of mathematical Statements
without having to admit that these Statements are about either
Forms or physical objects. It is perhaps not surprising that Plato
should have accepted intermediates, given bis respect for mathematics s the model of an . Plato always insists with
emphasis that mathematics does not refer directly to physical
objects; so once it was pointed out that a Statement like
"2+2=4" cannot refer to Forms because of the trouble with the
unique Form Two, intermediates may well have seemed the only
solution. It is hard to see how he could have analysed mathematical Statements in any other way, given the absence of set
theory for the analysis of number Statements, and the morecomplex
devices necessary for the analysis of geometrical Statements.
This sketch of the line of thought leading to intermediates is,
of course, not new. It goes back to Cook Wilson15, and has been
frequently repeated since. It should still be given emphasis here,
however, because it is the only line of thought leading to intermediates which is suggested by Aristotle's evidence. According
to Aristotle the intermediates are "many but eternal"16. The line
of thought just sketched would fit Aristotle's evidence very well. The
intermediates are the answer to a specific problem which arises
only with the mathematical Forms. I shall refer to this s "the
Uniqueness Problem". A Form has to be unique of its kind, whereas
mathematical Statements seem to refer to a plurality of entities,
and these cannot be identified either with Forms or with physical
objects. Hence intermediates are posited to be the objects of such
Statements.
It is important to remember that this is the sole line of argument
suggested by Aristotle's references to the intermediates. Elsewhere
15
16

In the article cited in n. 2.


An argument from the (Alexander 87. 3) seems to call into question
the claim that it is essential to the intermediates that they are not specifically
unique: "Besides, there will be a remarkable number of twos, if there are, all
different, the original Two, the indefinite two, the mathemtaical two we use
in counting (not identical with any of them) and then besides these the two in
countable and perceptible objects." $,
, , fj ,
, In $ $ . This
does not imply, however, that there is only one mathematical two, for if it did
then analogously the passage would imply that there is only one numerable
group of two physical objects. The Singular here is a way of referring collectively to the kind of number in question.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

152

J u l i a Annas

in Aristotle we find Information about the sort of mathematical


objects that the intermediates were taken to be. This information
has been usefully collected by Wedberg17. We learn, for example,
that the Mathematical Numbers are made up of ideal, undifferentiated units. But whatever we learn from Aristotle about the nature
of the intermediates and Plato's beliefs about them, there is only
one argument for their existence to be extracted from the
Aristotelian evidence. This is not to say, of course, that Aristotle
actually sets out the Uniqueness Problem anyvvhere, still less that
he asserts that this was Plato's argument for intermediates. What
I have tried to show is that it is a very reasonable conjecture that
Aristotle takes Plato's argument for intermediates to be an attempted solution to the Uniqueness Problem.
This conjecture is considerably strengthened when we look at
the grounds on which Aristotle attacks the idea that there are intermediates.
Aristotle's Criticisms
Aristotle frequently criticises the theory of intermediates. His
objections fall into three types.
The first type of objection comes to little more than the expression
of Aristotle's anti-platonism. Aristotle is consistently opposed to
platonism in the philosophy of mathematics (s distinct from
Platonism)18. Large parts of book M of the Metaphysics are devoted
to refuting platonism in the philosophy of mathematics, and to
establishing his own alternative. Since he has strong objections
anyway to Plato's theory that mathematical objects are unchangeable and eternal.real objects over and above perceptible
ones (Metaphysics 1076 a 1012), he has even stronger objections
to a theory that not only recognizes these but adds more. There
is an example of this at 991 b 2731, where he criticizes Plato's
theory on the ground that is needs these things. How do they
exist, where do they derive from, and why must they be intermediate between Forms and things ? As in all his general criticisms
of Plato's ideas about numbers, he does not really expect an answer
i? pp. 612, 65, 8491, 118120.
18
Platonism (with a small "p"), the idea that entities such s .numbers exist
independently of our thinking, is an appropriate theory of mathematics for
the inventor of the theory of Forms, but it should be shown independently of
that theory that Plato was a platonist (s can be done, I believe). For
platonism in this sense, cf. Quine, conclusion of "On What there Is", and
"Logic and the Reification of Universals", in From a Logical Point of View.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

153

which will satisfy him. He takes Plato's claims to be meaningless


rather than amenable to minor improvements. There are no facts
that Plato could point to that would impress Aristotle; his objections are to the type of theory accepted by Plato. At
1090 b 31ff. Aristotle complains that the Platonists fail completely
to give any account of the existence and Status of intermediate number. He means, not that they said nothing on the subject, but that
they faed to make their claims intelligible. Aristotle finds it
prima fade unintelligible to claim that abstract objects like numbers exists, in the way claimed by platonists like Plato and Frege.
He is therefore liable to be especially outraged by the claim that
there are mathematical numbers s well s Form numbers. This
basic clash of attitudes comes out most clearly at 1090 a 215,
where Aristotle is criticising Speusippus. He claims to find it understandable that someone who believes in Forms should believe in
numbers too, since numbers are thought of s a kind of Form, and
there are at least independent reasons for believing in Forms.
But Speusippus rejects Forms, yet still believes in numbers. This
is beyond Aristotle's comprehension. For him it is clear that numbers can only be said to exist in a qualified and derivative sense of
"exist" (1077 b 1517).
These arguments are not forceful. They simply confront one
general position with another, without giving us any means of
evaluating the differences between them. They do, however, make
clear the nature of Aristotle's main objection to the intermediates.
He regards them s implausible entities posited for insufficient
reasons and, further, a gratuitous addition to Plato's ontology.
Aristotle believes (s does Frege) that a reasonable analysis of
number should answer to our ordinary intuitions. He finds a theory
with three tiers of number, all of different logical type, counterintuitive, and so objectionable. In this he is surely right.
The second type of objections is internal, in the sense that it
shows that the theory of intermediates raises problems on the
Platonists' own assumptions. Twice19 Aristotle asks what the
elements or first principles of the intermediate numbers are to be.
Are they the same s those l the Form numbers ? In that case why
are the intermediate numbers not to be identified with the Form
number, since the Platonists suggests no means of differentiating
them ? Are they different ? In that case there is not only an extra
19

991 b 2731; 1090 b 321091 a 5. See Robin 106 and notes.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

154

Julia Annas

type of number but an extra type of first principle. This makes


rather a lot of first principles, s Aristotle remarks dryly. He might
have added that Plato's three tiers of numbers are even more
objectionable if they lack any connexion and have quite different
first principles.
These arguments suggest that for the Platonists the intermediate numbers had no independent rationale, but were introduced simply to solve a problem with the Form numbers, the
important question of their first principles being simply neglected.
Aristotle is objecting that this way of introduqing them does not
sufficiently explain how they can be an entirely different type of
number from the Form numbers. While this objections is presented
only for the intermediate numbers, there is an obvious parallel
for the geometrical intermediates, likewise unprovided with any
proper first principles.
It is the third type of objection that brings out most clearly the
way Aristotle sees the intermediates s being Plato's solution to
the Uniqueness Problem. In one passage20 Aristotle criticises the
Platonists on the grounds that they accept intermediates only for
mathematics, whereas in consistency they should do so for other
sciences s well. Plato would readily accept that if any branch of
knowledge is to have any claim to be an , its Statements
cannot be descriptions of the actual world; if true, they are
true independently of the vagaries of any particular physical
objects. But are the objects of the sciences Forms? (Some
Platonists argued that this must be so; their arguments are preserved in Alexander's commentary on the Metaphysics, 79. 3 to
80. 6.) If so, we are faced by the Uniqueness Problem. Biology cannot study the unique Form of Animal; it must study many (ideal)
animals. And some sciences, like astronomy, make essential use
of mathematics, and so inherit the Uniqueness Problem in its
original mathematical form. So it appears that we ought to have
intermediates in the non-mathematical sciences also.
This conclusion, however, leads to absurdities, which Aristotle
points out. Physics will study neither the actual universe nor the
unique Form of the Universe, but a third, separate universe.
Biology will study neither animals nor the Form of animal but
intermediate animals, distinct from both. Astronomy will study
neither the actual heavens nor the Form of the heaven, but a third,
20

997 b 1234. See Robin 105 and notes.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

155

intermediate heaven and to be a proper subject for astronomy


this will have to move. There will be intermediate cases of health,
neither particular instances of health nor the Form of health. And,
since the Platonists characterize the nature of a science by the
nature of its objects, there will be an intermediate science of
medicine, presumably to deal with them.
A passage at 1059 b 2 ff. makes it clear that the special force of
this objection is that it shows that while Plato only recognised
intermediates in mathematics, the argument establishing them
should have wider application:
It is hard to say, even if one supposes them to exist, why in the world the
same is not true of the other things of which there are Forms, s of the objects
of mathematics. I mean that these thinkers place the objects of mathematics
between the Forms and perceptible things, s a kind of third set of things apart
both from the Forms and from the things in this world; but there is not a third
man or horse besides the ideal and the individuals.

This argument brings out well the fact that the theory of intermediates has to assume that the mathematical sciences have a
special and priveged status in having intermediates. Aristotle
complains that no good grounds are given for this assumption.
Plato might reply that in Republic, books 6 and 7, the mathematical
sciences are distinguished from non-mathematical studies precisely
on the grounds that the former but not the latter "raise" the mind
from study of the actual physical world to study of an ideal subject
matter. It is not clear, however, that Plato's grounds here are sufficiently precisely stated for him to be able to defend the positing of
intermediate objects, which are ideal objects but distinct from
Forms. Aristotle's objection is limited and exact: intermediates
should be posited elsewhere s well s in mathematics, since the Uniqueness Problem is not limited to mathematics. Plato has no (recorded) answer to this precise Charge.
All three types of criticism are directed against the intermediates
in a clear and exact way. Aristotle is not protesting about the
nature of the entities introduced by Plato to serve s intermediates (he does that too,%but elsewhere); he is objecting to the
intermediate status of a class of entities distinct from both Forms
on the one hand and physical objects on the other. He insists that
these entities are unnecessary and unintuitive, and also that they
are not properly integrated into Plato's theory of number. He
further points out that the argument establishing them proves
far too much. Given unrestricted application it would result in the
11 Arcb. Gesch. Philosophie Bd. 57

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

156

Julia Annas

absurdity of having intermediates for every type of science. And


thc restriction to the mathematical sciences is not explicitly
justified.
We cannot avoid the Impression that Aristotle regards the interraediates solely s a solution to the Uniqueness Problem. The
"many but eternal" intermediates are an answer to one definite
problem arising for the mathematical Forms. Aristotle criticises
both the results of this answer in the field of mathematics, and the
illogicality of restricting it to that field. He sees it s a compromise
which is neither successful nor attractive.
So both Aristotle's criticisms of the intermediates and his
reports of them suggest that he sees them purely s an answer
to the Uniqueness Problem. Unless these references are all misleading and the criticisms all misguided21, it is reasonable to suppose
that Aristotle always takes Plato to be arguing for intermediates
in one specific way, s a solution to one definite problem. Only so
do Aristotle's criticisms have any force.
When we turn to Plato's dialogues, however, we find a different
story.
Passages in Plato's dialogues

The passages which have been taken to imply that Plato puts
forward a theory of intermediates are scattered unsystematically
through the dialogues. Wedberg22 has made a collection of the
relevant passages, and I shall in the main follow him23. The passages,
however, are of very unequal weight. I shall first deal briefly with
the inconclusive or incidental passages, and then at more length
with three important passages.
Inconclusive passages

At Euthyaemus 290 bd Plato insists that mathematicians


genuinely make discoveries; they discover , and are classed
with people like hunters who have tracked down something which
21

1 find this very unlikely, but the matter cannot be settled by the general
likelihood that Aristotle is reliable s a critic (which is controversial), but
only by detailed examination of arguments. In this case it seems improbable
that so many detailed arguments of different types should all be directed
against a single misconception.
22
In his Appendices B and D.
23
I shall not consider Seventh Letter 342 b343 b, in view of its disputed authenticity.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

157

they bring in and hand over. Plato's very realist conception of


mathematics is apparent here24: the mathematician's concern is
with independently existing realities, not 25. The
passage, however, is quite unspecific about the nature of these
realities. can quite well be interpreted s truths, rather
than objects; so this passage cannot be used to decide whether
Plato intends the mathematician's subject-matter to be Forms or
intermediates.
Theaetetus 198 ad is a passage where Plato is discussing the
problem of false belief, and he distinguishes the thought that 11
things are 12 things from the thought that 11 is 12. Here Plato
is clearly distinguishing numbered groups from pure numbers. It
is not so clear that, s Wedberg argues. these numbers must be
intermediates because they figure in computations, wh e the Form
numbers do not. Plato is concerned here with a straightforward
distinction between two types of number, which would onlybecome
confused if we had to bear in mind a subdivision of pure numbers
into Form numbers and intermediates. No such subdivision is
required by the course of the argument, and the natural assumption
is that Plato has not yet thought out the considerations that lead
to the making of such a distinction.
Phaedo 101 bd seems to be distinguishing numbered groups
from pure numbers. Socrates insists that the only adequate "cause"
of a pair of things being two is "participating in Two"26. "Adding
one thing t o another" will not do, because the "opposite" of this,
"dividing one thing into two", would provide an equally good explanation of the same state of affairs. Plato's use of "participating"
makes it clear that he is thinking of numbers s Forms, and the
rest of the passage shows that the demand for adequate "causes"
54

Wedberg seems wrong in claiming (p. 93) that "the philosophy of mathematics
underlying the short Statement in the Euthydemus is essentially that of the
RepuUic, and the Statement can be fully understood only in the Hght of the
later dialogue". In the Euthydemus what is wrong with the mathematicians
is merely that they do not understand the f ll signin'cance of their own
discoveries, and should "hand .these over" to the dialecticians. The Republic
comes to insist that there is something inherently defective with the procedures of mathematics even when successful.
25
Wedberg (p. 93) points out that this word could have any of several distinct
meanings in Greek nrathematical tenninology, and perhaps means "prooi"
rather than "diagram"; but the contrast with the "real objects" of matheraatics is clear in either case.
26
101 b 9c 9.

n*

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

158

J u l i a Annas

or explanations is to bc satisfied by Forms. What are the Forms


contrasted with? Wedberg argues that Plato raay have in mind
intermediates rather than numbered groups, that he is thinking
of "1+1" rather than the physical juxtaposition of two objects.
But Wedberg himself raises the difficulty, on this view, with dividing 1. Wedberg's conslusion is that
Socrates' criticism s a whole seems to be directed against any numbers that
are sets, inrespective of whether they are sets of sensible objects or of ideal
mathematical units.

It is possible to find intermediates here, but not necessary, and not


even very plausible. This passage (like the one from the Theaetetus)
can do nothing to establish intermediates.
Phaedo 74 bc is the notorious passage involving "the equals
themselves". To deal with this passage in the appropriate depth
would take me far beyond the scope of this paper. I shall merely
make three points. Firstly, the passage can be interpreted either s
referring to the Form of Equal or s introducing intermediate
equals, and both interpretations claim considerabel support27.
So the passage is not unambiguous support for the theory that
Plato accepted intermediates in the dialogues. Secondly, the
Phaedo argument makes essential use.of the notion of a perfect
instance of equality, whereas Aristotle's evidence does not present
the intermediates s -perfect instances. Thirdly, the Phaedo
argument is apparently not limited to mathematical concepts,
but is applicable over a wider area. No restriction to purely mathematical uses of "equal" is suggested. So this argument is not good
evidence for Plato's belief in purely mathematical intermediates.
There are two passages not in Wedberg's collection, in which
reference to intermediates is likewise not obvious, or is incidental.
At Cratylus 432 ab Socrates contrasts the logic of pictures and
images with that of numbers, and makes the perfectly sound point
that if you add anything to 10, or subtract anything from it, it
does not change slightly, but becomes a totally different number.
Here again Plato is concerned with pure numbers, and the
reference to computation suggests the mathematical numbers
27

See Geach, "The Third Man Again", Philosophical Review 1956; K. Mills,
"Plato's Phaedo 74 bc", Phronesis 19578; Bluck, "Plato's Form of
Equal", Phronesis 1959; Haynes, "The form equaty, s a set of equals:
Phaedo 74 bc", Phronesis 1964; Rist, "Equals and intermediates in Plato",
Phronesis 1964; Owen, "Dialectic and Eristic in the treatment of the Forms",
in Avistotle on Dialectic,, ed. Owen, p. 114 n. 2.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

159

(from what we know of them from elsewhere). But the passage


contains no Suggestion that these numbers are intermediate between Forms and things. The passage leaves the Impression that
Plato has not yet seen the need for a distinction between two kinds
of pure numbers.
Timaeus 50 c28 is a passage where geometrical figures "enter
and leave" the Receptacle; these cannot be identified with either
Forms or physical objects. This, however, is a problem proprietary
to the cosmology of the Timaeus, and since it would seem to raise
Problems in the case of Forms other than the mathematical ones,
this passage has no connexion with the need for intermediates s
objects of the mathematical sciences. The Timaeus items could on
reflection be considered s suitable candidates for the function
of geometrical intermediates, but this is obviously not in Plato's
mind. There is no trace of the Uniqueness Problem s an argument
for their existence.
This brief survey of these passages brings out three points. Firstly, the supposed intermediates figure in very different connexions.
The Cratylus line of thought, for example, is very different from
the Timaeus one. When the passage are compared it is clear that
there is no one single concern with intermediates that turns up in
all of them. The impression is rather that there are a number of
different problems which concern Plato at different times, all of
which involve reference to items which, if pressed, have some of the
features of intermediates. These references are all more or less
incidental and there is no indication that Plato was aware of, and
explicitly dealing with, the problems raised by casual reference to
such things, or took them all to be a single problem. So even if
Plato did have intermediates in mind in any or all of these passages,
they still should not be used s indiscriminate support for a theory
that Plato accepted intermediates. The different lines of thought
do not cohere into a single theory. Secondly, reference to intermediates is implicit at best. Even if they figure in an argument,
they are never its concern. Since they are at most touched on, none
of these arguments prove that they exist; so none of these passages
contains (even implicitly) an argument for positing intermediates.
Thirdly, none of the arguments in which they figure has anything
to do with the Aristotelian context of the intermediates. There is
28

Discussed by ROSS in his note.on 987 b 14.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

160

J u l i a Annas

no reason here to connect intcrmediates specially with mathematics,


and no hint of the Uniqucness Problem.
None of these passages seems to me to contains, or to licence, any
conclusions about the reason for Plato's belief in intermediates.
To use these passages to establish the theory is like trying to convict someone of accepting numbers into his ontology because he is
prepared to talk about adding numbers. Ontological commitments
are not revealed by a willingness to use certain Standard idioms,
or even by the assumption of what those idioms suggest. They are
revealed only by explicit and reasoned argument for what the idioms
merely hint at. Plato's casual use of language assuming the existence
of pure numbers that are not Forms therefore proves nothing in
this connexion. Hence these passages are very different from those
that can be taken to argue for intermediates.
The Republic and the Philebus. Two lines of thought

There are only three passages in the dialogues which put f orward
a line of argument which can be seriously treated s an attempt to
establish the existence of intermediates. These are Republic
509 d511 a (the Divided Line) and 523 c526 b; and Philebus
56 c59 d together with 61 d62 b. The relevant question is:
do any of these passages contain, even implicitly, the argument for
intermediates suggested by Aristotle the need for an answer to
the Uniqueness Problem ? The answer turns out to be, No; in these
passages Plato is following up two quite different lines of thought.
1. The arbitrariness of number-ascription.
Republic 525 c526 b is concerned with mathematicians and
the way they treat the "ones" they deal with in their subject. They
puzzle people outside the subject by refusing t o divide their unit
or one; if anyone divides it up they multiply it back again, "taking
care that the unit shall never appear to be, not one but many parts"
(525 e 34). Their numbers are such that in them the one is always
"equal every one to every other", admitting no difference and with
no parts. Such numbers can only be conceived by pure thought, not
grasped in any other way.
The Philebus passage (in its two parts) makes essentially the
same point, though the terminology is slightly more sophisticated.
The Republic talks about "the one" in numbers, but since there are
many such distinct "ones" the passage wavers confusingly between
Singular and plural. The Philebus has moved on to the concept of

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

161

a "unit" (
), pure units being straightforwardly plural but
having the function of the "one" or "ones" in the Republik.
In the Philebus the arithmetic of "the many" is contrasted with
that of the philosophers (and a similar contrast is indicated for
geometry and computation). The contrast is a slightly more developed version of the contrast between mathematicians and the ignorant
in the Republic. The many are content to count with "unequal
units", such s oxen or armies, and will count s two any pairs
of large and small things indifferently. The philosopher will deal
only with units that are all exactly equal, none of all the infinitely
many units differring from any of the others in the slightest
(56 d 9eS).
Both passages fhid the same kind of fault with what the ignorant
or the many do, When the many count, they take the actual objects
to be counted s their units, and these objects may be of very
dissimilar kinds. They count, "one ox, one army"30 and end up
with a group of two. This is unsatisfactory because the objects
counted are dissimilar and divisible: one army can be counted s
a number of men. Hence the numbers reached are less satisfactory
than pure numbers. The source of Plato's dissatisfaction here is
clear enough. Depending on how we choose to regard a group, we
can ascribe different numbers to it. We end up with different
numbers depending on wht we take s our unit for counting, and
we can "divide" an original unit into many by counting its parts.
This makes our ascriptions of number appear arbitrary and subjective31. Since Frege, we can solve this sort of problem by invoking
the distinction between object and concept. The different numbers
are reached by counting objects under different concepts. Plato,
however, blames the unsatisfactory nature of ordinary counting
on the nature of the objects of counting.. The possibility of ascribing
different numbers to one group of objects strikes him s a deficiency
in the nature of those objects. So he propose to replace them by
89

This increased sophistication may be dne to Plato's struggles in the Parmenides


with the concepts of one and unity.
80
The passage does not make it tquite clear whether Plato's point is that two
oxen make two, although they diifer (in size, for example), or that any two
units make two, no matter how diverse in kind, (e. g. one ox and one army).
The latter is the stronger point, since the fact that we can count under
difierent unit-concepts raises more puzzles for a naive notion of counting than
does the fluctuating bchaviour of physical objects.
81
And Frege says (p. 26 e), "an object to which I can ascribe different numbers
with equal right is not what really h s a number".

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

162

J u l i a Annas

more dcpendable objccts, in order to achieve satisfactory counting.


These depcndable objects are the indivisible units or ones counted
by the philosophers and mathematicians. These pure units are
exactly alike and indivisible, each being one and only one. "One"
applies to them all in a single fixed sense. When we count groups
of objects, "one" cannot for Plato apply in a single fixed sense,
because of the possible variability of the results. Satisfactory counting and calculating require a whole different rnge of objects
distinct from the objects around us that we count. These are pure
numbers, composed of undifferentiated pure units.
This line of thought leads to a platonistic cdnception of number
which can be recognised s the theory of "mathematical number"
criticised by Aristotle in book M of the Metaphysics. But it is clear
that it has nothing to do with the Uniqueness Problem, and has no
relation to Aristotle's criticisms of the intermediates. This comes
out in two ways. Firstly, the present line of argument concerns
counting and numbers, and has no obvious application to geometry.
It is true that in the Philebus Plato says that the same applies to
geometry, but he Sketches no argument for this, and the argument
about arithmetic does not carry over. The Uniqueness Problem,
however, applies in exactly parallel fashion to both areas. Secondly,
the emphasis of the arguments in the Republic and Philebus is laid
on establishing the existence of units and numbers over and above
those the ordinary man believes in, and the ideal nature of these
units and numbers; the question of their multiplicity is not stressed.
Similarly, Aristotle in Metaphysics M stresses the ideal and
"separate" character of Plato's mathematical numbers and the
units in them. The fact that there are many copies of each of these
numbers is not emphasised. Aristotle *does not argue against the
multiplicity of ideal mathematical numbers s such; he does, of
course, argue against the multiplicity of intermediates s such.
A further point is that Aristotle in Metaphysics M treats the
theory of mathematical number s one that Plato shares with
Speusippus. Since Speusippus rejected Forms but continued to
believe in the existence of mathematical numbers, it is clear that
this argument for mathematical number does not establish numbers
that are intermediates. The argument can be accepted even when
one rejects one of the types of entity that intermediates are intermediate between.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

163

2. The objects of knowledge.


The passage of the Divided Line (Republic 509 d511 a) is
so well known that it does not require detailed exposition. I shall
merely bring out very briefly the kind of line of thought that is
in question here.
By means of the Line Plato divides "the domain of human
knowledge", s Wedberg puts it. The upper part represents the
intelligible world and the lower the visible. Corresponding to these
are the faculties of and . further subdivision of
each part in the same ratio divides into $ and
, and into $ and . The most controversial
question about the passage is whether these divisions of the
faculties correspond toa division of their objects. If so, mathematical
intermediates would be required s suitable objects for the Operation
of , Forms being required for $. Whether one should
find intermediates here depends on the following type of consideration. The logic of the passage seems to require them, particulary the central role of the relation of copy or image to
original. Plato's characterisation of knowledge elsewhere in the
Republic, especially in the argument of 476479, ties closely
together the questions of what knowledge is and what the objects
of knowledge are. The Line is related to the long passage in book
7 where the Guardians' education is being discussed, and there is
a similar problem in the latter passage about the nature and Status
of the objects of the first stage of dialectic.
Fortunately it is not necessary to solve here the perhaps unanswerable c fficulty of whether intermediates must be posited
because of the internal logic of the Divided Line32. Here I merely
want to point out that the question is one of epistemology, posed
entirely in terms of the Republic's discussion of knowledge. Its
solution would be a solution of problems about knowledge and
its objects. This is quite different from the context in which Aristotle
discusses the intermediates. It is true that the epistemology of the
Divided Line has implications for the account of the methods of
the sciences in books 67 of the Republic, and this has an indirect
bearing on the Uniqueness Problem, or at least on some of the
considerations that lead to the posing of the Uniqueness Problem.
But the connexion is not direct, and seems not to have been in
a

~ There have been innumerablc attempts to solve this vexed question; some are
to be iound in the works cited in notes l and 2.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

164

Julia Annas

Plato's mind when hc worked out thc sime of the Divided Line.
In fact the character of the passage suggests that Plato is deliberately avoiding any hard-and-fast specification of the subjects of
.
Conclusions

It turns out that none of the passages in Plato's dialogues contains the same line of thought s Aristotle's. Some of them only
contain incidental or casual use of suggestive idioms, and no argument can be built on this. Even passages containing arguments
which could, in fact, manufacture items suitable s intermediates
have no direct connexion with Aristotle's concerns. They argue
for entities which would serve very much the same kind of function
that would be served by entities produced s a solution to the
Uniqueness Problem. But this does not show that the arguments
for them are the same.
So it appears that Aristotle is not talking about the intermediates s these figure in Plato's dialogues. Plato never, in the
dialogues, offers the argument for intermediates that Aristotle
sees s crucial.
This suggests a reconsideration of the traditional problem: is
Aristotle right when he says that Plato believed in intermediates ?
The types of solution to this problem which I mentioned all assumed
that the Aristotelian evidence about the intermediates either directly supported, or was in direct conflict with, the introduction of
intermediates in the dialogues. But if the arguments offered for the
introduction of intermediates are quite different then there is no
reason why Aristotle's Statements should support or conflict with
a passage in the dialogues where Plato's concerns are quite different.
Nor can Aristotelian texts be used to resolve any problem internal
to a passage in the dialogues. Thus there is no problem of how to
square what Aristotle says about intermediates with what Plato
says about intermediates in the dialogues. The two do not compete.
One can accept Aristotle's criticisms s cogent and yet deny that
they clash with what is said about intermediates in the dialogues.
One can also deny that what is said in the dialogues amounts to
the acceptance of a theory of intermediates without having to face
the daunting task of explaining away all Aristotle's references s
baseless or confused.
Still, the fact that Plato and Aristotle have in mind different
arguments does not by itself show that they are not talking about

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

On the "Intermediates"

165

the same items. It is reasonable to hold that they are both talking
about intermediates. To that extent the traditional problem can
be answered in the affirmative. We have seen that many of the
difficulties usually held to arise from this answer do not in fact
arise if we recognise the fact that the arguments presented are
quite different. But there remains the difficulty of the total absence of the Uniqueness Problem from the dialogues. Why is the
one argument Aristotle deals with one that Plato never uses?
Several possible answers to this suggests themselves. One is that
Aristotle is simply "rationalising" Plato, presenting him with a
single argument which Aristotle thinks he should have put forward, to replace the variety of inconclusive considerations actually
found in the dialogues. This cannot be ruled out, given the way
Aristotle often reinterprets a previous thinker in terms of the argument Aristotle thinks he should have put f orward, given his other
beliefs. On the whole, however, it seems unlikely in view of the
number and precision of Aristotle's criticisms of the intermediates
s the outcome of the Uniqueness Problem. It is not like Aristotle
to argue so painstakingly against a nonexistent Opponent.
Another possibility is that Aristotle is not talking about the
dialogues at all, but about intermediates s they figured in some
other work of Plato's. We have adequate evidence that Plato did
teach "unwritten doctrines" which dealt largely with philosophy
of mathematics. The care and rigour of Aristotle's criticisms suggests
not oral instruction but something like his own written account
of Plato's famous lecture "On the Good". This hypothesis, however, can well be regarded s extravagant, and lacks support in any
of the fragments of the "unwritten doctrines",
It is also possible that Plato did in fact have the Uniqueness
Problem in mind when he posited intermediates, but that this is
never stated in the dialogues, for a variety of reasons. Some of the
passages may predate the idea, and Plato may not have wanted to
put forvvard an explicit defence even in the others, possibly because
he was not happy about the ontological profligacy and did not want
to be firmly committed to them. Certainly, the important passages
where they figure aim to establish the ideal and separate nature
of mathematical objects, and Plato appears less concerned about
their plurality. It is also possible that he holds back a discussion of
the Uniqueness Problem s being too technical for a dialogue33.
33

At Republif 534 a Plato deliberately holds back a more technical account of


the objects of the various types of knowledge. It is worth noting that the

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

166

J u l i a Annas

However, whatever Plato's possible reasons for leaving the


Uniqueness Problem out of the dialogues, the fact remains that
this is what he does, whereas for Aristotle it is central. So although
Plato and Aristotle raay both be talking about intermediates they
are doing so in very different ways, and what they say about intermediates cannot usefully be directly compared. This conclusion
underlines a more general feature of Aristotle's treatment of Plato's
doctrines. Where what Aristotle says about a thesis of Plato's
does not square with the appropriate passages in the dialogues, we
should not assume at one that there is conflict and hasten to resolve
the problem on that assumption. When we examine the arguments
on both sides we may find that they are completely different34.

34

third Suggestion is consistent with the second, if one accepts that the lecture
"On the Good" may have been nearly contemporary with the Republic, and
may have treated more technically some of the things merely touched on in
that dialogue (e. g. the Form of the Good). Cf. K.-H. Ilting, "Platons ungeschriebene Lehren': der Vortrag 'ber das Gute'", Phronesis 1968. This does
not commit one to the more extreme thesis of an "esoteric" Plato standing
behind the "exoteric" Plato at all times (cf. Krmer, Arete, Gaiser, Platons
ungeschriebene Lehre, Wippern (ed), Das Problem der ungeschriebenen Lehre Plaions).
I am very grateful to Professor Charles Kahn and to the referee of this paper
for suggesting many major improvements.

Brought to you by | University Library Technische Universitaet Muenchen


Authenticated | 129.187.254.46
Download Date | 10/2/13 7:36 PM

Das könnte Ihnen auch gefallen