Bell, Ian - Demonstration in Aristotle's Metaphysics

Demonstration in Aristotle's
Metaphysics
Ian Bell
In recent years scholarship in Aristotle has focused extensively on the

role of the Posterior Analytics in understanding the structure and methodology of other parts of the Aristotelian corpus. In particular, a substantial amount of work has been done on the relevance of the Posterior
Analytics to Aristotle's biology, and it is now widely acknowledged that
attempts to understand the biological works must take the Analytics into
account.1 However, most recent scholarship implicitly assumes that the
investigation undertaken in the Metaphysics follows a methodology different from that developed in the Analytics. In so doing scholars are
following a tradition that goes back at least to Owen's influential
', which argued that much of what goes on in Aristotle's
major works is not the kind of empirical investigation described in the
Analytics but rather conceptual or 'philosophical' investigation using a
dialectical methodology.2 Walter Leszl and T.H. Irwin have applied this
conclusion specifically to the Metaphysics, arguing that it contains a
second-order investigation into the presuppositions of the empirical
sciences.3
1 See, for instance, Gotthelf and Lennox (1987), chs. 5-7. For considerations of the role
of the Posterior Analytics outside biology proper see particularly Bolton (1991) and
Anagnostopoulos (1994).
2 See Owen (1961), 83-92. The 'empirical' vs. 'conceptual' distinction is Owen's; he
calls the Analytics methodology 'empirical' because of its insistence that principles
be arrived at through (83-4).
3 See Leszl (1975), Irwin (1988); cf. Witt (1989), 25-31.
Brought toand
youscience
by | St. Petersburg State Univers
APEIRON a journal for ancient philosophy
Authenticated
| 93.180.53.211
0003-6390/99/3202 75-108 $18.00 Academic Printing
& Publishing
Download Date | 1/7/14 2:02 PM
76 Ian Bell
The most significant way in which the Posterior Analytics could be
relevant for understanding Aristotle's methodology in the Metaphysics
is through its account of scientific knowledge () based on demonstration (). A strong claim for the influence of the former
upon the latter would thus involve the claim that the science ()
of being undertaken in the Metaphysics, or at least a part thereof, is a
demonstrative science. There are at least three reasons to think that this
cannot be true. First, metaphysics is thought to be a science of principles,
and it is a commonplace that principles are undemonstrable. Aristotle is
quite explicit, for instance, that the principles of demonstration cannot
themselves be demonstrated. Some scholars Irwin, for instance take
this to indicate that metaphysics as a whole is prior to demonstration.
Second, it is difficult to see how the Metaphysics position that being is a
equivocal can be reconciled with the Posterior Analytics requirement that the terms used in definitions and demonstrations be univocal.
Finally, there is no explicit statement anywhere in the Metaphysics that
its investigations follow a demonstrative proof-structure, nor do there
seem to be any arguments in the Metaphysics that strictly conform to the
model of syllogistic demonstration that emerges from the two Analytics.4
In what follows I shall argue that these objections can be met, and that
the proof-structure for at least one significant task undertaken in the
Metaphysics is sufficiently close to that presented in Analytics to justify
calling it 'demonstration'. The first objection, I shall argue, is based on a
mistaken conception of what the tasks that Aristotle assigns to metaphysics have in common: this commonality results not from their proofstructures but from the relation of their objects to being and
substance. In Metaph IV 2, it becomes clear that one of these tasks is the
investigation of the per se attributes of being and unity. The second
objection can be met by recognizing that the nature of a equivocal
is found in its primary instance, which is not itself equivocal. Finally, we
shall point to texts in the Metaphysics and De Anima that state or imply
that demonstration is the appropriate proof-structure for the treatment
of per se attributes, and examine what characteristics of demonstrations
are emphasized in these texts. We shall see that one of the proofs that
4 I use 'proof-structure' specifically to refer to the part of a discipline's methodology

that indicates how the principles or premises of an investigation are related to its
conclusions.
Brought to you by | St. Petersburg State Univers
Authenticated | 93.180.53.211
Demonstration in Aristotle's Metaphysics 77
Metaph IV 2 would lead us to expect to be demonstrative the proof

that each contrary has only one contrary has the characteristics
emphasized in these texts. The absence of demonstrations that conform
strictly to the Analytics model is not peculiar to the Metaphysics: it is not
clear that there is any such proof anywhere in the corpus. Together these
facts suggest that Aristotle intends demonstration as the proof-structure
for the treatment of per se attributes in the Metaphysics, even if he is tacitly
abandoning some of the constraints imposed on the form of demonstrations by the Posterior and particularly the Prior Analytics.
I
We shall begin with an outline of the demonstrative methodology developed in the Prior and Posterior Analytics. Although the relation between these two books is controversial, Aristotle's characterization of
demonstrations as 'scientific syllogisms' (APo I 2, 71bl8) creates the
prima facie expectation that demonstrations will be species of the syllogism described in the Prior Analytics.5 A proof that clearly conformed to
the Analytics conception of demonstration would thus be a Pnor Analytics
syllogism that possessed certain additional characteristics peculiar to
demonstrations.
In its basic outline the model of the syllogism developed in the Prior
Analytics is well known. A syllogism contains two premises and a
conclusion; the conclusion states a relationship between two terms
which is proved using a third term as a middle term. The general form
of the universal affirmative syllogism (Barbara), for instance, is as follows:
C belongs to all B.
belongs to all A.
C belongs to all A.
One of the extreme terms (C) is shown to belong to the other (A) using
the middle term (). It is this form of syllogism that is most 'scientific'
(APo 114) and most useful for demonstration.
5 For recent discussion of the relation between the two Analytics see Barnes (1981),
which summarizes the earlier Solmsen-RossBrought
debate, to
andyou
Smith
(1982a and 1982b).
by | St. Petersburg State Univers
78 Ian Bell
Broadly speaking, demonstrations prove the existence of effects

through their causes, ultimately from first causes (APo I 2, 71bl9-72a7,1
10,76a37-bll). Just by itself this leaves open a wide range of possibilities
for the structure of demonstrations. In effect, however, the Posterior
Analytics is much stricter about what kinds of principles can appear in
demonstrations and what kinds of effects demonstrations prove to exist.
First, there is the subject genus ( , cf. 75a42-bl), which
is posited to exist and whose per se attributes (75bl, 76bl3) the science
investigates. Each science is one by having one subject genus (I 28,
87a38-9), whose signification and existence it assumes (76b5; cf. I 1,
71al5-17). Second, there are the attributes that are demonstrated to
belong per se to a subject genus. The signification or whatness ( )
of a per se attribute is assumed as a premise, but its existence ( )
that is, the fact that it belongs to the subject genus is the conclusion of
a demonstration (75a40; 76a31-6, b6-ll; cf. 11, 71all-17). Finally, there
are the common axioms (), such as the principle of contradiction, which are used by all the sciences to the extent that they are useful
in each (75a41-2; 76a37-b2, b!4-15; cf. 111, 77alO-35). A demonstration
proves that a per se attribute belongs to a subject genus, using the
existence of the subject genus and the definitions of the genus and
attributes as premises. The common axioms do not appear as premises
in individual demonstrations (APo 111). Thus the subject genus would
map onto term A of our Prior Analytics syllogism, the attribute itself onto
term C, and some definition which expresses the cause of the attribute's
belonging onto term B (cf. Aristotle's example, II8,93bl4).
What kinds of thing are demonstrable attributes? This is a matter of
dispute, and there are at least three different accounts in the recent
literature.6 All scientific truths are necessary truths, and a predicate must
belong per se (' ) to its subject in order to belong to it necessarily
(APo 16,74b5-8). The demonstrable attributes will thus be the predicates
that belong to their subject ' . In APo 17 and 110 Aristotle refers
to these attributes generically as ' (75a40-l, 76b4),
but also as ' (75bl), (75bl, 76bl5), and
(76bl3). In APo 16 (74b6-10), Aristotle implies that of the four
kinds of ' described in APo 14, two are relevant for demonstration and ' . In the first sense the attribute will be part
6 Ferejohn (1991), chs. 3-6, McKirahan (1992),Brought

ch. 7, Goldin
(1996), ch. 6
to you by | St. Petersburg State Univers
of the definition of the subject (type [1] ' ); in the second, it will
be defined in terms of the subject without being part of its definition
(type [2] ' ). Examples of the latter are oddness and evenness
with respect to number (14,73a39-40); according to Metaph IV 2 these are
of number (1004blO-12).
The relation between the various words used for these attributes is not
entirely clear. In APo I 6 Aristotle uses to refer generically to
both kinds of predication. This reflects its etymological origin in ,
which may be used to refer to the fact of almost any state of affairs.
Likewise, and are used generically at 76bl3-15. On the
other hand, Aristotle's usual use of is to refer to something
nonessential; the notion of a ' ('essential accident')
is almost an oxymoron. It is thus possible that the word refers primarily
to type (2) ' predications, i.e., to those where the predicate is not
part of the definition of the subject. This is suggested, for instance, by
Aristotle's brief treatment of ' at Metaph V 30,
1025a30-4. On the other hand, although there is no instance in the Posterior
Analytics where ' alone is used generically to refer
to all ' predications, Aristotle does appear to be using the phrase
to refer to all demonstrable predications in his discussions of the third
and fifth aporias in Metaph III (997al5-34). Thus, although we shall
reserve 'per se accidents' to translate ' , there is no
sharp distinction in Aristotle's use of the terminology.7
Finally, APo 14 demands that attributes that are strictly ' to
a subject also belong to it universally () and 'qua itself ( )
(73b26-32). Otherwise put, the attribute must belong to all instances of
the subject, and to that subject primarily (, 73b39; cf. b33). This is
best illustrated with an example. The attribute of having angles equal to
two right angles belongs to triangles. It is not ' or
to the genus above triangle, i.e., plane figure, because there are plane
figures to which the attribute does not belong. Nor is it to isosceles
triangles, even though it does necessarily belong to isosceles triangles.
Rather, the attribute of having angles equal to two right angles is
to the widest subject genus of which it is true that the attribute belongs
to all of its members (73b32-74a3). The attribute belongs primarily to that
genus, and is 'in a sense not ' ' to lower species of the genus to
7 Pace Ross (1924), 1.224

80 Ian Bell
which it belongs (74a2). As Aristotle emphasizes at greater length in APo

I 24, the subject to which the attribute belongs is the cause of the
attribute's belonging to it and its species (85b23-86a3). From this Aristotle concludes that the best demonstration of the existence of an attribute is a universal demonstration, that is, a demonstration that shows it
to belong to the subject to which it is . More generally, the
investigation of an attribute should demonstrate it to belong to the
subject to which it belongs .
We shall see that the investigation into the per se attributes of being
described in Metaph IV 2 shares many of the features of the proof-structure described here. At the same time, the foregoing should make it clear
why it is so difficult to find any proof that strictly conforms to the
Analytics conception of demonstration. It is difficult to see how demonstration as described in the Analytics could be flexible enough to describe
and explain any wide range of phenomena, if Aristotle's reference to
scientific syllogisms in APo 12 does refer specifically to the syllogism as
described in the Prior Analytics. We shall see that there is evidence both
that Aristotle wishes to retain the term 'demonstration' with the basic
elements recognizable from APo 17 and 110, and that he tacitly abandons
the requirement that such demonstrations be presented in the form of a
Prior Analytics syllogism.
II
Aristotle introduces metaphysics in Metaph TV 1 with the statement that

there is a science () that studies being qua being and whatever
belongs to it per se ( ' ,
1003a21-2). This science is distinguished from those that do not investigate 'universally about being qua being' ( v,
a24) but only about a part of being. If we are to read this declaration in
light of the methodology described in the Posterior Analytics, we should
expect Aristotle to be indicating two things. First, the science will study
being universally () and qua itself ( ). The phrase 'being
qua being' indicates the level of generality at which the science of being
studies beings: there are scientific truths applicable to beings not qua
some kind of being but precisely because they are beings.8 Second,
8 Cf. Stevenson (1975).

Aristotle's references to the ' (a22) should refer to

the per se attributes (' ) that are the Posterior Analytics
objects of demonstration.9
The difficulty, of course, is to show that the passage should be read in
this light in the first place. If Irwin is right that Aristotle is introducing a
wholly new kind of investigation in the Metaphysics, then probably the
apparent references to the Posterior Analytics are only apparent. In support of this view, Irwin appeals to Aristotle's characterization of the
Metaphysics as a science of principles and the prominent place that the
treatment of the principles of demonstration (or common axioms) plays
in Metaph IV. Since the axioms are principles of demonstration, it is
impossible to demonstrate them without circularity. Moreover, part of
Aristotle's treatment of the axioms is dialectical, and this dialectical
treatment appears to serve as a justification for the axioms.10 Irwin takes
the treatment of the axioms to be indicative of the character of metaphysics as a whole: it serves as 'a test for the method of first philosophy.'11
Metaphysics will constitute a second-order dialectical justification of
principles assumed by the first-order demonstrative sciences.
We shall argue, against views such as Irwin's, that what Aristotle
takes to be distinctive to metaphysics is not a second-order, dialectical
methodology but the relation that the objects of metaphysics all
have to being and substance. There are three kinds of thing that possess
this relation to being: the principles and causes of being, the per
se attributes of being, and the common axioms or principles of demonstration. Metaphysics thus has three tasks, each of which has a proofstructure appropriate to it.
The clearest evidence for our account of the tasks of metaphysics is
found in the first five aporias concerning the scope of metaphysics, and
9 This is the interpretation adopted by the Greek commentators on Book IV; Alexander assumes a demonstrative methodology in commenting on IV 1,1003a21-2 (in
Metaph 239.6-9), IV 2, 1004bl-8 (258.8-10), and IV 2, 1004bl7-26 (260.2-5, cf. 25-6).
See also Syrianus in Metaph 63.6-8,24-6; Asclepius in Metaph 246.6-9.
10 So Irwin (1988), 187-8; against this view see Code (1986) and Bolton (1994).
11 Irwin (1988), 179-80, see also 172-3, 196-8. Irwin recognizes in a footnote that 'to
claim that first philosophy is to [some] extent non-demonstrative is not to claim that
it is wholly non-demonstrative' (547 n. 2), but this does not seem to affect his
approach in the main text.
82 Ian Bell
in Aristotle's answers in Metaph IV and VI.12 It is clear from what
Aristotle assumes in the first aporia that metaphysics is somehow a
science of causes, and from what he assumes in the second through fifth
aporias that it is somehow a science of substance. Taking for granted that
metaphysics is a science of the principles of substance, the second aporia
goes on to ask whether it is also a science of the principles of demonstration (995b6-8, 996b31-2). Likewise, the fifth aporia asks whether the
science of substance is also the science of its per se attributes (995bl8-25,
997a25-6). In Metaph IV, 1-3 and VI1, Aristotle both confirms his initial
assumption that metaphysics is a study of principles of substance, and
argues for affirmative answers to the second and fifth aporias.
In Metaph IV 1, Aristotle maintains that there are certain principles
and causes that are ' to being and belong to it qua being, that
is, qua itself ( ).13 It belongs to the science of being qua being to
investigate these principles and causes (1003a26-32). In Metaph IV 2
Aristotle isolates the primary instance of being in substance, and so
maintains that the science of the causes of being will investigate the
principles and causes of substance (1003bl6-19). The relation of the
investigation of the causes of being to the methodology of the Posterior
Analytics is complex. In APo 1-2 Aristotle recognizes the existence of
scientific investigations into the and of substances, and at
one point even goes as far as to imply the cause of either a substance or
an attribute will be the middle term of a demonstration.14 When it comes
time to give detailed examples of such explanation in APo II8, however,
Aristotle restricts himself to the explanation of attributes. Moreover, we
have seen that demonstration in the strict sense explains the existence of
attributes by proving that they belong to subjects whose own existence,
as we have seen, must be assumed. If substances cannot be attributes of
some other subject, it is difficult to see how there can be a demonstrative
treatment of the of substances. The Posterior Analytics recognizes
12 I follow the enumeration of the aporias in Ross (1924) and Apostle (1966).
13 On this passage see especially Dhondt (1961). Cf. Metaph IV 2, 1003bl6-19, VI 1,
1025b3-18.
14 For the explanation of the of the subject see APo 1, 89b31-5, 2 passim
(Aristotle alternates between the explanation of substances and attributes). At
90a9-ll Aristotle writes that the cause for both substances and attributes is found
in the middle term.

that there must be a science that explains the of substances, but it
appears that the methodology for this treatment is not demonstration.
This much is implied by Aristotle's methodological remarks about the
investigation into the principles and causes of being in Metaph VI1:
All [the other sciences], marking off some being or some genus, conduct
their investigations into this part of being, although not into being
nor into their part of being qua being, and they do not give an
account of ; but starting from the , which in some sciences
is made clear by sensation but in others is laid down by hypothesis,
they thus proceed to demonstrate more or less rigorously the per se
attributes [' ] of their genus. Consequently, it is
evident that there is no demonstration of or of from such
an induction but that these are made known in some other way.
Similarly, they say nothing as to the existence or nonexistence of the
genus they investigate, and this is because it belongs to the same power
of thought to make known both and . (1025b7-18)15
Without attempting to address the many difficult interpretative problems posed by this passage, it is possible to see that in investigating the
of the subject genus to which attributes belong, one of the tasks
of metaphysics will be concerned with things that must be assumed by
the demonstrative sciences and hence are not themselves objects of
demonstration according to the paradigms developed in the Analytics.
While the investigation into the principles and causes of substances does
not constitute a simple repudiation of the Posterior Analytics, the methodology for such an investigation will likely go beyond anything espoused there.16
What is of greater interest for our purposes is one of the passage's
other implications. We have suggested that one of the tasks of metaphysics is a demonstrative investigation into the per se attributes ('
) of being. While the Metaph VI 1 passage seems to confirm
15 Translations from the Metaphysics are based on Apostle (1966). I have noted only
significant departures. I take 'from such an induction' to modify the clause 'there is
no demonstration ...'; cf. Ross (1924), i.352, Kirwan (19932), 184, Owens (19783), 288,
and Bolton (1991), 16; this reading is not crucial for my argument.
16 Metaph 17, 1041a21-b9 suggests that the explanatory structure for substantial
explanation is analogous to demonstration.Brought
See Bolton
(1995),
to you
by |452-8.
St. Petersburg State Univers
84 Ian Bell
that Aristotle regards the treatment of these attributes as demonstrative,

it also implies that it is sciences other than metaphysics that undertake
this treatment. Metaphysics, our passage suggests, is concerned with
and the of the subject; it is contrasted with the sciences that
assume these in order to demonstrate .
On the other hand, if there are attributes that belong ' and
to being, it is difficult to see to what science they would belong if
not to metaphysics. It is precisely this difficulty that gives rise to the fifth
aporia, whether metaphysics is 'concerned only with substances or also
with the per se accidents of substances [ '
]' (995bl8-20). The reason for doubting that metaphysics will
study the attributes of substances is that if it does, there will be a
demonstrative science of substance, but there is no demonstration of
hence (it is implied) neither of substance (997a30-2). If metaphysics
does not study the attributes of substance, however, it is difficult to see
what science will do so (a32-4). The implication is that substance has
that are ' to it, and that the science that studies
them must be demonstrative. The aporia does not doubt the existence of
a demonstrative science of the attributes of substance: it doubts only the
identification of this science with the science of substance.
Aristotle concludes Metaph TV 2 with the statement that 'it belongs to
one science to investigate being qua being and the attributes that belong
to it qua being [ ], and that the same science
investigates not only substances but also their ' (1005al3-16).
We shall argue that this is the conclusion of Aristotle's solution to the
fifth aporia: there is a single science of being and its attributes, and the
science that studies substance also studies the attributes of substance. The
passage thus retrospectively indicates how we should read Aristotle's
reference at the beginning of Metaph IV 1 to a science of being qua being
and its ' (a21-2). These should be understood not
simply as 'all the things that a science of being would study', but as
objects requiring a specifically demonstrative treatment and whose inclusion in a science of being and substance is controversial. We shall also
attempt to see how this position may be reconciled with Metaph VI1 and
the arguments contra in the fifth aporia.
The last task of metaphysics is that proposed in the second aporia, the
study of the principles of demonstration. The second aporia raises two
obstacles to assigning the study of the common axioms to the science of
substance: it is not clear why it is the science of substance that should
study the axioms (996b33-997a2), and it seems that the science of the
Brought
to you by | St.
State
axioms cannot be a demonstrative science
(997a2-ll).
In Petersburg
Metaph IV
3 Univers
Aristotle attempts to solve the first problem with the identification of the
science of being qua being and the science of substance. The axioms
belong to being at that level of generality: they belong () to all
beings and not peculiarly to one genus apart from the others, and so they
belong to beings qua being ( v) (1005a22-7). Therefore, since the science
of substance is the science of being qua being, it belongs to the philosopher to study both substance and the principles of demonstration
(1005b5-8). The fact that the axioms cannot be demonstrated does not
prevent them from being included among the principles studied in the
science (1005bll-1006all). Notice, however, that although the axioms
to beings qua being, Aristotle does not call them '
, or even just . This may be merely coincidental, but
it is also what one would expect if Aristotle is reserving '
for the demonstrable attributes of being and substance. The
principles of demonstration themselves would be known by some nondemonstrative means, perhaps dialectically, perhaps by . In the
latter case, the Metaphysics will follow the Posterior Analytics account of
the knowledge of the principles of a science.17
What common feature of these tasks requires that they all be assigned
to metaphysics? What they have in common is that their objects belong
to being qua itself ( ), to being qua being. There are causes (1003a2832,1025b3-4) and attributes (1005bl3-14) that belong to things qua being,
and the relation to being is true of all the principles of demonstration (1005a21-9). As we have seen, the Posterior Analytics requires that
objects of investigation be investigated at the appropriate level of generality. If horses and oak trees have certain causes and attributes not
because they are horses and oak trees but because they are living things,
the study of these causes and attributes will belong not to zoology or
botany but to universal biology. If they have causes and attributes that
belong to them not qua living things but qua substances, the study of
these causes and attributes will belong not to physics or biology but to
the science of being and substance.
If it is a relation to being and substance that makes something
an object of metaphysics, then it becomes unnecessary to suppose that
metaphysics is distinguished from the special sciences by its proof-structure or methodology. Metaphysics may itself be a first-order science of
17 With Code (1986), 344,1 incline to the latter Brought

alternative;
see Bell (1998), 115-23.
86 Ian Bell
a certain nature, and be distinguished from the other sciences by studying only causes, attributes, and so forth that are to that nature.
Because everything is in some sense a being, metaphysics will be in a
sense a universal science. However, it does not attempt to be a universal
science in the vicious sense ruled out in APo I 9: it does not attempt to
demonstrate the principles proper to the special sciences (76al6-25). The
causes expressed in the definitions proper to biology, for instance, are
to the explanation of biological substance qua biological, not qua
substance. Metaphysics may take an interest in explaining why giraffes
are giraffes, but it is not interested in explaining why they are giraffes.
Likewise, it will be interested in explaining why giraffes have the attributes that belong to being and substance, but will leave the
explanation of the attributes that belong to giraffes qua living being or
animal to biology and zoology respectively. One may take issue with
Aristotle's claim that studying the being and substantiality of a giraffe is
somehow different from studying the giraffe-ness of a giraffe, but the
claim is nevertheless Aristotle's.19
One may wonder why Aristotle makes no mention of metaphysics in
the Posterior Analytics, if metaphysics does not violate the Posterior
Analytics restrictions on a universal science. The short answer, as Owen
(1960) recognized, is that the Posterior Analytics does not recognize any
sort of commonality between the various things that may be called
'beings'. In the absence of any such commonality, there would be no
nature to which the objects of a science of being would be , and so
there could be no genuine science of these things. Any attempt at a
science of being would constitute the kind of science that Aristotle seeks
to rule out: a master science capable of proving all truths about all things.
18 Cf. Bolton (1995), 454-5.
19 See esp. Metaph IV1 and VI1,1025b3-18. At VI1,1025b 10-14 Aristotle suggests that
metaphysics explains . Whatever this means, it does not mean that metaphysics explains why a certain definition has a certain content: as Aristotle emphasizes
at Metaph VII17,1041314-04, there is no explanation of why, e.g., humanity itself is
humanity. A broader investigation into the reasonableness of Aristotle's claims
would require a lengthy treatment of the nature and causes of being; see Bell (1998),
chs. 3,6-9.
III
If Aristotle's account of the tasks of metaphysics does not rule out a
demonstrative account of some of these tasks, does his account
of metaphysics' universality? According to the Posterior Analytics, what
makes scientific knowledge possible is the universal (), and
what this in turn requires is the possibility of predicating the same thing
in the same sense of all of its instances (APo 111,77a5-9,1 24,85b4-22, II
13, 97b26-39). Although equivocity generates a kind of universality, it is not clear whether and how this universality can be sufficient
for demonstration.
Aristotle's account of the relation is extremely familiar in
outline, desperately obscure in its details and implications. Nevertheless,
two things seem to emerge clearly from Aristotle's descriptions of this
relation in Metaph IV 2 and EE VII 2. First, any equivocal will
have a primary instance: souls, bodies, instruments, and actions can all
receive the predicate 'medical', but the name belongs properly ()
to its primary instance, that is, to the art in virtue of which a man can be
called a medical man (EE VII 2, 1236al9-23; cf. Metaph IV 2, 1003bl6).
Second, what makes for a equivocal is the fact that the derivative
instances must be defined in terms of the primary instance, that is, the
priority in (or 'logical' priority) of the primary sense to the derivative senses. This is clearly stated in the EE passage (1236a21-3) and
implied by Aristotle's examples of equivocals in both passages.20
To use an example common to the two passages, a medical tool is defined
as one used by someone possessing the medical art, and not vice-versa
(1236al9-23,1003bl-4).
The model of equivocity presented in these passages thus
suggests that there is one instance where the nature under consideration
is found, and other instances that deserve to receive the name of this
nature owing to their definitional connections with it. Hence the opening
lines of Metaph IV 2:
The term "being" is used in many senses, yet not equivocally, but all of
these are related to something which is one and a single nature [
20 Cf. Owen (I960), 171-2 For Aristotle's definition of logical priority and posteriority
see Metaph 3,1077a36-b4.
88 Ian Bell
]. It is like everything that is called "healthy",

which is related to health by preserving health, or by producing health,
or by being a sign of health, or by being recep live of heal th. (1003a33-b 1)
Many things are called healthy, always because of some relation to health
understood as a state of an organism. It is not the case that healthy food
has the state of health in some derivative, attenuated sense: rather, it is
called 'healthy' precisely for its capacity to cause this state in its primary
sense. Likewise, a science of health will study tools precisely in their
capacity to be of use in bringing about health.
The same language appears in Aristotle's account of the derivative
instances of being. Of the derivative instances some are called beings
by being attributes [] of substances, others by being on their way
to becoming substances, or else by being destructions or privations or
qualities of substances, or productive or generative either of substances
or of whatever is related to substances, or negations of any of those or
of substances. On account of this, we say that even nonbeing is nonbeing. (1003b7-10)21
If we are to follow the model of the other examples of , we should

expect the nature of being to be found in substance, and the derivative
instances of being to be so called only in virtue of their definitional
relations to substance. Nonsubstantial beings are so called not because
they are beings and substances in some lesser sense but because they are
definitionally related to substance.22 To this one may want to respond
that if only substances share in the nature of being, then presumably
nothing apart from substances would exist. Aristotle's point seems to be
that this is to assume the wrong conception of being in the first place:
being is not the sort of nature that everything in the world must share.
If one assumes that it is, differentiation becomes impossible and Parmenidean conclusions must follow (III 4,1001a29-bl, cf. 998b22-7).
21 As Owen (1960), 173 n 23 points out, echoes of this list of derivative instances are
found in Aristotle's Metaph V treatments of unity (1016b6-9), contraries (1018b31-8),
potency (1019b35-1020a6), quantity (1020al4-32), perfection (1022al-3), and falsity
(V 29, passim).
22 Cf. Owens (19783), 264-75.

If the nature of being is found only in substance, then the primary

instance of being is something that is itself univocal, at least vis- -vis the
nonsubstantial instances of being.23 If the per se attributes of being and
substance are also univocal or at least have univocal primary instances,
then the univocity requirements of Posterior Analytics demonstrations are
met. In what follows we shall argue that Aristotle sets up his investigation of the per se attributes of being and unity in such a way as to meet
these requirements.
IV
The treatment of the per se attributes of being in Metaph IV 2 follows

immediately upon Aristotle's account of equivocity. The first part
(1003bl6-1004a2 and 1004a9-31) introduces sameness, difference, and so
forth as 'forms' (') of being and unity. Aristotle then explicitly raises
the fifth aporia and argues that its solution is to regard these as per
se attributes of being qua being and unity qua unity (1004a31-b26). The
scientific approach to this study is contrasted with dialectical and sophistic attempts to deal with the same material. Finally, Aristotle attempts to
show how his approach is continuous with his predecessors' accounts
of the contraries as principles of all things: all things are reduced to
contraries and these ultimately to the one and the many (1004b271005al3). The chapter ends with the conclusion we have alluded to: it
belongs to the same science to study both being qua being and its per se
attributes (1005al3-18).
Our discussion begins abruptly with the statement that it belongs to
generically one science to investigate all the 'forms' of being (
):
23 To this extent I follow Bolton (1995), 423-9, though I do not follow his identification
of the per se attributes with the secondary instances. Although the question whether
substance itself is univocal between sensible and suprasensible substance is important to a final evaluation of whether metaphysics can be a science, it is impossible
to address it within the scope of this article. If substance is not univocal, then
presumably another reduction to the primary kind of substance will be
necessary. For the present it is sufficient to point out that Metaph IV 1-2 is not itself
concerned with any sort of equivocity withinBrought
substance.
90 Ian Bell
For each genus of things there is both one power of sensation and one
science; grammar, for instance, which is one science, investigates all
spoken sounds [ ]. Accordingly, it belongs to
generically one science to investigate all the forms [] of being, and
the forms of the forms. (1003bl9-22)24
The role of this text is initially unclear. We shall argue that it is Aristotle's
introduction to his treatment of sameness, otherness and so forth, which
are the and per se attributes of being.25 The passage has been given
at least two other interpretations, however, as referring either to the
various kinds of substance or to the various categories. These difficulties
arise partly because of uncertainties about the interpretation of certain
key words and phrases. The notion of of being is unusual and has
no parallel outside this text and its immediate context (cf. 1003b33-4). It
is unclear whether should be translated as 'species' or 'forms'; since
the translation 'species' has certain definite and problematic connotations, let us tentatively use the less interpretive 'forms'. Furthermore, the
cryptic phrase has two possible translations: either
the one we have given or a more expansive reading like Apostle's ' ...
and it belongs to one specific science to investigate each ' of being.'26
Of these issues the one that can be resolved with the most confidence
is that of the intended referents of . At 1003b33-6 Aristotle refers to
sameness and likeness as of unity and hence of being. This in turn
links 1003bl9-22 to the treatment of these things in the texts that imme-
24 Our translation differs significantly from Apostle's, on the difficulties of translation

see this paragraph and the next in the main text. The Greek text from 'Accordingly
...' is: 6 ,
.
25 Cf. Metaph 1,995b23-4. This interpretation is supported by, among others, Ross
(1924), i.256-7, Owens (19783), 275-9 and n. 55; and Mansion (1958). However, Jaeger
(1957), apparatus ad 1004a2, and Kirwan (19932), 82 follow Alexander (in Metaph
250.32-251.6) in associating 1003bl9-22 with 1004a2-9 rather than with 1003b221004al. All are agreed that 1004a2-9 is out of place in its current position; the
disagreement is rather whether 1004a2-9 interrupts 1003al9-4al and 1004a9ff., or
1003a22-4al interrupts 1003bl94a9ff. For references to the various other interpretations see Leszl (1975), 241 nn. 25-7.
26 Owens (19783), 275 and n. 55, and Kirwan favour the translation we adopt; in
addition to Apostle, Ross (1924) marginally favours the expansive translation, as
does the ROT. Kirwan believes both are possible; see his note ad loc.

diately follow it (1003b22-1004al and 1004a9-b32). We shall have a better

idea of what kinds of things are of being after having examined these
passages.
Aristotle makes his way into an investigation of the forms of being via
a limited identification of being and unity. The motivation for making
this identification is that many of the forms of being are strictly speaking
forms of unity. 'Being' and One', while not having the same connotation,
correspond to each other just as 'principle' () and 'cause' () do
(1003b22-9). If something is a being, it is also one (and vice-versa); and
the of each thing is both essentially a being and essentially one
(b32-3; cf. VIII6,1045a36-b7). Hence there will be as many forms of being
as there are of unity (b33-4). The investigation of the forms of being and
unity will thus belong to the same science as the science that investigates
being and unity (1003b33-1004al). In particular, the science of being will
also study sameness, likeness, and equality.
Aristotle moves on to show that the science of being will also study
the opposites of sameness, likeness, and equality:
Since it belongs to one science to investigate opposites [],
and plurality is opposed [] to unity, and since it belongs to
one science to investigate also denial and privation because unity is
investigated in both ways, with respect to its denial as well as to its
privation ...; it belongs to the same science to know also the opposites
of the kind of unity we mentioned, for example otherness [to ]
and unlikeness [ ] and inequality [to ] and all the
others which are named either according to these or according to
plurality and unity [ ]. (1004a9-20)
Additionally, contrariety itself is investigated by this science, being an

instance of difference and so of otherness (a20-2). These things are either
named 'according to' () plurality or unity or, as in the case of contrariety, are named according to something else that is in turn named
according to plurality or unity (in the case of contrariety, according to
difference). Aristotle explains further in Metaph V 15 that quantity,
quality, and so forth are named because they are defined in
terms of unity (1021alO-12).
Since 'unity', like 'being', has many senses, and sameness, otherness,
and so forth are ultimately reducible to unity, these too will have many
senses. Nevertheless it belongs to one science to know all of them, since
they are equivocals (1004a22-5). Brought to you by | St. Petersburg State Univers
92 Ian Bell
Now since all things are referred to that which is primary, as for
example all things which are called "one" are referred to what is
primarily one, we must say that the case is similar with sameness and
otherness and the contraries; so that after distinguishing the various
sense of each, we must give a similar account of how all the others are
related to that which is primary in the case of each predicate [
] ... (1004325-30)27
The univocity required for a science of the forms of being and unity is
generated by the fact that unity and its forms are studied in their primary
instances.
The relation between the primary and secondary instances of
unity and its forms is analogous to the relation in the case of being.
What is one in the highest degree is that thing whose is one; other
unities are so called for their relations to things that are primarily one:
Most things are called "one" in view of the fact that they act on, or are
affected by, or have, or are related to, some other thing which is one,
but things which are primarily called "one" are those whose is
one, either by continuity, or in kind, or in formula; for we count as many
either things which are not continuous, or things which are not one in
kind, or things whose formula is not one. (V 6,1016b6-ll)
Similar relations hold for sameness, likeness, and equality: 'since "one"
and "being" have many meanings, all the other objects which are called
according to these must also follow, so sameness and otherness and
contrariety must be distinct in each category' (V 10, 1018a35-8). At
Metaph V 10,1018a31-5 and X 4,1055a35-8 Aristotle explains the
relation for contrariety in the same terms as he does for unity. A primary
instance is identified (and in this case, defined), and the derivative
instances are defined in terms of the primary instance. Just as the
secondary instances of being are intelligible only in relation to the
substance to which they are related, the secondary instances of contrariety are intelligible as contraries only in light of the contraries that they
have, produce, or are acquiring or losing.
Having introduced the forms of beings and shown how they are
themselves equivocals, Aristotle is now in a position to make the
27 I follow Ross (1924), 1.206, on the translation of .

connection between the forms of being and the issues raised in the fifth
aporia. We saw that the aporia asks whether it belongs to the same
science to investigate both substance and its per se attributes, but the
question is not raised only in these terms. Taken as a whole, the aporia
asks
whether our Investigation is concerned only with substances or also
with the per se attributes of substances [ '
]; and in addition, concerning sameness and otherness and
likeness and unlikeness and contrariety, and with regard to priority and
posteriority and all other such, about which the dialecticians are trying
to inquire, conducting their inquiry from reputable opinions []
only, to what science does it belong to investigate all these? To these we
must add their own per se attributes, for we must inquire not only into
what each of these is, but also whether there is only one contrary to a
contrary, (ffl 1,995bl9-27)
It is necessary to investigate both whether metaphysics studies '
of substance and whether it studies the things that the
'dialecticians' attempted to inquire into. In Metaph TV 2 Aristotle makes
it clear that in inquiring into the 'forms' of being, metaphysics is concerned with the objects discussed in the second half of the aporia:
It is evident, then, that it belongs to one science to discuss [sameness,
otherness, etc.] as well as substance (this was one of the problems we
listed); and so, it is the philosopher's task to be able to investigate all of
them. For if it is not the philosopher, then who will examine whether
Socrates and sitting Socrates are the same, or if a given contrary has
only one contrary to it, or what is a contrary, or the various senses of
the term "contrary"? And similarly with all other such questions.
(1004a31-b4)
The study of these objects belongs to the science of being and more
specifically, to the science of substance. More precisely, they are investigated as per se attributes (' ) of substance in its capacity as
the nature and primary instance of being.
Since, then, these are the per se attributes [1 ... ] of unity

qua unity and of being qua being, but not qua numbers or qua lines or
qua fire, clearly it belongs to this science also to know both the whatness
of these and their accidents []. And those who inquire into
these matters err not in the sense that
they do not philosophize, but in
94 Ian Bell
not considering , of which they comprehend nothing, as prior. For

just as there are ' of number qua number (e.g., oddness and
evenness, commensurability and equality, excess and deficiency), and
these belong to numbers per se and in relation to one another (and
likewise other ' belonging to solids ... ), so there are certain things
to being qua being, and it is the task of the philosopher to examine
the truth about these. (1004b5-17)a
In the first passage, Aristotle gives two examples of the sorts of question
that the philosopher will ask: it is the philosopher's job to investigate
whether Socrates is the same as Socrates sitting, and whether any given
contrary has only one contrary. We know from the fifth aporia (995b25-7)
that Aristotle considers the investigation into whether a contrary has
only one contrary to be an investigation into a per se attribute, presumably an attribute of contrariety. The beginning of the second passage
indicates how we should take the other example: to ask whether Socrates
and sitting Socrates are the same or different is to inquire into a per se
attribute of Socrates qua being and one, that is, qua substance.
The science of substance will thus be able to establish truths about two
kinds of things. First, it will be able to establish that certain attributes
for instance, sameness or lack of sameness belong to substances qua
substances (1004b5-6). Second, it will know the of these attributes
and be able to establish that certain other attributes belong to these
attributes, for instance, that each contrary has only one contrary (b6-8).29
Aristotle emphasizes the parallelism between this investigation and his
paradigm demonstrative sciences of arithmetic and geometry: just as
there are attributes that belong to number and solid, so there are
attributes of being and unity in their primary instance, substance.
Metaphysics studies the same objects studied by the Academic dialecticians, but as per se attributes of being qua being and unity qua unity, and
hence of substance qua substance.
28 In the context 'accidents' should be read as shorthand for 'per se accidents'; cf.
2,997a25-34
29 I take (b8) to refer to the of being and unity rather than to being and
unity themselves, although this does not significantly affect the overall interpretation.

Aristotle distinguishes his treatment of these objects from that of

dialecticians in two ways. First, the dialecticians do not recognize the
priority of substance (1004b8-10). It is easy to see the significance of their
failure to recognize this priority if, as we have suggested, substance is to
serve as the subject genus to which these attributes are ' and
. The dialecticians were arguing merely from 'reputable opinions'
(, 995b23-5). As Topics 11 puts it, dialectic argues from reputable
opinions, demonstration from premises that are 'primary and true'
(100a26-30). If substance is prior to these attributes as their subject genus,
it should be possible to reach conclusions about the ways in which
Socrates is the same and not the same as Socrates sitting by relying not
only on reputable opinions but on the sorts of premises required in a
scientific demonstration.
Aristotle goes on to draw a different contrast between metaphysics
and dialectic, this time emphasizing the difference in their powers:
Dialecticians and sophists put on the same appearance as the philosopher. Sophistry only appears to be wisdom. Dialecticians discuss all
things, and being is common to everything; but clearly dialectic embraces these things because they are proper to philosophy. Sophistry
and dialectic busy themselves with the same genus of things as philosophy, but philosophy differs from dialectic in the manner of its capacity,
and from sophistry in the kind of life chosen. Dialectic is such as to
probe [] concerning things that philosophy knows [], sophistry makes the appearing of knowing without knowing.
(WQ4bl7-26)x
Aristotle is clearly attributing to philosophy the capacity of knowing

truths in a way that dialectic does not: philosophy is . What
does Aristotle mean by calling dialectic ?
On one account, Aristotle is referring to Platonic dialectic: Platonic
dialectic is 'peirastic' and subscientific whereas Aristotelian dialectic can
be properly scientific.31 This interpretation seems unlikely. As Bolton has
emphasized, the label 'peirastic' is not intended as a pejorative but rather
30 Kirwan makes the most sense of the obscure sentence at b!7-22, and my translation
reflects his.
31 Berti (1996,125-7)

96 Ian Bell
to indicate both the limits and the philosophical uses of dialectic. Dialectic does have a genuine function within philosophy in the examination
and refutation of alternative philosophical positions. In particular, it is
possible to use dialectic to refute an opponent's position by showing
either that it is self-contradictory or that it is incompatible with what is
commonly accepted as scientifically true.32 Clear examples of such arguments may be found in Aristotle's criticisms of his predecessors' views
in Metaph I and III. These arguments do not, however, establish anything
except in relation to the views they are criticizing; they do not have the
capacity (, 1004b24) for establishing philosophical truths.33
What kind of arguments do have this capacity? Irwin suggests that
Aristotle is using the passage to introduce a distinction between 'pure'
and 'strong' dialectic. The former relies on generally-accepted beliefs for
its premises, whereas the latter operates from a more restricted, more
epistemologically reliable set of premises.34 In particular, strong dialectic
takes its starting point from the conditions for the possibility of scientific
knowledge. The possibility of scientific knowledge is not just any common belief; it is a belief that all scientific investigation presupposes.
Therefore, it counts among the restricted set of premises appropriate for
arguments in strong dialectic.
As we have seen, Irwin uses Aristotle's treatment of the common
axioms, particularly the principle of noncontradiction (PNC), as a model
for strong dialectic. Although it is prima facie plausible to regard Aristotle's argument in Metaph IV 4 as an example of strong dialectic, it is
difficult to see how 1004b25-6 could be pointing to strong dialectic in the
context of the Metaph IV 2 texts we have considered.35 Irwin suggests that
we should read Aristotle's insistence on the priority of (b8-10) as
a statement that metaphysics should study the 'presuppositions of the
32 See SE 8,11; Bolton (1990), 212-19, (1994), 327-30.

33 For passages contrasting dialectic and philosophy see Irwin (1988), 528nl. On the
relation between dialectic and philosophy see further Bolton (1987), 146-51, Smith
(1993), Lennox (1994), 58-64.
34 Irwin (1988), 174-5,185-7
35 I have argued in Bell (1998), 113-28, that Metaph IV 3-8 is not attempting a justification of the common axioms using strong dialectic, and that there too Aristotle's use
of dialectic is peirastic. But there is at least a pnma facie case for construing these
arguments as Irwin does.
special sciences' (176), but it is hard to see why Aristotle should be making this point in this context. There is no suggestion in Metaph IV 2 that
Aristotle is considering sameness, difference and so forth because they
are presupposed by the other sciences. Aristotle's stated reason is rather
what we should expect from our account of the tasks of metaphysics:
sameness, difference, and so forth are to being and unity and so
belong to the science of being and unity.
Metaph IV 2 and Metaph IV 3-8 are responses to different aporias that
introduce what appear to be two different tasks. The second aporia
makes it clear that the methodology for the treatment of the common
axioms is nondemonstrative. However, it is equally clear from the fifth
aporia that Aristotle expects a treatment of per se attributes to be demonstrative, and the Metaph IV 2 passages we have examined show that he
expects metaphysics to undertake an investigation of certain per se
attributes. Aristotle has already contrasted the epistemic reliability of
dialectic and demonstration in the Posterior Analytics (16,74b21-6 with I
19, 8lbl8-23).36 In its context, then, the Metaphysics contrast between
and is most naturally taken not as distinguishing
between pure and strong dialectic, but between dialectic and demonstration. We shall provide additional evidence to support the view that this
is the contrast Aristotle intends in the next section.
Aristotle finishes Metaph IV 2 with a brief discussion of the relation of
the contraries to the science of being qua being. The principles of all
things are contraries, and the principles of the contraries are unity and
plurality (1004b27-5a5). Hence, since the science of being qua being is
also the science of unity, this science will also be the science of the
contraries (1005a2-5). The passage seems to be developed mainly out of
a Platonic and pre-Socratic background (cf. 1004b31-3). Aristotle's emphasis on the contraries as principles is somewhat worrying, since in a
few places he seems to be on the verge of suggesting that the contraries
are principles even of (and thus are prior to) . At one point Aristotle
attributes the view that all beings and are composed of contraries
to 'nearly all thinkers' (1004b28-9); at another point he himself writes that
'all things' are either contraries or composed of contraries ( ,
1005a3-4). Aristotle is clearly concerned to express the continuity of his
36 As Irwin seems to recognize (1988), 470, there is no clear evidence for a distinction
between pure and strong dialectic anywhere
in the corpus.
98 Ian Bell
science of being with the investigations of his predecessors, which may

account for the appearance of being willing to reduce even substance to
the contraries.37 His indication that being and one are probably not '
(1005a5-ll), however, suggests that here too Aristotle ultimately
intends a reduction of the basic contraries (being and not-being,
unity and plurality) to substance (cf. Metaph XII7,1072a30-2). Being and
not-being are not treated as equals, nor are they principles of : in
the reduction being in its primary instance simply is , and
not-being is nothing more than a privation of , called a 'being' only
derivatively and in the weakest possible sense (IV 2,1003blO).
The chapter concludes with an unqualifiedly affirmative answer to
the fifth aporia. The science of being is a science both of substance and
its per se attributes.
It is clear, then, that it belongs to one science to investigate being qua
being and whatever belongs to it qua being, and that the same science
investigates not only substances, but also whatever belongs to substances [ ], both the
[attributes] mentioned and also priority and posteriority, genus and
species, whole and part, and the others of this sort. (1005bl3-18)
The introduction of priority, posteriority, and so forth in addition to the

topics already mentioned is unexpected, though Aristotle has already
mentioned priority and posteriority 'and all other such' in the first
exposition of the fifth aporia at Metaph III 1, 995b22-3. Although it is
prima facie plausible that such things are also considered per se attributes of being qua being, it is impossible to determine this just from the
text.
37 Apostle (1969), commentary 38 ad loc., suggests that this passage is dialectical.

Likewise, Kirvvan (19932), 85, suggests that the argument is ad hominem against his
predecessors.
In examining Metaph IV 2 we have shown that Aristotle aims to introduce

scientific rigour to the treatment of certain traditional objects of Academic dialectic by including them in the science of being qua being. In
particular, Aristotle insists that these things be treated as attributes that
are ' and to being and unity, and so as per se attributes
of , which is prior to them as their subject genus. It is possible to
shed further light on the methodology proposed for this investigation
by considering the proof in Metaph X 4 that each contrary has only one
contrary. When Aristotle raises the fifth aporia (III 1, 995b25-7), the
question whether each contrary has only one contrary is his example of
the problems involving ' whose treatment might
be supposed to belong to metaphysics; the example reappears in Aristotle's solution to the aporia at Metaph IV 2,1004b3. Aristotle's proof for
this example is thus our best indication of what he intends the methodology introduced in Metaph TV 2 to look like in practice.
The proof begins with a statement of the and of contrariety:
Since things which differ from one another may do so to a greater or
lesser degree, there exists also a greatest difference, and this I call
"contrariety". That contrariety is the greatest difference is clear by
induction [ ]. (105533-6)38
Aristotle goes on to show this by appeal to the extremes from which

generations take place. From his discussion he also concludes that contrariety is 'complete difference' (1055al6). Aristotle thus establishes both
the existence and definition of subject genus, making use of the Posterior
Analytics methodology of induction to establish the definition.39 Having
set down the and of the subject genus, Aristotle goes on to
prove a fact () that is true of it:
This being so, it is evident that each contrary cannot have more than
one contrary; for (1) neither can there be anything more extreme than
the extreme, nor can there be more than two extremes for one interval.
38 With the ROT I take im ' to mean 'that <contrariety> is the

greatest difference' rather than 'that there is a greatest difference' (Apostle). For
other mentions of induction in book X see 1054b33,1055bl7, and 1058a9.
39 Cf APo 110,76a31-6, b3-22, and APo 19 generally.

100 Ian Bell
And (2) in general, if contrariety is a difference, and difference is

between two things, then complete difference will also be between two
things. (1055al9-23)
Both proofs start from an element of the definition of contrariety to

establish a conclusion about it. Moreover, the definition is the cause of
the fact that is proved to hold of it. The fact that there can be only one
contrary to a contrary follows from the definitions of 'extreme' (which
carries the implication of superlativeness and thus singularity) and 'difference' (which carries the implication of duality rather than plurality).40
How closely does the proof follow the model of demonstration presented in the Analytics? The requirements of APo I 7 and 110 appear to
be met: the proof explicitly uses the definition of contrariety and implicitly assumes the existence of contrariety to prove the of an attribute
of contrariety. An appeal to the definition of unity is implicit in the
statements that there cannot be more than two extremes and that difference is between two things. In this respect the proof may usefully be
contrasted with Plato's dialectical treatment of the same question in the
Protagoras (332a ff.), which does not start from a but uses dialectical induction.
What is more difficult to see is how this proof could be formalized
into a Prior Analytics syllogism. A version of the first argument might be
quasi-formalized as follows:
The relation between contraries is (by definition) one of
complete difference.
All relations of complete difference are such that the relata
are extremes.
Therefore, all contraries are extremes.
But, all extremes are such that there cannot be more than
two for an interval.
Therefore, all contraries are such that there cannot be more than
two for an interval (which is to say that every contrary has
only one contrary).
40 The ' relations seem to be type (1) rather than type (2): the conclusion is
wholly implicit in the definition of contrariety.
Whatever this is, it is not a chain of idiomatic Prior Analytics syllogisms,

nor is it clear that it could be further formalized to give this result. Either
Aristotle is recommending a methodology that is very close to demonstration yet ultimately not demonstration, or the Metaphysics is working
with a conception of demonstration less rigid than that implied by the
two Analytics. The available texts do not allow us to answer the question
with certainty. I think, however, that the balance of evidence is in favour
of the latter conclusion.
The most significant pieces of evidence are texts in the Metaphysics
and de Anima that is, in works thought to be mature that recommend or assume demonstration as the proof-structure for per se attributes.41 An important short text is Aristotle's remark, made in the course
of inquiring whether there is a single method for the study of , that
there is a single method, demonstration, for the study of its per se
attributes ( , 402al5). Two more extended texts
emphasize the APo I 7/110 criteria for demonstrations: demonstrations
show per se attributes to belong to a subject using as a principle.
As we have seen, Aristotle writes in Metaph VI1 that 'starting from the
,... [sciences]42 proceed to demonstrate more or less rigorously the
' of their genus' (1025bl2-13). The same connection
between demonstration, definition, and the per se attributes of a substance is drawn in Aristotle's preliminary remarks on the definition of
the soul at DA 11,402bl6-3a2.
There is no suggestion in any of the texts we have considered that
Aristotle means to introduce some methodology distinct from demonstration as described in the Posterior Analytics. Aristotle is not hesitant to
distinguish his methodology from demonstration when he thinks the
latter to be inappropriate for the task at hand. We have already seen that
Aristotle does something of the sort for the principles of being in Metaph
VI1. Likewise, in PA 11 and Metaph IV 4 Aristotle introduces demonstration through hypothetical necessity (639b30-40a6) and elenctic dem-
41 See Metaph 992b30-3, 997al7-22, a30-6, 1025012-13, 1039b31^0a2, 1077020-2,

1086b34,1087a23; DA 402al5,402bl6-3a2.
42 As we mentioned, the passage is problematic in appearing to suggest a contrast

between metaphysics and the other sciences on this count. We shall consider this
problem in the next section.
102 Ian Bell

onstration (1006a5-12), and explicitly distinguishes them from Analyticsstyle demonstration.
What Metaph IV 2 and X 4 rather appear to attempt is to set up and
use a proof-structure that is as close as possible to Posterior Analytics-style
demonstration, and that possesses at least the characteristics associated
with demonstrations in our Metaph VI 1 and DA I 1 passages. The
example in Metaph X 4 is as close to a demonstration as anything else in
the corpus: examples of demonstrations in strict syllogistic form are not
more easily found outside the Metaphysics than within.*3 Thus if there is
anything in the corpus that Aristotle would regard as a demonstration
as his remarks imply there should be this argument should be an
instance of a demonstration.
If so, Aristotle's criteria for demonstration in practice are less rigid
than those imposed by the two Analytics. In particular, Aristotle is not
scrupulous about presenting demonstrations in the form of a syllogism.
Perhaps Aristotle thought (rightly or wrongly) that arguments presented
relatively informally in his scientific writings could be formalized as Prior
Analytics syllogisms. William Wians, for instance, argues that the Posterior Analytics 'gives a formal description of scientific practices which may
themselves remain informal in their patterns of argument.'44 It is also
controversial whether Aristotle always regarded the Prior Analytics account of the syllogism as describing the only valid form of inference that
could serve as the 'infrastructure' for demonstrations. Although the word
appears in various places in the Metaphysics, none of Aristotle's mentions of demonstration in the Metaphysics and de Anima (note
41 above) imply that demonstrations must take the form set out in the
Prior Analytics. Another possible sign of increasing distance from the
Prior Analytics is Aristotle's remark in Metaph VI1 that one may demonstrate per se attributes more or less rigorously ( , 1025bl3).
The other evidence in Metaph X is not as clear as our Metaph X 4 proof,
although Aristotle appears to be attempting similar kinds of proofs in
Metaph X 5 in his treatment of the relations between being greater than,
being less than, and being equal, and with the argument that intermedi-
43 In his 1969 Barnes claimed to be unable to find a single example anywhere in the
corpus; to my knowledge no one has convincingly identified one since.
44 See Wians (1989), 251; cf. Lear (1980), 10-11,Brought

Gotthelfto(1987),
you by195-6.
| St. Petersburg State Univers
ates are composed of contraries in chapter 7. Much of book X is concerned with establishing the definitions and hypotheses that are preconditions for demonstration. Thus chapters 1-2 define unity and chapters
3-4 those things that are defined in terms of unity. Chapter 8 defines
difference in species. Wians points to texts that suggest that once the
explanatory structure for a science has been determined and the principles of demonstrations established, the business of constructing demonstrations on the basis of these principles is relatively straightforward.45
If so, the absence of many written out proofs may not be surprising.
VI
It is now possible to clarify some details of Aristotle's argument in

Metaph IV 2 in light of our account of its project as a whole. Let us return
to the difficult passage at the beginning of the text.
For each genus of things there is both one power of sensation and one
science; grammar, for instance, which is one science, investigates all
spoken sounds [ ]. Accordingly, it belongs to one
generic science to investigate all the forms [] of being, and the forms
of the forms. (1003bl9-22)
One of the issues raised by the text was the translation of ' .
Many translators render this as 'species' of being. Given Aristotle's
ultimate identification of the ' with the per se attributes of being and
unity, it seems clear that these cannot be 'species' in any traditional
Aristotelian sense. This suggests that Owens and Kirwan are right to
translate as 'forms' rather than as 'species'. For the same reason, it
seems best to follow Owens and Kirwan on their reading of '
, rather than the expansive reading that attributes to Aristotle
the view that being and metaphysics have species.46 The phrase is better
read as referring to things such as contrariety, which are denned not
45 Wians (1989), 249-50, citing APo 130,46a22-7, and NE17,1098a20-6.
46 See note 26 above. For additional textual arguments for this interpretation see
Mansion (1958), 185-9.
104 Ian Bell
directly in terms of being and unity but rather in terms of something itself
defined in terms of being and unity, in this case difference.47
Why would Aristotle introduce the per se attributes as 'forms' of
being and unity? Owens and Berti see a reference to the Forms discussed
in Plato's Parmenides and Sophist.*8 Although this is dependent on controversial interpretations of these dialogues, these scholars must be right
in seeing some sort of reference to Platonic dialectic. As we have seen,
the fifth aporia is introduced in Metaph III 1 as raising two different
questions: Does the science of substance also shady the per se attributes
of substance? and, Does the science of substance study certain topics now
addressed by the dialecticians using arguments from ? The objects
of investigation introduced in Metaph IV 2 are not new: they are already
objects for dialectical study in the Academy. The Analytics makes it clear
that Aristotle believes that no merely dialectical treatment of a problem
can be scientific (APo 16,74b21-6 with 119,81bl8-23). Aristotle's position
against the dialecticians is greatly strengthened if he can show that a
treatment of these traditional objects of dialectic is possible using his
methodology. If such a treatment is possible, these objects must themselves be subjects with demonstrable attributes or demonstrable attributes of some other subject. Aristotle makes them directly or indirectly
attributes of substance as the primary instance of being, thereby also
underlining the priority of substance in his metaphysics.
Aristotle nowhere explains how the confident conclusion to Metaph
TV 2 is possible in light of the objections to a demonstrative science of
substance raised in the fifth aporia and in Metaph VI1. It is possible to
suggest a solution along the following lines. Metaph VI 1 is concerned
with the principles and causes of beings qua beings (1025b3-4). The
reduction of being to substance requires that this be an investigation
into the principles and causes of substances qua substances (IV 2,
1003bl6-19). Every substance has causes that are studied in metaphysics,
just by virtue of the fact that it is a substance. At Metaph VI1,1025blO-18
this characterization is given further precision: metaphysics inquires into
the and ei of substances. This investigation is to be distinguished from the special sciences, in which the of the subject is
47 See p. 91 above.
48 Owens (19783), 275-7 and notes, Berti (1996), 124-8; cf. de Strycker (1979), 53-7. This
would explain the absence of any other Aristotelian
of the
' .
Broughtuses
to you
by phrase
| St. Petersburg
State Univers
assumed and only the of the per se attributes is demonstrated and

thus explained (1025blO-14). The mere fact that an attribute exists (that
is, that it belongs to a subject) does not require explanation in the science
of being: it is explained in the science of its subject.49 The moral of Metaph
IV 2, however, is that there are certain attributes whose subject is being
and substance. These attributes will therefore be studied in the science
of being and substance. As an inquiry into substance, metaphysics is
distinguished from the other sciences by the fact that it provides an
explanation of the subject, whereas other sciences must assume the
existence of the subject to demonstrate its attributes. If substantiality it self
has attributes per se to it, however, metaphysics should also undertake
the investigation of these attributes. Otherwise put, the aim of Metaph VI
1 is to show not that metaphysics cannot have tasks whose proof-structure is demonstrative, but rather that it can undertake a task that the
other sciences cannot: an inquiry into subjecthood and substantiality.
VII
We have argued that Metaphysics TV 2 is intended to establish one of the

tasks of the science of being as a demonstrative treatment of the per se
attributes of being qua being. If this conclusion is correct, then one may
make a strong claim for the influence of the Posterior Analytics on the
Metaphysics. It is worth pointing out that this is not the only way in which
the Analytics can be seen to influence the Metaphysics. In particular, the
relation Aristotle uses to connect the tasks of metaphysics is itself
due to the Posterior Analytics. Even if one does not accept our conclusion
that Aristotle intends a precisely demonstrative proof structure for the
per se attributes of being and substance, the influence of the Posterior
Analytics is nevertheless in evidence in a proof-structure that resembles
demonstration in crucial respects. We have also suggested along the way
that the methodologies of the nondemonstrative tasks of metaphysics,
while going beyond anything presented in the Posterior Analytics, nevertheless do not represent a simple repudiation of the Analytics.
If we are right about Aristotle's project in Metaph IV 2, then it seems
that certain influential views about the character of the Metaphysics are
49 APo 124,85b23-7, Metaph VI1,1025b7-14

106 Ian Bell
incorrect. In particular, it is wrong to distinguish metaphysics from the

other sciences as a nondemonstrative science of principles as opposed to
a demonstrative science of per se attributes. The distinguishing character
of metaphysics is not a nondemonstrative proof-structure, still less a
repudiation of the Analytics generally, but the relation of its objects
to being and substance. While metaphysics' investigations into the principles of substance and the principles of demonstration will not be
amenable to a demonstrative methodology, the lesson of Aristotle's
solution to the fifth aporia is that the science of being must also include
a demonstrative treatment of whatever attributes belong per se and
to being, unity, and their primary instance in substance.50
372 Melita Avenue
Toronto, ON
M6G 2A3
ian.bell@utoronto.ca
Works Cited
Anagnostopoulos, G. 1994. Anstotle on the Goals and Exactness of Ethics. Berkeley and Los
Angeles: University of California Press.
Apostle, H.G., trans. 1966. Anslotk's Metaphysics. Grinnell. Peripatetic Press
1982. Aristotle's Posterior Analytics. Gnnnell: Peripatetic Press.
Aubenque, P, ed. 1979. Etudes sur la Metaphysique d'Anstote. Paris.
Barnes, J. 1969. 'Aristotle's Theory of Demonstration,' Phronesis 14:123-52
1981. 'Proof and the Syllogism,' in Berti (1981).
Bell, I 1998. 'Metaphysics as an Aristotelian Science.' Ph.D. dissertation, University of
Toronto.
Berti, E., ed. 1981. Aristotle on Science: The Posterior Analytics. Padua Editrice Antenore.
1996. 'Does Aristotle's Dialectic Develop?' in Wians (1996), 105-30.
50 I am grateful to Lloyd Gerson, Stephen Dumont, Brad Inwood, Michael Ferejohn,

and several anonymous reviewers for helpful comments and criticisms. The material in this article was originally researched and written while I was the recipient of
a Social Sciences and Humanities Research Brought
Council Doctoral
Fellowship.
Demonstration in Anstotle's Metaphysics 107
Bolton, R. 1987. 'Definition and Scientific Method in Aristotle's Postenor Analytics and
Generation of Animals/ in Gotthelf and Lennox (1987)
1990 "The Epistemological Basis of Aristotelian Dialectic,' in Devereux and Pellegrin (1990)
1991. 'Aristotle's Method in Natural Science/ in Judson (1991).
1994. 'Aristotle's Conception of Metaphysics as a Science/ in Scaltsas (1994), 321-54.
1995 'Science and the Science of Substance in Aristotle's Metaphysics Z.' Pacific
Philosophical Quarterly 76: 419-69
Code, A. 1986. 'Aristotle's Investigation of a Basic Logical Principle: Which Science
Investigates the Principle of Non-Contradiction?' Canadian Journal of Philosophy 16:
341-58.
de Strycker, E. 1979. 'Notes sur les relations entre la problematique du Sophisle de Platon
et celle de la Metaphysique d'Aristote/ in Aubenque (1979).
Devereux, D., and Pellegrin, P. (eds.). 1990. Biologie, logique, et metaphysique chez Aristote.
Pans: Editions du CRNS
Dhondt, U. 1961. 'Science supreme et ontologie chez Aristote/ Revue Philosophique de
Louvain 59: 5-30.
During, I., and Owen, G.E.L., eds.. 1960. Aristotle and Plato in the Mid-Fourth Century.
Gothenburg.
Ferejohn, M. 1991. The Origins of Aristotelian Science. New Haven: Yale University Press.
Goldin, 0.1996. Explaining an Eclipse. Ann Arbor: University of Michigan Press.
Gotthelf, A. 1987. 'First Principles in Aristotle's Parts of Animals,' in Gotthelf and Lennox
(1987).
and Lennox, J. 1987, eds. Philosophical Issues in Aristotle's Biology. Cambridge:
Cambridge University Press.
Irwin, T.H. 1988. Aristotle's First Principles. Oxford: Clarendon Press.
Jaeger, W., ed. 1957. Aristotelis Metaphysica. Oxford: Clarendon Press.
Judson, L., ed. 1991. Aristotle's Physics. Oxford: Clarendon Press.
Kirwan, C, trans. 19932. Aristotle: Metaphysics Books , , . Oxford: Clarendon Press.
Lear, J. 1980. Aristotle's Logical Theory. Cambridge: Cambridge University Press.
Lennox, J. 1994. 'Aristotelian Problems.' Ancient Philosophy 14:53-77.
Leszl, W 1975. Aristotle's Conception of Ontology. Padua: Editrice Antenore.
Madigan, A., trans. 1993. Alexander of Aphrodisias: On Aristotle Metaphysics 4. London:
Duckworth.
Mansion, A. 1958. 'Philosophie premiere, philosophie seconde, et metaphysique chez

Aristote.' Revue Philosophique de Louvain 56:165-221.
108 Ian Bell
Mansion, S., ed. 1961. Anstote et les problemes de methode. Louvain.

McKirahan, R 1992. Principles and Proofs Princeton Princeton University Press.
Owen, G.E.L. 1960. 'Logic and Metaphysics in some early works of Aristotle/ in During
(1960), 163-90. Reprinted in Owen (1986), 180-9.
1961. " / in Mansion (1961), 83-103. Reprinted in Owen (1986),
239-51.
1986. Logic, Science, and Dialectic. London: Duckworth.
Owens, J. 1978 . The Doctrine of Being in the Aristotelian Metaphysics. Toronto: Pontifical
Institute of Medieval Studies.
Ross, W.D. 1924. Anstotle: Metaphysics. Oxford: Clarendon Press.
Scaltsas, T, and Gill, M.L., eds. 1994. Unity, Identity, and Explanation in Aristotle's Metaphysics. Oxford: Clarendon Press.
Smith, R. 1982a. "The Relationship of Aristotle's Two Analytics,' Classical Quarterly 32 ( ):
327-35.
1982b. 'The Syllogism in Posterior Analytics I/ Archiv f r Geschichte der Philosophie
64 113-35.
1993. 'Aristotle on the Uses of Dialectic/ Synthese 96: 335-58.
Stevenson, J. 1975. 'Being qua Being/ Apeiron 9: 42-50.
Wians, W. 1989 'Aristotle, Demonstration, and Teaching/ Ancient Philosophy 9 (1989):
245-53.
, ed. 1996. Aristotle's Development: Problems and Perspectives. Lanham: Rowman and
Littlefield.
Witt, C. 1989. Substance and Essence in Aristotle: An Interpretation of Metaphysics V1I-IX.
Ithaca- Cornell University Press.


Bell, Ian - Demonstration in Aristotle's Metaphysics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bell, Ian - Demonstration in Aristotle's Metaphysics

Uploaded by

Copyright:

Available Formats

Demonstration in Aristotle's

In recent years scholarship in Aristotle has focused extensively on the

4 I use 'proof-structure' specifically to refer to the part of a discipline's methodology

Demonstration in Aristotle's Metaphysics 77

Metaph IV 2 would lead us to expect to be demonstrative the proof

Broadly speaking, demonstrations prove the existence of effects

6 Ferejohn (1991), chs. 3-6, McKirahan (1992),Brought

Demonstration in Aristotle's Metaphysics 79

7 Pace Ross (1924), 1.224

Brought to you by | St. Petersburg State Univers

which it belongs (74a2). As Aristotle emphasizes at greater length in APo

Aristotle introduces metaphysics in Metaph TV 1 with the statement that

8 Cf. Stevenson (1975).

Brought to you by | St. Petersburg State Univers

Demonstration in Aristotle's Metaphysics 81

Aristotle's references to the ' (a22) should refer to

Demonstration in Aristotle's Metaphysics 83

that Aristotle regards the treatment of these attributes as demonstrative,

Demonstration in Aristotle's Metaphysics 85

17 With Code (1986), 344,1 incline to the latter Brought

18 Cf. Bolton (1995), 454-5.

Demonstration in Aristotle's Metaphysics 87

]. It is like everything that is called "healthy",

If we are to follow the model of the other examples of , we should

Brought to you by | St. Petersburg State Univers

Demonstration in Aristotle's Metaphysics 89

If the nature of being is found only in substance, then the primary

The treatment of the per se attributes of being in Metaph IV 2 follows

24 Our translation differs significantly from Apostle's, on the difficulties of translation

Brought to you by | St. Petersburg State Univers

Demonstration in Aristotle's Metaphysics 91

diately follow it (1003b22-1004al and 1004a9-b32). We shall have a better

Additionally, contrariety itself is investigated by this science, being an

27 I follow Ross (1924), 1.206, on the translation of .

Brought to you by | St. Petersburg State Univers

Demonstration in Aristotle's Metaphysics 93

Since, then, these are the per se attributes [1 ... ] of unity

not considering , of which they comprehend nothing, as prior. For

Brought to you by | St. Petersburg State Univers

Demonstration in Aristotle's Metaphysics 95

Aristotle distinguishes his treatment of these objects from that of

Aristotle is clearly attributing to philosophy the capacity of knowing

Brought to you by | St. Petersburg State Univers

32 See SE 8,11; Bolton (1990), 212-19, (1994), 327-30.

Demonstration in Aristotle's Metaphysics 97

science of being with the investigations of his predecessors, which may

The introduction of priority, posteriority, and so forth in addition to the

37 Apostle (1969), commentary 38 ad loc., suggests that this passage is dialectical.

Demonstration in Aristotle's Metaphysics 99

In examining Metaph IV 2 we have shown that Aristotle aims to introduce

Aristotle goes on to show this by appeal to the extremes from which

38 With the ROT I take im ' to mean 'that <contrariety> is the

39 Cf APo 110,76a31-6, b3-22, and APo 19 generally.

100 Ian Bell

And (2) in general, if contrariety is a difference, and difference is

Both proofs start from an element of the definition of contrariety to

Demonstration in Aristotle's Metaphysics 101

Whatever this is, it is not a chain of idiomatic Prior Analytics syllogisms,

41 See Metaph 992b30-3, 997al7-22, a30-6, 1025012-13, 1039b31^0a2, 1077020-2,

42 As we mentioned, the passage is problematic in appearing to suggest a contrast

102 Ian Bell

44 See Wians (1989), 251; cf. Lear (1980), 10-11,Brought

Demonstration in Aristotle's Metaphysics 103