Vous êtes sur la page 1sur 88


being the record
of a seminar held in London, 1979-1982
This monograph is a sequel to Notes on Zeta, published in 1979
by the Oxford Philosophy sub-Faculty. The London Group, started by
G.E.L. Owen in 1976, and described by Christopher Kirwan in the preface
to Notes on Zeta, has continued its discussions of the text of Aris-
totle's Metaphysics in the Institute of Classical Studies in Gordon
Square, and we were encouraged, by reviewers of Notes on Zeta among
others, to publish the record of our discussions of Books Eta and Theta.
The form in which the material is presented is much the same as
before. No major revisions have been made; but I have rearranged
some of the material so as to bring together all the discussions of
a given passage of the text (and accordingly deleted references to
the dates on which the sessions occurred), and tried to achieve some
consistency of presentation; but some inconsistencies remain - for
example in the transliteration of Greek words.
The majority of the minutes of sessions are the work of Myles
Burnyeat, and a substantial number of others are by Bob Sharplesi others
were recorded by Lesley Brown and Alan Lacey. Apart from those per-
sons, the meetings were attended fairly regularly by Julia Annas, Bob
Heinaman, Gerald Hughes, Christopher Kirwan, Jonathan Lear, Geoffrey
Lloyd, Malcolm Scholfield, Richard Sorabji, Julius Tomin, Kathleen
Wilkes, and Michael Woodsi and most of them were presided over by Gwilym
A focus to our discussions was given by some characteristically
incisive and challenging Introductory Notes on individual chapters
circulated or tabled by Gwilym Owen. These have been included in
this Monograph in the appropriate place, as have been contributions
by Bob Heinaman, Richard Sorabji and Bob Sharples. There is also
included a paper read to one session by Sarah Waterlow, though, of
course, the full development of her ideas on the subject of the paper
can now be found in her Passage and Possibility (Oxford, 1982).
The meetings of the Group that discussed these two books of the
Metaphysics took place between May
our discussions of Book Theta were
1979 and November 1982. Thus,
almost complete when Gwilym Owen
died in July 1982. We should like to dedicate this Monograph to his
memory. The debt to him, as the person who established the London
Group and presided over it for seven years is only one of many that
we, like so many other Aristotelian scholars, owe to him.
The works of Aristotle are sometimes referred to by the following
An. Post. or A. PQ.
An. Prior
De Gen. An. or GA
De Gen. et Corr. or G&C
De Int.
De Mem.
Parv. Nat.
So ph. El. or SE
Posterior Analytics
Prior Analytics
De Generatione Animalium
De Generatione et Corruptione
De Interpretatione
De Motu Animalium
De Memoria
Eudemian Ethics
Nicomachean Ethics
Meteorologic a
De Partibus Animalium
Parva Naturalia
De Sophisticis Elenchis
Capital Greek letters refer to books of the Metaphysics unless
otherwise specified.
Unprefixed page numbers, as in 'l019a 10' refer to the Metaphysics.
Other references:
Aristotle's Categories and De-
Interpretatione, translated with
notes by J.L. Ackrill, Oxford,
Aristotle's Metaphysics, trans-
lated with commentaries by Hippo-
crates G. Apostle, Indiana, 1966.
Index Aristotelicus, H. Bonitz,
Berlin 1870.
Aristotelis netaphysica, H. Bo-
nitz, Bonn 1848-9.
Oxford translation
Ps. Alexander
The convention has usually been
Greek words without inverted commas.
Fragmente der Vorsokratiker, ed.
H. Diels and W. Kranz, Berlin.
Aristotelis Metaphysica, W.Jaeger,
Oxford Classical Texts, 1957.
Aristotle, Metaphysics
translated with notes by Christo-
pher Kirwan, Oxford 1970.
Volume VIII (Metaphysica, trans-
lated W.D. Ross) in The Works of
Aristotle translated into English,
Oxford 1928.
Commentary on Z (in fact by a
later hand) in Alexander of Aphro-
disias, In Aristotelis Metaphy-
sica Commentaria, ed. M. Hayduck,
Berlin 1891.
Aristotelis, la Metafisica, tra-
duzione, introduzione e commento,
Giovanni Reale, Loffredo 1968.
Aristotle's Metaphysics, text and
commentary, W.D. Ross, Oxford 1924.
followed of writing mentioned
l042a 3-24 The first sentence as placed strongly suggests we are
to have a summary of Z, but the next jolts our expectations. Where
was it e[pT'J'ta.l. that the object of enquiry is the causes and principles
and elements of substance? Not Zl Ross). Not El (Apostle),
which seeks principles and causes of 'tciiv Ov-rwv as including not just
substances but everything. r l-2 (cf. l003b 18)? A 1-2 fits better
still, offering several parallels to what is to come in Hl.
Further difficulties: (1) 1C42a 6-10 goes against Zl6 on parts
and the elements ('agreed by all' might mean 'agreed by all but the
speaker' but l042a 24 resumes talk of substances as if the list
had not contained controversial items). (2) It seems remarkably
bland to set 'tC fiv eTvcu. and side by side as cases of sub-
stance which are established by argument I which are arrived at by
consideration of what people will say under dialectical pressure.
It is going back to where we started out at the beginning of Z3, before
the hard work of Z was done. Still worse (3) to conjoin these with
genus and universal. If the latter also are cases of substance esta-
blished by argument ( it is not by Ar I 5 arguments in z, nor
even by his opponents
arguments in Z. For these do not urge that
genus is more substance than roo,, universal more than particular
( 1038b 7 is the nearest parallel but not enough). You really
have to go back to B for that line of argument.
sible to take xa.e6Xou "= yfvo,, 1:Wv
15 advance just one case of substance, not two.
[It was thought pos-
c!6wv so that 14-
If, however, two
cases are intended, there is a problem as to how a single argument
can yield both: e.g. definability selects universal over particular,
plus Forms (1. 16), but not genus over eT6o<;. Further, the two case
reading would break with the narrower use of x.ae6Aou as found in Zl3.]
More generally, (4) nothing is said to recall the challenge to Uxox(-
f..l.vov as substance (27-8 blandly accepts UXn with a justification
in terms of the actual/potential distinction to which again nothing
in Z corresponds). Nothing recalls Z 7-9. The hard work on essence
h"as disappeared from memory. The conclusion of Zl7 has gone for nought.
! 042a 3
But now, having got to tile point of being ready to consider chat
1042a 3-24 may not be a summary of Z (we even tried, without plausible
success, to make 1042a J-4 look ahead rather than back), one meets
the unmistakable backreferences L!.WpL<T't"aL (1. 18), 7!;Ept )J.lpout;; ?iv
!6etv (l. 20 ) . The second is especially telling because nowhere
else in the corpus is there anything like the discussion in Zl0-11.
Again, 21-2, denying that universal or genus is substance, fits 213.
So what we seem to have is (a) a summary of Z which (b) is not
the sort of summary that a careful reader of Z would expect. Possible
conclusions: (1) the summary is an connecting work (Andro-
nicus is known to have indulged in such). (2) There was a proto-
z without e.g. the critique of which adhered more closely
than our z to the 'keep all candidates but universal in play' line
'../hich predominates in H 1 (cp. the Hay 24 starts as if essence was
merely the next candidate on a list of equals, without recognition
of Z3's elimination of b1toxECI-1VOv/\SAT')). Problem: '.Yhat is then left
for proto-Z to C'1ntain? (3) Proto-Z = H: the editor got H and Z
in the -wrong order, put the hard work before the soft, and spliced
in the patchwork connection. ( 4) Z 'NBS essentially designed to remove
onesided overemphasizing of various candidates for substance, and that
done Ar leads off again on a positive note, not so much summarizing
z as reformulating the position he wants to start from after Z.
Note that some of these suggestions would have the consequence
that l042a 3-24 is no longer available as evidence for the 'be fair
to all candidates' interpretation of Z.
l042a 31-b 8 The argument is: is olxYCa. because it is
as shown by a I type examination of change. (Phys. I type be-
cause tiXXoCu.x:r1.<; is the model, not Ar's alternative model of motion,
which concentrates on the continuity of and is free of the
existential worries pressed by Elea.) Note that Wt;;
is not redundant: the subject in question is not the bronze
as such but the bronze as unshaped.
r"heir previous roles.
vUv - in 1042b 2-3 switch
fhe thesis at 1042 b 3-4 is not that substantial change is presup-
posed by the others, which would contradict 5-6, but that it entails
the others, i.e. any substance which comes to be is liable to the other
3 types of change.
Surprisingly perhaps, this thesis looked to be
Attention was called to M 1076a 8-10, where Ar says that the sub-
stance of perceptible things has been explained in two stages: matter
in the Physics, substance as actuality later ( OOn:pov ).

refers later to ZH9, but wl}ile a concern with substance as actuality
is the mark of H, it is conspicuously not a mark of z.
On the other
hand, Ml fits well as looking back to Hl (which itself refers forward
to M at 1042a 22-3), while HI in turn looks back to (1042b 8).
Could there have been at some time a course which went from (some of)
via H (e) to M?
The difficulty we had in relating Hl to z
would then be due to Z having been grafted on to H, with the help of
the one bit of decent Z-summary provided in Hl, viz. 1042a 17-22. (Fur-
ther evidence that Hl is patched together might be seen in the fact
that within the space of a few lines we are in effect twice (a 11-
12, 23-4) given the information that some people hold that Forms and
mathematicals are substances.)
The objection that H3 1043b 16-18
refers back to Z7-9 was met with the reply (a)
occurs in a passage usually thought to be highly
that the reference
parenthetical, (b)
that Z 1-9 are anyway to be regarded as having been pressed into z from
another context (see Notes on Zeta p.54). The question was also raised
whether the fact that 'substance as actuality' is not part of the wor-
king vocabulary of Z would prevent Ar using it to refer to his dis-
cussions in z.
The suggestion was noted rather than accepted, but it seemed to
raise interesting questions about how an Aristotelian 'course' should
be conceived.
For the more patching together of material we find,
the stronger the presumption that Ar is his own tailor.
1042b 5 6uot'v : what other change is such that yEverrL<:;;/tpBopci. tloes not
follow it?
None in Ar's scheme of things, but the rarefaction and
condensation of Anaximenes' air would serve.
b 7-8 The reference to V. I may be editorial, but perhaps a
I 042a Jl
1!)42b 7 ;!OfES )N ETA
linking to the might alleviate some of the pro h) ems of linking
Ht to Z.
l042b 9-25 The long list of differentiae at 15 ff. is structured as
follows: under which are subsumed L
(rhough xpC<rLc; usually contrasts ..Jith O""l>v6e:cn.c; - see Ross' note), 6e:cr)..liil
,,, J<).dcx:n 'to6'twv (concluding the subsection) (b) et<rEL, (c) xp6v<Jl
(d) 't6"'J' (e) the last section on xae'Tl 21-s summed up by Ohulc; b-xe:p-
ox;;'!/ lA.\e:C+L,which last therefore refers to the note and the less
(rather than excess defect) in qualities (cf. differentiation by the
and the less in Ar
s biology).
This open-ended list compares with Democritus
three geometrical
d.tfferentiae (unlike Plutarch, Ar makes no mention of the weight of
aroms). dS illustrated by the letters of the alphabet in Met. A4.
These distinguish kinds of atom and atomic .:trrangments, so only indi-
rectly microscopic types of thing, this he entirely relevant
to Ar
s discussion if he is concerned with something closer to real
rhan to nominal (linguistic) definition. is it part of the ordi-
nary speaker's notion of a book that it is constituted by gluing
sheets to one another to make a roll? If not, a more theoretical
rype of definition is indeed to the point. Democritus can tell you
of the atomic constitution which makes stone, or which makes something
soft enough to eat, but his story cannot differentiate hetween threshold
and lintel, breakfast and dinner, nor presumably between other, scienti-
'lll y more significant, examples.
l'l42b 25-l043a l Is a6't6 in l7 an objection to the account offered
hy )..;en in 'Aristotle on the Snares of Ontology', to r;he effect that
passage is giving an explanation of existence claims for specimen
.ingular subjects (a particular threshold. etc.)? It is an objection
if, oU66c;; being masculine, o6't6 introduces a different subject, viz.
r he \)).11 . But can pick up a masculine subject, and if Ar
<:HAPTER 2 1042b 25
analyzing what it means to say of a stone that it is a threshold, of
some water that it is ice, etc., he would not need a different
in each case ('Snares' p. 81).
Owen's story requires not only singular subjects but also a tens,ed
to avoid the charge that the analysis makes 'X tautological
(for if 'The ice on the pond is no longer solidified' is not self-
contradictory, 'The ice on the pond is {now] solidified' is not a tau to-
logical analyzans for 'The ice on the pond exists'). Some qualms
were felt about Ar giving no explicit indications either of his subjects'
singularity or of his relying on the present tense. However, the
perfect tense 7t1tuxvUxr6a.L etc., goes some way to ease both difficul-
ties. Ar is generalizing over singular statements such as 'The ice
on the pond exists', not analyzing the general statement 'Some ice
exists'. Thus at l043a 3 is specimen particulars, not species.
At the generalizing level there remains the problem that 'no ice exists
should be contingent, while 'No ice is ice
would appear to be self-
contradictory in a logic which has 'All A is A' as a theorem. But
NB it was a main thesis of 'Snares' that in the present context neither
'The ice on the pond exists' nor the generalization 'Ice exists' is
to be rendered, tenselessly, by the existential quantifier.
l043a 2-7 These lines encapsulate the difficulties of the chapter,
difficulties which come to a head when one inquires into the reference
of (a 3), 1:o61:wv (a 3), 'tOU'tWV (a 4).
Ross translated: 'We must seek in these differentiae (lv
what is the cause of the being of each of these things (-rol>-roov = thres-
hold, etc.). Now none of these differentiae ( ob5v "tol>-rwv is sub-
In the end we preferred this to the alternative of trying
to 'llake all three references to be to the threshold and other examples
of things differentiated by the differentiae. For on the latter rea-
ding abo-Ca. (a 4) means substance in the Cat. sense of primary substance.
and it is hard then to make sense of <Juv6ua1:,61J,vov or -tO &.vtiXoyov
lv tx60""'ttp.
The attempt to_ find an alternative reading to Ross' had been moti-
vated by a worry about its being implausible to have Ar recommend that
we look for substance tbc;; lvtpye:La lv = among the type of dif-
l04Ja 2
' 1hat c11ance 11..1s a tarted-up version
of Democritus of explaining real Aristotelian substances (living organic
things), not merely artefacts and such thin;s as the ice on the pond?
that at l042b 31 hand and foot, which hut for their incompleteness
would be proper substances, only get ln on an abstract promise of 'other
What would it take to redo the argument of 1042b
ll-43a 1 in terms of proper substances? For Ar to turn round in a
4 and say that o66Ev -to6-rwv is substance only seemed to make matters
What, in that case, is the point of 1042b ll-43a l, '>Jhich
must appear something of a digression if all it leads to is a lame
admission that we're still a good way off our goal of discovering the
c:,, lvtpyELC1 of sensible things (1042b 10-11)?
These worries can be alleviated conjointly by a closer under-
standing of the sequence of thought tn l043a 2-7, read in accordance
with the Ross translation. The limitations of the threshold type
of example are already acknowledged a 3: in these differentiae
(tv 't'OU-tat.C:) seek the a.f'tt.OV 'tOU erva.c. of these examples, although
of course (a 4) finding the 'ton of these examples is not
finding substance proper (for these examples are not - in the other
sense nf
'substance' - proper substances). Nevertheless, given that
is at-eCa. 1'oU eTvaL (a 2), finding the a.r-e1.ov 'toU e:!va.L of
these examples is finding what in them is anaJogous to substance (a
It is laying bare a structure which, when transferred to real
substances, will put us on the track, not of these illustrative dif-
ferentiae, but of the differentiae which are the object of our search.
The steps towards this general interpretation go as follows. l042b
l1-43a 1 i.s an essay on <ipx11l o!vC1L (b 32-3), one of its chief
lessons being that there are a good deal more of these than Dernocritus'
three (on the importance of nnmbers, see below). This turns out to
be closely relevant to the theme-question of the chapter, 'What is
f)Ua-Co &>c;; tvtpye:Ln of things?', when we are reminded (a 2)
rhat substance a('tCa. 't'OU e:lva.t. (cf. Zl7). !';ranted that (e;Lx.e:p),
it is clear from the earlier remarks (a 2) that it is to differentiae
!_hat we must look: to these differentiae for the crJ1:1. OV 'tOU erva.L
Jf these examples, to others for that of other examples.
(a 4) the saving qualification: none of these differentiae
are properly substance. (a) 'nor /nor even is
coupling of them', or (b) 'nor/nor even when coupled matter'
(Ross), this latter to be understood either (i) in terms of a particu-
lar bit of matter, or (11) in terms of a sort of matter such as gets
into the definitions of Zl0-11 and those here at 7-11.
does function as a technical term for 'T60e l.v 'tii>Oe at ZS 1030b 16,
3la 6, but it also applies to coupling generally. Against (a} (with
or without 'even') is the consideration that it is hard to see how cou-
pling the illustrative differentiae might be thought to improve the
chance of achieving substance proper. Against (b) (i) is the conside-
ration that it would involve switching mid-sentence to another sense
of to get in a denial that the concrete whole is o60""Ca.. Con-
tinuing then with (b) (ii) (which could, if necessary, bear 'nor even':
water thickened does look somewhat more substantial than thickening
by itself), how do we construe <ivdXoyov !v (a 5)? The sub-
ject is easily got from o()Otv 't'OU't'oov: the differentia in some specimen
case. This is not substance but it is nevertheless, says Ar, what
is analogous to substance (weakened by Bonitz to 'etwas ana loges',
tempting Jaeger to write 1:1. for 1:6, but Ale enforces 't6- see Jaeger's
Next (a 5-7) the analogy is spelled out (no need to be disturbed
by the fact that the case of proper substances is placed first): 'as
in substance that which is predicated of the matter is the actuality
itself, in all other definitions also it is what most resembles full
actuality'. So Ross, but how does he get the idea of approximation
out of (not eLd'XL<r'tC1 but) .-1'XL<r'tC1? It won't do to leave j.jd.'A..L<T"t'a
unsupplemented, for then Ar would be saying that you get the best cases
of actuality in the examples which are not proper substances. So
supplement as follows: in the other definitions (that which is predi-
cated of the matter is) J,!d:Ac.crta. (the actuality itself) ,i.e. as compared
with other elements in the definition it is what is predicated of the
matter which is most of all the actuality. In context this implies
that the item in question is the closest you will get to actuality
without the idea of approximation having to come into the meaning of
To sum up: '"e won't dignify every differentia with the title
0f substance or actuality (the et'?tep clause of a 2 is not convertible),
Jl'C::-J :N t.TA
but since the differentia is the a.ruov tuU el'.la.L , as is shown suffi-
ciently clearly by the threshold type of example, and since the of>crCa.
are seeking, vtz. o6aCa. We;, is a.l"tCa. "taU eTva.L we must
look for of>o-Ca. We; lvE'pyeLa. in the differentia which a definition dis-
plays as predicate of the matter.
1043a 7-11 This doctrine is then exemplified in some specimen defini-
tions, three of which (threshold, house, ice) reaffirm the relevance
of 1042b 11-lda 1. That being so, methodologically no doubt the chap-
ter is an example of Ar progressing from things yvWpqJ.a. +n.itV to things
Proper substances are what we want to understand,
but the structure whereby to understand them is more accessible to
us in familiar, not to say homely, examples like the threshold of our
house and the ice on the road outside.
1043a 12-14 So we come to the main conclusion of the chapter. There
is no one answer to the question 'What is substance as actuality?'
(nor a mere three answers as Democritus supposed), but as many as the
differentiae which our definitions connect with the equally various
types of matter. On the variety of the types of matter, note that
a lO-ll includes the high-low range of sound as matter in the definition
of Ca. But presumably we do not want to stop the same matter
connecting with different differentiae, as e.g. a stone can become
either a threshold or a lintel (l042b 19). Likewise, a given
0f actuality, such as o-6v9cnc; or (a 13), will admit of different
realizations (cf. a 10-11: 1-1n;t.c; ). liA.AT1
5,),.'X.n<;;: at a 12 is thus vague, but 1042b 31-6 shows Ar interested in
a systematic classification of differentiae under their most general

<i.").hll<: is vague in another again. The interpretation
so far defended would not like Ar to state his conclusion in terms
that every definition, including those of proper substances,
<Jill rely on a differentia drawn from the sensible contrarieties and
modes 11f etc., \Ve have been working with. Does 5.\A.o 'tl.
tWv e:lon1Jtvwv (a 14) imply that? [t need not. a l3's yd.p shows
C:HAPTER l 1tJ43a l2
that 13-14 serve to recap the grounds for the main conclusion of 12-
13, and the grounds are indeed to be found in the preceding analysis
of the illustrative examples. We can suppose that ... 'tWv 6' ..
6l gives a summary coverage of all the illustrative examples without
having to suppose that it thereby covers all the examples there are.
General points
side one takes
( 1) All this can be said without prejudice to which
on the question whether the essay on d.pxa.C "tOU eTva.Lat
l042b ll-43a 1 is about existence or the copula. We noted, however,
that the shift to the notion of what something is at 7-11 need be no
embarrassment to Owen's existence story. The conn2:ction between what
it is for a patch of ice to exist and what sort of thing ice as such
is is explicitly drawn at a 2: the olxTC displayed in a definition
of ice precisely is the o('tCa. -ro1'S e!va.L, what you get when you say
what it is for a patch of ice to exist (so 'Snares', p. 82). This
is a substantive thesis, and moreover it is a thesis that has to be
understood, as Ar would understand it, with some appropriate restriction
on the range of terms for which it is claimed true. For it is not
the case that the definition of bachelor tells you what it is for a
bachelor to exist; a bachelor does not cease to exist when he ceases
to be an unmarried man.
(2) The chapter began by throwing the emphasis
on the potentiality-actuality distinction, picking up on H1 1042a 27-
8. If this is to be the route whereby we will make some advance on
Z, or at least, more neutrally, if it is in terms of potentiality and
actuality that H is to make its contribution, one might expect the
distinction to do some work in H2. But does it?
We start off promisingly enough, with out thoughts focussed on
the recipe for honey-water, or what you have to do to bits of papyrus
to make an actual book, or such natural processes as the formation
of ice. We seem to be thinking, by and large, in of the physical
affections or operations which are needed to make matter into a deter-
minate something. Which both makes it reasonable to start from Demo-
critus and holds out the promise that we shall find work for the con-
ll)41a 12
r:ept of actuality which could not be done just as well by the notion
of form or shape. But it was not clear to us, at the end of the day,
that he couldn't have said it all with the matter-form distinction.
The more dynamic aspect of the potentiality-actual! ty distinction has
not - as yet - come into play.
l043a 14-26 A coda to the chapter. That there are two elements is
a definition, one on the side of potentiality and the other on the
side of actuality, is illustrated by the way some people emphasize
one at the expense of the other when defining and others find a place
for both. 66 which is why: it is because there are these two
elements that people define as they do (cf. 19-21).
at 12-13 explains this, and is thereby confirmed.
The conclusion
l043a 21-2 Archytas He is not credited with a theory of definition
(an early theory of definition which recognized both matter and form
would surely have featured prominently in Ar' s surveys of his prede-
cessors). but with accepting certain definitions which had both elements
in them. Are subsequent examples of still weather and a calm Archytan
(so Ross, Commentary p. 229 and DK 47A2)? Or should we think of some-
thing more mathematical such as 'a line is twoness in length' (cf.
1043a 34), which might be represented as an improvement on the traditio-
nal Pythagorean habit of defining things by numbers alone? It is
hard to believe that. if the cited definitions are his, they were of
interest to him for their own sake and not for their connection with
some wider thesis. ( as olxrCa. (a 24) is no worse than many
another Aristotelian casualness, though Ps.-Alex. took the precaution
Jf writing instead. A different sort of casualness is seen
in the duplicated of 1043a 2-3 - if we leave both in, with Ross,
tather than excise the second with Jaeger.)
Attent1on was called to DK 23A 23 (sequel in 58B 32 - cf. Burkart,
Lore dnd Science in Ancient Pythagoreanism, 47 n. 106) where Eudemus
reports that Archytas made 'tO &.6pLO"''t'ov, 'tb and the like
J.l'tCa. XLvfp-eu.u;: which was better than Plato's identifying xCv"'")'"Lc; with
these. This might indicate Archyta.s <)tiginal interest lO d<.:!fining
and yaA.fJvn. If calm is due to st lllness and evenness (cp.
1043a 24 Hith -rO Eudemus), then conversely uneven-
ness or indefiniteness of shape will '0xplain' motion - and the explana-
tion of motion was a subject nn earlier Pythagoreans had had
embarrassingly little to say (990a 8-12). fhere would not, on this
suggestion, be any need for Archytas himself to have aimed at distin-
guishing matter and form (contra Burkart, 47), even embryonically.
It would even suit Ar's context better to have a pair of theory-innocent
definitions which theory of definition can explain.
lU43a 21
l043a 26-8 The search is over. AWe;; (a 27) corresponds to We; ox
, etc.,
meaning 'in how many senses/ways it exists'.
1043a 29-b 4 Ar records that an a1nbiguity may lurk in words like 'house'
and 'line': :r"rl!-la.CvEL (29-30) is r:teaning, not reference, since the
evidence adduced is alternative definitions - 'A house is a shelter
of bricks and stones thusly arranged' vs. 'A house is a shelter'.
'A line is two in length' vs. 'A line is two'. So far dS H3 is con-
cerned, it is a question whether in a given case this ambiguity lurks
(cp. Zll l037a 8 and the discussion in 211 of the definition 'A line
is two'), but where it does, something needs to be said about the rela-
tion between the two meanings. On the supposition (possibly counter-
36) that 'animal' is ambiguous between 'soul'
dnd 'soul in a body', the two kinds of thing animal are so called
not in virtue of a single definition but Ws "'tpbc;; The two kinds
are so called in relation to a single thing. of us agreed that
this was best construed without importing a third thing besides the
two kinds of ':tnimal'. Rather, one of the two definitions presupposes
ur makes reference to the other. is which?
(A) 'Soul' is included
be thought to imply that the
in 'soul in a body' , and 1043b 3-4 might
use of '::lan' to mean just 'soul' is some-
,,-iir:ary use of nan' to r:".t:!d.n h-'3
c-an only understand
j'}St 'soul' if ,.,-e under-
-;tand this as secondary to <t,l,j der (by from its use
::.o mean 'soul in a body'. -whiCll ;__;;, t.her:!iore che in '0t; 1tpbc
(B) The claim that you do nJt ttnderstdnd c0mposite man = 'flesh
dlld bnnes animate in J. cerLain '.Hay' un1';3S _/Otl understand 'animate
in a certaln ..Jay', so that the prtJr :1.eantng of 'man' is just 'soul'.
r)bjection: that claim is the l.Jirn yuu .lo not undecstand composite
:tan unless you understand not che ..:ldim that you do nut under-
stand composite 'man' unless you compos it!..'! 'man'.
if fiJU must understand 'soul'. chat precisely shows that you must under-
stand nonct)mposite 'man', th011gh IOU not of co;1rse understand
it as the meaning of th1.t ..;orr:l.. F1\L-ther tJroblem: nut thi.i line
11f thought give us, '..lith validity, ..1 third :-:teaning of ';nan'
'flesh dnd bqnes'? Well, look dt J2lh 20 ff.: flesh and bones
1nd ,;!.' snch parts i.s 6vt-c0Y, :;ATlt:Ep 't,l.L tiUv O':f...\w\1 rWv tv
f:xOv-rwv xa.l yO.p J, \))._., \.ty't"O.\. XC' ... tO Er(sos n
Can we decide tetween (,\) il.i\l ( 8) by the in.}ependently puzzling
:-;tato2ment "l.t 37-b i bat Lhe of (ta:n'ta., 37)
lS fact irreif':vant to the -Lnquiry into substance? The
reason t;tven is that essence or 'tr,c: to', ,.,r\uch :rtusc. here ;;1ean
'is identical \.J'ith
, form and actuality. Does that indicate (a) that
because we are asking about substance we shall only be con-
<Yith 'man' as tne :Jmposite?
J.re .1sklng about :.>Ubstance as actuality ,_,,e
'man' as mt!aning form or soul? lhere
Or (b) that because we
shall he concerned
was some Lnclination to
narry (d) (.-\),(b) with (B). decision here.
\ ; hird not ;:,)nsLiered is hat 1hat W'e Uo not 'leed
LO settle for ')resent purposes l3 .;her:. her 0r any oLher Hord
is 1.:1 tdCt l_fl tne r,tarmer ,;<etcned. 2-4 as explai-
that '..Je have a !Jerfectly .J,OIJd . .;ccd for r:he form, \'iZ.. ',:;oul',
. ,nd '-:an' for the composite, ;o :Lc dnes :lot .Jn occasion 'man'
that 1C:
db 6 presupposes che
definltion of 'house' "' ':;ht:lr-,er nf bricl<.s'.
r,(h,;'S Ar himself Lhin:t r-ilar 1n 1 rnal L> ,l.'lbi guous La L.ll 1 ,-, 3 ?a 7
it is Sf)!Tie ;vho r:hink 1 h,Jt SJLH\.'!S is his ">oul ,md import
an ambiguity into the name 'Socr:tces' h Wuxf)again, l. 6).
ZlO l036a 16-17 lt is Ar himself wno the thought that 'dnimal'
is .J.mbiguous, but dS a hypothetical supposj t inn (NB xa.(,
1.7 - these
support (A)).
with this problem:
The passage remains to be considered, together
on the face of it the .1mbiguity thesis is simply
false, so v.hy is it believed, 'vhether !Jy Ar or another?
Or is it
less a thesis actually espoused than a consequence deducible (on Aris-
totelian principles) from the type of Platonist definition discussed
in Zll.
Here, as elsewhere in the chapter, ,,..,e face questions about
the relation between H3 and ZlO-ll.
Yet .1nother reading would be
what is doubtful is which is the right definition, not whether
the term is ambiguous.
(But then what would be the point of 36-7?)
The ambiguity thesis would commit one to view Lha.t 'There's a man
in the house' is 4-ways ambiguous - ttnless it can happen that a given
sentential context one of the t::"NO meanings of the term. This
possiiJility is '2xploitell in a p!lper by Lolx in !'!_ind,
.Jan. 1979.
4-14 We approached this with some dismdy, but eventually con-
eluded that perhaps its bark was uorse than its bite.
Ar's general
purpose seemed reasonably clear; the ctifftculties were those of detailed
lnterpretation. Ar seems to 11ake two main points: (a) Neither a
::r6v8EO'"I.C,: nor a iJ,tl;1.<; is simply a r:onjunction of the ingredients con-
cerned; when we have listed the ingredients, ,..re have still to specify
..;hat sort of a cr6v8e:rrt!; or )J.t'f;i.c; is tntended, e.g, in ratio.
fhe crUv8crLt; or ;nust not he treated as itself an ingredient,
either in the mh:tt:re ,,rhicb it itself i.s or a sort of second-order
mixture whose are the first-order mixture and the original
ingredients. 7he point is s:lmildr ;:hal '1lade 'lear the end of Z1.7
(l041.b 11 ff.), though did not 9urs1te this in detatl, nor whether
it bore on the relations between L. H .
The ;Ttain questions of r!etail that puzzled us were these:
Does l:x have the sdme meaning throughout the passage, 0r
it first '"'lean 'crHlsist 0f' (h 5) Jnd then adequately defined
1043b 4
in terms of (b 1 )? 'fhis Litter view enables che yd.p c2lause to
a reason for the one before it. Again at Ar is presumably saying
,10t that either a threshold or a iJOSition
COt1515tS in
the Other 1
but that reference to position is more important for understanding
threshold than reference to a threshold is Eor understanding position.
(104lb 23, incidentally, says that something could:1.'t be !:.x ( = consist
in?) jt1st one thing without being identical it.)
(2) Does (b8) imply that all the that fail to do
so for the same reason? If we are right under ( 1) we can presumably
say yes, at any rate so far as the ensuing threshold example goes.
't'Wv O:XN..uv presumably refers to the cases mentioned at 1042b 15 f f
(cf. 1043a 7-12). o!,O!: Ot} introduces yet a not her kind of case, where
a genus and differentia are in; but this sentence and the next
need a section to themselves.
(3) bl0-4 raises textual issues. Jaeger drops n olxrCo. at
bl2 (Christ dropped &xx as r..;ell, but Jaeger thinks the oU'te: clause
requires he also reads abo-Ca.,, "tOU'to at bl3 (with Bonitz), and
drops o6 at b14, with some mss.; Alexander seems to support Jaeger
in the last two cases. (Jaeger's apparatus seems to attribute contra-
dietary readings to x; thoughx is a conjunction of two mss., E and
J, we wondered whether the first nccurrence of '-x' in the apparatus
was a misprint. J has ob unambiguously; E's position is less clear.)
Ross follows Jaeger's text in his translation. hut in his commentary
(and in his translation as revised for the l-::cKeon volume) he returns
to the traditional text.
from Alexander
s support,
All '"'e could find in Jaeger
s favour, apart
was the awkwardness of iS referring back to
an immediate preceding -11 o6o-Ca. Alexander takes with
-rt)v \5\:nv, but Ross thinks it irrelevant to br1ng in a reference to
people who ignore the matter; we agreed, and thought that if tt;nt.poUv't'e:c;
referred to a new group of people, as Alexander's view suggests, it
...mulct need ot before it. We therefore interpreted as follows: 'nor
is man animal - two-footed, but [if people 3hould think he is, they
find that] something else Ls needed, if these (animal and two-
footed) are matter - which neither 1s nor consists of an
element, but substance, which such peonle would be ignoring, mentioning
unly the matter. So if this (additional element:.) is responsible for
and this is 511bstance, they would not be stating the substance
Since Ar himself does not think of animal + two-footed as
the matter, we wanted to make the sentence, from AX\.d. 1'1. onwards, coun-
terfactual, despite the indicative 6e:t; the solution given would be
not Ar' s but one forced on anyone who thought man was dnimal ... two-
We thought e:( 'taU6'\SXTJ perhaps justified such an interpreta-
1043b 14-23 The ground for treating these lines as parenthetical is
that Wo"te in 23 related to 4-14 (so Ross ad 23-5); but there may be
questions to raise about this in due course.
The context of the passage was discussed in the light of R. Heina-
man' s paper
s Tenth Aporia'. Two main theses of that paper
were taken up: ( 1) If the E !Oo, is perishable, at least in the case
of artefacts and other cases not excluded, then it is not the species
but the substantial form which is referred to; for Aristotelian species
are eternal. (2)
If the is pecishable, it is individual, not
11niversal (general, shareable, etc.),
There was some question, first, as to whether Ar in the present
passage does actually assert the perishability of the forms of arte-
What he asserts is that these forms are not xwpLCT"t'a.( and not
Heinaman argues that this means they do not survive
the destruction of the composite, i.e. they are perishable.
His chief
grounds are (i)
cp. with K2 l060b 23-8; to which it can be objected
that strictly speaking that passage says only that perishability is
a consequence of not being or 'M.p& -rb CJ6voXou not that the
two things are one and the same; (ii) the claim that the target here
could not be Platonic separation (existing apart) since after ZB (cf.
esp. l033b 20-1) and Zl3-16 Ar could not say it is 'not at all clear
yet' (1. 19) whether there are Platonic forms of perishable things.
(ii) was challenged by reference to HI, . ..;hich does not regard the
'{Uestion of Platonism as wholly settled.
One could suggest that a
reference to Platonism could be quite in keeping with Ar 's present
agnosticim on whether forms are eternal or perishable without process
l)f perishing.
If forms are eternal,
they can separately
-::-0. 'Uvd., as the Platonist Hants, hence (contraposition) if arte-
facts it is clear that they do not exist separately Jtnp6. "tt\ 't'Lvd., in
1 041b 10
l 'J4Jh
Lhose cases they must be perishab1,_!, lOt (the same inference
as K2). ( [ f the clause in l. 20
be xwpLcr-"ta.(, it looks to be tautological:
specifies which forms cannot
those forms cannot be sepa-
rate cannot exist besides the particulars.
Better. therefore,
to construe 15ou not as specifying lvCwv but as epexegetical to xwp
'except that it is clear that the substance of some perishables cannot
be separate, that is to say, cannot exist besides the particulars,
e.g. house'.)
But the antecedent of (l) and (2) can stand as Ar's own Hithout
this passage. So what of the theses themselves?
In connection with ( 1) it was emphasized that the eternity of
an Aristotelian species is Heraclitean rather than Platonic: one man
succeeds another without any single entity enduring. Is this the kind
of eternity which Ar here denies for artefacts and leaves open for
other perishable things?
It may be doubted, yet to doubt it is not
to dispute the truth of (1) but to agree that Ar's topic is forms rather
than kinds of species.
Discussion of (2) took us i.n two directions. First, can the
opponents of individual forms f .lnd :1 satisfactory sense for this and
passages (e.g. ZlS 1039b 20-7, A3 l070a 15-17) where forms are
and are not or come to be and perish in a 5pecial way?
Second, we
looked at one of the key passages that have been adduced on behalf
of individual forms.
As to the first, some at least of the opponents were anxious not
to be stuck with defending, on Ar's behalf, a tame version of Platonism.
The Aristotelian principle 'A universal exists if something instantiates
it' should not be taken to assert a mysterious biconditional connection
between two indistinguishable states of affairs, the existence of the
universal and its instantiation. The existence of a universal just
is its instantiation, it exists just insofar as it is instantiated
somewhere. From this point of view its ceasing to be instantiated
in a particular thing its ceasing to exist there - though it may,
of course, still exist as instantiated somewhere else.
lhe opponents
nf individual forms should not be required to explain how Ar can both
propound the above principle for the existence of nni.versals and, compa-
r_ibly with that, say that (universal) forms are and are not -as
there were two separate tasks difficult to reconcile. Talk of the
being and not being of forms is part and parcel with Ar s dedication
to the proposition that universals do not exist except as realized
in this particular individual or that. Consequently, the perishability
of the forms of perishable things is not a difficulty for those who
would view these forms as universal (in that they are predicable of
multiple parcels of matter), but belongs with a committed Aristotelian
understanding of that view. If the instance is perishable, that is
the end of that realization of the form, and the form is nothing 7ta.pd.
its realizations here and there. This, of course, is compatible with
Heraclitean eternity, but perhaps it will be thought an objection to
the above that it allows for eternity and being/not being to hold of
the same thing.
As to the second, AS 107la 27-29 is a favourable text for indivi-
forms: 'The causes of different indi victuals [ sc. of the same
3pecies] are different, xa\ xa\
while in their universal definition they are the same'. Opponents
must press hard the objection that you and I may be siblings from the
same x LvT')a'ov and insist that Ar does not actually assert that
of the various causes (as opposed to the total causal story) is diffe-
rent, so as to leave themselves room to insinuate that the form may
he the same so long as there is a different matter (note the feminine
a-ft, l)J.ft :: different to supply the required causal difference.
Alternatively, talk of your form and mine may be admitted on condition
that it is construed as 'the form of you as a composite being', i.e.
as identificationally posterior to the composite, not as prior to and
explanatory of it. To the objection that form as universal is not
needed to explain coming to be, since another concrete individual is
all that is required, the reply was made that the same argument would
apply to form as individual.
Another passage considered was A 3 l070a 13-17, but we deadlocked
,Jn whether the items such as 'house without matter', which are and
;tre not, were to be taken as individual or as universal. So no help
here on the preceding statement that the form of ho115e does not exist
the composite.
-----------------------l(ll!!i'Ol:;-------:--------:---------------.... -..
l04lb 23
In 23-32 qnly 2.J -d r_o ,\ntlSthenes.
28-32, which was once commonly used to father the Dream
theory on to Ant., is Aristotelian in language and content and its
allowing definition for composites could not be the consequence (ilx;-r',
28) of an d.JtopCa. about defining anything. This 1vas accepted from
Burnyeat 's paper in ?hronesis 1970, together with (i) a disinclination
to think it necessary to follow Jaeger's emending of 26-l [xa.C]
<6p(<TCl.CT6CLL 6' o6), (ii) Ross' translation for xaLp6v, 'timeliness'
as against 'point' or 'plausibility', so that Ar should not himself
endorse the d.xopCa. or the grounds for it. Less dpproved was Burnyeat' s
understanding of xotov as part of the (deriving
from the Socratic -r(-Jto'tov contrast), not a concession of some sort.
The objection here was that 'Silver is like tin'seems a striking example
which should have some part to play. whereas (m 8urnyeat 's reading
any non-definitional descriptive statement would serve. 'Silver is
like tin' does not say -r( l:cru, but unlike, say, 'Silver is found plen-
tifully in Cyprus', it could be an (imperfect) instrument for getting
someone to attach the name 'silver' to the right thing (note 6 Liidl;a.L,
27). In this sense it might be the nearest one can get to definition.
The next question was how l043b 23-8 fits with tJ.29 l024h 32-4,
which specifically names Ant. where H3 has the more diffuse reference
'the Antisthenians and similarly uneducated persons'. Perhaps Ant.
was enough of a paradox-monger for there to be no call to make a con-
sistent position out of the two passages, but if one does try, two
problems arise at once: (1) A 29 can be read as saying that (so far
from definition being impossible) only a thing's proper definition
can be said of it, nothing else; (ii) even if X6yoc:; is broader than
definition, only one X6yo.; is admitted for each thing, so: in the silver
case, either that "X.6yo< is 'It is like tin', which would mean excluding
a.ll other comparisons, or it is not, which would mean that &29 disallows
the very thing that H3 allows. Any solution must be such as to explain
further, 'Nhy in consequence of his thesis Ant. was committed to the
impossibility of contradicting and practically to the impossibility
0f falsehood ( 1024b 33-4). The suggested solutions we discussed can
be distinguished by what they take to be the 11nit in f.<P' lv6<; ( 1024b
J 3).
'.-\t- ft:;R ,
(A) The units are e.g. silver, Socrates, 8:
roughly, su bj ec t s
for description, about loihich the thesis is that only one description
This fits Ant. into the context in 629, 'double' is
2's own \6yo;, hence something else's X..6yo<; \vhen you say '8 is double'
(b 35-a l).
'8 is double'.
Thus understood, Ant. allows '2 is double' and rejects
The impossibility of &.v-n.).tyeLv would then follow
either because 'not double' is not the of 2 (would negative de-
scriptions be the X6yot; of anything?) or because e.g. 'treble' is the
of 3, not 2, so that 'not double'/'treble' cannot be meaningfully
applied to 2.
is that this
And if not meaningfully, nor falsely.
The trouble
coherence with the rest of .629 spells incoherence with
For 'Silver is silverish' (or something of the sort) should be
an acceptable of silver by (A) and 'Silver is like tin' should
fail to say anything about silver.
(B) The units are states of affairs (cf. nho 10),
e.g. silver's being like tin. dv-rLXyEav would involve saying silver
i.s not like tin, which fails to describe that state of affairs at all,
even falsely.
This makes the remarks about Ant. somehwat digressive
to A29, cued by the thought (31-2, where perhaps the parentheses should
be removed) that a false is not the '\6yo<; of anything. Ar 's point
is that if you take that thought the wrong way, Ant's olxet'oc;; the-
sis results.
Consistency between !129 and H3 is achieved because while
excludes all false statements, leaving true ones intact, H3 excludes
a subclass of the latter, viz. definitional truths.
One objection
is that (B) allows an acceptable X6yot; to be as long as you
',.Jish, which was thought not to be a natural reading of the rubric ev t<p'

(C) The units are essences (cf. l024b 29) and the thesis is that
there is no room for between rival definitions, nor such
a thing as a false definition.
This is outr_ight inconsistent with
113, fits less well than (A) with 629, and does not yield the general
rlenial of falsehood indicated prima facie in the text.
Note that it would be compatible with (A), supposing (A) ran cope
-vith H3 at all (see above), to take \6yov ua.xp6v (l043b 26) as any
\ . .5yo<; longer than one word, rather than in the specialized proverbial
:-teaning (cf. Ross ad 109la 7, Burnyeat p. 113, 115) 'evasive verbiage
such as slaves tell to cover up failure to rio the job assigned to them'.
' 9
I Jh 2 3
l.Ydb L8
l043b 28-32 28's Uxrt.' has to be understood in conjunction -..;ith the
Wo--te: of 23 and its reference to 4-14. Burnyeat's account was accep-
ted: the trick is to relate ii.xrt' to the preceding sentence's main
clause Bxe:L xat.p6v, not to the subordinate d.Jtop'a. and to see
that the clause 30-3 (which shares the same consequent with We--t'
for which reason Burnyeat punctuates with
after in 30) clarifies the connection.
comma rather than a colon
Thus: the moral of .4-14
is that we must recognize a certain complexity or predicative structure
in a definiens. Just this complexity or predicative structure is
what puzzles the Antisthenians and makes them say a definition cannot
achieve 1ts goal. Their is timely as focussing the very struc-
ture Ar wishes to affirm. The consequence - the consequence, that
is ( r7!:e:p X'tX., 30-3), of this predicative structure in which one ele-
ment stands to another as matter to form - is that only complex items
can be defined.
If this is right. it confirms that 14-23 is parenthetical. The
only difficulty is that 14-23 is mentioned in the terminal summary
to the chapter, 1044a 11-14. But the summary (puzzlingly enough)
refers 9_tg_y to 14-23 and to 104 3b 32ff., not to 1043b 4-14, 23-32,
so it remains that 14-23 is parenthetical to the chunk of the text
into which it is sandwiched.
What are the Kpoi't<t (cf. <tOLCl(pE'<<t, 35)? Perhaps all we
can say (as with Wittgenstein) is that there must be some. Definition
comes to an end somewhere (and not usually with a category) - cf. 35-
6. Of this we can be certain even if we cannot give examples of inde-
l043b 32-44a 11 The points of analogy between ob<rC<tL and are:
(1) Both are divisible until you come to indivisibles; (2) Neither
will suffer subtraction or addition without loss of identity; (3)
Both stand in need of a principle of unity, something in virtue of
which it is Ev l.-x. -n:oA.XWv (text of 1044a 3 hard hut sense clear); (4)
:ieither admit of more/less. We are of +J xa"t'i\ -rO eTboc; abo-Ca.
l04Jb 32
(l0-1) or 't( e!vnL (44a 1): in what sense of &pL9J..1,6.:; is this
being compared to Jaeger worries about the text of 34. Ross
p. 231 mistranslates 33-4 as 'If numbers are substances, it is in this
way and not as assemblages of units'. Keep J..i.Ovdbwv and translate:
'If substances are in a certain way 6.pL61J.oC, it is in this way, viz.
as (numbered or numerable) complexes of elements, not as some say as
collections of abstract units.
are what we call numbers, and this the Platonists) substance/es-
sence is not (cf. 8). But substance/essence is, as has been seen
and referring back) in a certain way a number of elements
which is one common use of the Greek (cp. in 34
with the same phrase famously in the definition of time, 219b
5). So we are not to ask 'Is one of the substances, for instance,
8?' nor to seek analogies between substance/essence and e.g. 8, but
to appreciate that one may, with some justice, say that is
- provided one takes it in the right sense of not in the
Platonists' sense of '(abstract) number'. For the preceding discussion
has made it clear why (q>a.vEpov xn\ OLo'tL 32-3) substance is in
a certain way a number of elements.
We further discussed Burnyeat 's contention that Aristotle is here
maintaining that substance is a numbered collection of elements, rather
than that by which we number. Against this it was argued that Aris-
totle is concerned to draw an analogy between substance and number
as that by which we number.
(i) Of the four analogies he states (ibid. p. 4), (3), that a principle
of unity is required (1044a 1-9) l! something which he asserts elsewhere
of !!umber, not in the sense of a numbered collection (Metaph. A 9 992a
l; M 1082a 15 ff., 20 ff.). We agreed that we found it paradoxical
that he should assert this even of number-by-which-we-number (see be-
low); but he apparently did, and, it was maintained, there is no evi-
dence that he asserted it of numbered collections (where also we found
it difficult to see in .1hat way a principle of unity could plausibly
he required).
(ii) Analogies (2) (l043b 3ft.1044a l) and (4) (l044a 9-11) are applied
elsewhere to number rather than to numbered collections. at Metaph.
t.27 1024a 12 ff. and Cat. 6a 21 ff, respectively. It was agreed that
l043b 32 .e;TES ON ETA
this point was weaker, in that these analogies '.vould apply also to
numbered collections (and indeed the point was raised: if the threes
of which one is no more three than another 6a 22) are not numbered
collections, what are they?). But
(iii) in the context of the dicussion in H3 as a whole, the point
that a principle of unity is needed for substances is of more importance
than the distinction between numbers and numbered The
important point about the analogy between substances and numbers is
not the of elements in a given essence, but the fact that there
is something that unifies them.
Difficulties were r<'lised against this interpretation. It was
felt awkward that the first analogy (1043b 34-36) applies "!!re easily
to numbered collections than to numbers. The analogies between sub-
stance and number were felt to be rather weak to justify the assertion
that substances numbers in 1043b 33-34, even though this is quali-
fied by i'l'.oH; and nc:. e:r-xep in 33 was not felt to express much of
a qualification, especially as it is followed up by the non-hypothetical
statement in 34.
1043b it was argued that q>a.ve:p6v in 32 and in 33 referred
forward rather than back, l044a 7 being advanced as a parallel in the
case of the latter.
1044a 2-9 The sense given by the various corrections in a 3 is con-
firmed by 1044a 7. It would in any case be odd to say that the number
seven (e.g.) was a principle of unity among the things it numbered.
Ihe Platonists referred to in l043b 34 have a different under-
standing of number, and so can't, in Aristotle's view, meet his demand
for a principle of unity (cf. M7 l082a 20 ff.); but does he himself
have an answer to this demand, and does he have to have an answer of
his own in order to make the objection against the Platonists? H6
l045a 7f. refers to the problem again, but the discussion that follows
is concerned with substances rather than numbers, and it was suggested
that xa.t dpL8uo6<;; might even be
gloss inspired by the
present passage.
At !17

l5ff. the )nly possible <:lllswers the

Platonists are allowed are -'rpf], and ?to-Le;, none of
;eems 11uch help.
unity could be derived from the operation of counting; Aristotle may
hold that, as time could not exist if there were 110 souls, so other
numbered things could not exist (as numbered things) unless there tvere
souls that could do the counting (Hhether they ever actually count
this particuJ ar set of numbered things 0r not).
But this is no help
\.Jhere a principle of unity is needed for numbers themselves, as the
act oi counting presupposes the number-series.
[s the number seven
a unity in virtue of itself or in virtue of something else?
The first
of these alternatives is not paradoxical, for Aristotelian forms, too,
are not unities in virtue of anything outside t hemse 1 ves.
lt was
pointed out that, since numbers .::tre, i.n any case, ,1bstractions, the
..rhat unifies them co11ld not be discussed in the same terms
as apply to concrete things.
1044a The composite substance may admit nf the more and less,
although substance cannot. This appears to contradict
':'a...'::... 3b 33 ff., ,,....hich si.mply :t.sserts that does not admit
of more and less.
fhe suggestion was made that the present passage
might be explained by the doctrine of tl1e 1mperfect mastering of matter
hy form (de Gen. An. 4.
767b 8ff.); i..s the suggestion that woman
is no less Clv8pW7to<; in form than man is, though less so as a o-UvoAov.
It was objected that the distinction between male and female scarcely
constitutes a scale; but IJ.aXXov ".<a.t 11't-rov need not imply that.
1\ristotle (lac. cit.) a lesser degree still of mastery of matter by
form produces .;hat is but not J]uman at all. [Alex. Aphr. man-
tissa 168 f. argues 'that 'llale and female 1o not differ in e!Ooc; -
which means species, but also form, since it is contrasted with acciden-
tal differences due to matter, 168. 33 f.l
Alternatively, it was
o:;uggested that e:r:rte:p should not be pressed, and tnat this clause was
d passing reference to snme minor obiection to the doctribe of
3b 33 ff . an objection which Aristotle fe-t need not be spelled out
l!l44a 2
l044a 13
1044a 13 l"fi(,; E(t; 'tbv d.vayUJYi'lc: refers to the view of Aristotle's
opponents (1043a 33 f., cf. Zll 1036b 12); Aristotle is not claiming
that he himself has achieved, or even proposed, such an
List of passages supplied by R. Heinaman
fwo definitions: Meta. l033a l-5, b24-26, Z lO-ll, l025b 30-l026a
6, l043a 14-18, l064a 19-28;
6, 23-25, De Anima 403a 29-b 16.
A term can refer to form or composite.
194a 1-7, De Caelo 277b 30-278a
De Gen. et Corr. 321b 19-22; !1eta. I035a 6-9, 1035b l-3, l043a 29-
b 4, l076a 13-25, l037a 7-10, l033b 17-18, l035a l0-17 (cru:l.:l.aj)-f) );
De Caelo 278a 13-15.
A term refers to form alone:
l033a 29, b 9, l033a 33, l034b ll, l035a ll (cf. 14-16), l036a 17,
18, l032a 18, 23, l037a 7-8, l033a 27-28, l035a 9 (>.hl>.ou), 1035a
10 (<rV:I.).apf)), l036a l De Caelo 278a 13-15; cf. De Gen. et
Corr. 32lb 22-23, 33.
Further discussion of beginning of H3
We returned to the issue of the ambiguity of certain terms that can
indicate either the form or the composite and Heinaman' s contention
that this is parallel to Aristotle's usage elsewhere and not in itself
anything unusual. It was suggested that Aristotle had arrived at
this position through concern that substance should be the subject
of predication, and the difficulty of applying this to the form.
Aristotle's point at the beginning of H3 is not the difficulty of dis-
tinguishing the cases where there is ambiguity from those where there
is not, but rather the need not to be misled by the ambiguity in the
many cases where it exists - the only counter-examples that occurred
to us being 'soul' and, in a non-substance category, 'hollow' as opposed
to 'snub'. But in the context of first philosophy the problem does
not arise ( l043a 37) as we are clearly concerned with form rather than
with the composite (i.e. (b) on p.l2).
Reservations were expressed about a sharp distinction between
meaning and reference in a 29 ff., on the grounds that. even if in
both usages 'house' (e.g.) is referring to the house, the difference
in meaning indicated by concern with definitions does involve a refe-
rence in the one case to the form, in the other to the composite.
Discussion concentrated on the claim in 1043a 37 that the two
which is primary. It was conceded on all sides
usages are
that understanding (e.g.) house included understanding its form,
i.e., in this case, what it is for (though there were problems as to
whether this did not, in some cases (e.g. soul) involve understanding
the matter too; see below). But, it was contended, this scarecly
justifies Aristotle even in saying that the term 'house' can mean the
form as opposed to the composite - which he certainly does say, and
we took him to be expressing his own view; still less, therefore, would
it justify his regarding this meaning as the primary one.
Is Aristotle saying (I) that a shelter made of bricks and stones
is a house because a shelter (a permanent shelter, to exclude cloaks?)
is; or (II) that a shelter is a house because a shelter made of bricks
and stones is? (1) seemed excessively Platonist to some, though not
involving any suggestion of transcendent forms existing apart from
individuals. The point was made that in ordinary usage 'house' cer-
tainly applies to the composite, rather than to the form; but is Aris-
totle necessarily bound by the order of priority implied by normal
usage? It was suggested that the point of 7tp0<;; Ev might simply be
that 'shelter' appears in the definitions both of the form and of the
composite, while the matter appears in that of the latter. The
form of house in any case differs from soul in that the latter can
exist only in one sort of matter, the former in several - though we
debated how far this Aristotle's view; does xaC in l043a 32 express
an alternative, or does a house generally, for Aristotle, involve stone
(for the foundations) and bricks (for the walls)? Is a structure
made entirely of wood a house for him? If the form of house exist
in several alternative sorts of matter, it seemed doubtful whether
<me could the composite 'house' at all, as opposed to defining
a brick house or a stone housei such a definition would perhaps take
form 'a shelter made of suitable matter', but 'suitable' is redun-
l044a 29 :JOIES iJN
ctant as otherwise it wouldn't be a shelter, so we're left with a shel-
ter made of matter'.
The difficulty was raised that, in the case of soul, the form
(not only had to exist in a particular sort of matter, flesh and bones,
but) could not be defined without reference to matter; the definitions
of perception and most of the other soul-faculties would involve refe-
rence to matter, as for example in the case of anger (de Anima 1. 1
403a l6 ff.). Aristotle, it was maintained, was concerned not only
to resist undue emphasis on matter to the exclusion of form, but also
vice-versa. We differed in our view as to how great a difficulty
this was for (I) above; the reference to matter is indirect, it does
not have to be the matter of the same individual, and flesh, bone,
etc. have a formal as well as a material element (de Gen. et Corr.
32lb 20 f.).
Forms in Aristotle - Universal or Individual? (Note by R.W. Sharples)
l. What follows is an attempt to clarify some points relating to the
claim that we may represent Aristotle
s thought more accurately by
speaking not of form as universal, possessed by all members of a given
natural kind, but rather of individual forms in each member of a natural
forms identical in kind and differing only numerically. It
may appear from what follows that the difference between this and the
opposed view is one of terminology rather than of substance; in which
case the question becomes one of which terminology is the more true
to Aristotle's thought. My primary concern is with the interpretation
of Aristotle, rather than with the philosophical merits or demerits
of the view which I am attributing to him; a Stoic type
of nominalism, for example, may well be more satisfactory than Aris-
totle's position however interpreted.
2. in what follows is intended to suggest that the forms of
individual men (for example) differ other than their pecu-
liar characteristics, such as Socrates' snub-nosedness, are accidental
rather than essential, attributable to matter rather than to form,
and outside the scope of (scienti.fic) knowledge. L n sense one
might indeed speak of the specific form 0r of man as 'l!niversal'
(henceforth: universal*), .:Is opposed to speaking of forms that 1.<1ere
individual* in the sense uf including peculiar characteristics (and
hence differing not only numerically). It is not with the universal*
- individual* contrast that I will be concerned. The question must
remain open for the moment whether there is any other meaningful sense
that can also be given to the universal -individual contrast.
3. In view of (2), while the view for which I shall be arguing may
derive support from the identification of the form of a living creature
with its soul and the fact that we speak of a number of souls where
a number of creatures is concerned, it would not seem to provide un-
problematic support for any doctrine of the survival of individual
souls when their bodies have perished.
4. Individual forms (which are however universal* rather than indivi-
dual*, since they differ only numerically) will, it is argued, be poste-
rior to matter, in that they can only be individuated by reference
to the matter in which they are embodied. To refer to individuation
here, however, seems misleading. The forms of concrete things are
precisely themselves principles of individuation; it is only by refe-
rence to the form of man that we can say that a certain amount of matter
here constitutes three, rather than two or four men (and hence, of
course, Aristotle's concern that it should be things which are real
unities that are substances and have essences in the primary sense
- Metaph. Z4, Zl6). The role of matter is not so much to individuate
as to pluralize. However, it may be argued that it is by reference
to universal form, rather than to individual forms, that we count indi-
viduals: so that, while universal form will be prior to the concrete
individuals, individual forms will be posterior to them and to matter.
5. I his however seems to suggest that universal form is something
else over and above the collection of individual f0rms, which is cer-
tainly not Aristotle's posit ion. We do not need both universal forms
and individual forms; the question remains whether it would not be
Aristotelian to dispense tYith universal (not universal*) forms,
and simply say that there are three men here because there are three
instances of the form 'man' in Lhis matter.
1044a 29 NOTES UN C:TA
6. Specific
forms or essences, it may be argued, while admittedly
are not in themselves universal rather than individual.
The definition of 'man' states the essence that an individual man pos-
sesses, and it would not be any the less applicable if only one man
existed. It is purely accidental, and dependent on matter, wheter
the form is exemplified in more than one instance (though it must be
exemplified in at least one if it is to exist at all. Of course.
in the case of perishable individuals others of the same species must
have existed in the past for this individual to exist; but this is
still a dodo even if it is the only one left.) This is the position
argued for by Alexander of Aphrodisias (Quaest o lo 3), and it seems
accurately to represent Aristotle
s thoughts; one reason why we cannot
define individuals as such is that a definition may equally well be
exemplified on one or many occasions (Zl5 l040a 33 ff.). But this
state of affairs seems best represented, not by speaking of a universal
form1 but by saying that there is an individual form of man which may
or may not be repeated in an indefinite number of instances Does
Aristotle ever refer to forms, in his own view, as 'universal' (rather
than 'universal*')?
7. Knowledge, for Aristotle 1 is potentially of the universal, but
always actually of an individual (Metapho MID l087a 15 ff o) o This
mind of
that, even before a house is built 1 the form of house in the
the builder, if he is already actually planning the building
of the house, is the universal form of house, but an instance of
the form of house - not indeed brought to actuality by being embodied
in matter, but individual nonetheless in that it is the form of house
that he is considering at this time. Universal form, this line of
thought would suggest, exists only potentially, not actually.
8. At Z7 l032a 24 Aristotle speaks of the efficient cause,
case of begetting, as -/i <'L'tll. 'tb eTboc; l>.eyO!JlVT] <p(xnc; -IJ
in the
here must refer to the form, rather than to the concrete individual,
t 0 r he continues a.\S-rn Ot lv Consequently this seems to be a
reference to a form which is identical in kind with that of the off-
spring; but to refer to it as identical (only) in kind suggests that
it is different in another respect, namely, numerically.
that in AS l071a 27 should be taken in a strnng
This suggests
sense, rts indi-
eating that all the causes - material, formal and efficient - are (gene-
rally speaking) numerically different i.n the <:ase qf different indivi-
duals (even
accidens, be
though the efficient causes of
one and the same individual).
two individuals
lt certainly seems more
natural to read 107ta 27 in this way, rather than as saying 'the causes
of members of a single species are different (numerically though not
in kind), in that one and possibly two of them are different numerically
even though the third, the form, is the same'.
9 0
But will the individual forms not be posterior to matter,
on it for their being (above [4])?
Here it seems necessary to examine
the sense of being:
to talk about being, simply, is misleading and
was Parmenides
mistake 1. 3 186a 24 ff.), and talk of mere
rather than of being-something,
may be anachronistic. The
individual form in this man certainly will not be dependent on matter
for its being an instance of the form man, rather than horse; nor yet
for its being the form of Socrates in the sense of including his indivi-
dual peculiarities, for it is not individual* in that sense.
It de-
pends on matter for its being the form in this individual rather than
that, though identical in kind; but since it is purely accidental to
any instance of the form man that it the form of this individual
rather than that, is this problematic?
It is true that, if there
were no matter here suitable to form a man, there would be no form
of man here; but (i) this is to speak as if spatial distinctions were
more fundamental characterisations of matter than are the forms that
occupy it, which I take is not Aristotle's position; and (ii) it is
equally true, on any view, that if there were matter anywhere suit-
able to form a man, the form man would not exist at all, which does
not seem to be felt as problematic.
In conclusion, it should be stressed that nothing in what has
preceded implies any form of nominalism; the distinction between essence
(form) and accident is a real and is present even in the case
of a species with only one member.
This is Alexander's position (cf.
Quaest. l. 3, already cited) and Aristotle's too (cf. ;-1etaph. 215 1040a
Tony Long suggested I consult A.C. Lloyd's 'Aristotle's principle
of individuation', Mind 79 ( 1970) 519 ff., which I did not previously
Professor Lloyd expresses approval of speaking of the forms
in individuals of the same species as the same in species but differing
numerically - 'it L; a useful as well as a plausible way of speaking'-
!lnd holds that Aristotle jg indeed speaking ln this way at AS 107la
27 ff. ()21-523).
But he brings two nbjections against this approach
as not strictly accurate (522):
firstly, it involves confusing the
form and the universal, and, secondly, the matter that J.iffe-
forms are in is either the matter of man taken universally, in
which case it is always the same - flesh and bones or else it is
the particular matter of each individual, in whtch case it is identical
with the form, the distinction between form and mdtter in indivi-
dual being secondary to the existence of this individual as a composite
and the result of abstraction.
'3oth objections can I think be met if we assert that the distinc-
tion between form which is universal* and accidents which are to
matter is even if the universal (t:J.Ot the universal*) is post
accidents which differentiate this matter, this instance
of the form man from any vther, Jre to be contrasted with the form.
(This presumably commitG Aristotle to the indiscernibility o two in-
stances of the same 3pecific form without even any accidental diffe-
but that is not perhaps so implausible, if accidents include,
spatial and temporal relations.)
If your essence, the cause of your being what you are, does not
include any of the accidents due to matter, your being, in the sense
of what you are, presumably does not do so either; in other words,
what you for Aristotle, is simply human, not snub-nosed human,
or even (cf. above p. 2) male or female human.
This may not be so
Lmplausible: it is the essential characteristics, not the accidental
nnes, which are involved in the doctrine of the production of one indi-
'lidual hr i'lnother of the same species, and it is the form, not the
..Jhich the individual':; growth and development endeavours
to realise. (Alex. Aphr. uses the argument that mar1 and woman come
from chc same seed to argue that they do r..ot difier 1.-n form,
168. 24 ff.; though Aristotle's doctrine of heredity does take account
,f --De ueculiar characteristics of individuals, too, 4,
:6/b 24 ff.)
HJ l043a 31 ff. distinguishes between the and the frwm.
suspect that in cases where a form h_as to be in a certain t.vpe of
matter - as is the case with human soul, ,1nd mav not he the
case with house, cf. above - the universal* form, ',..:hich is the being
of the individual, of which I've spoken Clbove earlier in this note,
is to be identified with the former rather than the latter. Flesh
and bones after all have a formal as well dS a material eLement
Gen. et Carr. 321b 20 f., cf. above), and though they are material
by contrast with the purpose they serve, they can stiLl be distinguished
from their actual material embodiment, involving accidents. After
all, when we think of 'man' the concept ,;hich enters our intellect,
the form, may - must? - include the flesh and bone, but not actual
material flesh and bone. And if the 'form' of which I've spoken here
and earlier in this note is equated with Aristotle's here, this
leaves me free to support (II) of the alternatives stated in the minutes,
rather than (I) for which I actually argued, in cases where a thing
necessarily involves a certain type of matter, the distinction between
the formal and the material element in the xot.v6v (which still excludes
the actual matter of an individual, and the accidents) will still be
of use for purposes of analysis - particularly for showing 1t has
to be this type of matter if this purpose is to be fulfilled - but
only for this purpose. That the importance of soul, as opposed to
universal* man, should be reduced to this ls not perhaps very surprising
in the context of Aristotle's view of it.
l044a 15-25 We were dis inc 1 ined to follow Ross in seeing a reference
to prime matter, as traditionally cc,ncelved , 1n tx 't'oU a.O't'o\3 . :rtpW-rou
( 16) or Tf!V <pW"tTJV ( 2 3) . First, lan :rtpW-tTJ U\11 in 18 should
undoubtedly be excised for the reason that 1tptirtT) would
here have to mean olxE(a., the neclrest not the furthest 1.-n the series.
Second, Ar leaves it open whether what you would find at the furtl1est
point in the series is one or several (several the disjunction
'> 14-l l
) 10 Lt:S ()l'J :TA
.:cttdd he tdutological if rhe could be as many as you il.kc):
rhtts in 16 must have L:te same sense as in '!Wv a.b'tWv We: n;;t:nU'twV.
t'ut there cannot be several prime matters dS prime matter is traditio-
LLall! conceived. Ar is saying: suppose evBrything derives in the
end frr1m some one originative 'first' stuff, e.g. water, or from several
such, e.g. air fire and water, still the Lmportant thing for
explanation (cf. 32 ff.) is the oCxe(a.. UAn He does not commit
himself to either version of the ultimate derivation story, since his
cr,ncern is to insist that you should not answer the question 't( lcJ'tL
xo":ri 1n Presocratic style by 'Like everything else, it's earth, ,.;rater,
, but should give the specific or proximate matter (cf. 1044b
)-]), The Presocratic style of reply is not sufficiently explanatory.
;'ht> sense of Jtptlrnw \J\.nv in 23 is then determined Ly 16, as Ls Low<:
t rnU-ra. lx 'taU cd.noU 19-2(1.
and bile start off ( as parallel exampl:::s, on the
>fl.m>-. as it were. And they remain such, despite the suggestion
that phlegm might Jerive from bile, becdU3e that 1s Lube under-
.. uJ{d in the light uf CaxWt:; yO.p x-rA.., 23-5. are two sense in
.vlnch :\ m,JV derive from Y 15 l13tter. ln the ftrst sense Y is situaled
',)n u1e rudd to' X - whether !.t ts the immedi.ately preceding olxt:Ca.
1',\n, one level d'.Jwn (the in the case of phlegm), 1)r a lower
il:'Vel (the sweet in the case of phlegm). In the seco11d Y. has
to be resolved into its elements, which are then reconstituted as X.
r t-'lJssibly \Ve >?;O no lower for the xpUn11 UXTJ or &pxTi here than the
cf. 18-20.)
I3 this a promise of alchemy? Can anything be got from anything?
cveryth1ng comes from the four elements on Ar's scheme, and con-
, rihutes to the whole when resolved back into Lhem (1075a 23-5).
:ut no Jl)uht he 15 nor !<now talking practical chemistry.
:, 44a !5-Jl. having (!mphasized the need to g;o the 0CxCa. Ut..TJ, Ar
,lJ\o.' L<dnparcs three cases: ( 1) same :llatter ciitferent products; (2)
1:tr-rent l'li:ttter necessitating different products; (.J) different matter-
ln (.25-7) the differencA is due to the effi-
<lue to the tor:n'!
does, for as 31 shows he is thinking of the form as the efficient cause,
and properly so, because of course the same carpenter could he the
maker of both box and bed.) In case (3) (29-32) the sameness also
is due to the efficient cause. We searched in vain for a type ( l)
example in the sphere of natural things. Natural things would seem
all to come under case (2) (27-9), where some specific matter is neces-
sary for the product in question, as e.g. metal for a saw, and so limits
what the efficient cause can do. Even here, but most evidently in
cases (1) and (3), the olxe:Ca. \S).Tl is not sufficient on its own to
determine (explain) the product.
l044a 32-b 5 Explanation illustrated. The XO.'ta.f.J."f!vl.a. are given as
the material cause of a man, but no help is offered with the problem
of how that can be the same as or continue as a part of the flesh and
bones composing the final product.
l044b 5-8 The 'kinetic' matter of the heavenly bodies has come up
earlier at Hl l042b 6, cf. also l069b 26, It is ob -roL0.6'tT!V, not
earth, etc., as in the cases we have been talking about, but it is
of course still a (special) kind of stuff (not a potentiality
for change), with properties such as visibility. Perhaps the heavenly
bodies have to be made of a special kind of stuff if their potentiality
for change is to be restricted just to local movement. Powers do
not define or determine a kind of stuff (since we all possess the capa-
city for local movement), but they limit what it can be, as tn case
( 2) above.
This led to a digression on what Ar means by saying that Socrates
and Cal lias differ in the matter. For surely initio they have
the same potentialities for change just as they have the same form.
Admittedly, not all the potentialities can be realized by a single
thing, which means Socrates and Callias must realize them differently.
But here the unavoidable reference to 'by a single thing' :nakes the
proposed explanation circular. What 'difference in matter' precisely
has to explain is how Socrates and Callias are two rather than one.
:044a 25
: .44b )
So the problem is not solved by taking matter as potentiality for change,
:1ence difference in matter as difference in potentiality for change.
Is it solved by taking difference in matter as the distinctness of
two parcels of matter, given that each of them is living a human life?
[f this parcel here is human and that parcel there is human, you have
r.wo humans if the parcels are distinct. Objection: difference
perform that trick, why fixate on material composition? E.g.
if this parcel knows French and that doesn't, they are two persons,
not one. Partial reply: many distinguishing characteristics that
one might think to cite (e.g. this is red-haired/ snubnosed/ 5' tall/
a carpenter, that one os not) presuppose for their ascription that
we have already picked out the subject as one whole human being.
we left it, but NB this whole discussion presupposes the use
:1 f criteria on the side of form to identify the parcels of matter dbout
.;hich we are wondering whether they are one man or two. So it is
1 nt 0 questi..on of matter deciding what is a man and what is not.
8-20 We move to (what we would call) natural events, such as
an eclipse. There is no matter which i.s the matter of 3.n eclipse.
There is a but it is a full substance, viz. the moon which
:3uffers the eclipse. (Elsehwere Ar prefers to let substance count
us \JXT} to the attributes.) There is an efficient cause, the Earth.
rawc; (Ross: presumably) there is no final cause (an important text
for the limits ofAristotelian teleology). With the formal cause there
is a complication. You can say 'An eclipse is a deprivation of light'
( sc. by the moon - did Ar notice that this is an instance of the 'snub'
phenomenon, where the subject must be specified but not put into the
formula?), but it is unclear unless you add the efficient
cause 'hy the interposition of the Earth'. Why unclear? That it
4as not sufficiently illuminating for explanatory purposes seemed more
likely thanthatit issimplyincomplete. With sleep the unclarity to
he dispelled concerns which is the first bit of the animal to undergo
Or rather, which is the first bit of the animal (perhaps the
heart) to something - what? ( 't'C 't'O xd.8oc; 18-19), perhaps a
stillness - in virtue of which the whole animal undergoes sleep?
U'44b R
But the efficient cause has to be sought as well, presumably to be
incorporated into the formal.
9eneral contribution of chap, 4 H2 began by saying that there was
no dispute about :Jfxr-(a. 'b1toxe::C1-1.E\1Tl xa.t ill<:;. \S),:n. so that it remained
to analyze ob<JCa. We; lvlpye::l.a.
So why go back to \5ATJ in H4?
and H4 both begin by telling us not to forget something, as if to intro-
duce a PS, which in the case of H3 is explicitly said not to be relevant
to the main project ( l043a 38-b l).
Perhaps some relevant connections
will emerge by the end of H6, but it cannot be said that Ar does any-
thing to alert us to them.
E.g. he fails to say, as he so easily
could have done, that the matter we are comparing with actuality has
to be the correctly specified ol"XeCa \SAT).
We had the now familiar
feeling that Ar is patching in material originally put together for
another context.
We also thought it noteworthy that the brightly
conversational express.ion va.( dX\6. occurs twice here (16-17, 19) but
otherwise according to Bonitz only in the Moralia three times,
and that such examples of va.C as Bonitz 1 ists are all from early works
plus Z9 10J4a 17.
l044b 21-29
'Since some things, such as points and forms, are and
are not without genesis and phthora ... '.
Comparison with earlier
passages, especially 85 1002a 32-5 xat xat
obx lvl>f')(Het OU'<E y(yvwBa OU'rE q>8E!pEcr8a. o<e
O<e Oe obx OUOU.<;.
EJ l027a 29-30 o' d<rtv &pxaC xal OLHCt YEYT]'riL xa.l <j>8etp'riL CtVEU '<OU
yCyvEcr8et xal q>8E(pE<T8a, q>etVEp6v.
H3 l043b 14-16 dvd.yxT] 07) <ahT]v [sc. tTlY oroCa.vj n <:Lt0.ov Elva i\
,p8a.p-rt]v tlve:u -ro'U !J16E Cpe:rr8a.L xa.t 5:veu -roU yCyve:ai3a.L
'Tiakes it clear that wjth 'are and are not' we must understand 'at nne
time ... at dnother time'.
Aristotle is not denying that the rhings
..., c'! d'' . D[0 f'SS
. .
>11'.1! t .. !d'
') "'"T ' ' '
.Jl ,, . ., ,.hC'm
r tho . .\m<i.: "
<111' 1_<; .nade ln 135 f\[
; j ..
1, 1_ "''
l I ! '\ , " , 'f l , t !) t' ntr ! t..'Il'.itlc.:s v..itt tt
' - ' i. ,' i_ ' fl I K. t i I '
' '
1 1 ' ' v ,- ':\'/0 e . .,. the
-, ' r r, , lL .' f -1 ._ t
l..:r '\ i._.
; r in ;f regress
'l'-1'1 ;rl<ii:--; fn: .-pAopd).
lnt tl-tl .111 ,m,__,
. . ..
t- ,-, J
l:;trlier r IH; (_()1\'.<::rse
has been
no douht
Ar. held it to he a two-',<ty Implication.
about matter Ls i.ntr(Jduced as ct cornll3cl, it :>urclv :tn lnt"gra.l
part of the posHion sketched in prect;riin)'!; linPS.
2 dp11ri.1i about rnalrer, IJotentiality dnd
The theme of oppnsitcschauging into une ttt<Jt'ber pruvidcs
the - s.Jmewhat tenuous- cor.nection the first half <lf the chapter.
If X is A, and B L_; the kva.v1:Cov ot -\, l'> '{ there-
foro as well as 5uvd.fJ-E4 A?
2 examples: X ht1dy, A healthy,
:;ic k
:..JZJ.ter, wine.
vi r. l'he 'lti 1.'Il in
li.ne:s whcrn a dist.inction is dr<Jwn i.n the wa)._, ln .hiLh X 1s he
ot dso /\ and l) t<.:-;pe( tLvely: x.a.8'
xut xntU 'tO E[&oc, and ::ta.'tO. O"'tEprl(nv v.at riw nnpO.
So X is potenti.ally B, but onlv with rhe quaL.fi.ctr ion given.

1()44b v .. 5 C'Jntinues with the X-A-R '3Clll!!ne, hut the
example ts dropp!d i.n f::1vnur of X--li;i.JJW: (presumably
because this, like but tl ke hPalthy sick. is an Lrre-
f!TSihle process, a point to be made, V.'i.th<ut qudllficatton, later).
The apori.a
i.s: Why do we not say that A i:.; the md.tter 'Jf, r;r potcn-
L \.ally, B, given that
\e do nol
Sdy this en vU rJ36) l' xa.-ta. nl
1c; stri.ctl; spedking .1 rpBor)n nDr 1,t
-IS) tioe '.-,'lll<', 'Ill d water
ich tJi,:-
nt f rhe L_'i.lnilr1 ,(Jn is !Jt"Psum-
t h<"lt
:045a 2 NOTES ON ETA
night is potentially day? And what is the UAT} (X) here? O.f}p presumably).
3-6 We took the point here to be, not, as per Ross, that both
changes A-B and B-A, require a 'return to \SA:n', but that only the B-
A change does (i.e. vinegar to wine, corpse to living thing).
XQ\ Bou we agreed that could
mean 'one into the other' but what does oU-tw mean? It can't refer
back to the day-night example, since that .!:._! symmetrical. It must
refer back to Xa."tlz. cp6opd.Y changes, and the point would be that when
A's change to B is Xa.'t'& ,eopd.v B' s change to A, if at all possible,
must be via the
General comments on HS A bitty and unstatifactory chapter containing
no new material of any great interest. The first half could be said
;:o continue H4' s discussion of bXLx-t, of>c--ea. or to connect with H3 via
the discussion of things which are and are not without genesis. On
whether :i.t presupposed Z, we thought yes; at least that it presupposed
28 on form not being generated.
1045a 7-8 The backreference: HJ l044a 3-6 suffices for both the unity
of numbers and that of definition. No need to bring in 212 as well
(Ross). However, Zl2 l038a 34-5 concludes a 'first' answer to the
llllity of definition problem, implying that there will be a second:
:wt because the first is tentative but because it deals only with de-
finition by division ( l037b 28, 34, Notes on Zeta ad lac.). Possible
responses: (1) look for the second answer within Z (e.g. second half
,,f Z17 following on the puzzles of Zl3 1039a 3-23); (2) H6 is the
answer and ZH are after all a unity; (3) H6 can have this role
,;ithout impugning the chronological hypothesis discussed earlier (cf.
:1ote on l042a 31-b 8 ad fin.: z could be written with an eye to H
if H pre-existed and Ar thought H6 a satisfactory enough solution to
the problem outstanding at the end of Zl2.
i 7
Does the xa.C in l. 11 say even in hodies' (Ross Ox. tr.,
implying that bodies are the wedkest examples of unity), 'in bodies
too' (Apostle), or simply that the cause of unity in bodies is e.g.
contact (Ross ed., p. 237)?
Perhaps better is thdt it should emphasize
a particularly evident instance of the general thesis.
was offered as an example of something unified by stickiness. For
cf. 227a 1517.
Definition now enters as a species of unitary thing, the
explanation of its unity being that it is the definition Q_f something
Does Ar always run the explanation this way round or does
he sometimes explain the unity of the definiendum by that of the defini-
tion? Zll l037a 18-20 is indeterminate on the issue.
Z12 1037b 24-7
looks more threatening, but can be read as stating that the unity of
the definiendum is a necessary requirement for the unity of the defini-
tion (cf. Notes on Zeta ad loc.), lvhich would not reverse the order
of explanation.
l045a 14-20 Ar then asks for the explanation of the unity of !vBpw-
1tov. Is this (a) man (the form) or (b) a man (the individual)?
Ross has (a), Apostle (b).
The first clue is that the unity of the
thing in question will be especially jeopardized if one posits Forms
Animal and Twofooted ( 15-20).
Admittedly B6 1003a 9-12 does argue
that the Theory of Forms makes Socrates many things, viz. himself +
man + animal, but that consequence is drawn from the explicit premise
that on the Theory of Forms the general terms 'man' and 'animal' each
Whereas the H6 problem is supposed to arise
't'66E xa\
especially if (15-16),
not only if, there are Forms. c.Jote also the
reading a.6'tod.vSpW7t.oc; for a!>'tb. 0' &:v6puntoc; in 17: someone decided that
the right answer is (a). 17-19 makes better sense with (a): if Man,
the thing that individual men are supposed to participate in, is the
t ..m things Animal and Twofooted, it will be these two things, not a
single thing Man, that individual men participate in. 19-20 simply
generalizes this: Man will not be one thing but more than one. Perhaps
J , i 1. \ '
(as >lme hoped both here dr,.J l.1cer J r::ne dtL::>wer co Lhe quc::.t1on Hhat
nakes man a unity will also -;hm; ,,;hat makes ..1n individual man a unity,
but that aspiration not rely on lb) la) here.
I,.Jhy is it so nnfortunate that man should fall fipart into a conjunc-
t l._Vfi? This chapter assumes the unity dnd louks co account for it.
doubt a full j'lstificdtion of the 'lssun-.ption would take us deep
into Ar's belh:fs about essence. What ;.;auld he J;,tike of the idea
that the essential predicates of a thing might be independent one of
another? 1,.,1 0 uld he wonder whether that absurdly, that one essen-
tlal predicate would cease to huld did not?
[he criticism ul the way people tl$Ually Jo definitions
3eems charitable than H2 1041a 14 tf. The second ... bt
,_,f 23-4 <auld be read either ')ne vf these is viz. the i5A.,
is Suvd1-1E1. ', marking a separate (bttt of course equivalent)
d.Lstinction or t"..JO cutllpon::!nts WLtlli.n a definition. But does everything
1,::oally dS clear :\s .-\r c laj_ms unce the distinction is made?
:-\ll i.s we,ll so long as we as this chapter does, to the simple
plus one differentia. Gut H2 t042b 24, 28-9, envisaged
1 plural1.ty )f differentiae tor sor1e cdses. fhe four 'dements are
.Jef1-ned each by 'l f)3ir nf ?or more discussion see
on ad Zl2.
l o 4 5 a -.lQ::J_

At first sight a to a 1uestion '..ltlich is irrelevant
-with a choice earlier uf (d) over to the foregoing,
,lbnul potentiality und actuality not in a definition but individuals
1 hat c...o\e to be, Continu1ty lS rescued, however, by playing down
tne potentiality/ cJr rl.eleting the comma arter Jtod'taa.v in 3l.
lctuality .1nalysis of rtefinit:ion, the aext question is, 0r better per-
ltaps, nur original question as to the cdu.se of L:nity becomes, What
; 5 , he. of what is potentially something be1.ng so actually?
JlPrfectly general applying to the components of definition
'ell "lS to an1 other pair rela7:.ed as potent1al to actual. Which
1'.l..:st 1on 'a,ide f!'nm the .nonng cause in Lhe
case of things that come to he' (so Ross Ox. tr., Apostle).
only in the case of generable individuals that there is a moving cause
It is
to be set aside.
This case is then illustrated (31-3):
there is
no other cause (sc. than moving cause) of a potential sphere/ball being
an actual one.
Sc. and not even a moving cause in the sort of case
we are interested in.
Cp. I045b 22 '<l..Tjv e[ "tL XLvT\<mv.
No further
explanation of unity is required than the potential-actual relation
This is clearly the moral, though the construing of 1. 33
presented problems:-
-ro'ih' seems to pick up a.L't'I.0\1: no cause is needed for the cause
(explanatory factor)
is just the essence of each.
Ross offers two
references for 'each':
( 1) the potential ball and the potential man
(so Alexander), to which he objects that the example of man occurs
too far back (in any case the example was man, not a potential man);
( 2) the potential
ball and the
actual ball, i.e. to
be a potential
just is to be potentially an actual sphere and to be an actual
sphere just is to be the actualization of a potential one.
we have qualms about both these items having an essence, it is perhaps
just another way of putting Ar 's point to say that in the end there
is really only one essence, that of ball.
The above treatment of 31-3 as slightly digressive was challenged
by reference to 1075b 34-7.
There the unity of the compound of matter
and form is as much in the centre of focus as the unity of numbers,
and the efficient cause may seem less detachable.
But we still do
not want an efficient cause of the unity of numbers.
So I075b 36-
7 can be read: nor is it possible to say (sc. what makes these one),
unless you say, as we do (sc. in the appropriate cases), that it is
the moving cause.
There nothing to say in the other cases, and
nothing more to say in the case of examples with a moving cause.
Still, perhaps 'digressive' is the wrong label for l045a 31-3.
It might be better to regard the ball/sphere as a nicely perspicuous
concrete illustration from which to extract (by setting aside the moving
cause) a general answer to the general question of 30.
For this pur-
pose a ball or bronze sphere (suggested by 0 26)
would be better than a mathematical sphere.
1045a o3
If lJATJ von't'Tt is genus (S(J Ross) these lines spell out
the moral just mentioned, transferring the role of potentiality from
the UX.n a.la6Tl't'ft in the generable compound to the generic element in
the definition.
other component.
hut so far no more.
Each plays potentiality to the actuality of the
There is an analogy between genus and matter,
Alternatively, \SXTl 'VOT)'t'fJ is mathematical dimensionality (as at
1036a 9). o!ov 6 X1ix).o<; <TXiiiJCL t?<CJ<&Oov, if Jaeger is right in deleting
it, may be a gloss by someone who took this view. In that case a
stronger thesis could be in the offing: the generic element in a de-
finition is the kind of matter 0 which instances of the definiendum
are composed, whether it be (as a circle is such and such a plane
figure) or a.[aihJ'tfJ (as ice is frozen water or a ball a spherical piece
of bronze). The genus on this view (christened the B-1-R thesis by
:larjorie Grene in 1974, after Messrs Balme, A.C. Lloyd and
Rorty) really is (a kind of) matter, not just analogous to it. H2
1043a 14-16 reads nicely on this view, but will all the examples in
H2 come to heel? Biological examples are another issue. It may be
that Ar thought that embryo horses and embryo donkeys start from one
;;md the same generic equine matter, but can we suppose all that to
lie behind the present text?
1045a 36-U There is even less of a problem about the unity of items
which do not have a material component to be unified with an actuality.
They are immediately one. <S1tep lv is the number 1, so lv u
is 'exactly what some unitary thing is'. Perhaps also: essentially,
as opposed to accidentally, one thing (cf. l86b 17, 31-4).

are these items? Not just the categories if b 1 reads 'each

is one, iust as the categories are 0-xEp Ov u (no comma after a se-
cundum). Just the categories if b 1 reads 'each is one, just as it
is also !'5xep Ov n The second option seems to be required by
the fact that the sequel maintains a close parallel between and
r'Sv .. S0 o(,ee:v\. -t06'twV in 5 is 'none of the categories'. The only
where the scope broaden is the 0L6 sentence in b 2-3. This
need not refer just to defining the categories (Ross says really they
no definitions, but cf. Notes on Zeta, p. 6). It could be, rather,
chat the fact that the categorie-s nre immediately v and Ov explains
tnese latter are not part of any definititln: they are not ultimate
heynnd the categories (1.6).
L\P l'FR A
1045b 7-17 The <hopCa. (cf. 7-8) at the beginning of the chapter and
at 1045a 25, 29 is an d.xopCn about the unity of definition. What
question about unity does participation answer?
Zl2 considers and
rejects participation by genus in differentia as an answer to the unity
of definition, but the issue there is whether, having used participation
to explain how a white man is one thing, you ca11 extend it to the pro-
blematic case of two-footed animal.
Similarly, we thought, in H6
participation, a'UvoucrCa. and the rest come up not initially as answers
to Ar's question about the unity of definition, but as answers to ques-
tions of the form 'What is it for there to be a 0 (a case of knowledge/
health/a bronze triangle, etc.)?'.
Thus the question to which Lyco-
phren answers <:ruvoU<TCa. 'to\5 l1tCc:rra.oea.L. xa.\ tuxi'lt; is 'What is l7tL.CJ"tftiJ.TJ ? '
(9-11). Cp. 'what is "<O l:;'i'jv I byLa.(v&Lv?
xa.l 6yLdn.; (11-14).
Especially revealing formulations are Tbv
xa.).xov eTva.L -<p(ywvov;- xa.).xon xa.l "<p<ywuov and TC "<o 'eu-
xov eTva.q - hL'I><:wda..; xa.l ).eux6"tTJ"to<; (14-17). To the
question, 'What is it for there to be something white?' the participa-
tion of something in white is recognisably an answer, using participa-
tion in the ordinary Platonic way, as earlier at 1045a 18, as a relation
between particular and universal.
Ar 's point is diagnostic: it is
because people don't think about the unity of definition in the right
way that their answer to the stated question is participation. It
is an unsatifactory answer to the question it is intended for because
it simply raises, and the Platonists cannot solve, the further questions
'Well, but what is the explanation of participation? What is participa-
tion?' (8-9). As at the end of H2, Ar 's diagnosis of other people's
responses confirms his own. lcrt'a.L. at 13 shows he is elucidating not
Thus what people don't see is that you do not need an explanation
for the. unity of potentiality and actuality (b 16-17). not seeing
this, they postulate, for the cases they are asking about, a bond to
hold the two together (participation, , etc.). The inadequacy
of their answers confirms the Aristotelian thesis that the unity of
potentiality and actuality needs no explanation. Which in turn con-
firms that no explanation is needed for the unity of definition (the
case Ar is asking about), once definition is seen in terms of actuality
and potentiality.
l 1J45b 7
l(l45b 1 , 1 r::s (:TA
Whereas participation is u1;missed on the that its propo-
nents have a problem to explain it. implying a debate
)TI the issue, nothing ts said by "t-Jay of critici.sm of rrUvde::cn.c; , oi>v-
6e:O).I.oc;, cruvoiJO"Ca. eLc, doubt they are thou,sht r.o display their
own inadequacy. crUv9e:cYI. c; and 6lO).I.oc; in t2 were differen-
tiae/actualities (cp. 1045a 13), not hands actuality and poten-
tiality. Lycophr0n is further discussed at l85b 25 ff.: his
proposal to eliminate the copula lcrt&. in favour of e.g. 0 ).e-
'X.e:6xurra.1. fits well enough the idea that his worry about the bond
between subject and predicate, whjch from Ar's present point of view
ts just one case of the issue about the unity of potentiality
i045b 17-23: textual problems The grounds gives in his appara-
tus crit.icus for enclosing <J"'ts.. 'tl. in the special brackets which
signal 'added by Aristotle' no more than that the end of the papy-
rus roll suffered tattering. Much more puzzling is A.6you tvoxo1.bv
( a in Ar but cf. 'I<OLE! 1045a 14ha,\ 6 Lo.cpopd.
Ab). Suggestions included: (i) xal I>Laq>op6. means 'i.e. the
differentia', this being what Ar hi.rnself uses in H2 and so Hhat would
in fact satisfy their search for a way of unifying potentiality and
actuality. (ii) The words were added in text or margin by someone
Hho thought that X.6you tvcnto(ov the genus. lv1:Xtxe1.a. (17)
is no problem, as it occurs already H3 i044a 9.
17-23: C"0ntent On 21 see note (ln l!J45a 30-3; on 25 <5ee note
<m l04'5i1 30-b 7. [hese are straight summa:ry of Hhrtt the chapter has
established. dpTJ1:ClL ([8): it i,een but not in the terms
used hi?re and provided we understand 'ta.6"tb to mean that form
-'l.nd matter are one and the same thing, i.e. thC'Y constitute a unity.
occurs in the 'proximate mattert at 1035b 30 and
1 in the view of Ross ad lac.) l069b 36, while H4 115ed :!:pW't'Tt for non-
pt<)X.imate 11atter (see note on 1044a So no serious problem
,:ere. 1'.:1-20 compares asklng t1Jr an explanation of 'Nhy
fF.R r,
and actuality are one with asking for an explanation of why anything
is one. No answer is needed, no separate account of the unity of
a thing has to be added to the account of the thing itself, because
the explanation of the thing is at the same time an explanation of
its being one.
We concluded with a brief discussion of Jaeger's thesis
that H6 as a whole is a later addition by Ar (cf. Aristotle p. 199
n. 2, apparatus criticus ad Met. 1037a 20; cp. Notes on Zeta p. 95).
Our feeling was that without chapter 6 H would amount to very little.
This is
the chapter where the distinction between potentiality and
actuality does the work we have been expecting it to do since H2 (See
note on 1043a 12-14 ad fin.).
l045b l/
Note by G.E.L. Owen
in 6 12
As in some other chapters of 6 (4. 1015a 13-17, 11. 1019a 11-14, 16.
1022a 1-3), so in 612 (l019b 35-20a 6) focal meaning makes an express
appearance late and almost as a postscript in the account of
5 uva/t6v, and here too does not provide complete coverage of uses mar-
shalled in 1019a 32-b 14 (cf. Kirwan ad locc.). But the model seems clear:
- source of change in another or (in the same thing) qua other
(l019a 15-16: ll-tal>oA-t'i
includes stopping something, 1019a
- having dunamis
(implied at l020a 2; cf. 1019a 33-35, 1046a
- subject to something else having dunamis
( 1020a 2-3)
over the first
- not dunaton
( 1020a 3, developed earlier in 1019a 26-
33 and 1019b 5-10, 10-11, where homonymy is threatened in
b 6-10 if a steresis cannot be a positive hexis; hexis apa-
theias at 1046a 13; 1046a 31-35 on stereseis; and 1003b
8-10 for introduction of stereseis, apophaseis and
in focal analysis of onta)
(1020a 3-4. He can Dunaton
- having dunamis
in some special way
inspire us, but only on Fridays? No, the prime example
is 'well', 1019b 13, 1046a 16-19: see below on 9 1)
It may seem that, while dunamis
(or its formula) turns up intact in
all the dunaton-definitions when spelt out,

does not, viz.
not in dunaton
where the having of dunamis
is subject to unspecified
'llodifications. But the case is parallel to the analysis of philia
tn EE VII, esp. 1236a 7-33:
'\L ,,
- \J!Sh/desire for the 50uJ ,;:>d tile j.Jl!:dsant \1Zr1b lrl-19)

wish/desire for the the ,t.:,ent 30 conditionally
pleasant? l235b 25-26, i2 -15)

- wish/desire for the pleasunt for (and so seemingly
/lOOd, 1235b 26 -29)
The formula of r._hilia
r-e..tppears tn the other (1..;'0 (l236a 20-21, reading
with Bz. Sus. et omn. post.; cf. l046a 15-16, 18-19, 1077b 3-

but ..;ith dlrect qualifications.

(TfJ ..::omponent expressions, to
be ..;ure, the qualification in is to the whole expres-
sion 'having

i but in neither js it an cxtenlal governing

as tn

or the definitions of in subordinate catego-
ries.) Wi.thout such 'lUalifications 'good' and 'pleasant' are to be
understood haplOs (1236a 7-10), but does not appear in the
recurring formula of

66v;H<;(Ouvo.-t6v in 9 1
There is no sequence of

corresponding to dunaton
in !J. 12,
hut it is on

that 91 builds its account (l046a 4-19).
;ource of change in another or (in the same thing) qua
other (l046a 10-11)

- source in patient of change effected by another or (by it-
qua other ( 1046a 11-13. Hut is riunamis
We are not to extract, monkeying with an inflection,
'source ... of change ... another or( ... itself) qua vther' .
Rather, as in dunamis
(1046a 14-15), after 'by another
or (by itself) qua other' supply 'viz. by a source of change
(something which is/has dunamis

-settled state of not being affected for the worse or dest-
royed by another or (by itself) qua other, viz. by what

(1046a 13-15:
ts >!xplicit)
here the last complement

As in dunamis
bllt adding 'well' (in some grammatically
1ppropr1ate form) to exPresslon5 of acting/being acted on
( l046a 16-19. 'These dunameis'
Lo .dl the three: but
;,e introduced wtthotlt :J.bsurdity
.1tfected well for the worse',
in 1046a 16 seems to refer
h1)W can the qualification
tn riunamis ?
'Not being
affected for the worse or
l<:stnJyed by a or v,o0d :hange'? The k.1i in l046a
refers back to l046a 15-16. 1.Jhere the ev1dently

Perhaps rewrjte 'taU l(a,).c]c:; IJ.il 1td.cr-

:(ELV 'X""t\.r0 mark a SpP.cial1y ;;<lmirilble rr.!SJ3!ance to harm.)
l>jOT r:S UN fH ETA
(5, 6) correspond .., 11 ffi,:ic;ntly tn the :.,horthand

of 1::.12. 1046a 6-9 insists that the relation between derivative and
primary cases is more than m:J:re similarity, which '.-muld produce homo-
nymy. (Contrast EE VII l236a 7-33 explains by focal
meaning without invoking similarity and EN \'Ill 1156b 19-21, 35-1157a
J, which invokes similarity without using focal meaning; and explain
as you will.)
What is the str0nger relation bP.tween primary and deri-
(1) A case of in a derivative sense coincides directly with
one of

viz. ousia. (ii) A case of iatrikon in a derivative

sense seemingly requires but need not coincide wi.th one of

viz. iatros. (iii) A case of hugieinvn in one derivative sense ('pre-
servative of h.') requires and may coincide with one of hugieinon
in others (e.g. 'sign of h.') rloubtful. (iv) A case of philia_ in
a derivative sense does not expressly require one of philia
. (v)
cases of dunaton and dunamis in derivative senses seemingly require
may coincide with some of dunar:on
, dunamis
with dunaton
if the qualification 'partly' or inadequately is allowed). So here
is no regular requirement, logical priority does not entail natural
1045b 27-32 The backreference: e:tpma.r.
xpcirtor.c;; "X.6yor.t; to Zl in particular.
to ZH in general, lv "tott;
z 1 asserted ( 1028a 35-6), as
rz did not (pace Notes on Zeta, p. 6) 1 that the account of substance
will enter into the account of the derivative cases, as is implied
here at 31-2 (cp. 46a 15-16). Other backreferences to ZH in 8: (i)
1049b 27--9 'It was said lv 1tp\ o!xrCn.; [cf. ep\ -roil
l'l:p<irt(o' Ov-toc:. I+Sb 271 that everything that becomes {is becoming)
becomes ( i.s becoming) something from and by the agency of
something, and that this l..s the same in form', referring to ZB l033a
24-8, b 29-34a 5; (ii) back in 1 i045b 35-46a 1, "'hat is most useful
for the business in hand is evidently the notion of dunamis - paten-
tiality taken from H6, esp. l045a 20-33: that what we expect to
have discussed, which is why Ar warns us the discussion is to be post-
tJ<JOed until 6. All this suggests that ZH existed as a unit, "uegin-
ni.ng with Zl (tv l7-9 be already a proper
part of Z; which is compa(ible Y-'ith the hypothesis of H's pre-existence,
us discussed earlier.
in the strictest
.-\[" j)!"\ T'llses Llrsl 'I) di::;cnss
sense, viz. which l,as reference to at
l048a 25-6 he says he has done this.
!046a 2-4 promises, secondly,
to treat nther sorts of in distinctions r .. o be m<1de about ener-
geia; 6 1048a 26-7 uses the same expression for '"hat he is about to
In general the contrast is mar-kerl by the dominance of dunamis
in the earlier chapters and of '!2:1namei later (thus i.n dunamei at
l048a 32 (2), b 10, 14, 16, and in 87 at l048b 37, 49a 1, 5, 6, 8-
9. 11. 16. 17. 21, 2 l, 49b 2). In both sections dunaton occurs:
in the first in connection with dunamis, in the second (e.g. 1048a
27, 34, b 6, 14, 16, 49a 4) in connection with In '.:18
But in the
r akes over as a general ter-m ( 1049b 5, 0, l050a 3, 9).
first section at 93 l047a 24-47b 2 (note 1047b l) and the con-
sequent 94 he enlarges the scope of in anticipation nf the
second section.
had before us Owen's comparison of 6 12 and 91. li 12 is fairly
forward. It offers a focal analysis of dunaton, with dunamis occur-
ring in each derivative case. One could p;o on to define

on the basis of but Ar does not actually do so, even though
he had distinguished the corresponding senses 0f dunamis earlier in
the chapter (1019a 15-33) and had defined senses of therefrom
(33 ff.). In 8t, by contrast, he proceeds from the noun dunamis,
allowing the verb ( l046a 5) and the adjective to tag along.
A new development is the suggestion ( 16-17) that the xa\Wc;; modification
applies to all of and is not j11st a case on its own as at
6 12 1020a 3-4. fh..i.s gave rise to some discussion:-
He are first introduced to the case as the sense of 'can' in which
rme may deny that a lrunk man staggering and slurring can and
calk ll\12 1019a 24-6), Ihe rhoHght lS that ther-e is a sense in which
it may be said that he can do these thin!<;s (for there he is doing them)
<tnd therefore, i..f it is also correct to denv that h8 can, lt must be
Ln a different sense of can , viz. 'can properly' nr something of
the sort. rhus the adverbial aodifier xa.AWc;, xa.-tl\ etc.
)h 14
1046a 4
occur in analyzanst not in the ordinary language analyzandum. The
drunk's walking and talking is an example of a denial of dunamis
fied. /J. 12 1019a 26 .already indicates that dl1namis
can be similarly
modified, e.g. perhaps 'This land can be cultivated [sc. with a fair
level of success I'. But '"'heredS 1:!.12 deals with this before going
on to .. , &.xa.eeCa.c; at 1019a 26, 91 reverses the order in such a way
that it is hard not to think that &.xa.BeCao::; is included among these
buvd.J..!1.<.; at 16 and covered by ll.Od'}OU.I. f} 1ta.8e!v at 17.
an example be of an adverbially modified
What would
Start, as
before, with a denial: 'this plant can't stand feast' - it is not
that it is dead in the morning but that it does not look at all happy;
thus in one sense it can stand frost (it does not die), but not in
the sense in which we say of a plant which still looks fresh and bloo-
ming that it can stand frost. We also thought of the distinction
between 'combustible' (can be burned) and 'tnflammable' (can easily
be burned).
Nevertheless, it is a question whether, for any of these cases
), an adverb in the paraphr.:tse is sufficient proof of a
distinct sense of 'can'. Perhaps they belong with 'There's a man
for you [ sc. a real man]' and should be explained in terms of what
a speaker can do with the ordinary sense.
Another doubt abnut Ar' s present project was whether adding a
reference to dunamis
or, if need be. the account of actually
helped to elucidate dunamis
or dunamis
. Being changed is of neces-
sity being changed by some agent, but why put that into the definition
of the capacity to undergo or to resist change?
1046a 6-9 fhe cases 'Which are excluded to avoid homonymy can scarcely
be understood except by reference to 4 12, just alluded to. Two cases
seem to be envisaged. (a) the geometer's :se of for 'square',
explained 0!-10<6-rmC nv< (46a 7) or xa.-riJ. L<ETO.q>op<iv 1612 10l9b 33-4)
[Alex. liet. 394, 34-6: it is 0 66va.'fal. f1 11::\.eupd., similarly Anon.
in Plat. Theaet. 27, 31 ff .}. (b) certain things we say are Ouvo't&
"<.'l.t a.OUva.'ta. 'tif e:Tva.t ?tWc; 11 ).Jh e:Tva.l., which seems Lo refer t0 what
l:.l2 1019b 14 calls Ouva:tl'L o6 xn'tO. C<o!tt.::lin ar"} possible
and imposslble not 1n
not being the case in
'Tletry - at t::. 12 l019b
virtue of a capacity but in virtue ot being or
the manner discussed - with an example from geo-
22-33. But the discussion itself is full of
see Kirwan
On (a), note that the likeness in
terms of which it is explained rests on the use of 6Uva.crea.,;
the mathematical sense is not derived via likeness or metaphor from
a non-mathematical one.
1046a 19-29
On the question tvhether the dunamis of poiein and cor-
related paschein is the same or different, cf. III 3 on the ques-
tion whether the kinesis in teaching-being taught is
the same or diffe-
The latter, after familiar paradoxes,
is solved by arguing
the change is the same but under different descriptions appropriate
to the two parties involved (202b 19-22: i..e. not identical kuriO's).
This change however, is located in the patient, whereas the active
dunamis 0f our present chapter remains in the agent.
\S\11 as O.px-ri (23) does not mean it is an
is a passivity.
agent but that it
at 216 l040b 15, in view ofA 3 1070a 10-11,
we first thought it better to afterq>6a'et. , as
heLonging to natural formations (Aquinas supplies complantatio, grafting,
which seems neutral); but editors recur toGA 773a 2, where

are connected with 't'tpa.'ta (773a 3-4) and with some non-evident as well
JS evident <iv<b:11pa. ( 77 3a 13).
This would lead us to believe that
ilt l040b 15-16 ailuq>IJO"'I.c; could be taken as xT!poxnt:.
(But we were not
moved to excise )Jl] at 1070a 10).
Here ln 8 there is no suggestion
of malformation, and this with A may make the transposition at l0
15 (nnt recorded in the Notes on Zeta) persuasive.
!G46a 19-35 Incapacity.
To be compared with i.\ 12 l019b 15-21.
'tL-\i'iER i l
l046a 36-46b Ross does not annotate the lines, Lut two pages on
them will introduce the chapter.
of logvs/ logon echon: (a) EN
Cumpare some familiar 0ccurrences
13 also contrasts (1102a 30,
h 14-15, 17, 24-28) and to logon echJn (1102a 28, apparently identified
with l102b 14-15) with another element in the soul, to alogon
(1102a 26-28).
11) and (lt02b 28-31) of which the first is a dunamis and
marion of the soul (1102a 34, b 4-5). To logon echon, both in its
prime sense and in its derivative use of the obedient ( 1103a
1-3), is considered in terms not of gunamis but of and a rete
(!lOla 9-10). So there is no consideration here of as providing
a dunamis tOn enantiOn, as there is in 8 2 (l.OI16b 4-24). The
enkrates and akrats are equally praised for the it: ( 1102h 14-
16); what provides the opposition is t.he horma.!_ which
defeat prohairesis (1102b 16-21).
(b) It may lJe reflection on this lumping of all logos under good logos
that prompts EN VI 13 to stress orLhos log(JS (1144b 23-28). V 1 ll29a
ll-13 has already pointed out that is and unlike hexeis,
are of opposites (just as in Met. 9 2 1046b 7-14 is the
of opposites); so EN VI 13 insists on the orthos which turns the
into a hexis (or at least a well-directed piece of rPasoning) confined
to producing one opposite (e.g. health. for the doctor). This does
not of course imply two types or parts of a logos, any more than the
:;kill in hurting-or-healing is a different sk.ill from that of the depen-
dable healer: the of the first can be channelled into the
hexis of the second.
(;:) If this ts so, the logos which is embodied in meta logou
,:1 is the unrestricted implied by the contrast
wtth orthos in r:N VI 13 ll44b 27-28. 9 2 does not indeed speak
oi but later in 95 1048a 13-15 Ar. proposes qrexis or prohai-
C?Sis as factors which determine che rational in one of its
t direct ions. EN VI LJ on the other hand does n("_ "'-peak of h_1_2_8.Q_
dunamis tOn enantion; but earlier in VI 12 1144a 6-11 Ar says
t!"lat phronesis and ethik! arete respectively make the means and the
dim (}rtha, and then says that of the fo'J.rth part of the soul. the tJ1rep-
-:_!--ls:_O_f.!. there iS !10 such for this part has nothing Which it iS
in its power to do or not to Jo.
The other three parts in this con-
text - epistemonikon, 12..s.ist iko_!! and - are therefore by
t rast rational dunameis in the sense given in 92.
So EN VI 12-13
and Net. 8 2 are compatible and imply the same scheme, and both are
more sophisticated in their account of than EN I 13.
But at one point 82 may seem less sophisticated than EN VI 13:
meta logou This recalls the insistence of EN VI 13 that
is a not merely kat a ton orthon logon but meta tou or thou
l.OJ>.211 (ll44b 26-27, 30).
It does not merely conform to, it C!mbodies,
the right logos.
component. But
So in 8 2 rational dunameis have logos as an essential
subsequently at A2 l046b 22-23 Ar reverts to kata
logon; and similarly in AS he uses kata logon as well as meta logou
(l048a 2, 13; l048a 3) in discussing the operation of rational dunameis.
But this does not mark the greater sophistication of EN VI 13.
Ln both 02 and Rs Ar is careful to use logou of the dunameis but
i_.;_ata logon of the dunata possessed of such dunameis.
The logos is
a defining component of the rational dunamis but not so, or not directly
so, of what has the
This may be because .-Jhat is now dunaton
:nay subsequently l<)Se the_dunamis and therewith the logos, or vice
versa ('by forgetting, or some accident, or the lapse of time', 93
l047a 1); but the cannot lose the logos. Alternatively, con-
->ider the examples of kath' ho given in 6 18 1022a 14-17: that in virtue
qf '...rhich a man is good is the good itself (here called the eidos and
:>usia: ps. Alex. supplies handier examples, statue and man, 414. 31-
32); that in virtue of which a thing is coloured is the natural primary
possessor of colour, the surface.
Thus that in virtue of which a
thing is dunaton enantiOn is the logos or logon echon which primarily
houses the dunamis tOn enantiOn.
It may he objected that the
is not the primary_ dunaton, since it Ls rather than has the dunamis.
But this may be met by (a) referring to the good-itself in 1022a 14-
16 (property an<! prime possessor?), or (b) observing that the
'lre themselves described in language appropriate to agents, e.g. at
.q5 1048a 5-10, or (c) recalling the interchangeability of and
_logon echon, the latter hein.g evidently a possessor 0f the relevant

1-2., etc.: Ln EN I 13 Ross consistently translates
':-"ltional princlple',
doubt dVotding 'reason' part of the old
l\J46b 1
debate he cites in his note on the tr. Jf EN J '3 1U95a 10. In EN
VI 13 he translates it 'rule or r<J.tional principle' (l144b 29-30),
and orthos logos 'right rule' (l141b 23, 26-28). In VI 13 logos is
indeed put into the plural, 1 i.ke episteme, in reporting Socrates' views
(1144b 29), but Ar. keeps to Lhe singular e.g. in identifying
phronesis with at 1144b 27-2&. In 9 2 and 95 Ross
tr. logos 'rational formula', though he lr. loe>on echon at 1046b 1
as simply 'rational'.
If logos is put on a footing with threptikon-phutikon-orektikon
in EN I 13 and with phronesis in EN VI 13, it is natural to translate
it 'reason' or '(power of) reasoning', with the possible addition of
'about ... ' for a dependent e.g. in 92 1046b 8-13. It is
not a 'formula' (or 'rule' or 'principle') which uses denial and removal
to exhibit the opposite; it is reasoning about e.g. health which must
mark it off from its enantion or is, though health must remain
the basic idea (92 l046b 10-15).
l046a 36-b}_ What does mean? Thert?: are two main views, that
it means rational principle or rule (Ross), and that it means reasoning
or power of reasoning, sometimes
subject (see end of Owen's note).
specified as being about a certain
We preferred the 1 at ter. One
might ask how a power of reasoning about something could &TJA.oiiv things
(1046b 8, 14): doesn't Ross have the advantage here, since a formula
could already 'exhibit' something, without needing to be put into words?
But to exhibit opposites seems to need rather a bunch of formulas,
and tt seems better to say the power of reasoning exhibits them via
such a bunch. Similarly reasoning about health could exhibit both
health and disease, which a single formula couldn't, as it rioesn't,
so to speak, carry its negation in its pocket. Nor need we be troubled
by the thought that, with \.6yot; as 'reasoning'. A6yoc; 6 a.i>'"r6t;; at b 8
ought to be replaceable by something like rpp6vry:nc; 1'1 a.b't""ft which would
sound odd: b 7-8 shows a shift from reasoning as a power to reasoning
about something, i.e. a specific exercise of the power. How could
knowledge have a power of reasoning (b 17)? Well, it could 'have'
(in an etiolated sense) powers of about a subject.
llAPTF:R 1.
In fact, since is often interchangeable \.Jith \6yov Exov, b 17
could have had --rijl \6yot:; Eiva.c. as easily as -rq, A6yov 'XLVa Of course
we are not saying A.Oyoc; never means formula (at 1032b 2-6, where it
is again associated with opposites, it seems to mean form); but ..;hen
it does mean formula an 61t6(pcunc; gives a different formula, which it
doesn't here (b 13: incidentally we saw no significant distinction
between t:7t6q:Kl.O'LC: and d.1tocpopd.).
At b Alexander reads 1tOLll'Hxa.t xa.l 11t (xat without
a!: recc.), which separates the crafts from the sciences.
We preferred
the standard 7t:Jt..TJTLxat with the preceding xnC being epexege-
tic. We thought Ar didn't bring in 7tpax-rc.xat l7tLO"'t"-r,)..la.c. here because
he was concerned with 0:\Ao lv U\Xql cases, and also the txc.<Tt-r,-
f.lO..L are concerned to oppose doing to not doing, not to oppose doing
something to doing its opposite.
On the xa.-rn \.Oyov I A6you question (b 2: j..l-rd, , b 22: xa-cd )
agreed in the main with Owen's note, though we thought the point
mj_ght be simply the one about Xct-rd, unlike 1-lE'td., not applying to a
defining component, so that 618 wasn't needed.
We noted noncommittally
the view of Hardie (Ar's Eth. Theory, pp. 236-9, preceded by J.A. Smith
[CQ, 1920]) that Nic. Eth. says not that j..J.E't"d.. gives a stronger condition
than x.a.-cd.,
one which
but the reverse, and that there are two types of virtue,
is both -.:a.'td and iJ.E"td , and one which i.s merely ;...i't<i, the
\l_rthos logos.
1046b 3-24
1046b 13: ''va signifies the way in which
knowledge of, say, health also involves knowledge of diseases.
1046b 15 The premise introduced by lxe:( doesn't seem to be used in
this chapter. But it is relevant in 9 5, where the thread is taken
up again from after a diversion in 93 and A4, and it is effectively
repeated at 1048a 9. The point there is that somet.hing is needed
to decide which way the rational potencies act. and 5p;;L c; and JtpoaCpe::-
are offered. This could have been said in q 2, and could per-
i D4fla 36

lQip')b l:) 'WTES THETA
1aps have been 11sed against tl)e '1egarians in 9 3, Lhough they might
just deny that any deciding had to be rl.one. The C.pxh x L v'licrf;wc;; of
b 17 and b 21 doesn't seem L') do the trlck, and seems to be simply
the A.6yoc; itself, if <ipx1'! at b 24 has the same reference. A" 2 also
doesn't raise the analogous quest1on about the nonrational potencies:
,;hat triggers them off? - though here the trigger may be something
external, as something is heated by being pushed towards a fire.
{Indeed the nonrational potencies too can be potencies of opposites,
though only per accidens, as in the rather unfair example of hot gas
working a refrigerator, or as when healing, especially Greek healing,
involves hurting.) Anyway when the heating 5'6vn)JLC:: is triggered it
produces heat. Analogously a cational potency when triggered should
produce both the relevant opposites, which, as b 15 says, is absurd.
Something more is needed, which doesn't come till 95, while 82 turns
to elaborating rational at b 20 ff. One suggestion was
that b 20 itself provides a role for the l1teC clause by answering it:
the opposites are united by a single A6'(oc:: which shows how powers
for opposites are pvssible despite the l1teC clause. But it was pointed
out that if the l:xsC clause is relevant at all it should affect the
nonrational potencies too, in so far as the heating 66vnf ..Hc; covers
cold as well as hot, albeit per accidens. b 15 f f. would anyway be
awkwardly expressed if the above suggestion lay behind it.
What does 1tpbc; 'tO nb't6 refer to? Two suggestions: ( i)
1\n object we are operating on, e.g. a piece of paper we are making
black or white. The soul, starting from one and the same i.e.
\6yoc;, sets both the opposite processes in motion, connecting them
with a view to putting now one now another (or: putting one
or the 0ther) of the alternatives (black and white) into the same piece
,1f paper. (ii) The same thing as aJ<o <Lb-rijc; <ipxn<: refers to (cf.
':he equivalence of 1tp0c; ttv and in focal meaning, and other
,onnections between 1tp6' and The soul, starting from the \6yoc;
(.as above) sets the opposite processes in motion, -:::onnecting them with
to that same d.p):'..,C, or )..6yot; only not (o6x 6J....l.o(wc;:
c\1oo<>es nne :Jlternative rathr!r than the other). ( i) seems to
involve taking as obx t'f!-la, which would seem to require some-
thing like l:v )JfpEL rather than o-uv<i+aou. Also (i) doesn't explain
how the asymmetry ( 6jJ.oCwc; applies to the A.6yoc; itself (b 20),
even granting that there is an asymmetry in the sense of a difference
between 'the processes themselves (blackening and whitening). ( i)
also reads a lot into 7tp6<;;, and one would expect rather the dative
for 1tp0c; -,;0 We therefore inclined towards (ii), re-
ferring for o6x to b 9-10 (f.'ll"l-"l.ov) and b 12-3 HVIL );
the asymmetry there will be that, whatever it may be, which is involved
in ( (ii) might read more easily if CL)J<PUJ is taken only with
a-uvd.+-a..aa., xLvf,o-eL having no expressed object; but this would give
the clause an awkward structure.)
EJAb read 'pb<; for "Pb <;
We noted en passant that at b 22
l046b 22-3 similarly evoked two views. (a), in the spirit of (i)
above, says the rational Ouva.'td. do opposite things to the nonrational
Ouva:td. (taking 'totc; Ouva:totc; with 1tOLEt'), while (b) says the rational
6uva't0. do the opposite to what the nonrational 5uva'td. do (taking 'tote;; ...
Ouva:toi:'t:; with 'td.vav'tCa.). (a) treats the object of rational action
as passive Ouva'td. and seems committed to calling all passive Ouva'td.
nonrational (if not indeed vice-versa); it deals with this by saying
that passive powers that seem to be rational (e.g. the ability to be
taught by one method in preference to another) are really active powers.
(a) perhaps has the advantage that the fJL9- clause follows on more
naturally, the subject of 1tEpLfxe:'tac. being the nonrational Ouva.'td.,
if we can think of them as being encompassed in their capacity as pas-
sive 5uva.'t({ by the ).6yoc; (the power of being whitened is, being non-
rational, a different power from that of being blackened, but they
are united by being both subject to a A.6yoc; which decides whether to
whiten or to blacken). Also 'td.vnv't'Ca seems a rather strong
phrase for (b) - though perhaps it is just a stock phrase meaning 'acts
differently'. For (b) the subject of 1{pLtXE'tal. is the rational
Ouva'tci, which is awkward unless one can taken the 6 L6 clause as paren-
thetical with the yd.p referring back across it. On the other hand
(b) fits the chapter as a whole better for the nonrational powers must
be being thought of as active because of uta tv6' at b 6: this
l046b 2l
l_l)'t6b 22 :;JTES ON THETA
won't apply to the passive powers if paper can be blackened as well
as whitened, (We did not discuss thE: nominative reading of AbJ:
,...Cn d.pxt].) We all agreed, however, that Ar is showing that in
s()me sense there is the same <ipxf! of opposites.
l046b 24-28 In the first clause (24-26) d.xo)...ov8E! can be taken
in the sense of 'is implied by' , but in the following clause ('ta.U't";J 6'
obx &.e:C, 26) this seems difficult - in most cases it isn't just
that the ability to do something doesn't always imply the ability to
do it well; it doesn't it at all, though it may sometimes be
accompanied by it. We considered four possibilities:
a) = 'accompanies' throughout, and ri C is to be understood
in the first clause.
b) &.xaA.ouBe:;t "" 'is implied by' in the first clause, but is understood
only as meaning 'accompanies' in the second.
(c) Aristotle is thinking of cases where to be able to do something
at all does imply being able to do it well. But in such cases the
qualification 'well' is redundant.
(d) It was pointed out that, in ordinary speech, 'isn't always implied'
might be natural enough, though inexact, for 'isn't ever implied, but
sometimes is the case'.
But what are these remarks doing at this point in any case? They seem
to relate to e1 l046a 16-19, but not, apparently, to anything earlier
in 92. We concluded that they are a note which had been placed here
for want of a more appropriate place; or, perhaps 1 an answer to a point
made by one of Aristotle's audience?
Note by G.E.L. Owen
The Megarians:
(to school it now appears that Diodorus Cronus did not belong:
Sedley, Proc. Camb. Philol. Soc. cciii (1977) 74-120; so Sorabji,
I Cause and=-__t
I) Q
11)46b 29
64, 106.
But Diodorus may rear a head on one interpretation elf 1047a
13-14 sub infra,
l046b 29-30 X can F only when X does F; when X does not F X cannot
f. (i) For values of verb 'F' cf. 1047a 28-29: they include not
only kineseis and staseis but einai and gignesthai and their negations,
and these were introduced at l047a together with the quite general
kinesis kai genesis at l047a 14.
In other words 1 although 93 begins
with examples (building, seeing) that would suit the account of dunamis
in R 1-2, as a whole it covers all forms of and so
belongs with A4.
As takes up from 92.
(ii) Notice that Ar is interested only in the form 'cal>'jdoes',
not in 'does"'*-an', which is not in question.
(iii) X can F at time t ...... X does F at time t: this brings out
a possible ambiguity, between can jump-at-t and can-at-t jump.
The arguments in l046b 33-47a 10 about house building and seeing etc.
seem to need only the second formula:
X is not now a housebuilder
because he cannot-now house build.
(b) The comparable construction
with pephukenai (see under !. below) seems rather to want the first
formula, is naturally apt to F-at-t. (c) At l047a 12-17 Ar uses
against the Megarians the idea of can-now F-in-the-future; given the
choice between and can-now F, this is more naturally an appli-
cation of the
X can F-at-t:
But the Megarians can surely not allow the
for them X can-at-t F stands in mutual implication with
when X has the power, it is the power of doing what
X does at that time.
Confusing the two may be one source of their
thesis, and the confusion is not met by amending 'doing what X does
at that time' to 'doing what X could do (sc. has the power to do) at
that time', which threatens circularity or regress,
activity is entailed by capacity .
So concurrent
l046b 34-36

(and so for all

l046b33-34, from !'. and g:
X is a housebuilder only when X is house-
l046b 36-47a 2:
(i) X has an art; X has learnt and acquired that art
l '::tob \f)
(ll) X does not (sc. aheady a.v{uir<::d)
X has lost the art, bv some
pathos .. ... time, the obJeCt surviving
'..'hat ubject? The art of building? But i;r the Hegarians this, even
if t:-:ey concede g_!.. j s temporary and particular.
The troubles of Z 7-9 are upon us.
The !!cuse in house-
Ross proposes 'the
form ot the house', but will the l1egarians accept the theory?
Alex. 571. 6-11 proposes the but these are not eternal (and
cannot be reduced to survive' v.:ithout reduplication)
from and ?._: stops he does not
art; whenever he starts he
How Lhen does he (re)acquire it?
Viz., from ( i), ,..1here does the
l,:arning occur? (The they jib at g_, 1.vill presumably
= what can be perceivr.>d)
perceptibles, e.g. cold,
only when perceiving is going on. (Presumably accepted by the Megarians.)
echein e.g. perceive, e.g. see)
X Joes not have sight, apt_to have
it, .1nd Nhen it is naturallY apt to have it, and in
tut is blind.
(i) For hon t in 104 7a 9 J.1eger- cites 104 7b 29, l048a 1, and
CJ.i.>;ht have Lited 1022b JO, 1046 a 13-34.
: still
Othepvise the banal 'and
fnr hate pephuke cf. [llZ 10l9b 17-18, 12 l012b Zl-29, 9[ l046a
rhe second passage explains that to lack 3ight is not blindness
t c:ery though in general l.t is natural for the dnimal to have
tiere tht! qualification seems to attach rather r,o the
t:han to the (cf. the at L01.9h 18); the .mqualified
belongs <lt all (Jges; what does not belong is e_<::2__hukenai
from P and U:
The same man will __ rnany times a
lThe Xegarians may re1ect 1_: as ill-formed or false.)
capacity ( P.ster;menon dUl}-ameOs) -) adunaton
1 l046a 29-10.
?erhaps needlec;s, but see 1,n LJ47a 13-14 below.
12-14 What is udunaton genesthai nclt:her is nor ' . .Jill be - for
this is what 'adunaton' ;neant
and where?
(a) Perhaps . .;hat 1s is that tdunaton genes-
thai either (i)
implies ur ( i i)
( ii)
gives us Dindorus; ( i) seems sufficient.
firstly, where has Ar said either? ;.lot e.g. in !J. 12 0 r 81 l046a 29-
Secondlyt why should the Hegarians accept either, when they Hant
only present performance or non-performance to have Logical connections
\.Jith or adunaton? (b) Perhaps then it is that adunaton :
implied at 1047a cf. l047a
far the argu-
ment has Leen conducted wholly in terms of Q.'.:!. and apart from a
different at 1046b 36.
Or (c) it is that -== estere-
menon duname'Os, taking '!_ as a two-way implication:
cf. 12 1019b15-
22, 1 l046a 29-35.
Cut neither (b) nor (c) is important to the
Jnless either s_tere"sis or a"?.:!!!amia is illicit 1 v 1 mported as im-
In any event \.Jhere will Aristotle
then tird his 1;1hich :Jill reject on grounds given under
_!_047a ll=J_?_.L_ll_J._?_, and\'!: What _ __
__ there is no __
The Hegarians.
We had before us Owen's notes; '"hat follows should
be read in conjunction with them.
The thesis under discussion is
stated at 1047a 29 f.:
x can only 1p if* he is .:p-ing, and (the equi-
valent), if* x is not <p-ing x cannot 'P.
(The quest ion was raised
\.Jhether the position of )..L6vov in 30 was not rather odd, hut the sense
in any case clear.)
Six possible interpretations of this thesis were distinguished:
X dnesn't have the ability to (I{) while not r.p-ing),
<The Law of
If x is not '-ing at time t he has thrown away the 11pportunity
,)f at t.
(the 'irrevocability of the ,_f, perhaps
'1e Interpretatione 9 19a 23. Perhaps:
n0t, at t, able to i.p at t.)
has not 'tf' hut '..Jhenever:
')!1 11)4/a
Lf x does not 4' at t, x is
ct. f"Jrther und
'\ l a 1-'
L [f x '.PS at times, :"" L1cks abl J it) Lo .;> n 0ther times.
: if x does not '='at certa1n tir'les, he held ttw abillty to
IJ) dL those The tlwsis nC
4. ff xis nr)C ,;.-jng nmv, he 1ar_ks r:he al)11i.:--y :'l ; 'ne recr)verq
the ability- in r_he way in . .Jhich someo!le i:H1 t a builder can't
h..1ild a house until he learns <-o b:Jild, r)r .1: blind man c,m't until
lns t-J se:: is restorer!),
lf x is not tp-i.ng rhJ:., he
a man is sitting now, h:o. can't up here now), [':'his seems
at first sight chan :., C11t is d-::rually stronger. infers
U1e absence of the ability at present from !"rqrperformance of the .:1cti8n
at present. 1-Jut ledves 1nr:ac.:t the aor'Tlal ;iew tf];;1t abilitif:'s 'lldY he
d'.:qu 1 red (jf lo.::;t. 5 hrJ'iev<:!r implies (bel o.,.J';
;_mplies that I can't get 1.1p 'th(o!re ,1nrt ._ht:!n' (? in Ll.e 'rwxt 'f!G:nent',
1...f there were such a thing) I musL be sitting at- the, moment',
U' ..Jhich case I c.Jn't p;et up dt one Rtter tilat, ,:nd rrn. Given
rhe infinite cf time the sonsequenL absence ot 'next
moments', one Lndeed th:;,.',_ 6 i3 av0ided hecduse, starting
this simply shows that the expression and now' further
ana 1 ys is .
fi. f x: js notf.P--ing now,
( L. e.: he now lacks the
it 1.<:: not ncJssible that r:--er wil.l
ahility l_q T, and lack that
1.bili ty. C E t)e 1 ow on 1 U 14 )
)[ these, 1--2 ':tppe-'l.r to l:Je tr.uLsms, and 3 ts <1 1 _st a thesis
':.hat many have maintaj ned. \ri.'-itotle in this chapter first of all
pr.et the ;-.tegrtrian thesls i_:l ">'-"t\Se !,, and (fr,Jm l047a 10)
lmulying 1:1. fhe lll!tidl sLdlement sEems ,:cm.r':F.lble <.!ith a11y c)[
, t ( h'Jt ror- vit h '',
'if' does olOr-, that DFi t h':' Hegarians
;1:-mselves lntend 4, or is it '1n i.ntecpr, :A.ru:n LliJ:- Arisrotle has
lorced on to th,em in order to i.-?
lmbiguit? Jf t!le r:hesis may reflect confusion Llett...'een
:f 'Plitt, not-p05Sible not-tp-;:;!;t-r (1-i above).
; f 'D at t, not-p:nsible-at.t not-ltl (4-6 above),
rs, tf Lts tmplic.dti.on:; are ,,.-Jt 'l'lt 1
But did the Megarians themselves - whoever they were - make this
confusion, or has Aristotle foisted it on them?
If they did only
themselves intend the former (1-3), it was noted that 1 and
are so
obvious that there seems little interest in stating them; \vhich leaves
3 It was also remarked that Aristotle discusses the distinction
between 'if you are sitting, you cannot stand/be standing' and
truism 'you cannot be-standing-if-you-are-sitting' at 4 166a
24, as an example of the fallacy of composition (and cf. ibid. 20 177b
22 ff ) without suggesting that it makes a special point about possi-
bility, or that it derives from a particular school; he is alive to
the ambiguity both there and in 1. 12 281 b a. [This ambi-
guity is slightly different from the one above; for here it is 1-2
that are
most naturally covered by the latter formulation,
3 as well as 4-6 to be covered by the formulation 'if you are sitting,
you cannot stand/be standing.]
The Megarians' point, it was suggested, was that at any one moment
there are only two states of affairs that may obtain; either you are
<p-ing, or you aren't.
So no room is left for 'possibility'.
is reminiscent of the 'Reaper' paradox, given by Ammonius in de Int.
131. 24 ff., with no statement as to its origin, and also mentioned
by Diog. Laert. 7. 25, 44 (interest of Zeno and Stoics generally in
it; also Lucian, SvF 2. 287. Sedley links it with Diodorus' Dialectical
school; PCPS no. 23 (1977) 98 and n. 135).
l-3 are most
naturally stated in terms of possibility, rather
than of potentiality or ability, unlike 4; cf. further below.
It was remarked that, if 3 was intended in the thesis rather than
or 2, this could have been made explicit if we had had
rather than 6uva.m><tL in 1046b 29-31.
1046b 29-33 can be taken in any of senses 1-5 (though it may be
felt that the example at b 30-32 is already tilting the balance towards
4, because of the most natural way of taking 'cannot build a house';
has Aristotle added his own, loaded, example to the initial
Me gar ian
the sense of ability, rather than possibility, which
already implied by 66va.m>cu in b 29-30?). A definite
thesis? Or is
leads towards 4,
shift to sense 4 comes at 33 ff., ..,ith the additional premiss 'being
a builder is being able to build' ("' 'if one is not able to build one
,1't a builder') where 'ability' foas the force it has in
b 34 (ur
; ,""j be Cl:l.Swer;d in follows).
preferred oUO' as ou't doesn't
N. B.
(i) refers to 'i species
)f \f!Ehl , which may also he immediate; <_ii) )..flAT) forgetfulness
,'IS a characteristic of some individuals, xp6vql refers to an inrerval
"t time 'lfter which anyone mi'Sht expected to l!avR forgotten; (iil.)
loss of capacities with old age; (iv) xp6v!f refers to loss
1f 3.n art by failure to practise j t, where it ,,1ight not always be natu-
.1i. (in Greek'!) to refer to .f.9..E_gett.l!!ai (,) tbe loss of arts thrClugh
nJtural disasters - but this is not relevant where it is loss
hy the individual that is in questl.on.
__ ?: The point of this is clearly to some other way in
,.1hich one might no longer have an drt: but '.vhat is the ;n;piiyua.?
The form of house (quite generally). No wRs felt over
Aristotle introducing his o...,-n in argument l'!.gatnst the t1egarians.
:1tere t:tight be difficulties if the only forms that there are are indivi-
hal forms which pass in and out of existence; but there is also the
Hfficultv nf (:f. (ii) below).
r 1.) the art nf building; the form of house ln the mind of this builder,
.r. in the mind of builders. The mtght \..rel 1 A.ccept that
! he -1rt nf building tn general, as opposed to man's possession
it, ts eternal. But what of Aristotle's belief in the loss of
1 ts in natural disasters and their subsequ0nt rediscovery. so that,
tnr a. t.ime. t:here wo11ld be fnrm of h0use in the mind of nnv builder
'r. anywhere else)? 'Need-for-shelter' might still exist, even if
'l)t ri-te of 'shelter-made-of-stonec;-and bricks'. But we inclined
1ther to the v1.ew that such loss of the art:=.; is irrelevant here. the
dint hei.ng nnly to contrast the relative permanence nf the art with
"lao's loss of it; d.C is cl.n for
1 !he P.Xtra premiss is required 'if -=ometh.i.ng is ho-c Lt j s
1naoie nf being perre1ved dS not'.
' 11r' hIe - ct. 1 n a ').
perceived as hot is hot, or equivalently (II) if a thing is hot, it
is perceived as hot. But Prot agoras' position seems to require the
converse of (II), (III) if a thing is perceived as hot, it is hot.
Howevert, if 'cold'= 'not-hot', and 'not perceiving as hot' 'per-
ceiving as not-hot', then it follows from (II) that, (IV) if a thing
is perceived as cold, it cold. We noted, though, that reduc-
tic-argument does not have its conclusion spelt out clearly, the refe-
rence to Protagoras apparently being thought sufficient without further
explanation, and we contrasted it in this respect with the subsequent
argument at a 7-10. At de Anima 3. 2 426a 20 Aristotle argues that
it is not correct to say that a thing is not perceptible if it is not
being perceived; but that turns on the distinction between potentiality
and actuality, which is what is being denied in the Megarian position
as represented here.
7-10 Why blind many times a day? whenever one blinks, we
decided. But there is no analogous process in the case of deafness;
so one will be deaf, perhaps 1 whenever there are no sounds to hear.
We agreed in preferring Ov <1:p6'xov> in a 9.
How would the Megarians reply? Their position can only be defen-
ded either (i) by claiming that they never intended pnsition 4 in
the first place (cf. above), or (ii) by defining 'blind' in such a
way that not all 'incapacity' to see involves blindness (and analogously,
in a 33 ff., defining 'builder' in such a way that not all 'incapacity'
to build involves not being a builder). Some qualifications are al-
ready ruled out, however, in a 9. A blind person will be 'a person
' . .Jho wouldn't see even if ... and if ... and if .... but isn't a mole
(f.22 l022b 26), etc. etc.'. We agreed that it would be legitimate
for people who maintained position 3 to employ counterfactuals in this
way, and even for those who maintained position 6.
L047a 10-14 This introduces a new point, the denial of change (position
0 above). It wasn't envisaged in a 7-10 that the men would always
be blind; and N.B. l:S'ta.v in 1046b 29-JJ. 6 is implausible if stated
104 7a 4
in !:erms of ability - it would be required that a man who isn't <P-
1 ng now
why not?
doesn't have the ability to and 'Hill never recover it; but
But it is implausible enough tf stated in terms of impossi-
bility; tf p is impossible, it ,.jill never happen. And the shift from
ahility, or rather its absence, to impossibility is expressed at 10-
ll. Aristotle's conclusion here, it was suggested, results from his
illegitimately importing a tenseless type of impossibility (it is impos-
sible, tenselessly, that x will at any time) into an argument which
negan with a tensed statement about potentiality or ability (if x is
not -ing x does not have the ability to q>)
In 11, we preferred "(Lyv6)J\IOV the argument requiring 'what is
not (now) in process of coming to he'. In 12-13, that what cannot
c0me to be is isn't obviously false - it might be that it cannot
now come to be because it already. 'is' is therefore being taken
as equivalent to 'has come to be' (eternal existents not being in ques-
tion), and 'cannot come to be' means 'cannot come to be at any time'
The reference to 'being or being-about-to-be', excluding any refe-
rence to the past, in 13 recalls Diodorus' definition of the possible
There are three possibilites: (i) Aristotle has Diodorus' in mind
here - unlikely on grounds of chronology, though not, we thought, to
be entirely excluded on those grounds alone. But there is no reason
t 0 identify Diodorus as a Megarian ( c f. Owen's notes). ( i i) Diodorus
had this passage of Aristotle in mind in formulating his definition.
(iii) The similarity is not a matter of influence in either direction,
but simply reflects the fact that the past is in a sense necessary
.1nd unchangeable. It would be implausible, in some contexts at least,
say that a thing was possible because it had happened in the past,
1f it '.Jas certain it would never happen again.
_'c::)4'-'7'-'a'-=.c13,_--'1c:.4 The tense of lofu.;.a.LYEV we felt, doesn't indicate a refe-
renee back, but is analogous to that of "tO "t'C erva.L. It must mean
'implies', rather than expressing an equivalence, if 1:ve are to escape
tr.e position that whatever doesn't and won't happen is ipso facto impos-
Could this be avoided if "t'OU'to = not 'what neither is nor
be', but rather 'that of which it is false to say that it is or
\.:i 11 I ' ?
lt is false to say that the diagonal will lle measured,
one interpretation of ') it isn't true ro say of a C<.)n-
li ngerlL
Sl'u-hattle that it >von't occur tomorrow.
But this won't work,
because, with such a view of truth-v.'llues, 'not true' doesn't
irlply 'LJlse'.
(Boethius, in 2 215. 6 ff., refers to people
\.Jhn tr,uk Aristotle to be saying that both 'there will be a sea-battle
tnnorruw' and 'there won't be a sea-battle tomorrow' -1re false, but
r<-:J<:>.cts this interpretation.)
lr was pointed out that the imperfect lmiJJnLvt:v indicating a gene-
t tlly nccepted truth would have been easier if Aristotle were not,
at this very point, introducing dO implication of d:.06va."t'ov that had
been in question in what preceded.

Compare the second premiss of Diodorus Master
:\t gument, d.OUva/tOY IJ.i-1 &xo\:.lu9c:t'v;
the sense of rixo'A'Jt7e:tv is
here expressed hy the combination of lQ,_, and S'o""'tUL - the latter being
.JH rather than a temporal future.
Having stressed that
t -;\. is different tram Aristotle has to ind1cate what ::.u.;u':6v
Rut ls this meant as a definition of 'possible'? If so, it simply
\ei:ld5 immediately to the need for a definition of 'impossible', and
u1 looking tor this we are in danger of being forced down a Hintikka-
type road, the impossible being what never happens.
However. in the
next chapter we have an example of the impossible, that which can be
proved to be impossible because it leads to a logical contradiction
(104/b 9-12; cf. ps.-Alexander 575. 4 ff.).
Not there just a matter
\lt what never happens.
Anxiety was never expressed about the extreme
;:enerality of 1047a 25-26; but we decided that this was quite right
the possible cannot have impossible consequences, hm.;ever remote.
The general question '.vas raised, how far Aristotle's protests
against the !1egarians in this chapter (and N.B. that it is only a vague
,r J'' in l046b 29) were motivated by anxiety about possible deterministic
Lmplications of his own position; cf. 'necessary' in 6, 14.
n:; separates the possible from the actual by introducing qualifications
(a fJ-7, a 14-21); there is a similar concern to separate the potential
.1nd the actual in A3,
[s (he Megarian thesis interpreted in the ex-
0 7
i !)4/a 1 J
} I )4
ttme lorms and (6) (cf. p.hL <Wove LO it easier to dis-
.e.' may relnforce his own view that potentiality
Ls distinct tram actuality, t)y implying (in (6)) that denial of this
1n.0lves the denial of all change.
20-h2 Explanatiun nf how term connected ({)UV't't6E!llv'Tl)
.nth ur1_ginates; link hetween tv!.I)YELa. and xCvwtc;. N.B.
.c;1l. (= in 3li implies that there are other examples of XLv1)crE.tc:
,_hat are 8vpye: (perhaps in non-technical sense?) but not lv'te:)..txe: La.t
Tt '..tas that 1047a 34 .... fluld have given Brentano a
hetter example of intensional objects in Aristotle than does the doc-
of perception; cf. also 'the house this man is now
l n Z7-9. If 32-33, by givLng a consequence of 32, is intended to
it as a reason for 31 - i.e. if the force of yO.p in 32 extends
Lhe cla11se aS well- the implication might be that motion _h!!
is not predicated on non-existent things; 1.16\LO"'t'a. being
carried nver from 32-33. But why should motion not
l>e predicated of the non-existent? Readiness on Aristotle's part
r-o that the Centaur in this picture is brown, but not that it
is. galloping? But the implications for other art-forms will be diffe-
:1-:nt; this man playing Oedipus is walking, but he isn't blind. And
"1 B. "tt.\lat; in 33; are the predicates that admissible for non-exis-
' in fact to mental predicates, like those that follow
t n 1'd If so, the connection of thought in 30-35 as a whole seems
rather loose. (Is J0-31 explaining, not just the origin of the term
hut it is linked to lv-c.s\EXEta because SvtpyELo is
to x(vT)(Tt.t; to Ov""tu, and Ov'tct to l:v-re-
If so, the more necessary v.e suppose 'in particular' to be
n r his line of argument, the more awkward would be a concession in
2-33 that there are many other predicates, as well as motion, that
.to not apply to non-existents.)
it pu1nted out that the examples in 34 are of thinking, rather
l!an of perceiving, only real objects being perceptible. Hon-existents
de not the only things that c3.n 't move (change): points and forms
an t ', find nor can the !:nmoved Hover, which is in fact the :;u-
preme example of (Ross thinks Coxe:t in 32 is intended to
take care of this last point ) The argument, in any case, turns on
popular usage; N B. the 3rd person plural in a 33.
Only some of the things that are not, are potentially: explai-
ning be in actuality' in the previous sentence. (But note,
in connection with the Principle of Plenitude, that only two cases
are recognised here; things that are not and cannot be, and things
that can be and will be - in this case, or in some case?)
by Sarah Waterlow
'What always is cannot not be' This is one of Aristotle's stran-
ger positions, as is the associated doctrine that nothing is possible
that is not actual at some time. In !_Je Caelo I 12 he purports to
demonstrate the first of these, and that there can be no doubt that
he is entirely satisfied with the proof. He unfolds it there with
an air of total assurance, and there is no sign elsewhere that he ever
came to doubt its force.
not to say embarrassing.
but ludicrously inept.
To modern eyes this is incomprehensible,
The reasoning seems not merely inconclusive
it appears to turn on one or another puerile
fallacy. I shall argue here that this is a false impression. If
I am right, the De Caelo argument is more subtle and also more cogent
than is generally supposed.
Comment has mostly centred on the passage 28lb 2-32, with parti-
cular criticism directed at lines 18-23. Aristotle is regularly ac-
cused of a fallacy here of division, to wit of illicitly transferring
the modal operator so as to convert 'It is impossible that X should
always be and ever not be' into 'If X always is, it is impossible that
X should ever not be'. This simple diagnosis is unconvincing. It
is true (as I shall argue) that he does here perform a move that is
easy to misconstrue as a fallacy of the above pattern; but it can hardly
l04/b 2
te tne fallacy ttseif. ror in the lillmeuiately preceding passage ( 15-
17), in connection a different example, ,\ristotle's avoidance
of pr::::cisely this lype of mistake Is so deliberate as to almost
to an explicit warning. It is impossible that a man should ,.;hen sit-
ring be standing, but this rioes not entail that ' . .Jhen he sits it is
L'Tlpossible that he should be standing. Fttrthermore, if anything in
this part of the text is clear, it is that for Aristotle the crucial
difference between the two cases turns on the terms 'always' and 'for
an infinite time' (here treated as equivalent), whose absence somehow
leaves rofJm for an assertion of possibility which their presence seems
The fact that for him the omnitemporality ..1f a state
of presents a special bar against the possibility of 1ts oppo-
:.ite is left unexplained by the hypothesis that the trouble here is
1 simple fallacy of division.
To speak, as I just did, of trouble' with the argument will
turn out to have been rtui.te inappropriate on the view to be proposed
here; although it is perhaps not altogether inapproprtately ambiguous
15 to whether Aristotle here is the one j_n trouble, rather than some
1f his critics. He for one I hope to make plausible) knows what
he ls dning, at least to the extent 0f avotrling any text-book fallacy
And the fallacy of division is not the only lapse of that sort that
a quick glance at this oassage seems to discern. Hy point is that
..;hat he is here qu1te knowingly engaged in is a somewhat elaborate
performance governed by principles not only different but logically
Lndependent. however much the smoothness of Ar.istotle
s execution may
\.u:!ep us from attending to them singly. Of these, the one most properly
a formal nr analytic (as opposed t0 a principle,
is che hy ...,Mch he tells 115 he applies nnd withholds the term
1 hl e': Lhe rurP"nt r::eaning of '.;hlch T shall (for L1ck of other
_ l c<:>.s) assume to ftxei by its use in accordance with the rule.
With one important f11Jalification, this rule is the nne frlmiliat
13, 34a 25 ff., which reappears at e.g. !1etaph. 9 3,
24-28 and J, 104/h l0-11. 'The possible is that '.Vhich is not
nrus-;ary, but iVhich if we supp0se it the case has no impossible con-
c>equences'. (In the An. Fr. passage Aristotle '3peaks of 't'O lv6sx6-
l1:vov, tn I 12 )f -r'> 6uvr:t"t6v, and in 93 he uses the terms
It will be clear later that 6uva:t6v is more suitable
for the De Caelo context, although not for reasons of logic.
proof in I 12 depends on a mataphysical principle concerning
and the wording of the formal rule for 'posssible' simply facilitates
the assembly of these two elements of the argument.)
It is more useful
to regard the Prior Analytics statement as a criterion rather than
a definition, not only because the circularity does not then matter,
but because it is as a criterion that Aristotle needs it in a number
of passages where it figures. These (which include Metaph. 9 3 and
De Caelo I 12) are passages where he is concerned to distinguish and
support the distinction between what is impossible and what is merely
false. A definition of possibility (whatever that might look like)
would be useless for that end unless it were also a criterion, and
a criterion that fails to be a proper definition will still do what
is required in the context. However, for the criterion to be effective,
we have to be able to recognise some impossibilities (the impossible
consequences) straight off without a criterion. We also have to be
able to tell straightaway in at least some cases whether something
from something else. In De Caelo I 12 the self-evident
touchstone of non-self-evident impossibility is the entailment of an
explicit self-contradiction: the same thing both is and is not dt
the same time (cf. 28lb 22-23) . Whether Aristotle means his rule
always (or even ideally) to take just this form is a question I shall
leave aside. Nor shall I here consider what sort of modalities are
involved in the statement of the rule itself (as opposed to those esta-
blished by use of the rule). This is an important question for any
overall account of Aristotelain modality, but it will not affect the
present task of identifying the man ingredients in the proof of 'What
always is, cannot not be'.
The special feature of the rule for 'possible' as it appears in
that proof is an explicit reference to 'another time' (281b 17-19;
cf. That is, a false statement, that X (actually sitting)
is standing, is possible iff nothing impossible follows from supposing
it true at another time. We may find this addition perplexing; but
this, I suggest, is because we are accustomed to thinking of possibility
and its modal fellows as absolute. The contrast is with 'relative',
t.-:(1 'rh- is not i.Jetween 'categorically asserted' and
',,n-,ln-hypothesis'. Aristotle !Tlakes the latter distinction at 3-
IJI3' -1ls0 makes Lt clear 1.n 9 ff. that his rule i.s primarily intended
lO just tiy categorical assertions of possibility and impos-
,;; 1 hili r y*.
But what is thus justified is, t believe, a relativized
There is not soace to develop in full the case for this
lnterpretation. Briefly, this ts the argument: the supposed
truth of (false at .r:_) to a time other than !]_ (For if nothing
impossible .Jere found to follow from counterfactually supposing 'E'
Lrue at :_ itself, surely that would prove 'e' possible just as effec-
t:vely - and in that case one need not mention any time rather than
.Jny othet".) Answer: The 'other-time' reference not otiose if
1..and, think, only if) we take it that Aristotle is telling us to
r;xa1n1ne the consequences not of 'E' alone, but of 'E' co sidered to-
1Nith the situation in which '.e.' is false: in other words, of
the conjHnct.ion of 'g' with 'g_', where 'g' describes the state of af-
fairs in which we are prompted to ask whether 'e_' is possible or im-
passiblP.. That, of course, is when 'E.' is false (or taken to be so),
since if we rhink it true we don't need to ask that question. The
p0int is, that unless different time-references are assigned to '.e.'
,J.nd 9.', rhe conjunction just mentioned would inevitably entail an
iiJlpossibility, no what 'E' may be. The test has no chance
0f proving any 'e.' possible (given that it is false) unless 'E.' is
::;JJnonsPct at another time: assuming, that is, that the relevant
are not those of '.e.' alone but of On that as-
sumpt1on. the test is not a test (the game is lost in advance) unless
times at"e differentiated. That is one way of putting the matter,
".Jhtch is especially relevant when what we want is a real criterion
f<)r !01Jbtful possibilities. Rut the essential logical point could
:tlsn he pUt <lS foll.Ol..IS: SiVen that
iS false at _.r:., it iS impOSSible
:-hat E 1le true at t (impossible, that is, given, or relativelyto,
r.he actual state of things at !:_). That holds for any '.e_'. and is
...nat o\ristotle has in mind when he says at De Int. 9, 19a 23-24: 'What-
r.>ver ts not. !wcessarily is not it is not'.
rn rhe last quotation I take (claiming no originality) 'necessarily'
qo':ern rhe first 'is not', and the phrase '!lhen it is not
Fnr dS such used in another relevant contrast, see below.
when the necessity obtains What we have is: given that it
is necessary at !_ that -E./!. This makes sense if the consequences
to be vetted in the suppositional test are deduced from a conjunction
in which one conjunct says how things are at !_, when '_e' is false
For if the necessity of (or the impossibility of is rela-
tive to that state of affairs, it is reasonable to assign to the neces-
sity itself a time: that of the state of affairs relative-to-which.
Now it is not the case for all 'E' false at that some absurdity fol-
lows from conjoining a description of things at with the supposition
'E.i!.*' ,provided that !* When there is no absurdity, it is pos-
sible relative to the way things are at ! that Hence if, for
instance, !_ is the present, it is possible now that E./!.*. E.g., it
is (present tense) possible that X (now sitting) should be standing
at some other time. But if there is the possibility of his standing
at some other time, there is the possibility of his standing as such
(not only of his realizing that general possibility now). Aristotle
says at 281b 15-16 that X h. 'tt)v 66v<J.I.LLV of standing, and according
to this interpretation he means the present tensed quite literally.
The literal sense is appropriate if the modality is relative to a his-
torical state of affairs. Moreover, what is possible/impossible rela-
tive to, or at, one time may not be so at another. Thus at !_ is is
impossible that the man in our example be standing at !_ (though not
that he be standing or at some time or another). Thus at !_ it is im-
possible that the man in our example be standing at !. (though not that
he be standing at some time or other); but at some other time !_+ (it
is most natural to think of it as prior to !_) it may have been possible
that he should stand at !_ (whether or not he actually does so then).
There is more to be said about the concept of possibility as rela-
tive to fact and temporalized. But here I shall only call attention
to the light immediately thrown on the accusation that the argument
of De Caelo I 12 rests on the fallacy of division. In terms of this
sort of modality a move is in general admissible which verbally re-
sembles a case of that fallacy, but which in those terms is valid.
We have Sl: 'It is impossible that E.l!. and -E_/!_'. In any system
Sl entails S2: 'If E./.!. then not -E./!' But in terms of relative
possibility Sl also entails what we cannot get in jllSt any system,
namely S3: 'If e/t then it is impossible at !:_that The appea-
ranee ;_n S3 of an impossibility-operator as part of the consequent
is exactly what ..... e should expect the logic of temporalized relative
modality to generate from the premiss Sl. But what is not thereby
generated is the transference of the operator that introduces Sl itself.
For the one which appears in the middle of 53 is a different though
resembling operator. It says 'impossible at !:. ', while the one in
S1 says simply 'impossible', i.e. 'impossible without temporal restric-
tion', or 'impossible cbtA.Wc:;' (cf. De Int. 9, 19a 26). Whether in the
Aristotelian context it is better to think of this strong operator
as timeless or as omnitemporal I shan't consider here. The point
i.s that if by fallacious division it were being transferred, then from
S1 we should have not S3 but S4: 'If .E_/!_ then it i.s impossible without
restriction that This entails (if it does not actually mean)
Lhat there is no time when '-.e./!.' is possible. Thus, e.g., it is
not nnly impossible (given that /l), but always was impossible.
Aristotle warns against confusing S3 with S4 at De Int. 9, 19a
26 (except that there he makes the point with 'necessary'). His alert-
ness to the fallacy in that passage does not of course prove that he
did not commit it in De Caelo I 12 while trying to demonstrate that
Nhat always is, cannot not he. There are independent reasons why
we cannot take it for granted that the thinking of the two passages
harmonizes (see below). However, it is difficult to see how, if he
had been in a mood to commit it in t2e Caelo I 12, he could have been
so confident there as in fact he is that logic supports common sense
in allowing us to say that X' s actually sitting at !_ is consistent
.;ith the possibility at t of his standing (sc. at some other time).
ro explain. A person who overlooks the difference between S3 and
S4, ahd who reveals his carelessness by, for instance, inferring 54
from 51 'Nhen he is only l!ntitled to 53, is guilty of an error that
falls under the general heading: failure to distinguish the use of
simeliciter from its use qualified by some restriction (where'a.' is
any term). One special (although still quite general) case of this
"':rror occurs when the restriction and non-restriction are
L.e. when 'a.'/!:. is confused with'a' without temporal restriction'.
This temporal case in turn has various possible instances: e.g. 'a,'
may be the operator 'it is impossible that', and it may be a non-modal
proposition 'E' or '-E'. I make the asumption that someone who avoids
the temporal restricted/unrestricted confusion in one instance is likely
(although not of course bound) to avoid it in another in the same con-
Now the wrong move from 51 to S4 is not only a fallacy of clivi-
To diagnose it as such requires no particular awareness on
the critic's part of the logic of temporalized relative modality.
But that move is also (and given that awareness can be seen to be)
an example of confusing 'it is impossible at ;' with 'it is impossible
without temporal restriction'.
Seen that way, the move is not from
S1 to S4 direct, but goes via S3, which is wrongly replaced by S4.
Another wrong substitute for 53 would be SS:
'If then it is im-
possible at _t. that -.e. without temporal restriction (or: impossible
at !:. that -.E_ ever). For 'e' read 'X is sitting' and for '-.e.', 'X is
standing'; and it is clear that in De Caelo I 12 Aristotle is denying
that Sl yields S5.
Hence I conclude that he is unlikely, in the same
chapter, to have made the parallel mistake od moving from Sl to S4.
(We can see that it is parallel only when we realize that relative
modality is at work here,
It is not parallel if regarded simply as
a fallacy of division, as is usually done.)
One of Aristotle's objections to the Megarians in 9 3 re-
inforces the view that for him the inference from Sl to 54 is wrong
in the same way as an inference from S1 to SS would be.
When both
are seen as mediated by S3 (innocuously reached from 51), then they
can be viewed as presenting analytically distinguishable effects of
what would in fact have been a block refusal to bother about the dif-
ference between the temporally restricted and unrestricted.
this block refusal, Sl (via S3) yields S6:
(and never has been) possible that --e_ ever'.
'If e./!:_, then it never
S4 and SS are merely
different aspects of this.
Thus, e.g., if X is sitting now, it is
for ever excluded that he should ever do other than sit.
Hence change
is impossible.
This is one of the paradoxes Aristotle derives from
the Megarian view that only the actual is possible ( l047a 10 ff.),
a fact which has often seemed puzzling; the modern determinist holds
that whatever happens has to be as it is, but he does not deny the
fact of change.
But Aristotle's criticism is apt if the Megarians,
7 5
like n
m, are think.tng 1 n terms of relatlve modality (hence, like him,
the inference tram Sl to S2) but, unlike t1im, refuse to care
ab0ut temporal rcstr .tctiuns.
ln that case they then get straight
to S6.
l<.'hether or not thls is hnw they proceeded, it seems
that Aristotle thought it
I must return to the main argument.
Aristotle's rule is: Given
that it is pos.:>ible 3t !:_ that -.e_ iff nothing impossible follows
from & -E./!:_*' (where 'q' represents the facts at
I have argued that with this governing his reasoning,
that reasoning
is unlikely to be marred by the so-called fallacy of division usually
alleged. lt is, however, another whether the rule can sustain
his desired conclusion. By itself the rule certainly does not entail
any more than it entails that '-E' is possible at only if it
tr..te at some other Lime.
In fact, it is not at all clear how it
even ilpplies to the case in which Aristotle is interested
The ques-
t LOTI now to be answered by means of the above crj.terion is: Given
that r. always. is it possible that -E_ (ever)?
But the criterion as
1 t
dve stated it. on the basis of Aristotle's treatment of the sitting/
tanding illustration, is concerned only with relative
(') facts at Jne or another time.
It seems that Aristotle is
hoping to extend it to a case ..1here the given is (and is given as)
It also seems that in proceeding thus he pays no atten-
to the lor;!;ical complexities introduced by the term 'always' that
'Here absent in the original example.
',.Je may ..1ell feel that he might
have been a little more helpfully explanatory on the legitimacy of
extending the rule to cover this apparently very different sort of
Hut without demur he declares that the suppositional test which
;-he sttting man's possibility of standing also shows the omni-
being's !..._fi!.pOssibility uf not heing (or ceasing).
tt may seem that in terms of relative possibility this
For a moment
works. For
(-...-rhere '(l' '->tates the tacts as Lhev are - facts ..;hich
'llilke it talse the the conjunct1on of 'q ah1a.ys' with the supposi-
'Ji '-p_' implies an i:npossibility.
referred to 'another time', this can lmly mean
If the supposition
a time not included
:nrl.er ''allvays"'. which is absurd; if it is not thus referred, then
1n referred to a r-1me within "J.Lways",
which case the
. \F fER 3
conjunction entails a straightforward contradiciton.
This at any
rate is Aristotle's argument (2Blb 18-25).
Even if we allow that the modality here is relative to fact, not
ahsolute, we sense a trick.
This feeling begins to take shape as
soon as we recall that the relativity entails that the modalities are
We are entitled to ask: is it impossible that -e. (given
that always :e)? The question shows what is wrong. The answer has
to be: when the given is given.
That is to say, to reach his conclu-
sian Aristotle has to treat the omnitemporal fact as if it were, so
to speak, actually given all at a blow.
For only so can he argue
that there is no time beyond the time of the given to which the supposi-
tion of its opposite may coherently be referred. But, by the same
token, there is not time at which the omnitemporal given, considered
thus en bloc, is given.
When all time is over, then it is given,
i.e. no-when at all.
But it is only when it is given that the impos-
sibility obtains which his logic seeks to prove,
Hence that logic's
true result seems to be that it is never (no-when) impossible that
-E given that always ! What he wants, but (as it now appears) is
totally unable to justify, is the relative impossibility of '-.e' at
each and every moment Qi time, as distinct from its pseudo-impossibility
(likewise relative). In any case, if such a pseudo-moment were not
absurd, '-.e' could be shown possible now by coherently referring its
truth to the pseudo-moment.
Even pseudo-moments beyond time cannot help Aristotle here.
Forgetting these now and considering only moments, or, if we prefer,
finite periods, Q.f time, we can say this: For every moment or period
!:_, (a) it is the case that E/!:. and (b) there is a t* -= t such that
the supposition that
-pJ'-* together with a description of the facts
at t entails nothing impossible. That is to say: it is always the
case that .e. anq always possible that -E Aristotle's logic cannot
show that this might not be true. One 'the given' is taken bit by
successive bit, there can be seen to be an infinite abundance of 'other
times' to which to refer the negative supposition without incurring
contradiction or any other apparent absurdity.
Thus the argument of I 12 proceeds as if the author sees in 'al-
,;ays' none but a collective, or perhaps [ should say holistic sense.
But how could anyone be oblivinus of che rlJ:stribut ive sense in terms
of which the proof gets nowhere? Or if for his reasons Aristotle
chose to ignore this, how could he expect readers to connive? (Remem-
ber that his immediate audience would have come to this passage undis-
tracted by morages of the fallacy of division.) Soon I shall begin
to develop that aspect of Aristotle's position provides him (so
I shall argue) with a respectable, even if to us not particularly agree-
able, answer to this question. But let me first prepare the ground
with a general observation relevant to any diagnosis that locates the
mechanism of the argument in some fallacy or confusion occurring within
the passage we have so far considered, namely 28lb 2-25. Sheer thought-
less equivocation over 'always' would fall under this head, as would
the muddle alleged by Prof. Hintikka, whose explanation has not yet
been mentioned. Briefly, it is this.* Hintikka is rightly dissatis-
field with the 'fallacy of division' verdict. It is not that he denies
that the fallacy occurs, but he sees it as symptomatic of something
else. The underlying problem, he suggests, lies in the suppositional
criterion for possibility as used in the context of Aristotle's view
of time. Hintikka does not say with any precision what aspect of
this view gives rise to difficulty. He holds simply that an obscure
metaphysical pressure emanating from this quarter inhibits Aristotle
from supposing '-E.' true at some moment 'in the actual history of the
universe' (Hintikka's phrase) when he knows or believes that the actual
facts at that moment would render '-.e.' false. Thus inhibited, Aris-
totle cannot even entertain as possible what he takes to be never true,
not because his criterion proves it impossible, but because he cannot
rationallY. frame the supposition to be tested. sha 11 not try to
reconstruct the sort of thinking that might conceivably land someone
in the position of denying himself the use of the very criterion that
he himself recommends. For if one holds that '-e' is possible only
if it is eventually true, one must wait to know of the truth before
asserting the possibility; thus supposing it true and seeing what fol-
lows is a waste of time: it is an insufficient guide to possibility
before the truth is known, and an unnecessary one afterwards. However,
it is enough to say here that only a very confused thinker could have
heen Lerl to this position from Aristotle's publicly expressed views
'!!APTER 1
about the nature of time.
This is not intended as an objection to
Hintikka, who ts very far from maintaining that Aristotle's thought
about time and modality are wholly clear and coherent.
Perhaps indeed
we should expect from him rather more than Hintikka seems to,
At any rate' Hintikka 's account of Aristotle's inference from
'always' to 'not possibly not' faces a common difficulty.
If it is
due to confusion about such topic-neutral matters as his own criterion
for 'possible' i or the difference between supposing something to be
the case and asserting it to be so; or, for that matter the varying
forces of the word 'all' - then Aristotle ought by to draw his
crazy conclusion
which it holds.
universally, without restriction on the field for
Thus, for instance, we should reject the existential
interpretation of 'what is, cannot not be', favoured by Guthrie (Loeb
tr.) and Stocks (Oxford), since this is far too narrow. We need:
'For all E_, if always E. then not possibly -_e'.
But this formulation,
which for convenience I have several times used in the preceding para-
graphs, is open to a serious difficulty. The proposition entails
that for all _E, if it is possible that -.e. then -eat some time. Thus,
e.g., if it is possible that --9. (i.e. that g) then --9. (or g) at some
time, for any q. But this conflicts with one
of Aristotle's plainest
views expressed elsewhere (e.g. 9, 19a 9 ff.), to the o::!ffect
that some possibilities may go for ever unrealized.
This is not only
it is also logic, since in some cases if one of a set
common sense,
of contraries is true no other member of the
set can ever be. Thus,
for instance. it
is possible that this coat be cut to pieces, but in
fact I mean to keep it until it wears
out: and even if I don't, it
1s still possible that I shall.
Hintikka hopes to resolve the conflict
(inevitable on his account of the modal-temporal connections) with
the idea that it is only general possibilities that have to be realized
in one. or another instance, but not in any particular. Thus, the
possibility of my being murdered fortunately requires not that I shall
be, but only that some creature somewhere meets with a sticky end.
But this cannot be reconciled with pJ':._ Caelo I 12. since Aristotle is
clearly arguing that if any individual always is, g_ cannot be (see,
e g 28lb 32-33):
which entails that if the individual's not-being
is poss1ble, the possibility is realized in that \'ery same case.
Jhall below chat this l3 not JHt ,Jf ltne wlch statements such
:s that i.n 9 (see also Metaph 3), s1nce there is indeed
1 "" the qr.ope of the proof in De Caelo I 12. But the
P'Ji.nc of inmediate tmportance is that in the passage .,e have so far
(28lb 2 ff,) no grounds for restriction appear.
This, if nothing else, justifies ns in steptJin)l; aside from what
... 1ne might call the logical nucleus nf Aristotle's argument in the hope
of a hetter The answer to this and other problems is near at
hand: have only to go to the preceding Lhapter. Here Aristotle
has a lot to .say about the possibillty that is entirely ignored by
iHnt ikka. Perhaps this is because he is at the same time, and in
r.he terms, talking about what we should call'capacities': which,
it may be, are at present of no great interest to the formal l\)gician.
:\t :1ny rate, Aristotle stJrt'3 by introducing various senses of 'in-
v,enerable' and 'imperishable', ending up with a pair of master-senses
iefined in terms of the impossibilitY of beginning and ceasing. These
t.n turn he "immediately spells out as :neaning the impossibility of not
being and then being, and versa (28la 16). After this, being
-lnd not being, and the possibility and impossibility thereof, are not
ment toned again until the beginning of Chapter 12 ( 28la 28 ff.).
'1eanwhile, Aristotle takes an apvarently sideways look at certain other
such as for lifting and for .;alking, where for any given
:::ubject there is an inherent limit to thP. exercise. fhe essent tal
p<1 int as far as Aristotle is concerned, is that such possibilities
1r caiJacities ought to be defined with reference to the limit or maximum
)r fullest realization (28la 8; 10-12; 11+-15; 18-19). This is not
"O much concept1al analysis as concept-cr.nstruction, although not un-
hy cnmmon The man who can lift up to two hun-
ired pounds is tu be thought of :ts exercising a specifically different
apacity from the man who can lift unly a hundred and fifty - even
,'hen they each lift the same '.Jei.ght. We might say that the millie-
aire's rlonation of a mite is indeed a c;pecifically different act from
r- hat nf the pauper .Yidow, and i.t is worth observing that this does
'lOt 'iepend on a difference in the agents' intentions, which after all
he the same. So it is not on that account unreasonable of Aris-
t Jt1e :-o nrrJcPed, as he does, tn ipply his ,.,)nstrurtion to cases \>Jhere
intention and volition play no part. What he is concerned with is
the extent of power, of whatever kind and whatever the agent, but in
particular with one general sort of power, namely for 'being' and 'not
Now one might well ask 'Why this sudden interest in maximal perfor-
mances?' - seeing that he is engaged in a general discussion of possi-
bility or capacity, and for many capacities the question of a maximum
logically cannot arise. Aristotle, however, has no reason to disagree
with this point. For he says here not that all capacities, but capa-
city in the principal sense ('tO xuplws 6uva.'t6v. 28la 19), must be de-
fined with reference to the maximal exercise. Now, capacity in the
principal sense is for him, presumably, capacity for being, where being
is also being in the principal sense, i.e. being q:> for some value of
tp falling under one of the Categories. Accordingly, when at the start
of Chapter 12 he resumes the discussion of possibilities and impossi-
bilities of being and not being, he takes care to say that he means
possibilities or capacities (and their opposites) for being 1l or not
'!' for some categorizable '!' (28la 30-33). Nor does he signal any diver-
gence from this in the rest of the chapter. Hence 'is' and 'is not'
must be understood as copulative throughout, with the omission of the
complement functioning positively as a complement-variable Thus
the existential translation is wrong; but so also is the propositional
translation which equates 'being' with 'being the case' or 'being true',
since not everything that is the case can be expressed by a predication
in an Aristotelian Category. Here, then, is the restriction sought
earlier on the scope of the doctrine that what always is, cannot not
be; and it fits perfectly with the De Interpretatione example of a
possibility never realized (and also with the remarks in Metaphysics
E 3). Possibilities for coming to be and ceasing to be (as being
cut up for a coat) presuppose possibilities of being and not being
(or being not), but are not to be classed amongst them, since coming
to be and ceasing are themselves not ways of categorizable heing (al-
though, again, they presuppose them).
Thus it is those capacities or possibilities that are primary
and central for Aristotle's logic and his metaphysics that require
to be specified Ln terms of a maximal exercise. The plausibility
_____ -.; ___ .. ______________________________________ .,.,_
of this position depends 1n (Jn wh.-1t he is nere prepared to count
as a Category, and this we are not .:.old. At Jl-33 he tails off (as
so often) with 'etcetera' after mentioning che traditicnal first three.
At any rate, by now the discussion has moved to a high level of abstrac-
tion where these primary cases are gathered together under a single
logical pattern. '..Je are no longer to be concerned with different
types of limit such as he was at pains to point out in the preliminary
illustrative passage 28la 7-27, where, for instance, he explains that
in some cases (e.g. vis.i.on) it is the degree of minuteness of objects
that measures the magnitude of capacity, If, now, we consider all
and only categorizable being, the only dimension so universal as to
apply in every case is, of course, time. Thus his primary possibili-
ties are for being q> for some maximum duration. In other words, it
belongs to the very essence of what it is for a given thing to be cate-
gorizably q> that it should be so for some definite period which can
at least in principle be specified: and this temporally determinate
<.p-ness is what it is the essence of the corresponding possibility to
be a possibility (For the linguistic viability of taking the
duration terms to qualify not Ouva.'t6v but tts complement, refer to
C.J.F. Williams, Religious Studies I, 1965, pp. 95 ff. and 203 ff.).
shall not dwell on the 'Iletaphysical considerations that for
Aristotle make this a solid starting-point. It is enough r.o $3Y that
despite his sometimes almost obsessive insistence on the eternity of
the universe and of natural kinds, as well as of s0me particular enti-
ties and processes, Aristotle is as far removed as it is possible to
be from those who would explain these and any other phenomena by what
we may call 'metaphysical inertia'. \.Jhate"er substantially is, is
for J1im governed throughout its spatial and temporal beings and doings
by an lnternal princtple of unity that finds expression not in :1ny
repetitive 3ameness going on .Jnd on indefinitely until suppressed or
redirected from without, but in an inherently hounded pattern of acti-
vity whose continuity is more than simply freedom from interruption.
Indeed one might say that the absence of interruption is not even its
necessary condition either, since there is nothing of a definite nature
to interrupted unless there has already emerged some project whose
identity was intact despite sudden removal of the conditions suit-
able for it to be fully displayed.
In most though not all
cases such natural patterns develop through series or cycles of empiri-
cally distinguishable stages. The diversity of these is consistent
with the overall unity precisely because it is more than mei"ely consis-
it is actually necessary, if that pattern, whatever it may be,
is to succeed in being produced.
It follows that each stage, no less
than the whole sequence, is inherently bounded as to its duration.
For if the raison d etre of any one condition is its contribution to
some necessarily successive whole, then that condition must be so to
speak spontaneously ready to bow out to its successor after whatever
is the appropriate time. The actual mount of time, whether for stages
or whole sequences, will depend on the object concerned (cf. De Gen.
et Corr., II 10, 336b 9 ff).
All this sets the scene; but for his current purpose Aristotle
needs one prop which even the most arrantly teleological metaphysics
cannot really provide, and all he can do is mime it into being by what
is indeed a kind of logical mime. His argument requires, and its
premiss claims (28la 28-31), capacities for negatives as well as for
positives, and that both alike be defined by reference to a maximum.
(Throughout, not being and being not-cp are treated as identical.)
Here if anywhere would locate the sophistry of the argument. In
this respect it is secured by nothing more than the verbal appeal of
a totally schematic symmetry.
weight-lifting cannot help here:
The original analogies of walking and
indeed, it is best to forget about
For what on earth could be the limit of my capacity for not-
What does it even mean to speak of ?___ capacity for this,
as distinct from the various capacities for positives whose exercise
is excluded when I am engaged in lifting? This, we have to say, is
only paper-concept-construction. Teleology will not make intelligible
the proposition that is intrinsically times as to the duration
of its not-tp-ness irrespective of what that positive characteristic
might be from which it eventually returns to being
But it is on just this eventual return either way from one limited
contradictory to the other that Aristotle's proof depends. That as-
sumed, it follows that whatever has the possibility of being tp and
also of being not-11 will inevitably realize whichever of these is cur-
rently unrealized.
That presupposes that it oorh its possi-
bilities, but this is never questioned.
Another question on which
Aristotle has nothing to say here concerns the status of the 'it'.
Where is a predicate 1n a Category uther than substance we may think
ot" the subject of change as an individual substance, and the alternation
may be supposed to continue for as long as the substance exists as
locus of the contradictory phases.
Thus the alternation need not
be endless, since not all substantial individuals are eternal.
where is a substance-predicate whose instances are not eternal, it
must be thought of as applying to something other than the individual
q>-thing as such- presumably, therefore, to the matter.
The reason
is that, by Aristotle's construction, the subject can cease to be e.g.
a man (so that the individual man would perish simpliciter) only if
it can not-be a man (or be a not-man); but the possibility of not being
il man here means a capacity, and actually not being a man means exerci-
sing that capacity.
We therefore need a subject of which it makes
to say that it wi 11 eventually exercise its capacity for being
other than a man; and evidently Callias as distinct from his matter
is not such a subject.
Where the substance concerned is a portion
of one of the four elements, then since these are subject to transfor-
mation the corresponding substance-predicate holds of some-
thing possessed of the capacity to be first fire, then not-fire, etc.
That 's0mething' would be no empirically knowable stuff.
Thus it
that the position of pe Caelo I 12 commits Aristotle to the postu-
late of 'prime matter' in like the scholastic sense.
rtny rate the cyclic transformations of the four 2mpirical elements
provides him with a rare example where lt Ls not absurd to :iew negative
'-;tates as intrinsically timed ln parallel wi.th theJ.r contradictories.
-:..1.nce the mutational sequence is it is presumably roughly
true for him that the 'something' not-fire ..mulct be not-fire for an
<tll.otted period: this being the of its periods of being the other
elements in turn. But even so r-he negative period is only timed
s1m of positives, not
dut now what about the aimed-for conclusion that what alHays is,
':annot not be?
-1s tt stands.
It may seem that this does not fallow from the position
For something al-wavs (I) rines not
i{ \fTI':R 3
cease to be on account of being intrinsically geared to cease within
a given time, this hardly entails that it might not cease through some
other cause, Nothing has been said to rule out interruption ab extra,
or so it seems. The question of interference is one that Aristotle
has not handled as effectively as he might from his present position.
However, what we have now to notice as contained in that position is
a claim yet more extraordinary than anything so far uncovered.
possibilities for being (sc. categorizable) are, he says, essentially
for a maximum. But if something is always q>, then it is tp: hence
it has and exercises a ,-capacity. Hence in its case too there must
be a temporal maximum. But it cannot be a finite period, since then
the object must become not-cp, which contradicts the assumption. It
must then be an infinite maximal period - 'greater than any that might
be suggested and lesser than none' (28la 33-b 1). Since time is in-
finite, 'for always' entails 'for an infinity of time'. Since anything
that is tp exercises a capacity for a temporally determinate exercise,
anything that is always q> is in a mode both temporally infinite and
temporally determinate. The exercise fills a period, whose peculiar
property is that neither the whole nor any part of it can be repeated.
It is not merely that Aristotle's argument will require this ama-
zing concept; the very text proves it indisputably present. have
inferred its presence from 28la 33-b 2 taken in conjunction with 28la
10-12 and 18-19. But even more decisive evidence occurs later, at
283a 4 ff. Here Aristotle is arguing against the possible view that
something might be <P or not-<P for a time infinite in one direction
only (in which case even if it has always been and always will be !p
there is at least the possibility of its not being so as from some
moment of time). Aristotle rejects this on the ground that a time
infinite in one direction is not rleterminate, hence not a suitable
stage for anything's <P-ness to occupy, since t.n every case being
is for a determinate time.
'Each thing has the capacity to do or
to suffer, to be or not be, for a determi-
nate time, either infinite or of a speci-
fiable amount. In the case of infinite
[in the strict sense, i.e. both ways]
time there is such a capacity because
infinite time is in a sense determinate,
'lS being that than which none is greater.
But the nne-way infinite is neither in-
finite nor determinate.'
The last sentence shows that he is not altogether happy about
calling the infinite 'determinate'; the second to last, that he is
even more unwilling to withhold this title altogether; and the first
shows why: if the first is true, then the second too must be true
and meaningful, since otherwise nothing could be said to be or to do
etc. for always.
Logically, what supports Aristotle here is the fact that 'Always'
provides an answer to the question 'For how long?' that is no less
definite that 'For a minute', 'For a decade', etc. By contrast, 'For
a time infinite in one direction' covers any number of different lengths
of time, inasmuch as more time will elapse counting, say, forward for
ever from the Battle of Marathon than from the death of Socrates, and
less time stretches from the infinite past to the former event than
the latter. If it were not so, how could one happening to be earlier
or later than another in the infinity of time? (It is true that 'from
the battle of Marathon onwards' unambiguously specifies one particular
1.ndefinite length, and 'from the death of Socrates' another: but these
phrases cannot begin to be considered as names of periods, any more
than 'the week beginning last Thursday' names one (as distinct from
'a week').
His metaphysics too allows a foothold for the notion of Always
as a period of totality in time. The basis of this lies, I suspect,
in the fact that he is not in this context concerned with time in the
abstract 9 'universal' time, but, as he says, with the time of some-
thing's doing or suffering, being or not being. Aristotle is certainly
not clear about the distinction, and I shall not try to explore its
significance here. But something that can be briefly said is this:
if in general the period of a state's duration is prescribed by an
inner principle, an Aristotleian 'nature', then that state has a certain
temporal bounded ness independent of external circumstances. Under
'external' include the temporal 'environment' of a finite state,
consisting of states that precede and states that follow. The point
is that it is not simply because something else comes after it that
an inherently bounded state finishes when it finishes. After a given
stretch its time is up, whether or not the time is not yet up for other
things. A symphony finishes because !_! finishes, not because other
\P fi:R 1
happen afterwards, although of course they do.
Some things.
perhaps not all things, depend causally on pre-existing conditions
c.J.nd on lJToceses bound to outlast them.
But where there is no such
causal dependence, so that what occurs requires no temporal environment,
the absence of such an environment would not in itself affect the inner
determinacy of the temporally uncontained condition,
whose time can
therefore be
seen as the time it takes to itself,
This is
.1 time which, when compared against the periods of transient conditions,
can nnly be said to be 'greater than any and lesser than none'. This,
it might be objected, does not make it a which is what Aris-
totle's argument requires, True, it is not a quantitative maximum
(which would be meaningless); but it is still the maximum necessary
for the complete realizat.ion of the capacity concerned. (Cf, 281a
Mov xctl 'tTJV hEpoxT)v TTjv 56v"'J.i'v;
see also 18-19.)
t:oherent or not, this is Aristotle's position.
We may now spell
it (;Ut as follows.
Since capacities and their exercises are specified
by the maximum, should not say simply that x is '4> (or not--<+') for
a certain time, but that X is for-that-time-r.p, or for-that-time-not-,,
thus, e.g., it is always -r.p or finitely-r.p (I use 'finite' as a variable
for specific units). 'Finitely' and 'always' qualify first and
ioremost the predicate (the complement of 'possible'), and only deri-
vatively govern sentences constructed with unqualified predicates.
It might even be better to say that the duration-adverbs occur in the
instances as parts of predicates or complements otherwise incom-
plete, thus
in the sense in which an expression containing an unbound
is incomplete, (But this is a misleading comparison if it
Sllggests that 'always' functions like a universal quantifier.
it is logically on the same leve] as variable-fillers ltke 'month-
long' and the rest.)
Now, for X to be al..1ays. -r.p takes time: all
Throughout, the same capacity is heing exercised, just as the
man lifting a hundred pounds exercises his one limited capacity on every
pound he lifts,
Hence throughout e'Jer:y finite period of time, and
J.t every moment, it ls true of X not merely that it is <p, but that
it is always-(fl, It is always engaged in doing what is defined as
t._:eding the whole of time to do.
fhis is a Jpecial kind of always
:!ling the same thing, and its special properties are what legitimize
,\ristotle' s refusal to acknowledge the distributive sense of 'always'
n'.ned earlier. For if it is true for 3ny one short time that X is
then how could it be consistent with this to suppose X ever not-
Formally, one might put the point by saying that since 'always'
is part of the predicate (as distinct from a temporal quantifier on
a sentence with the variable one cannot bring to bear the appara-
tus of quantification theory that portrays so beautifully the varying
scopes of 'all' which the word 'distributive' warns us not to confuse.
However, the temporal quantifier 'always' does also have a place, since
we 3 re entitled to say that at .!:n, and at and indeed at every
X is always
The omnitemporally given is itself (although not
in the same way) omnitemporal. Thus even if take it bit by bit,
... is thus successively given is a truth that covers in a sweep the
'..Jhole of time. Hence it covers also those parts of time that lie
uutside whatever finite bit we momentarily consider as given. For
;_hat reason, the situation at every moment is such that relatively
to it, an absurdity follows from supposing that what is given then
ls not given at some other. Hence 'what always is, cannot not be'.
This take to be the mainspring of Aristotle's proof, in which
case the formal charges against it must be dropped. But a challenge
arises when we consider in concrete terms the analogical basis from
which this reasoning was launched. As we know, a capacity need not
be realized to the full, from which it would not follow that it is
:<ot exercised at all. A man who walks ten miles when he can walk
thirty still, we may suppose, doing (so far as he does it) the kind
nf walking a thirty-miler can do. Why then should "-" not suppose
, hat X is for a while always -q> in the sense explained, but then has
its state interrupted? After all 9 not even everything finitely-cp
is actually permitted to reach its natural limit. Logically, .it .Jould
that the longer the natural span, the mere vulnerable to premature
'-'xtinction. (It is generally agreed that Aris cot le would not want
1 cone lusion of I 12, which is as an ar8ument from universal
first principles, to depend on implicit special considerations concer-
the physical objects which he in fact believes are eternally doing
_..;hat they do, i.e. the heavenly spheres whose position and substdnce
puts them beyond disruption.*)
So it seems that being even
if in fact for always, is not inconsistent with the possibility of
becoming not-q>.
have not been able in the densely packed later pages of Chapter
12 to find Aristotle meeting precisely this objection:
but think
he can.
A man who does not walk to his limit on one occasion still
has the capacity to exercise in full, on another.
If his walking
only fifteen miles somehow meant that we could no longer regard him
as having the capacity for thirty, then if capacities are assumed to
be constant we should have to admit that thirty was never his maximum.
Now suppose that X is interrupted while always-q> so that it ceases
to be q> altogether.
It becomes but this being not-q> must be
for a determinate time, either finite or for always.
It cannot be
for always, because nothing could begin to be always-11: if it begins,
U is too late to exercise to the full the capacity of which that would
be the exercise.
Hence we cannot say that it has or ever did have
the capacity for being always-not-cp.
For a capacity in the abstract sense does not really exist if
the circumstances are such as to entail that 1t cannot be fully reali-
and if capacities are constant, our grounds for saying that it
does not exist are grounds for saying that it never did.
Hence if
X becomes not-<p, this must be finitely determined.
In that case it
will return again to being <p .
But not to being always -cp, by the
preceding argument.
Nor 9 by the same argument. could it be thought
then to have the capacity for always -q>.
Hence it never did have
or exercise it, even for a short time; which contradicts the assumption.
The coherence in some important respects of the position in
Ci!_elo I l2 with Aristotle's statements elsewhere only serves to throw
into relief one extraordinary discrepancy which has not so far been
mentioned. His account of the infinite in Physics III is in direct
.-:onfl let with the idea governing the De Caelo argument. Indeed, our
Hm resistance to the notion of the infinite as a totality owes much
to Aristotle's classic disposal of it in the Physics.
*Thus Aquinas' appeal (Comm. ad lac.) to '.omnia natura appetunt esse'
is inadequate. He derives from it 'unumquodque tantum est quantum
f2.2Sest esse', but this depends on non-interference, which in general
is a contingent matter. He does not, I think. show how the terms
.,f the current argument alone (as distinct from cosmological require-
::ents) exclude interference in the case of 'what ts'.
'It turns out that the infinte is the opposite
of what people say, for it is not that which has
nothing beyond it. No, the infinite is that which
has always something of itself beyond it ... that
of which, in taking a specific amount (l(O."t<i ?too-Ov
we leave always something more to
take. Whereas that of which there is nothing
beyond is complete and whole.' (206b 33 ff.)
This is as much as to say that any infinity of time, one-way or both-
ways, is as such incapable of functioning as the field of a temporally
determinate exercise. Is Aristotle's past self among the 'people'
of the first sentence? This belongs with a bag of questions about
relative datings of different parts of the Physics and different parts
of De Caelo. But on any view of the chronology there is a problem
of reconciliation; for Aristotle appears never to have withdrawn the
conclusion of De Caelo I 12. If, as seems more likely, III
6 ,yas composed later, how could the earlier findings still stand?
If Physics III 6 came first, how in De Caelo I 12 could he have failed
to defend explicitly the conception of a determinate infinite (which
would have made things easier for the interpreter)?
not defend it, how give the argument we have?
And if he could
Perhaps, however, a reversion to 'what people say' needs no public
justification. And as for a change from that holistic notion of the
infinite, this too perhaps would have left Aristotle still in possession
of adequate grounds for the essential position of De Caelo I 12
For the argument does not need to he couched in terms of 'infinite
time' at all.
between what
The phrase is a natural one to use to point the contrast
holds for a measurable period and what holds for always
provided he retains this contrast, he has all the premisses he
needs, and the naive equivalence of 'always', with 'for an infinity
Jf time' can be quietly dropped - as indeed perhaps it The as-
sumption must
Jf a whole.
of course still stand that 'always' denotes some k.ind
To call this 'infinite' in the sense of III 6
i.s to emphasize the impossibility of assigning a specific quantity;
but that impossibility was never in question. However, the impossi-
bil1ty does not, I think, make it absurd to regard as a whole what
is also in that sense infinite. Only it is not whole that it
is infinte, nor infinite a whole. If this represents Aristotle's
more considered vtew, then we should expect to find no mention of 'infi-
nite time' in his later statements of the De position, but 1mly
and <i(o,ov.
And thts is exactly what we do find, so far as I
have been able to verify.
(See e.g. Ill 4, 203b 30 (where the
reference to infinite body does not affect the present point); De Gen.
II 9, 335a 33-14 and ll, 338a l-3; Metaph. !18, l050b 7 ff.
(cf. ibid. 6, 1048b 9 ff. for a statement of conclusions reached in
III 6; Metaph. N 2, 1088b 23-25; and, for good measure, E 2,
l026b 27-28; K 8, l064b 32; and De Int. 9, l9a 9 and 35-36.)
To conclude. Aristotle has not shown or tried to show that what
is always cp might not have been otherwise in some sense of 'm1ght'
that we at any rate seem to understand.
The possibility which this
t.;ord conveys is, so to speak, unavailable except from a standpoint
extraneous to the entire history of the actual universe, given that
it is a fact of that universe that X is always <.p.
The contradictory
is an option perhaps for God, although not even for him on some concep-
tions of divinity, but certainly not for anything embedded in the actual
order of nature.
So if X's not betng cp is in some sense a possibility,
it is not in any sense a capacity of an actual subject:
unless we
regard it as representing a divine capacity.
But in that case it
would hardly make sense to test it by referring the corresponding sup-
position to 'another time'.
In itself, Aristotle's time-relative idea of possibility seems
perfectly coherent, especially since in itself it does not entail that
possibilities (particular or general) are in fact realized.
For that
he needs the additional notion of properttes defined by inherent tempo-
ral spans. His handling of this in De Caelo I
us as strained to the point of absurdity. The
12 may well strike
consequent holistic
treatment of 'always' is an obvious target of suspicion.
At the same
time, though, a clear understanding of the way in which temporal maxima
enter into the argument certainly helps to salvage Aristotle's reputa-
tion for logical sanity.
Hintikka has maintained that in connecting
as he does the omnitemporal with the necessary, Aristotle is dominated
hy a 'statistical model of modality'
(Time and Necessity, pp. 102-
The difference, then, between 'necessary' and 'possible' would
he simply the difference of 'always' from 'at some time'.
Thus 'neces-
sary' :.lairns no more than the continuance ad iHfinitum of some state
of affairs which had it been interrupted would properly have been called
'contingent'. It is not it is that makes whatevr:-r is necessary
neccr,sary, but the fact that ,,..,hate';er it is, it simply goes on and
But pe Caelo 1 12 :.>hmJS Aristotle as far from holding this gro-
tesque position as any of his admirers could hope to find him.
difference (in the central cases) between the necessary and the possibly
otherwise reflects the difference between specifically different ways
of being <;>
The limit of duration is the differentia. being Ill is
the genus.
To suggest that they are to each other as shorter
and longer (much longer) occui"rt:!nces of the same attribute matches
the suggestion that the lion dUd the mouse, being both of them animals,
essentially larger and smaller versions of the same creature.
doubt the De Caelo argument will cCJnt inue
to seem as a lien to us as
ever it did; but perhaps by now we can see that it is the thrust of
his metaphysics,
not the failure of his logic, that takes Aristotle
beyond our company in this chaptei".
Discussion of paper by Sarah Waterlow
In connection with the argument that infinity unbounded in both
directions is in a sense (de Caelo I 12 283a 9) it was pointed
out that in the case of an infinity bounded in one direction only (the
starting (or [in1shing) point has to be specified if it is to be defined
at al.l; on the othet: hand a finite period of (e.g.) five years is de-
!:ined as to its length at even if the starting point is not
But does it :nake any difference uhen an infinite period bounded
1 nly at the beginning starts, provided t:hat it is infinite thereafter?
L.. was pointed out that is reluctdnt to accept the inclusion
')f one inf 1 n1ty in another lll. 5 204a 20 ff.; where t.he assump-
tion that the parts of an infinite must themselves be infinite results,
:t t!bserved, from the fact that lt is the infinite as a substance
1_:-tat is under discussion, the infinite characterised precisely by its
1.nfinite; 204a 26.
It follows that an act1al infinite substance
-.ust oe indivisible, as in the o
, md, Ji.nce a proof of n1e infinity of the of prime numbers
.-as available in i\ristotle's t1.me, he :-;hnuld 1:':i''e aJlcno1ed fn tl
e i.n
elusion of the infinite series of t->rime numbers in the infinite series
of integers.
Would time exist if there was simply one rotating heavenly sphere
(that of the fixed stars), and no stationary earth at the centre?
As it is, time is indicated by the relative motions of the different
spheres, in any case; but the main argument of Physics VIII is conducted
in terms of the motion of the first heaven.
.!.! there is no absolute
frame of spatial reference, the existence of motion, and hence of time,
seems to require at least two motions differing from one another.
But an absolute frame of reference is suggested by the argument for
the earth's being at the centre of the universe; the centre would be
the centre even if the earth was not there (de Caelo 2. 14 296b 6 ff.,
cf. 297a 9 ff.).
It was suggested in the paper that the requirement that a possi-
bility be realised at some time applies to being in the various cate-
gories, but not to the case of the cloak which can be cut up, because
being cut up is not itself a case of being that falls under one or
another of the categories; rather it is a of passing from categorial
being (in this case, being a cloak - a substance?) to the corresponding
not -being.
It was observed that this applied well enough to the first
few categories; but there might be problems with the category of e.g.,
paschein, and certainly with that of time.
Aristotle is not so much concerned to make the logical point,
'every possibility must be actualised at some time', as to make the
metaphysical point that any unexercised capacity will be exercised;
and in any case more directly concerned here with the corresponding
implication, 'whatever is always is necessary'. Hintikka 's use of
the expression 'Principle of Plenitude' is doubly misleading; for (a)
it relates more to metaphysical and theological issues than to the
purely logical ones in terms of which Hintikka tends to interpret Aris-
totle; and (b) Aristotle's concern too is metaphysical rather than
logical, Hintikka.
There is a difficulty in finding plausible examples of something
that has the potential for two opposed states and alternates between
them. Possible cases include the changes connected with the seasons;
but there is then the problem of identifying the subject (which will,
moreover, it was pointed out, be a general problem in all cases of
being something and not-being something in the category of substance).
Perhaps it is the earth that is alternately hot and cold, in summer
and winter?
(Another possible example would be
this water
t ing between hot and cold
- cf. Boethius, in de Int. ed. sec. 238.
ff. Meiser _ except for difficulties over 'this water'. Alexander
of Aphrodisias, Quaest
1. 19 - and cf. also 2. 15 - discusses the
notion of alternating possibilities, but with a marked absence of ex-
amples. In ibid. 2. 20 he suggests that it applies (i) to prime matter
taking on the forms of the elements in succession, and (ii) to the
formation of homoeomerous compound
bodies from those elements, and
their subsequent dissolution again, but not at any higher level.)
Perhaps Aristotle's thought is rather that the matter of which a man
is composed has different potentialities at different stages of his
life; cf. VIII. 1 252a 11 - there is always order and proportion
in what is natural.
It was also felt that it would be rather implau-
sible to suppose that predicates come
in families of incompatibles;
is a picture of men spending their lives in alternating between standing
and sitting a very appropriate one?
Further, it was pointed out that potentialities like those for
standing and sitting do not fit very well into the framework of the
discussion of maximal capacities in ch. 11,
which it had been argued
was the context that should be borne in mind in assessing the argument
of ch. 12.
In answer to this it was suggested that the basic argument
has already been stated by 28lb 2; what follows is Aristotle's explana-
tion of the argument, and in the course of this he
employs stock ex-
amples of possibility without concerning himself unduly as to whether
they are suited to the context of the argument as a whole.
It had been argued in the paper that, if a thing with the capacity
to-always-cp ceases
i.t must either always-not- cp (which is already
e.xcluded) or not-cp for a finite time (in which case it would then have
to h again, and (by the same argument), to do that for a finite time).
flUestion was raised, why is
for-all-future-time not-<Ping?
does not allow such capacities.
there not the third possibility of
The answer, that Aristotle simply
In connection with 'maximal capacities'' i.t was asked whether
iiAP fER _I
a man w.th the capacity to lift 500 lb also has the capacity to lift
400 lb?
Answer, on this doctrine, no; for to have a (maximal)
of lifting 400 lb excludes having the capacity to lift 500 lb. So
if this man lifts 400 lb, he is not exercising a capacity to lift 400
lb, but partially exercising his capacity to lift 500 lb.
Is it neces-
sary that each such capacity should be maximally exercised at some
time? - no, except in the case of a capacity to which is
exercised in full the only time (i.e. all time) that it is exercised.
Interest in maximal capacities, it was argued, leads Aristotle
to talk about the capacity to but in ch. 12 he makes the
further claim that a thing that has the capacity to always-cp, always
has the capacity to always-cp.
(What would he make of a myth about
a being who in principle possessed immortal life, but was told that,
if he transgressed, this immortality would be taken away?
that, in real life- or in nature- things don't happen like that.
But note too that such a story inevitably involves the possibility
of an infinite bounded only at one end; at the end, if the being had
always existed but lost its immortality; at the beginning, if the begin-
ning has not always existed, but retained immortality.)
At 28lb 5 Aristotle distinguishes between what is possible or
impossible or true or false, hypothetically on the one hand and haplos
on the other.
The examples given for what is hypothetically impossible
are things which, on usual assumptions, are mathematically necessary.
With the example of the triangle compare, perhaps, Physics II. 9 200a
16; the sum of the angles of a triangle is two right angles since the
straight line is such-and-such.
The Euclidean proof of the point
depends on assumptions about parallels.
So, if we make differing
suppositions, the angles will not add up to two right angles.
is the corresponding supposition in the case of the diagonal?
that the same number can be both odd and even.
But what, then, will be examples of things which are impossible
if even mathematical examples are related to assumptions? Per-
haps, the things which are in fact impossible - such as the diagonal
being commensurable; so that 'hypothetically' will mean. not 'cteriva-
tively', but 'on an unreal assumption', and hapl;s will mean, not 'sim-
ply', 'underivatively', but 'relative to something else which is itself
assumed to be true' .
Does Aristotle then recognise a.E_I underived
necessities? The principles of the sciences? - hut perhaps they are only
Hll1l relative to our assumptions about space lAnd, it was n0ted,
., 1 our inability to form intuitions which do not correspond
lilt.. '''dY LILI.ngs really are would not be a limitation; the correspun-
dence between our cognitive capacit Les and the way things really are
l J tlJt a.n arbitrary matter.)
l'he motive suggested in the paper for the relativisation of possi-
11iUty 1.;a 5 the fact that what is true is ipso facto possible, so that
vf whdt
of whdt is false is the only interesting one, and in the ca3e
ts false the test by supposition must apply to a situation
ut_!wL the present. Do similar motives apply in the case of the
rhey presumably do, if the necessary is that of which
1.he ,_onl Ladictory is impossible; it is by taking something that is
true, and of which the contradictory is therefore false, and considering
what the implications would be is the contradictory true that
whether a thing is necessary or not. Just as some things
,1 ro ;"tlways impos.<>ible, so, presumably, some things are always possible;
tr i 3 1 r 11 r Aristotle) always possible that some human being or other
>hnultl be standing.
Ts 'dlways' itself a modal notion, for- Aristotle? It is true
1 !lut. [.,r him, the capacity to-always-cp is a sort of capacity whlch
annot he possessed by something accidentally; there is a possible
iJ.nl'!;tt 1f ,_ilcularity here. tn some cases, the impossibility of some-
lhing derives, fur Aristotle, from things which themselves cannot change,
irl thp whole of 1eal time - which is the only time he concerns
!L observed that, on this interpretation of Aristotle, there
,,., 1 11 hP 111 any things that are necessary for him but nnly universal au .. i-
inni S f 'll 'IS that there are no golden mountains, for instance. (But
l. 'mithl!t'H, '!lume on Existence and Possibility', Proc.
I.;HO But it was emphasised that this is not how A.ristrotle
f"<>S rhe rnatter; he would accept the suggestion that modal state"-
simply functions of assertoric ones, a thing being
"irnnlv it is always, as a matter of fact, true.
.hist\Jtles modal notions may seem weak; it does not [QlllW
he w<\Jld have shared this view of them or regarded them, in cnn-
Translation by R.W. Sharples of part of Ps.-Alex:ander's comments on
2. [AlexanderJ
in metaph. 5 74. 6-575. 17 (on 9 4 1047b 3); ed.
M. Hayduck, CAG 1, Berlin 1891.
574. 6
Having said that those things are possible which
are not, but are able to come to be, [Aristotle}
says that 'if what has been said is possible in that
it follows
-that is, if something is said to
be possible, in that it is possible for it to come
to be and being actual follows for it
- it is
not true to say that something is possible, indeed,
10 but will not be I and will not be able to come to
actuality. For those things are said to be possible
which are able to come to the actuality of that for
possibility. But if someone
which they have the
says that something is possible, indeed, but is not
able to come to be and result in actuality, the impos-
sible escapes us and it is unclear what it is.
For we say that a thing is possible [if}, when it
is not but is posited as being, nothing impossible
15 results; and [that a thing is] impossible [if],
when it is not but is posited as being, something
impossible results.
But if someone says that
something is indeed possible, but will not be or
result in actuality, how are we to know the difference
between what is possible and what is impossible?
For example, when it is said to be possible for the
log to be burned, if someone says that it is indeed
possible for it to be burned, but it will not be
burned, and the same person says also that it is
indeed possible for the diagonal to he measured,
20 I but it will not he measured - how will this differ
from that, when both are said to be possible [though}
they will not be [the case]? Accordingly, the log
is combustible and the dL.tgonal .neasurable equally;
but if this is so, it escapes us what the impossible
lS, Hhether it is the diagonal's being measured or
the log's being burned. So, [Aristotle] says, the
person who does not reck,1n or discern what the i!Tlpos-
sible is, and [what] the I difference {is) between
it and the possible, that man would say that 'nothing
prevents it that something, for which it is possible
to be or come to be, should not be, now or in the
But if he knew the nature of what is
i.mposssible, he would not have said so uncautiously
and readily, [both in the case ofJ the diagonal and
[in that of] the log, that it is po5sible for the
0ne to be measured and it will not be measured, and
for the other to be burned and it will not be burned.
Rather, he would have proclaimed }in the case of}
the other that it is possible for it to be burned
and it will be
of statements J
burned. But
are what would be
former pair
by the man
who is ignorant of the impossible; this however is
cLear from what has been laid down (and it has been
laid down that what is able to come to be will come
to be), namely, that if we IJOsit that what is not,
hut is able to come to be, is the case, nothing im-
35 possible will result. I For jf, when you are sitting
but are able to walk around, we suppose, when you
Jre sitting, that you are walking around, nothing
impossible will result from the supposition.
5 75. I i Suppose, however, that this is true, but some-
one says that this man. who is sitting, is indeed
.1ble tn ..Jalk around, but t ..;ill not walk around, and,
similarly, that the diagonal too is able to be mea-
->dred, but will not Oe measured. lf, then, the
latter case is not different from the former, just
.Js in the former case nothing impossible resulted,
so (will it J i.n the latter. - _!3ut, if
1 it
is granted that the diagonal measured, it follows
an odd number is equivalent to an
even one,
is utterly impossible.
And this results
from not seeing the nature of the impossible and
the difference between it and the possible, and,
as a result of this, saying of the impossible that
it is indeed possible, but will not be the case.
So it is clear from this that, if something is pos-
sible, it will indeed be the case.
And this
impossibility I - I mean an even number being equal
to an odd one - is not [a matter of) being contrary
to the supposition that says 'let it be granted that
the diagonal is measured'. For if it were cant-
rary to the supposition, something impossible would
follow also in the case of the man who is sitting
but is said to walk around. [But] there the conclu-
sian is false, but not impossible, while here it
is both
and the
false and impossible.
I impossible were the
So, if the false
same, there would be
no difference between the two suppositions, that
which supposes that the man who is sitting is walking
around, and that which says that the diagonal is
But since }the false and the impossible J
are different, the suppositions are different too.
The commentary on books E-N of the Metaphysics attri-
buted to Alexander is certainly not authentic in
its present form, although it is uncertain how much
material from the original commentary is incorporated.
Cf. Hayduck's preface to CAG 1; K. Praechter in GOt-
tingen Gelehrte Anzeiger 168 ( 1906) 882-896; Meraux,
Alexandre d
Aphrodise (1942) 14-19, and id. 'Aris-
toteles, der Lehrer Alexanders von Aphrodisias',
AGPh 49 (1967), 181 f.
9 4 l047b J.
t ead
in that', with J howe,,er has 11,
rurther A'u has -rh .Juva."tbV 'if the impJSSible
is '..Jhat has been said'. Zeller quggested buvn-rbv
'if, [a::.j has been said,
l that [ i.:3 possible un which nothing impossible fol-
Ln ..'s'; cf. below n. 2Y5, Ross (above n. 264) ad lac.
and Hintikka, Time and Necessity 107.
0::94. It seems very unlikely that this interpretation accu-
iately teflects the intention of Arislotle's words
i'.l s.
- whatever exactly they were (cf. last note. Ross
that 4xoAou6t can 0nly mean 'in so far
it is 'or is convertible
.;ith itj, comparing 32a 24; Hintikka 107
the latter). But [Alexander's} interpretation
dues commit him to what would appear to be a strong
... ersion of the principle of plenitude, the principle
that every possibility must eventually be realised;
ind this has a bearing on certain features of his
911bsequent discussion.
Fct this definition ot the possible cf. ;\ristotle
A_!!.:___!'_r_,_ l. 13 32a 19 ff., l:i_etaph. 9 3 1047 a 23 Ef.,
q 4 l047b ff. Cf. Hintikka, ib1.d. 30 E., 154
ii. 12, 22 ff.
:leta ph. 94 1047b 8 E.
this Llearl:; asserts that Lhe principle of plenitude
01ppl1es to individual cases (cf. above nn. 13 f.).
;'he c.Jntrast with the diagonal would have been ade-
llJateiy expressed by 'it is possib.ie for it to be
burned, even ii it is not' . It. might therefore
be thought that this was simply a 3lip. But c f.
below, 574. 33 ('it has been laid down that what
is able to come to be will ..:orne to be
), and nn.
294, 300.
seems to be required after in 575. 2;
alternatively, one might add O'tt. before in 575. l
('that, as this man . , so similarly the diagonal,
too .
CE. Aristotle l. 23 41a 26 f.; T. Heath,
A History of Greek Hathematics, Oxford 1921, 91.
The proof is given by Euclid 10 app. 27; summarising,
lf a/b is the ratio of the diagonal to the side in
its lowest terms, a
= 2b
, so that a
and a are
even. If the ratio is in its lowest terms, b is
therefore odd.
Then a
= Zb
::::1 4c
if a is even, let a = 2c.
so that b
= 2c
But in that
case b
and b are even, b is therefore both odd
and even, which is impossible. Therefore it is
impossible to express the ratio of the diagonal to
the side on its lowest terms.
This does not of course follow; all that is clear
is that some of the things, which will never be the
case, are not possible either. The present statement
could indeed be interpreted only as saying that what
is possible for a type of thing will happen to some
one of the type in question; but cf. above, nn. 294,
297, and the summary of this discussion in the sequel
(575. 20) - 'having shown that what can come to be
something will also come to be that for '..Jhich it
has the potentiality .. '.
l047b l
by uwen
i'1etaphysics 84: potentiality and actuality. with not a hlnt of power
or potency
!U4lb 3: E & A; .J ('earliest extant ms. of the Metaphysics', Ross
Intra. clv) 'appears to preserve the true reading', viz. 11 (Ross ib.).
dxo\ou6EL suggests a lacuna, and ps. Alex 8) supplied tvep-
yT)c"a.L. 'Nhich is not to be had from 8 3; Zeller supplied d.E>Uva't'OV iJ.>TJ
JxoA.ouBe!, which shows due disregard for the accenting of TJ in r: A .
Ti seems far likelier: if the impossibility of some consequence of
negating p is what constitutes the possibility of p, or at least (if
we find a less apparently circular definition of 'possible') follows
from the definition, we must presernve Lnpossibles against the sugges-
tion that they are possible but will not occur, if this is a general
move vs. tmpossibles.
!04/h 5: Not '!.f the (intended) result is that' (but): 'to
the intent that', standard with verbs of speaking-to-a-purpose, e.g.
oracles and
his motive:
Lt won't be;
votes, requests and persuadings. The inter locutor has
in any case of alleged impossibility he insists 'No doubt
but (since we entertain the possibility even in denying
it) it could be'. He might have continued 'After all, arguments can
be upset'. but he does not:
which standardly means 'not taking into account' and not 'not accepting
the proof'. He is a simple, to be inv<-'nted. 81.a.cpe\rye:Lv: escape,
run away: his prime object is not to kill them but to eliminate them
from any suggested case. He is perhaps invented as the extreme anti-
'can never entails realization, non-realization never entails
-on not'.
On the necessity of past, present and future cf.
:a) EN VI 2 1139b 5-11, where what cannot be deliberated about is
t hin.Q,;S
vs. future; (b) De Int. l8a 28-34, ..;nich distinguishes
are and (?) have come about from future cases (and if it
for the one might still thlnk of quantified vs.
11 nquantified universals); (c) Rhet. 1418a 2-5: Ov'ta. f1 Ov'ta. coupled
.nth Y-v6ueva. as v, the future, and as r.ecessary; (d) De 1:nelo 283b
inP rr:R 4
12-14 which couples the present with the future as agalnst the past
as the field of dunamis. Our 1047b 13-14 belong with (d).
l047b 13-14:
on the allocation of time-connectives which exercised
us in 9 3 cf. SE 166a 24-32 on fallacies of rrUvBEuL' (here distinguished
In conjunction 'The non-writing can
write' may signify ' . can write-when-not-writing' or ' ... when not
writing, can write'. The argument of l047b 8-14 requires
a present possibility which is not realized; the present has not become
104 7b 17: 'There is nothing to prevent its not being capable of being' .
Dunaton einai has gone together in e.g. 1047b 15.
1047b 14-26: B) - (poss. A B)
104/b 26-30: (poss. A_,poss. B) -+B) II
On II: There are two parliamentary seats in Kilmarnock. If a Conser-
vative is allowed to stand for one, there must be a Socialist standing
for the other. Pass. A .... pass. B. But if the Conservative is elected
to the one this does not entail the election of the Socialist to the
other. More generally, of two alternatives the possibility of one
entails the possibility of the other; but the realization of one pre-
eludes the realization of the other. The 'exactly when and how' of
1047b 29-30 do not affect this. What has Aristotle in mind? I can
drive my car out because I can open the garage door; this depends on
driving - opening, but this is I.
The defence of I is a direct application of the axiom that possibles
cannot imply impossibles; it relies on the assumption B). Here,
then, alternatives and Scotch voting are ruled out. How would Ar
deal with alternatives? De Int. 9 19a 9-22: In general, in what
is 'not always actual you find what can be and not-be (not cannot),
and consequently what can come to be or not come to be; and there are
plenty of plain cases of this. For instance,
into bits; and it won't; it will wear out first.
this coat can be cut
On these same terms
it was capable of not being cut into bits; for there would not have
been the possibility of its wearing out first if it hadn't been able
to be cut into hits ...
i 04 7b 10
lJ47b l
.Y,. the reading !)f rr.e oldest MS J, seemed preferable to
'.,Thich was felt to require a :-;upplement - eith.cr lAlexander j 's 'tO vEp-
ril;:ra.L (574. 8i but not (.0 be had frcm d 1) or Zeller's &.E>6va.'tOV f-L)i,.
n could indicate that Aristotle .... as worried about the circularity of
nis 'definition' of possibility in 9 3, dnd so TNas keeping open the
possibility 'lf a better definition, if one could be found. 'tb
'JQV will be understood as subject of Axo-x.au6e't: or if (what has
been stated) (at any rate) follows (from the definition of the possible)'
- though this was felt to be S()mewhat elliptical.
'that, the actuality of which entails no impossibility' (cf. 9 3 L047a
Werre:: Owen and Kneale "" 'if the (intended) result is that';
Hintikka (and pseudo-Alexanaet, and the Loeb), 'the result 'i.Wuld be
that'. On the latter view. the argument is that anyone who denies
plenitt1de, who says 'this is possible but will not be', has no criterion
for distinguishing the possible from the i.mpossible; once you allow
chat one thing that won't happen is possible, there can't be any im-
possibilities. But Aristotle does provide such a criterion in what
Eollows; the fact that something impossible is implied (l047b 9-12).
On the Owen view, it was suggested that, since Aristotle i.s
11 ow supposed to be asserting that 'this is possible but will not be'
1s not true in any case, it is odd that we have 'it isn't true to say
:) with the intended result that q'. rather than 'it isn't true to say
(p and q)'. or 'it isn't true to say (p implies q)'. For p, 'this
Ls possible but will not be'. isn't, itself, false in every case.
Lt was suggested that EC1tetv might be followed by an accusative and
infinitive, meaning 'beca,tse' .1ud Wc:rte: 'therefore'; but the 'AxrtE
:.>P.emed odd, and duplicated -roU1:-r;1. Perhaps t'Ix:rte not in the original
but someone added it they failed to see that Ota. meant
':"e.cause rather than 'that'? .:Jut the cor!"uption would have to be
arlier rnan ps.-Algxander.
Ahy, in any case, should anyone suppose that 'this is pos-
t,le but will not be' i:nply that there are no impossibilities?
t'n the Hintikka view of (':(J"te. this is a belief Aristotle himself holds;
1n the Owen it is an imputation he is concerned to
l047h 4
rlecause of a supposition that whatever can be significantly stated
is possible? Or, is it that, as Aristotle in 9 reduced the Megarian
position to absurdity, so here he is resisting a i'legarian attempt to
reduce his own position - that there a difference between the pas-
sible and the actual - to absurdity by exaggeration?
(Note though
that Aristotle in A
had only argued for the possibility of
is possible and not happening', not of 'this is possible and will
not happen; e:Iva.a. in l047a 21 is present; so such a Megarian attack
on Aristotle would involve an extension of his position, as he had
exaggerated theirs.)
What is the sense of &t.a.cpe:Uye:Lv in l047b 5; 'disappear altogether',
or 'slip through', 'get away with it
, 'successfully masquerade as pas-
The latter, weaker, sense seemed to suit the general sense
of the word better, but we felt that the former, stronger sense is
at any rate implied. e:Yvcu. depends on d.OGva:ta.. It may seem odd
that someone should assert that a thing is possible, but will not be,
with the intention that the impossible 'should get away with it'; less
so if what we have here is Aristotle's report of, and react ion to,
the attempt of an actual or fictional opponent to reduce his (Aris-
totle's) position to absurdity.
i1ansion suggests that the Owen interpretation of Ox.Tte not only
means that the passage doesn't support Plenitude, but actually means
that it opposes it.
We felt that this was questionable; it isn't
actually ruled out that it might as a matter of fact be the case that
the only things that never happen (in any case?) are those that are
(Hintikka himself elsewhere does not seem to assert more
than that what is possible will happen in some case, not that it will
happen in this case; Time and Necessity 100.)
It was also suggested that the point of the denial that one can
say (in some sense, at least?) .'this is possible but will not be' might
be that it is a contradiction in terms. a denial of the very notion
,,f the possible; note that it is 'this is possible but will not happen',
not 'this is possible if it doesn't happen'.
(l047b 3. Zeller's CuvCL't"ov M6v<11:ov cho;>.ou6et is attrac-
tive at first sight, as the consequences of postulating the occurrence
of '..rhat ls possible seem more to the point. in view of l047a 24 ff.,
b 9 tf., than the (o,ercautious?) statemeilt that ,.;hat has been satd
may be an implication of the possible, be co;1vertible with it. rather
than being identical with it. (So Ross r)n the MSS readings, and Hin-
tikka). But - cf. above on 9 3 - A."X.o\,)1Jdetv is Diodorus' term in
this context, not Aristotle's; has Zeller been led .1st ray by the appa-
rent parallel with the Master Argument'q formulation here?)
i1ight the point not be, it was that to allow assertion
of the form 'this is possible, but 1vill nnt be', does away with the
impossible, i.n that it removes what Aristotle's opponent regards as
the criterion for distinguishing between the impossible and the
possible - namely, the non-nccurreuce of the former? flut the trouble
this, ..;hich amounts to taking Uxrte in the Him:ikka rather than
in the Owen sense, is that the conseqnence in 5-6 is clearly one which
Aristotle himself sees as following from the assertion uf 'this is
possible but will not be' - on the Owen view. from some assertions
of it, not assertion.
It was also suggested that Aristotle's point might be that we
cannot simultaneously assert of a thing (i) that it is possible and
(ii) that it will not be, because this involves a contradiction; to
say truly that a thing will not be implies - cf. the Sea-Battle - that
it is impossible. And this might be felt to lead to the disappearance
0f the impossible in the (rather Pickwickian) sense that everything
that did not happen would be impossible, so that the impossible could
lonlil;er be distinguished from the possible hut counterfactual.
"1ut this seems rather tortuous. ro take acco,Jnt nf the point about
the Sea-Battle, it might be suggested that Aristotle's posJ.tion is
that, \.Jhile we can't say 'this is possible but will not be' of cases
1 ike the commensurability of the diagonal, we can say j t in general
of cases llke coats which can be cut up but won't he, though we can't
assert it of any ,e_articula:r:_ coat, :Jince J t is stilJ 0pen whether it
..;i.ll be cut up or not. 0n r-he other hand (i) we do ha'Je -ro6C here;
and (ii) in 9 l9a 14 Aristotle .... ays 'i.t is possible
t.his coat will he cut up, and ("" .<ind yet; hut he uses xa.C, simply)
1.t won'r he'.
'not trtking it into
account', rather than 'not accepting the proof'.
The opponent asserts
of any case that, even if it may well never happen, it is possible
(? since we entertain the possibility even in denying that it will
ever happen); he isn't making the sophisticated claim that arguments
to prove impossibility can always be upset, but simply doesn't consider
Lhe arguments at all; a simple-minded character, invented perhaps as
the extreme oppos1te to the Megarian of 93 who held that that
didn't happen was possible.
(Or perhaps he is the Megarian of e 3
going to the other extremes to show that, even if Aristotle can ridicule
his position, he can ridicule Aristotle's as well.)
can be construed in three ways: (a) 'that the impossible (in general,
as the concept) exists', with 'tO ci66va."tov as subject of Elva.&.; (b)
'what is unable to exist', with e:lva.L depending on &.E>6va:rov - cf. "tO.
e!va.s., 1047b 5; or (c) 'that (the commensurability of the
diagonal) is impossible', with d.66va:tov as complement after elva'- ,
111d 'tO going with the whole clause (in which ca.::;e it is redundant,
as was pointed out). (a) and (b) indicate that it is a general point
about the impossible that is ;}t issue, while in (c) it is a specific
point about the diagonal; in an attempt to salvage our respect for
the opponent, it was pointed out that the argument 'the fact that vou
reach a rational value for 2 by the method of successive approxi-
mations doesn't in itself prove that there isn't a rational value'
has force against an attempted proof of the incommensurability of the
Jiagonal by the method of exhaustion, even though it doesn't against
the reductio, which the opponent will then be disregarding.
But it was felt that there was little evidence for such a specific
point in the context; and, however -rO &.E>6va.'t'ov elva.L is taken, the
shows that the issue between Aristotle and his opponent is
CC)!JCerned with impossibilities in general, not just with the tncommen-
surability of the diagonal. The opponent is right to real1se that
wn-occurrence does not of itself lndtcate impossibility; but he is
wrong to refuse to allow any other criterion.
'_f_1_47b In de Int. ':1 18a 28-34 and 3. 17 1418a 2-5 Aristotle
:.dsses the present with the past, distinguishing it from the future
-;hich is still open (Eth. Nic. VI. 2 ll39b 5-ll doesn't mention the

41b 4
in this connect1.on dt <Ill, simply contrasting the past ::tnd
the future.) In st_e Caelo I t2 2d3b 12-14, on the oLher hand, he seems
tu regard the present as Jell as the future as the fielJ of possibllity,
unlike the past; and 13-14 here seems LO Sl.tggest rhe same, what is
not now the case not r_herefore be1.ng impossible. Admittedly is
dmbiguous; the point could he either {i) 'it 1.s false that you will
get up here and now, but not impossible
, or ( ii) 'it's false that
ynu re standing now, but you could have been'. But, against (i),
on ..:hat grounds can ..;re say 'it is false that you will get up here
and now'? :-toreover, the fuller version in de I 12 28lb 9-11,
yap fEUCo4 obx &t, suggests
that here too 1t is the of present counterfactuals (ii)
that is in question; and l047b 8-9, too, indicates that it ts the non-
realisation of a present possibility that is Ln question, as well as
that of a future one.
fhere l$ thus an apparent discrepancy between this passage and
l 12 on the oue hand, and and Rhet. on the other,
on the question of the possibility of the present (It was noted
that Diodorus Cronus relates possibility to the present and the
and that he may have been influenced in this by this chapter.) lt
was pointed out that the present, when considered closely, tends to
into either the future of the past, and that Aristotle may
have been influenced by different, equally natural ways 0f looking
at the matter on different occasions. But, more convincingly, the
sense in which the present counterfactual is not impossible here may
be different from that in which the present dctual is necessary (and
the present counter factual imposs.Lble) ln de int. 9 (cf. L9 a 24) and
ln I 12 28lb 8-9, though not in 9-ll itself, the con-
trast that is drawn is one hetween d:Jt.\W<; falsehood and 6.7tXWc; impossi--
it ooen that what is false may be impossible hi l i.ty; which leaves
''"', some qualified sense. [f so, 1.t ..;as felt that Aristotle in 13-
l4 here should not just have contrasted 'false' and 'impossihle
, but
>hOlild have spelled out of i:npossl.Ollity he was reierring
C() 0 It was f1'.Jinted out, too, that the possilnlity, in some sense,
r h 2 counterfactual should extend to past counterfactuals as '1ell
ro present ones: past <iS well as present ones could be de-
<'JlAPTF;R 4
as accidental rather than as nP.cessary.
th'lt the perfect yeyovtvnL in 1047b 10 could possihil ity
in the past; however, it was felt that the past implications of the
tense were not clear, and that the pairing of ETvcu and ye:-
y')vvcu was intended rather to accommodate the distinction between
permanent, omnitemporal truths (like the incommensurability of the
di.agonal) and those relating to things subject to change.
(But are
there any omnitemporal unrealised possibilities?)
We also disc,lssed the relation of this discussion to tlte 1ilstinc-
t:ion heU-leP.n powers and possibi1ities.
It was f":'lt that more was
for the assertion of a present possi.bi1ity than [or that of
a power; one might have the power of standing without its being possible
f0r nne to be standing - if, for instance one was tied to the chair.
But in the present passage, at least, it is clearly held that the
fact that the opposite of one
s standing is the removes the
poc;.qibility of one's standing.
The contrast between powP.r and possi-
hili tv thus did not seem relevant to the discrepancy betw>en 1 hi.s pas-
sage and and Rhet.; and there was no need to j nt roduce powers,
as opposed to possibilities, here or anywhere else in q 4.
[t W<'\S further noted that the point at l047b IJ-14 I trP.ated
in terms of
the positioning of temporal adverbs, and of th-=- fall<Hy
0f composition, at Soph. El. l66a 24-32.
l_0_47b Aristotle here argues that
(I) "*B)
ThP argument takes the form of a !'_eductio; assuming that B is_!!?_.!:. pos-
sihle, it is deduced from this, by means of the converse of 'nothing
impossihle follows from what is possible' - viz., 'that from which
i.mposstble follows ls itc;elf impossible' - that A is not
possi.hl e either. Bnt why, in 20, is the suggestion 8 is tmpos-
sible put in the past tense, as if referring back?
It was suggested
!-hat lrrrw Or, &.6Uva:'tov (sc. -tO B) in 20 should he to 17,
)!l indePd hetnp; somewhat repetitive; but the double f(J"'tw in 17 would
t hr-:-n hA awkward.
So it was suggested that lc;-r;o Erh tO A in
17 IR might be a corruption (in capitals) of an original lc:rr:,,l OT,
i and that EcJ'tw 01, &.56va:tov in 20 mtght be an Intended correc-
7b 1.:0
tion of 17-18 that had been misplaced.
Even so, it was pointed out. that Aristotle has all the ingredients
of a direct proof by 20, without needing to employ a (for,
fromO A and 'nothing impossible follows from what is possible',<> B
is already implied). And the whole argument is of little value as
a 2roof of 'what follows from the possible is itself possible', since
it does is to derive the truth of this from the principle 'nothing
impossible follows from what is possible' (cf. e J l047a 24-26).
yllp in l047b 16 indicates the start of the proof.
third E r ua.l. depends on Ouva-tbv ; cf. e.g. b. 15.
B, = A.
In b 22,
In bl7, the
-rO 1tpW-tou =
l047b 26-30 Aristotle here asserts the converse of (I), namely (II)
(C,A -l'CB)
An attempt was made to save him from this by suggesting that in 26-
27, as in 22-24, the (complex) apodosis precedes the (complex) protasis;
but it was generally felt that this was grammatically impossible.
in 27 introduces not che protasis of the entire sentence, but the
protasis of the conditional which, as a whole, forms the apodosis of
the entire sentence.
The trouble is that (II) is plainly false, as can be seen by con-
sidering cases (like that of the coat which can be cut up or burned
in de Int. 9) where the realisation of one possible alternative pre-
eludes that of the other. The possibility of a Labour candidate being
elected may imply the possibility of "l Conservative candidate being
elected at the same election, but it is certainly not the case that
actual election of the one implies the actual election of the other.
Nor do the qualifications in 29-30 help. In the case of the election
it is a matter of the possibility of each being elected in the same
time and in the same way; and the only way in which the qualification
could help, would be if the only way in .-;hich a thing were possible
. .;ere that in which it actually happens, t>,1hich involves identifying
the possible and the actual and saves (II) only by making it a tau to-
logy. Furthermore, with regard to the problem created for (II) by
-1lternative possibilities, it was pointed out that Aristotle himself
in the very next chapter refers to potencies which may bring about
o:::ither of two opposed results ( e 5 1048a 10).
It ,...,as pointed out
that the mere fact that there can be a potency for healing which is
a potency for harming does not of itself entail that potency
for healing implies a potency for harming; but Aristotle clearly holds
that the potency for healing by the use of reason, at least, also im-
plies that for harming in a similar way.
i.Jhat then are the cases in which (II) does apply?
It was sug-
.;;ested that they are those in which B is a necessary prerequisite for
A, and in which that is the reason for asserting() B; if for example
A = house, B = foundation. This amounts to saying that B)
implies (A ..... B) in precisely those cases where (OoA has itself
been derived from (A ... B); which is hardly surprising.
And indeed
<TntJ,a.!ue:L in 29 may indicate that Aristotle has in mind precisely such
cases; CT'r'jj.J.aCvE&. = 'this is how I got to it'?
(t was pointed out that &.vd.yxT} appears in 26, while it does not
appear in the corresponding clause in 22-23. However, it t.;as felt
that &.ud.yxn throughout 14-30 applies to the relation of implication,
rather than qualifying the terms involved (unlike 6uva-r6v and d66va:tov);
3nd it was not clear that any contrast was intended between implications
qualified by <ivd.yxn and those not.
i il4 7h 16
l. 'Jisible at times ..;hQn it __ being seen?
wonder if l0413a 15-20 is relevant r.:o a '.:ertain debate on the
nature of possibility. It Hill t.:, i.f it is to Le rearl as saying
that a thing has its 6Gva.J.H' or.ly at times when external preventing
factors are absent. If this is 3pplied to like visibility,
r1r. ccmbustibilLty, then Aristotle \:ill in effect be siding with the
Stoics against Philo 1n the ensuing debate. :\<;cording to Philo, it
is possibLe for a piece of wood at the bottom of the ocean to be burned,
similarly a shell at the bottom of sea is perceptible, in virtue
of the 'bare fitness' (psile epitedeiotes) of the subject. Most Stoics
disagreed, and said that a second condition must
,-1e hcve a possibility: the absence of external
be fulfilled before
in allows Philo's view, '"hile in effect sides with
r:he Stoics.
[n fact, both sides are for there is no one answer to the
qne-:;tion, The answer can vary according to whether we are talking
0f a possibility of being q> 'd or q>-abi..lity, according to what is,
,\nd according to the context of discussion. F.xample: a piece of
.;ood hundreds of feet underground: the intervening earth would normally
.,_Je thought of as removing equally lhe possibilities of it and
Uu.rning it. On the other it ,.,rould more often be called combus-
tible than visible. '..Thether it was called visible would depend on
!:he context of discussion. Lf the context were distlngui.;hing between
_ 1tell Lglbles and visibles, it .wuld rank as visible. If the context
one of a slght-seeing holiday, it would rank a:;
r>.ridence in Necessity, Cause dnd Blame 78-9. Philonian examples:
)hiloponus An. Pr. i69, 20 f.; Alexander in An. Pr. 184, 12
196, 1. u:3ed in later
mti1uity, is interpreted by Gwil Owen in an a1T!lost oppoaite sense
1:s t:ne absence of interfering ccnditions. I should be interested
whether it could not be interpreted as still having the
")hilon:i.an sense. (A.C. Crombie, __ N.Y. 1963,
)] . )
L. l'1urus .2..E Philoponum, De :1'..l!lcU:_ (Rabe) 146, 10-13;
\lPxander (?) Quaest1.ones 18, p. n, ll r.--:.cknowl<:."dg;ements t:.o
,:,arnLos and 12 ff.
1 j l
2. a thing vis.1ble dt c. there dre not beings who can
,\nother crmtroversy, 1 . ..rhich very much interested Aristotle, might
perhaps be seen as a special case of the first one.
Would !:here be
knowables, perceptibles, or countables, 1f there were no ensouled beings
to do the knowing, perceiving, or counting? (The absence of such
beings be seen as a special case of an external obstacle.)
Aristotle has at least 4 discussi.ons, in probable chronological
(i) Cat. 7 7b 33-Sa 6: YBs, there would <>till be knowables and
without animals.
(ii) IV 14, 22 3a 21-9: Reverse verdict: nothing would be
ccuntable (so there'd be no time), if
rhere were no souls.
(iii) 0. 5, l010b l0-10lla 2: Agrees: perhaps (i:aw<;) no a is-
!_:h,eta, if no ensouled beings, but there
,;oulr! 3till be the substrata, '.-Jhich give
rise to perception.
(iv) !Je Anima III 2, 426a 15--26: :-lore complex: perceptibles in
their active qtate exist only during
-1ctual perceptlon, but insofar as merely
pntent:Lally active, exist \.Jhen perception
is not actually going on .
People have been so amazed hy the claim about ti.me in (ii), that
they have sought to put other constructions on Ar's words.
But Ar's
mistake here becomes more intelligible when the difficulty of the issue
about possibility is recalled.
His successors could not agree on the issue.
(a) Roethus Jisagrees ,.,ith (ii) and (iii) on countable and perceptiblJ
(b) Alexander agrees with (ii)
(c) Slmplic.1us agrees with (ii)
dnd disagrees with the rival view in (i)
1. Details in P.F. Conen, Die leittheorie des Aristoteles, Munich
1964. 156-69.
2. Boethus, a__E. Si.mplicium 766, 17-19 (= Ihemistius, i!!..__!'_b_y__.
163, S-7), and 759, 18-20.
1. Alexander, Simplicium in Phys. 759, 20-760, J.
4. :;Lrnplicius in Ph....Y2. 760, 33-"161, 5.
5. :JimolicilJS to R.W. Sharples) 196, 12 and
"l()TF:S Ti-iETA
3. If so, visibility (the passive power) CJn exist without sight (the
(]Ctive power).
Might not something, in suitable contexts, be as a count-
able, or quantifiable, or visible entity, even thOltgh everyone had
permanently lost the power of counting, quantifying, or seeing? If
so, it constitutes a warning about tne close connexion that Aristotle
makes in e 1 between the power to 1> dnd the power to be cp'd, although
doubt if it actually contradicts anything he says in 8 1.
rhomas Aquinas tries to an anaLogous example.
Does God
suffer a change and a loss of power when somerme loses her virginity?
For he is unable thereafter to prevent her having lost it; yet before-
hand he had the power to prevent the loss. Thomas' reply is that
both before and after the loss God has the power of prevention. It
is the retention of virginity which has become impossible, not God
'NhO has lost a power. Once again, on Thomas' view, there is a lack
of correlation hetween a power to prevent and a possibility of being
I don't think this can be right. Admittedly, when the virginity
is lost, the change does not take place :!:_.!! God, but still something
about God has changed: he has lost a power.
would compare a lethal
to lt.
Surely, it does not remain lethal, if we have all become immune
Admittedly, the change does not take place !.!!_ the virus, but
something the virus has changed: it has lost its power to kill,
because we have lost our power to he killed.
In this example, unlike
the first ones, we have the close correlation between active and passive
powers that 9 1 leads us to expect.
D0es it help that the last change took place in the passive parties
(the vulnerable humans), whereas in the earlier PXamples it took place
in the parties (the counting, quantifying., or humans)?
I don't know, but there are examples of the npposite sort. E.g. if
'Ve (the active parties) permanently lose the power of surmounting some
nbstacle, that obstacle (the e_assive party) ceas>2s to be surmountable.
2lo the question when active and passive powers are correlated is com-
plex. >1y point is: not always.
l. fhomas Aqninas, I, q. 25, a. 4.
t )_4
0:.. L2

Juspect, pending correction, that -.: c in ,q '5 rrn\'ed
influent1al was in a cc:rtutn TsL1mic
<reated the w-0rld with a lwP,inning.
':lrgument for 1-;nd's having
At '.iny T'Jte, the fir:n
';tep nf that argument is that, where there are .JlternJtive possibili-
ties, dS it is that tht=:>re are tor rhP location of the physical
'-'orld, there must be a whil..h decidP.S wh1ch possibility is realised.
Thi.s pattern of argument i.s PlUch discussed by fslamic scholars. Is
it not very close to l048a 10-11 '?
Where there is a capacity to produce
opposite results, for exdmple, the ctoctor 's capacity to kill or cure,
there be some other factor which dec1des ..inai to kurion), and
this factn is desire or choice.
Does 5 .
5 has been, dnd may easily be, taken ta imply determtnlsm.
the analysts r>f 'can' 'ls 'will, if .. ' i1as been reinvented by modt.:rn
.3oft determinists. ,\ga:in, Jadkko Hintikka has, for different reasons,
cited q 5 along 1.Jith 0ther p.-1ssages, .JS .;ho1ving that 'in one of his
moods ar least he rAr1 thus comes fatrly close o Diodorean cteterminism.'
Finally, Elizabeth Anscombe Clt'2S a 5, to . how that ,\ristot:le beli.eves
lhat what lS caused is necessitatect.
-;hall 'lerely summarise here the arguments in NCB 52-3;
136-1. to show that determi ni ;m , '3 not implied by 5.
Ar is
t_alking about c.;pecial cases, when says AvO.yxn ( HH8a 6 and a 14).
He ls not, for example, talking about e.rerything should call action,
but dbout actions springing from a rational <.::apacity like medicine. Hin-
tikka certainly allows that Aristotle Ls talklng dhout soecial cases.
In 97, 1048b 37-1049a 18, ,\r dist-inguishes the -1bility of a seed,
'lnce implanted in t_h0 'VOmb, to f)ecume a !luman r.etng from the mere possi-
bility !:hat ::>orne pJece of elt:mental "arth ..;ill one day turn into d
bronze statue. Surely 85 does rtnt. 1;nply that, if that
o:arth does indeed t11rn into CJ: SLitU, ever:; ">t"li!e in the process will
IJe necessitdted?
,: .
L,wayni a..E_. I /4, 5th argument
-1nd ae_ .. \verroem (' r. H.:\. Lfson, The Philosophy ')f rhe
. -U7-8). -----"
Hintikka, r"cme and Oxford t973, 2Gl-2; l957,
:t .-\nsc0mbe.
r1L;C! l'Jll, 2.
l.>2cture, C:tm-
.;n fES u:.--.1 THETA
I expect that Aristotle would have imagined he was well clear
,__)f determinism inter alia because of his clause 'when the conditions
are right'. doubt, he would think that coincidences often play
a oart tn whether conditions are right, and in Meta ph. E 3 he believes
he can prove that coincidences are not necessitated.
To get Ar looking more deterministic we would need to put 9 5
together with some of his other ideas. Thus an efficient cause is
that whence comes the origin of a change II 3 and 7), and change
is defined III l) as involving the activation of a If
we put all rhese ideas together, it looks as if all efficient causation
involves the activation of a while ( 95) are activated
of necessity, once the conditions are right.
we are closer to Anscombe 's formula, that what is caused is
necessitated. But I am sure from his other remarks that Ar had not
put all these other ideas together. And, even if he had, it is still
aot logically implied that whatever is caused is necessitated. For
it 1s not ruled out that may sometimes be activated and produce
an effect before the conditions have actually become necessitating.
The effect would then be caused, but not necessitated.
1Q_47b_l_l-_-35 Jl-33 dtstinguishes 1 t [lJLenl.y, (a)
1_t1e tnnate
(uuyyvt'Ov), \b) tnose resulr.ing fro.n habir.uatil1n, (c) r_hnse resulting
frl)rn learni.Dg.
31-15 assert t:1..1t (b) ind (c) em ()nlv he possessed
.JS the result of previous activity, \.Jhile this doc!S not apply to (a)
or to rasstve powers (not mentioned in the classification).
\Oytf in 34 picks up u.aef]m._ in 13.
:1ay 110t he ':J()TUe other ex-
3mples too, not falling under (b) or (c), previous activity is
required? -those under (a) indeed are innate, hut 15 the classification
in 11-33 exhaustive as far as active powers are cc>ncerned?
ample, the power of begetting is natural, but is not possessed birth,
if C""l)YYE:'JWV has that strong sense.
Further, what is the previous activity t0 in JJ-34?
it a previo'JS occurrence Pf the 2dme activity as that for which the
resulting potency is a potency, or not?
cvith the treatment
nf t.:;Et.S in II. 1 would that it is (and cf_ 9 9 lU49b
on the other h,1nd, the l').gical of 1ctualtty co potentia-
that actual:Lv presupposes th1:;> power to t.p, in wnich
r:he activity that pr\3cedes the power ro c.p must be 5(1 me other acti-
Alternatively, we m1ght dppeal to the contrast doing
something and doing it : ... ell; it 1...s :,y that we the power
t:o sing well.
l047b l5-l048a 15 The point of reference to soul at 5-6 ls clea-
rer if this is rE>ad in conjunction w1th A z - which is relevant to
the suggest1on that A 3-4 are i.n some dn L11 terruption.
is in a soul ilnd is of opposites.
Btlt what is the force of
Ul 3 J: is it :;imply 'opposites; like health J.nd (A 2 1046b
7), or does the expression cover 'lll contradict11ries, all cases nf
1c- t ng -or -not- qr ing?
'there are rwo dr14uments t:h(lt could
:1ave used. correspomHn\S t0 those n.Jo '-.lays >Jf r tva.v-rCwv. (A)
rational potencies are CIS can _r1r()duce either
nckness or health, \lecause. Ln cl vay, une hothi and so
r3tional potencies
L-,n .;nuld result.
cannot. ne a....:tualised aut;matically, <'!S a contradic-
1 B) 1t lS an ;bserved 0f rational behaviour
1 /h j5 :!'rES ;N
1L . o has the power not-to-cp. It was senerally agreed that (A) is
t:he argument Aristotle uses here; cf. the contrast with tvOc; ltOLT'J't't.XT)
in 8. Worries were expressed over the ddequacy of (A) to cover all
rational potencies; but it is claimed that it does in 9 2 1046b 5,
1nd 7-8 here suggest that the conclusion is intended to apply to all
rational potencies. In r 4 Aristotle expresses contradictories, as
i.n (B) by rva.C 't'E xa.t ) ..n; rather than by though Ev6.v't'La. could
rover (B) in the sense of 'incompatibles'. (It might also be remarked
that (B) would involve a more obvious begging of the question against
determinism - though for Aristotle, presupposing that human actions
are not predetermined, there would be no question here to be begged.)
Rather stmilarly, in 11-12, it was asked tvhether a distinction
..... hould not be drawn between, e.g., (i) the decision whther to do some
ioctoring or not at all (corresponding to (B)), and (ii) the decision
'...rhether to use one's medical skill to cure the patient or kill him
(corresponding to (A)). But it was observed that, while we might
_;peak of a decision not to do any doctoring at all on this occasion,
Lt was doubtful whether we could speak of a decision to do some docto-
ring independent of a decision to cure-by-doctoring or else a decision
lo kil (ii), corresponding to (A), has the advantage
For argument that it throws greater emphasis on the decisive
role of and than does (i}.
cO<; in 6 picks us the qualifications in 1047b 35-1048a
2, these in turn picking up 9 4 1047b 29-30 (which has a bearing on
the relation of e 5 to 94); tc in a 6 corresponds We;; OUva.-
at a 12, and this in turn is expanded in a 14. So in 6 we should
ranslate 'in the manner in which they possess the potentialities',
lr rerhaps 'in accordance with the qualification which apply co the
!1ot::o:ntialities which they possess'.
Does this discussion have deterministic implications (cf. (5)
'Jr Sorabj i 's notes)? It was agreed that d.vO.yxT'J in a 6 must express
necessitas consequentiae rather than consequentis (cf. dvd.yxT) in a
extra premisses are required, and Aristotle might not have rea-
lised the deterministic implications. Sorabji outlines a positive
ar-"!ument; it can be further strengthened by an appeal to the necessity
,,f the past as Irrevocable. One :-n.Jy compare. in the case of rational
1] 8
potencies, EN VII. 3 1147a 25-28; there the existence ut d setti..t:d
"'tpoa.Cpe:cn<; seems to play a similar part to that which be played
hy the necessity of the past in the more general an'!;ument cuncerning
irrational potencies too.
Even if the deterministic implications J.le realised, Sorabji sug-
gests an escape route; from the fact that X's capacity to <p will be
actualized given condition a, b and c, it does not follow that it may
not <p sooner, when only a and b apply. Against this it might he Stig-
gested that this only shows that X had the capacity to q:l given merely
a and b; and in that case it follows from 5-7, where irrational paten-
cies are concerned, that it must as soon as a and b (only) apply.
So far, however, we only have 'whatever happens is not
'only what happens (at each given moment) is possible'; see further
below. It was observed that 13-15, too, do suggest that, whenever
potency is actualized, there must be something that necessitates
its actualization. For otherwise, if we could suppose that an irratio-
nal potency may somtimes arbitrarily be actualized before its actualiza-
tion is necessitated, why should we not suppose that a rational potency
too may sometimes arbitrarily be actualized in one 1c.;ay rather than
dnother, even without the presence of desire to 11ecessitate its acttiali-
zation in one particular way?
-\nd, more generally, it seems a rather Heak objection to deter-
min ism if Aristotle had realised the deterministic implications,
'Nhich he probably did not - just to suggest that things may sometimes
happen sooner than they have to; one may compare Hintikka's suggestion
(i.n 'Aristotle on Modality and Determinism') that Aristotle tries to
deterministic equation of the possible and the actual merely
by pointing out that the realization nf some possibilities is inter-
rupted (in a predetermined way, for all that the argume11t says?) before
it is completed. ('Aristotle on Modality and Determinism', 59-79.)
,(upCwc; in 1048a 12 refers to in 10 (1.;hich raises the old
problem of whether the strongest impulse is that most strongly feltor,
rautologously, that which in fact prevails).. xupCw<; night suggest
'',,lit hout any misapprehension about the circumstances (etc.)', but even
iesires accompanied by misapprehension lead to action. fhe reference
in Eudemian Ethics II. 6 1222b 21 to d.px_aC which are fi.rst sour-
of motion was compared.
l i 9
Sorabji (.l) :ju5gests Ln.tt Lhis expre;:.ses a vLew of possibi-
L1tv (I) sLmilar to tnat held by the if is prevented
oy factors, the possibility uf its c,ccurrence is removed.
nf cJlHSe, the way in which this Ls inteLpLeted by the Stoics will
1e different from that in whi..:h it is interpreted by Aristotle, since
'.hey assert universal causal determinism as an explicit thesis, and
'e does not.) We inclined, ho,.,ever, to interpret these lines rather
lS expressing (II) a view of possibility like that of Philo as long
,15 -1 thing is prevented, it does not happen, but that does not mean
hat the possibility of its doing so, too, removed. This means
Jldt the commentators who attribute a view of type (I) to Aristotle
CAlexander in An. Pr. 184. 10-18, Philoponus in An. Pr. 169. 21-23,
Lf. also Philophonus in de Gen. et Gorr. 302. JO ti.) are in error -
' f at least they have this passage in mind; they are speaking about
nasaive possibilities, whereas the present passage is concerned with
1ctive poLencies.
In support of (11), it was that 18 dues suggest that (e.g.)
'\)ower of burning' is being restricted to 'power of burning dry stuff',
__;o that the power of fire to burn wood is Hot removed just because
his piece of wood happens to be wet, rather than that (I) the power
LS only present if the wood is not wet.
uly should Aristotle make such a point?
The question was raised,
Answer, toresist those, real
Jr imagined, who try to argue away possibility by pointing to impedi-
and so tend, in the extreme case, to restrict possibility to
vhat actually happens when it happens, like the Megarians.
This, indeed, was the second main argument in support of (II)
nd against (I); the combination of (I) with the implication of 1047b
t5, that possjbilities are actualised when and only when all
.ondltions for their realisation are fulfilled- that, as soon
s t.Je have all the necessary conditions (including absence of preven-
and, tn the case of rational potencies, presence of desire to
r1ng about the particular result), we also have sufficient conditions-
vl tl tend to push Aristotle into the Megarian position. (Indeed,
1nl.:>:r to push him into it, Hintlkka, (Time and Necessity 201 f.)
"llpiwls 1. wo premisses; \a) possibilities realize lhemselves
:1nt orever,ted from doing so (dcri'Jed from ':) 5 l048a 6-
\:-', .L.I. '
; and g 7 l049a 5-9, 13-1'5), (b) that tt ..1 i.::> ,,,Jt rea-
lised there must be some external f.1Ctnrs preventing it frnm being
realised (so far equivalent to (a)) 'or,
making it impossible for it to be realised'. (;-tv italics throughout).
On the other hand, '..Je >?;rant th;:lt 'necessary' 'not possibly
not', it at once follows from 'whenever a potency ls actualized, there
must be something that necessitates its actualization' (above) that
'whatever happens could not possibly happen', and hence that 'nothing
other than what happens is possible'. In 'Aristotle on Modality and
Determinism' Hintikka clarifies his position on the presence of
preventing factors does not remove possibilities, hut it is
the question Nhther there are any total possibilities other than those
that are actualised which is relevant to the issue of determinism;
18 ff. In de Motu Animalium 4 699b 17 ff. Aristotle is prepared to
c;ay that 1.t is impossible (in one sense of 'impossible') for us tn
see the men on the moon, even though they are not <Jf nrcessity invi-
which might seem to suggest (l) rather than (II).
f<>.ren with (I) \le might find an escape Loute fnr Aristotle by in-
sisting that the absence of any dctual is not equi-
'Jalent to the presence of sufficient conditions fnr the realisation
(and the Stoics may well have used some Sl\Ch Bnt it ' . .Ja"i
pointed out that, even so, there are certainly m1ny occastons 1.Jhen
a builder is prevented by something from builrti11g; so if we escape
t:he implication that he loses his <::1pacity to build whenever he is
not actually building, we are still forced t) hn1cl that he may gain
and lose his capacity to build many times a Jay (whever it rains, for
furning to (II) then, it was pointed cut thdt Lhis requires Jt0t'i:v
or understood in 15-l6i i.t is l:HleP.d in the ( Hor..Jever,
if it is retained, Exet &e dw 6Uva{JLv) 't'oU '11"a.p6v-roc; -rr1U
xot WOt lx6v'toc:;: 1tOLELv might have been clearer.) Further, lll) requires
caking ob Ouvfroe;'ta.L in 16 as 'he w1.ll 110( able, altho11gh he retains
the capacity'. (One might try to U:i.s ;,y :.-tyc11g that, while
prevention by external factors does not rne the ah-
--;crce of other necessary conditions- those 1.nternal to the pa-
t - rioes; but la) this is implausible and lb) 18-20 snggests that
r he absence of e.xternal preventing fact<Jrs is i:1cJudcd L!t 1he other
1'lalifications, rather \han being cr1nrraJted '>Jith tbtm.)
' '0 '!l
:.-sa ')
[Alexander} in Metaph. 577. 33 f .. 1eLt.nitely takes 16-21 in the
Philonian way: E C ya.p 'xe: -roU 1tod')01U, Ot tl?t6 'tt.vo<;
ob JtOLfJaEL. fn 16 he expands EC Ot l.ifl as e;t 6 Exet. 'ti)v
06voi-!LY, however, ; and he does interpret 15-16 as expressing condi-
t icns on which even the potency is no longer present; one doesn t have
the power to cure a man if he isn't ill (577. 27 f.).
16-21 presumably applies to rational potencies too, though it
is in the context of the rational ones that it is introduced.
Are things visible if there are no beings who can see? (Sorabji
It was doubted whether Lhis in fact a special case of the
first issue about possibility in Sorabji' s notes; is it entirely natural
:o regard the absence of anyone who can see as a hindrance to a thing's
heing seen? (We must now be talking of passive powers.) The discus-
si.on in (iii) is in terms of a.Ccr&rr":6v rather than of &uva.'t6v; the con-
nection may not be as strongly felt as that between 'perceptible' and
'possible' in English, and the point in (iii) may be one about the
systematic ambiguity of words in -'t6v rather than about possibility.
Might a.Cafhyt6v in (iii) not be taken as 'perceived rather than as
If (iii) is not therefore a str3ight assertion that there can
no perceptibles without perceivers, we are left with (11) as the
only assertion of this - and (b) and the first passage in (c) are com-
ments on this. We cannot argue in the case of (ii) that
,rJeans 'r.ounted' !"ather than 'countable', because of the contrast ii "tb
But may the
point not be a special one about 'counting' rather than about percep-
t ion? - we may have five trees, as a matter of fact, even if there
re no men to say that there are five, but we ..:annat have a quintet
'Ji trP.:Ps HI t.he absence of someLone to .:aunt them, just as, it might
be claimed, if there were no rational beings nothing could be 'late'
'lr a 'weed' (though we were not all happy with this).
<10 men, '.olould anything last for exactly one year?
If there were
But Aristotle's
rlaim is stronger than this; it is not just that notlling would be coun-
ted, but that nothing could be. We observed that the contrast in
of meaning between 'has the possibility of being counted' and
'has the capability of belng counted
may enable English L.O express
subtle distinctions which tile Greek Cdllnot. Je ctlso not-:'d that there
is no suggestion of a distinctiGJlbctween rhe \li different
predicates, like the suggestion dbove '(nuntctble' be d 3 pe-
cial case, in our discussion about !Jossibiljty 1n l04Ha 15-21.
Can an active power exist __
Sorabji (2) (paras. 2-3). What is Lhe pm;er is supposed to have
lost, in para. 3? 'The of X from losing her
ginity'; or, more strictly, 'the pnwer ilf hi:"inging it <Jbout that it
is not the case that (tenselessly) X lnses i1er virginity'. But did
he ever have this power, or did he only have
the power of preventing
X from losing her virginity, provided she lla$n't already done so''?-
Hhich power he retains. (To avoid the t0mporalqualiftcarton 'already'
,lne 1flight suggest rather 'the power to prevent A her virginity
as long as she has it'i but that Ls either e'luivalent or tantologous.)
God has the power. at each time, to that chv; table alwdys re-
tains the colour it has at that tinte (he the power,
Hhich is :.;hy .,;e can paint a table ct different coloc1r, but that doesn't
mean he doesn't have it). Has he lost -"!HV power lf, once we paint
brown table green, he no lon5er tFJs the ld keep it brown
for evermore, but only the power (1Jhich ile ,:ltJn't h.'!Ve herJie) to _k_eep
it for evermore?
Against this it was objected that, if ...,e are too Late, we do feel
we have lost a power - or is it rather that 1ve l1.1Vf! rnissed an opportu-
nity? God will then suffer no Loss of in the case of X's vir-
ginity, but he will have missed his opportunity df preserving it.
What about Sorabji 's example of the vlrus? Can '-'e ::1rgue that the
0nly power it ever had was the power t') kill peopie \,"hO 1.vere susceptible
1 1) it (which need not be tautologous, if . .;e c.Jn ;;i.'e a speci.fication
of .vhat made people susceptible), tlnd that it n:>taius r-hat power, even.
as a matter of fact, thece has heen a (rlecisive .-tnd permrPJent)
change in our make-upwhich :rreans th<-It
fvreseeable future be, susceptible 1 o it!
1 s llJW, ,Jr Hill Ln t
olr ;S Ultr dttentlon not
rather directed, in such a case, lo the f.Jct rhat t1te .,iJns previously
l1;ls the power to kill rnany ar most tJ'-"!nple, and nmv rlo"'s nut?
in any 1:ase denotes a u::ndencv, r1.ther : !mn a capacity;
:ir:s may retain the capacity to ":f rvrr:- 1:; person who i
.-1 l
} a l J
_;usceptible to it, Out we :night not rilercfore be prepared to say it
is iethal. True,
:an no lcmo;;er kill
even if the make-up of us all changes so that it
any of us, we might say that it has the potency
o kill i.n the
1gain soo as to
sense that it could killus if our structure changed
make us susceptible to it again; but in that case any
virus might be lethal, if we rould specify the change in us (however
improbable) that '..Jould give it the power to kill.
i048a 20-ll Is &.<pa.LpEi''tnl. middle or passive? One would expect the
,Jassive (cf. 1048a 33); there are parallels for cicpa,l.pe'toiKu in the
middle in Aristotle, Lut not in such a use. Nor is the use of
to parallel in Aristotle. So delete 'v1.n, and take "t'Wv - 1tp00"6v-
twv as genitive absolute? But the general sense is not affected in
J. case.
.!..2-24 This, being concerned with limitations of that of which
1 power is a power, may provide further support for interpretation
(II), rather than(l), of l04Ra 15-21 (the 'Philonian' rather than the
')toic; see notes ad. lac.). Un either (I) or (II), it was argued,
there is a problem over how we can know what a man's capacity is -
the Megarian position of 9 3 csLapes this by limiting capacity to
is actually happening. But Aristotle would no doubt reply that
'le can have a t,eneral of the relevant conditions
;r(!, dnd can thus know or not particular men have particular
apacities particular circumstances. We; at anv rate.
l048a 25-JO 86 is conce-rned not wi rh C)l.JvdLLEI.c; to i.'S x1.vry::ra.1. and simildr
potencies, as in 9 t-2 and
to what is lve:pye:Ccr. This
9 5, but
with \-Jhat is btJv6.ue:l. as upposed
picks llp e l l045b 35-l046a 4;
1048a 28-30 apologise, as it were, for the rhgression. -rp6xov 't'l.v<i
in 1048a 29 takes up the 'lC.a.N]c; of 8 1 1046a 17, rather than the qual if!-
cation of A 5 1048a 17-18.
1048a 30-1048b 3 Aristotle gives examples ot the potential and actual.
The argument is one by analogy (36-37; read -r.q; in 37, comparing l048b
7, with Jaeger), which is a type of l1ta.ywy-ti (35-36); but not all
is a matter of analogy - not that from species to genus. for example.
'Analogy' in Aristotle is always a proportional relation, A:B:C:D (or
\:H:B:C). In 13 [Alexander l - but it might be from the genuine Alexan-
der supplies this seems right (the feminine shows that
the r-eference must be to a line, ), ilayduck, the C_IA.G
editor of Alexander, was Hrong to print as it it ;ere an actual
quotation from a text diffC>ring from rJurs. ( i!'
579. 6).
One case of the potential-actual disti11ction is !Jetween matter
and the torm-matter compound (a 32-.33, b2-3). It '"'as noted that the
distinction here is only d relative one; Lhere lS no j\lggestion, as
there is in 23, of the removal of a_"!:._l_ of the substrate,
leaving only 'prime matter'. The reference to 1'\.Va. \S\11v in l048b
9, too, fits with this.
',!hat is it, then, that i1ll th examples have in cu1nmont fhe '.mod
is r;otentially a Hermes, but the whole line is not potent idlly a half.
then, t:he point is Lhat the Hermes ts the '-"Or:ld potentially,
dnd the halfllne the Hho1e? l3ut one can hardly that the actual
knower is the potential knower potentially, one might :;ay
c. hat the actual knowledge was. So perhaps the cu:nrnon feature is that
in each case the way in which A is f) ;nust he -lUaiified:
'V)W, but it cnn1rl cc,me tu be B (cf. Ci.v Ouva.'tO<; ry, a 34).
A isn't B
rhar_ ..mod
lS a (in i'l way), that whole has a half in it (in a way), that
has knowledge (in a wav). .:. l(J48h 12--13.
l048a 30
Worries were expressed about the of knowledge (a 34-35);
is Aristotle right to suggest that a man who knows something, but isn't
considering it at present, knows it is a different way from one who
'I hate
It was remarked that with 'know' we don't use the continuous
like 'I am knowing', while with 'hate' for example, we can say
Mrs. Thatcher, and I hating her from ten till ten-thirty'
if that is when I saw her on the television so that my hate was fully
actualised. Would Aristotle's point be more acceptable if we render
by 'understand'? One may understand something, but yet
r.ot understand a particular exposition of it if one is distracted.
True, if a man is asleep and we expound Pythagoras' theorem to him,
we wouldn't say 'he understands it, but he isn't understanding it now';
but that is because, when he is so obviously not listening, to say
he isn
t understanding is superfluous.
l048b 4-9 Jaeger argues that d.l.fiWpLO'}.llvTJ in 5 cannot mean
guished, as one member of the division' with (that would rather
require 6 and hence reads <+!> and takes the term
to refer to the production of the actuality by its being separated off
from the potential (cf. l048a 33, and B 5 l002a 24). But, (i), <-/)>
that is a restriction to one particular type of
rather than an explanation of actuality in general which
is 1.,rhat is wanted here; and (ii). it is indeed the case that not all
cases 0 f actualization are cases of separation (cf. the knower in a
34-5; though it was pointed out that there might always be the tempta-
tion to stretch the language so as to say that the actual, in general,
in' the potential), but ..,hat we want here is a reference to actua-
tity in general. So without<+!> must mean, not indeed
Jistinguished in the division
, but
marked off in the division'.
Jaeger Is interpretation required the nominative ed:tEpov J..16pLOV ; .even
on the alternative interpretation the nominative seems much more natu-
ral. What would the sense of the dative oe: one part
, 'in the
Trt h 7 Hermes ts not said to be in actuality by being in the wood,
but rather by If the first alternative, lv is
,_aken in this way it can be restricted to those cases Nhere 'in
really appropriate, the other alternative, "tCO<; "'t'oiJ'to to <:uver
the rest.
l048b 9-17 Ross supplies 11 in ll, Jaeger a whole clause, e.g. 'tL
an anacolouthon, the construction switching from
former involves
'tb d:u
supplement avoids
If with no addition were tolerable as a dative depen-
ding on Cl.AM.oc; the following datives could he explained by attraction.
The sense is in any case clear; Lhere is a contrast between, on the
one hand, cases like those in ll-12, vhere what is said to be poten-
tially is at some time unqualifiedly so, and, on the other, cases like
those of the infinite and the void to v.;rhich this does not apply.
ra\ha. in 12 = etc.
In 13 understand, with not xa.t d1tAW<; d.A:r,8e:6e:o--
8a.t. 7tO'tE but tc:rt't. ; for the former would so have to be understood
in the second cL1use, i.n 13-14, and is not an example of d.11:AWc;
It is, in nther t.wrds, the tirst (d the two clauses
in 13-l4 that is the operative rme; in the case nf is seen the
first sense as well as the second one is ,wa1lable, but in the case
of the infinite and the void this is not .'>o.
the statement in 14-15 that the lnfinite
What is the force of
exists potentially yv1.&oe:1.?
That we knc'w that further division will
always be possible; not that we know that 'there is a possibility of
..my number of divisions', but rather that we know that 'for any number
of divisions, there is always a possibility Jf more' - and presumably
for better reasons than that have always found it so tn practice
11p to the point .Je have now reached
then is the corresponding claim about
i.n rhe dividing process.
the \'uid? Perhaps that
we can
always (at least in theory) reduce the amount of air in a container
heyond the point to He have ii.lready reduced it, but never produce
a perfect void. It t.Jas remarked tha.t A,ristotle rejects the void,
in IV, on physical grounds, whereas the point c1hnut the infinite
livisibility is a mathematic une; but, does Aristotle distinguish these
types of reasons in the way do, and is the infinite divisibility
of not a physical AS well 38 a C11nceptual pol11tl
',:4Sb g
With 14-15 may compare tite contrast between what is separable
SpYtp and what is separable only A.6Y<tJ In 15-17 'tO 6 is
to be taken as a ::;econd p.Jrallel to -tO i.J.TJ {llto"-.E:CJte::Lv 'ti"jv Ot.aC-
rJEULv; the structure is chiastic, in 14-15 'potential not in way A
rut in way B, in 15-17 (the .,explanation) 'warranted (Hope's transla-
tion) as potential by B, not by A'.
1048b 18-35 These lines are bracketed by Jaeger as a later addition
by Aristotle himself. Jaeger's Qrgument that 35-36 refer to the first,
rather than to the second half of the chapter seemed weak to us as
a reason for supposing a later addition, and we weren't happy about
the relation between the theory that 18-35 were a later addition
Aristotle himself and their umission in EJ; if there were really two
recensions going back to Aristotle himself, we would expect signs of
this elsewhere. But, on the other hand, if the omission in EJ were
purely mechanical (due to the loss of a page, for example) it is odd
that it corresponds so exactly with a section of the argument.
'are among things concerned
with the end', though one might have expected 7tp0c; rather than 1te:pt.
J reads -cb (O""X.va.C vE . v and deletes f) lo-xvCl.()Ca. because otherwise
it would be implied that <J"Xya.crCa.is the end of lcrxva.Cve.v, whereas
in 29 Ca"';(vCl.()Ca. is described as a xlvT)c:n.c;. [IJ""'X.va.!vE!.v is transitive
in force. Bywater reads i1 [<TX.vCl.()(a.; but that seemed to us too obvious.
-r:a.U'ta. in 21 == -rWv Jv E'rr'tL in 18, a.i,..ra. in 20 being
-1n example of these. There is a shift in the use of "Jtpa.;u;, which
cTlCludeS actions directed tOWards ends in 18, but iS limited tO those
in 21 (with a qualification - but see below) and in
'l (without a qualification). This u3e of the same term both for
1 and for one of its species, however, 1s characteristic of Aris-
rotle. difficulty was felt over the claim in 18-19 that no action
vtth J. is itself an end; contrast the argument that one end may
subordinate to another, being desired for itself and for the
1Jrther ::nd. in Nicomachean Ethics I. 7 . Is this, rather than a worry
. bout t shift in the application to 1r.:pli.!;u; the reason for the qualifi-
--uon in 21-?.2? ft was noted. though, that the idea that lvtp-
t' n
are more valuaole Lhan vi)O"E Lt; is found elsewhere in Aris-
totle, as in the treatment of ')E:I,L,lp(a. in racomachean Ethics X. Does
,-,:lpa.c; in 18 have the sense of 1LX.oc;. or is there a distinction, 1(tpa.<;
referring to the of the action rather than to that for the <>ake
of which it is done, in the sense of what is most valuable about it?
For - although Aristotle's argument here would not allow it -the gcJal
of an action is not always the most valuable; if one plays chess at
all one must play to win, but that does not mean that one must value
the winning more than the playing. In other words, 'for the sake
of' is ambiguous, though Aristotle does not realise this here. We
compared the later Stoic distinction between the skopos (aim or goal)
of a virtuous action, and the telos which is the action itself (Long,
Phronesis 12 ( 196 7) 78 f.).
Does Aristotle have the notion of a mathematical limit (cf. his
remarks on the squaring of the circle in Physics I. 2 and
11)? Yes, he does, but the limit is not part of the sequence. A1-
though we cJ.nnot reach B by completing an infinite number of intervening
we cdn still reach B; but B is not any point in the series
0f infinite repeated divisions of the remaini.ng intervening space,
and similarly a circle is not any rectilinear figure, however many sides
the rectilinear figure might have.
l048b 23-35 Jdeger's supplements in 23-24 are not needed. 26: lf
the perfect were not compatible with the present, the actuality would
have an end - once it had taken place it could not st i 11 be taking
place. X!.vfp-e:Lc;, unlike lvtpye:La.L, can be done incompletely; and
(Ethics) they can be done quickly or slowly. Is Aristotle's point
just that the present is <;;ompatible with the perfect for tvtpyELa.L,
or that the present the perfect? The former would avoid diffi-
culties over the fact that one cdn assert xe::xCvrrtoL at the first moment
of a change, too VI. 6 ); but 'has moved' (simply) is different
from 'has walked from A to B' or 'has built a house'.
'is walking
fr\lm A trJ B' doesn't rule out 'has ' . ..Jalked from A to B' (at some other
time), hut we can't infer this, and he certainly hasn't completed this
.Ja lk. 32-J/.j.: ...;e preferred to take l'ttpov and 't,_, a.U't6 as objects,
with Penner, r,lther t:han as subjects; Clnother possibilitY, 'are not/
; li.:OHh ! A
l048b 23
are the same thing' would be easier with infinitives in 33-34, and
are the same thing' is weak. for 'one implies the other' (the relation
is not reciprocal, for 'has seen' doesn't imply 'is seeing').
as object in 32-33 answers the objection that you can't be walking
(from B to C) and have walked (lrom A to B).
1048b 37-49a 18 It was thought that e 7 Is discussion of when one thing
is potentially another might shed light on the issue we debated earlier
(see on 1048a 15-21) as to whether or not the circumstances on which
the actualization of a capacity depends belong inside or outside the
specification of the capacity. Thus, is a piece of wood at the bottom
of the sea (a) visible (because 'visible' means 'can be seen if circum-
stances c obtain') or 'b) not visible (because 'visible' means 'can
be seen' and at the bottom of the sea circumstances C do not obtain)?
If this is the question,
87's answer seems to be (b). Neither a
quantity of earth nor a seed is such that it can be(come) a man (49a
), so 'can be(come) a man' must mean not (a) 'can be(come) a man
if circumstances c obtain' but (b) 'can in present circumstances be-
(come) a man', with 'in present circumstances' elucidated in 49a 5-
18 as the obtaining of all the required bar the efficient cause (exter-
nal in the case of artistic production, internal in the case of natural
Given these circumstances, nothing prevents the efficient
cause (the artist's wish, if he has it, the seed's inherent nature)
initiating the process of actualization, in the way it would be preven-
ted if some further change was required before that process could start,
as e.g. earth must be changed to bronze before it is ready to be(come)
a statue.
that it could be true in this way that nothing prevents the
bronze being made into a statue if the artist wishes (all is ready).
even though in fact it won't be - won't be either because the artist
doesn't wish it or because
interferes to prevent it.
some outside agency (earthquake, sabotage)
The !at ter is not what Aristotle means
to exclude by
'nothing prevents it', or else he lapses into the thesis
that nothing can be unless it will be!,
Second thoughts suggested, however. that the above exposition
is attached to the wrong question.
The talk in this chapter is not
of having a SUva.IJ.I.c; to be F but of being 6uv<i1-1e:L F.
The quest ion
is not 'When does something have a capactty to he(come) a man?', but
''..Jhen is something potentially a man/a potential man?'.
lf there
is a distinction between these questions, it t:ould presumably be true
0f earth that it has the capacity to be(come) a man (Ar is not denying
that the earth has the capacity it has), tmt false that it is

In which case the chapter offers no guidance on the general
issue of how much goes into the specification of a 66va.!-LI.t;
The thesis
is still not that a thing is F only if it T..Ji!l he F, but rather
that if something is 6uvd.tJ,e:1. F it will be
pr.Qvided that things go
r)n from here as normal and (in the case of artistic production) provided
also that the artist wishes. Ar assumes that we can say that things
are set up to proceed to a certain outcome Ln the normal course of
Even if we cannot spell out the proviso in detail, we know
in a common sense way what sorts of thing would count legitimately
as an abnormal interference with the normal 11f events.
l049a 18-b 2 lxe:Cv1.vov is a lexical device for picking out the tE;
(in one sense of that term:
lhis chapter (cp.Z l033a 7 ff,) identifies as what is potentially the
subject \ve describe as lxe:Cv1.vov.
Thus y is x-en if x is potentially
y. It follows that if there is a y which is not x-en, for any value
of x, then for that y there is no x which is potentially y, i.e. no
x which is the matter of y.
In that precise sense y is prime or first
matter (49a 24-7):
first in an ascending series such as fire-air-
earth-wood-box, last in the same (purely illustrative) series going
is not
This is not the traditional prime matter nr tntally neutral
Nor do the conditions given require that y he unique: if y
reducible to anything else, it does not fr;J1,Jw that everything
is reducible to y; the terminus of one <;eries nef.:d not he the terminus
nf all. The only 'traditional' use of iS\T] in ,\r is tvhen he <;peaks
of Plato's xWpa. as \)\.,.
Accordingly, to say that prime matter, e.g. fire, is not 't66e: -n is
J:i l'HETA
not to say that it is a stuff without qualities or attributes, but
that it is not a reidentifiable something. Is this compatible with
the talk at 49a 24 of 'toOt 'f0 >;6Aov or does Ar mean that it is only
at the terminus of a series that you no longer have i.Je pre-
ferred the first alternative. 49a 27-36 is a quite general contrast
between subjects which are 't66e: n and subject as matter. What 23-
4 offers is a proportion: this wood is to this box as wood to
We took here to mean: without the sort of qualifica-
tion (concerning intermediate changes) that be required is 'earth'
stood in place of 'wood'; this results directly from the information
that a box is (22L given that a statement of the form 'y is
x-en' is equivalent to 'x is potentially y' and the conditions already
laid down for the latter. Now, this box is not 't6Se:. reidentifiable
as the same box, because it is this; rather, because it is a box it
is reidientifiable as this again. Correspondingly, then, if this
wood were "t60e:. tL, it would have to be so because it was wood, not
because it was this wood. But it is certainly not the case that it
is 't60e because it is wood - that is the point of 49a 27-36. There-
fore, it is not -t60e: "tl. It can only be called this wood because
it is the wood of this box, which is "t'- in its own right. In
other words, the notion of 'this wood' is derivative from that of 'this
In connection with tnis it
fire is described as the most form-like
was remarked that elsewhere
of the elements (De
,-:;orr. 2. 8 335a 19; Meteor. 4. 1 379a 16, the other elements are matter
fur fire).
liJ49b 2-.3 dnnounces the completion of the topic introduced at the
beginning of the chapter.
l 32
Note by G.E.L. Owen
Priority in 9 8
1049b 4: such refs. are generally taken as pointing
is not in Bz. Ind. 626b 30-31). 6 11 distinguishes
to 6 (but ours
Qualified priorities (cf. 1018b 29-31), 1018b 9-29, (a) in place
(nearer), (b) in time (earlier, where the contrast with spatial priority
is stressed, b 15-19), (c) in change (nearer to 'prime mover' in sense
of e.g. 242a 53, 242b 72, cf. Met. 1050 b 4-6, not the proximate
mover of 243a 32, 1049b 26), (d) in dunamis (more powerful), (e) in
order (which could be but is not generalised to cover the preceding
cases as III is in 1019a 11-12).
II Unqualified priorities, 1018b 30-1019a 2, (a) for knowledge, viz.
(a.l) by logos, e.g. universals over particulars or a sumbebekos over
the composite (cf. Z 1030b 12-13), (a.2) for perception; (b) as proper-
ties of such prior things (pathe, corrected to intrinsic properties,
1019a 1).
III (not expressly connected with the qualified-unqualified distinc-
tion, but apparently given focal application to I as well as II in
1019a 11-14):
Priority by nature and ousia, viz. where A can exist
without B but not B without A.
as parenthesis.)
(Ross,Jaeger, Reale take 1019a 4-11
Z 1 l028a 31-b2 distinguishes priority in logos, knowledge, time;
but it apparently identifies the third (I (b) above) with III; and
although it prima facie distinguishes II (a) from II (a.1), it is effect
conflates them by connecting II (a) with ti esti.
and not taken up in e.
g argues that energeia is prior to dunamis
Not a neat
A In logos (1049b 10-17) and hence for knowledge (1049b 17), thus
recognizing connection in II (a) -(a .1) as formally z 1 did not, and
ignorning (II) (a.2) and (b). Examples are the logoi of oikodomikon,
horatikon, horaton, though at III 1 20la 9-b 15 oikodomein-oikodo-
mesis is a kinesis requiring definition by dunamis (as even horan is
in such causal explanations as De Sensu 438b 2-5); no doubt Ar is thin-
king here of building as an achievement, not the process whose com-
pletion cancels the oikodometo'!_ 20lb ll-12).
(ii)-(iii) below.
Contrast on B
In time (1049b 17-IOSOa J), sophisticating on I (b) by distinguishing
hetween numerical and specific identity in the temporal items:
(i) actual indtvidual is preceded by (same) potential individual
but this by (other, but specifically same) actual individual which
is the proximate agent of change (ctr. I (b) on proton);
(ii) ( l049b 29) 'x can build' entails 'x has built'. First,
'hence' has nothing to do with B (i) or A, and must be an independent
application of B. Secondly is building an achievement here (implying
that when building his first house the builder does not have the abi-
lity) or an ongoing activity any part of which is building (as kithari-
and the proximity of B (iii) would suggest)? Neither exactly
suits III 1;
(iii) further (alla) if a process is going on some process must have
been completed, e.g. if x is learning x has learnt (something): if
building is an analogue, this takes the second option under B (ii),
conflicting 201b 11-13.
C In ousia (l050a 4-b 34)
'less strictly' (cf. 1050b 6):
by having form which through coming later in time is prior
to the potential (no more than a conjunction of A and B (i)?),
(1. b) ( hoti, l050a 7) the of is its end or purpose,
viz. the form, even when the end is realized some kinesis/ergon or
some product (cf. (i.e));
(i.e) (eti) matter is potentially the form to which it tens (conjunc-
tion or generalization of (i.a) and (i.b)?).
Nothing yet which specifically takes up A III; but now
(ii) 'more strictly (lOSOb 6), in that '../'hat ts possible is capable
of either being or not-being, and what is ecernal must exist without
this possibility. (On the extension of all dunameis to a capacity
for opposites, ctr. e.g. e 5, see the half-defence at 1050b 33-34:
does an irrational power include the power to be absent?). By applica-
tion of III there cannot be potentiality/ contingency without actuality,
hut this need not entail Plenitude; merely that there cannot be contin-
,;;ent beings without non-contingent heings.
Analysis of De Anima II. 'd7a 16 by R. i!elnJrn.1n
After distinguishing between the first level potentiality. first
level actuality and second level actuality of knowledge, Aristotle
417a 30 Both of the first are potentially knowing
31 but the one <becomes actually knowing) by (a] being changed
through learning and often
32 changing from a contrary state [Ot.c\ ua&ftm;wc; xcrt
but the other < becomes
actually knowirig) [b] from possession of sensation
or grammatical knowledge, tve:pye:tv &e:t<: 'tO
Being affected (-rb "Jtd.cry_et.v) is not simple either, but [l] the
nne is the destruction by the contrary, [2] the other rather
the preservation of the
potential by the actual and the like, in the way
potentiality is related to actuality; for what has
the l?tt.<r't'fJ!-J.TJ becomes 8e:wpoUv, which either is not 0\A.ot.o'ikrea.c.
(for the advance is into itself
and into actuality), or it is another kind of alteration (<iA.-
Ao L
So it is not right to say that what thinks is altered when it
just as (it is not right to say that) the house builder< is
,1ltered >when he house builds. Now
10 (I] into actuality from potentiality xaTa 'tO voaUv xat
11 is not correctly called teaching but something else.
12 But {II] that whlch from potentiality is learning and taking
knowledge (TO 6 'lx Oui.IOIJ.e: L Ov'toc; IJ.ai.l8d.vov xa.t f..<J+lad.vov l"Jtc.O'T'ft
13 by the agency of the actual and the capable of teaching
14 either should not be called affected as was said,
or there should be said to be two kinds of alter-
15 ation [A] <<
\6 xal[B] TT]v h:\ xa\ thv 'p(xnv.
: l)
Dlscussion began with introducing his ,Janer analysing the
rreatment of transitions from potential to actual in De Anima I. 5
4l7a 30-b 16. (His nf L:; attached. as are also
Owen's nntes on e 8 which He had hefort:! US,)
After much discussion of the Anima passage we arrived at the
following views. The contrast between (l) and (2) is not the same
as that between (a) and (b). That (mtrast hetween {a) and (b) is
the fam1liar contrast between frvm first potentiality to
first actuality (a) and from first .1c.tuality to second actuality (b).
The contrast betwen (1) and (2) then makes the further point, which
ts the central concern of the rest of the passage, that in the context
l1f sensation and thought we have to understand 'being affected' not
in the sense of being destroyed by the contrary (1) but rather ln terms
-1f preset"vation (2); both (a) and (b) falt under (2) rather than (1).
That (2) rtoes t"elate to (b) as well as to (a) is shown by the fact
that at 5-7 it is the from first actuality to second actua-
l:.ty that is used to lllustrate (2); note that hnth (2), at S-7, and
(B), at 14-16, ctre descrLbed as not in the strict sense
of the force of c.nange to (cf. 'into
l t self' in 6, 3.nd also 16). Compare Aristotle's remarks about the
acquisition of knowledge at VII. 247b 14. It is true that at
'1llb 2-1 the contrast is between two H<ly<; of 'being affected', only
IH1e of .;hich is alteration in the strict while at 14 it is not
'being affected' in the 3ense :ither; but this is not a real
riifficulty. (II) picks up (a), and (I) presumably (b).
',.J!_l_y is ( L) not an a.pt descriptio:1 of (a)? Arter all, in (a)
itself le.Jrning is 1iescribed as c.h2nge frum :-t cuntrary state; and at
465a 22 learning Ls ,i_escri.bed as the of ignorance
knowledge. lt is true that {a) ls qualified tv -n:oA."\6.XLt;. But
t:he central roint is that ln en.:: case ul learning actual i.s
- as for exampole, the colour Ls in the case
,f a change from green to red. The from water tf) wine ls not
t,:.:1tT'JCt1.ve in the same way as :-.hat trom '.nne t_,l vinegar (H 5 1044b
29-35); Out learning ts an even more positi'Je chans.!;e than that, since
realises his nature most fully when he has understanding.
The cant rast
with 'preservation' in (2) shows that it is 'destruction' that is the
important word in (l).
The point is not so much that ( 1) is incompa-
tible with (a) as it stands; rather ( 1) warns against a particular
misinterpretation of (a).
It is not the potential that is destroyed in (1), but it is the
potential that is preserved in (2).
Is it preserved by being actua-
Or is Aristotle claiming that the general potential for lear-
ning is enhanced by each particular act of learning?
For the potential
for learning a particular piece of knowledge is not preserved by its
actualisation; on the contrary, once I have learned this theorem I
no longer have the potential for learning it.
On the other hand,
it seems rather odd to refer to the actualisation of a (particular)
potential as destroying it.
The contrast between and in e 6 1048b 18 ff.
is not directly relevant to the present passage, in spite of De Anima
3. 43la 5-7.
Housebuilding and contemplating are linked in 417b
5-7 i the fact that it is not an alteration of the house builder (9)
does not mean that the building of a house is not a xCvT)CTLf.: which cannot
simultaneously be happening and have happened.
For what has knowledge
to become 6e:wpo;sv (417b 6) does not involve an alteration, because
it becomes, precisely, knowing through the fact that it contemplates
There is, it was suggested, a parallel between Aristotle's concern
in this De Anima passage and in the beginning of 9 8, in that here
he is concerned to show that 'alteration' in a sense also covers cases
where a thing's own nature is fulfilled, and there he is concerned
to shown that 'potentiality' also covers what changes itself ( 1049b
l049b 4-12 The initial reference is to 6 11 (cf. Owen's notes). The
txEt clause in 4 serves to indicate that a previous discussion is being
presupposed. Is there a problem over the contrast between (i) paten-
tiality for changing or being changed and (ii) potentiality for being
X? 1049b 4-12 in isolation suggests we are dealing with (i), but
examples of (ii) appear at l049b 19 ff.
On the other hand, ylvEcn.'
is a type of xCv'T)Cf"l.t;; and is there any difference between the capacity
of becoming a man a'1d thP.: capacity of being a man? It is true that
a house doesn't have the capacity r:o become 1 hn11se; but that difference
not seem relevant here.
l049b 12-17 The priority of the actual to the potential in
The &\ivC!lJ.L c; of
7tp<i.l'tc.u<;; & uv11:t6v
10491) 5 is
in 13-14?
here replaced by -rb C uva:t6v. What is TO
The p(Jwer to cause change 1 l04Sb 35;
Ross); the power to change something else (1(J49b 6-7), contrasted with
other types of potentiality in b 16; or, equivalent to -rb xup(wt; &uva-
't6v for example visible things qua invisible ( 1 5), not qua dogs or
Is the point that the A6yot; 0f 'visible', for exa1aple, presupposes
:\,,jyoc; Q!_ 'seeing', nr just that the A.6yor; of 'visible' includes
a reference to 'seeing', includes the term 'seeing'? The former;
cf. tov l\Oyov tn b 16.
1 ion or more loosely?
Jhould t...Oyo<; then be translated by 'defini-
And does in 17 indicate temporal
priority? The order in which we come to Know things may not he the
same as that of the logical dependence ot one thing on another in fact;
hut here X.6yo<; and yviilO"'-<: are linked - 6 tl the former is a sub-
cl"'..ss 0f the latter, "'tnd in Z 1 they are apparently conflated (cf.
Owen's notes). But r.eed there
quiring knowledge of seeing and
:an the-y not l:e 9imultaneous?
be any temporal priority between ac-
acq,J::._ring of visibility
( lf)!.9b 17 ff. of course argues that
the actual i.s prior to the potential in tiMe. not that the knowledge
of the actual is prior to the J<nowledge of the potential, which would
be the point here.) If dc:es not re.mporal prio-
rity, 16-17 is not a new point but the repetjtion nf what has preceded,
'Nith the addition of yvWcn.t;; the structure of the section is p (12-
13) because q (13-16), therefore p (16-17), an ARA structure, thrmgh
t-1-)i.s fs tJbscured by .Jaeger's and pareucheses.
The next section. trom b 1/, draws a the specific
1nd the individual; to which does 12-17 'loply? To ..;pecific
ties, certainly; and to tndividuals, in 90 far 1s there -1re of
indivi!]11-'lls as such at all.
L\nc'R H
l049b 17-lOSOa 3
The priority of the actual to the potential in time.
Again, it is the potential and the actual that
are discussed, rather
than potentiality and actuality.
Aristotle is concerned by the fact
that the priority of actuality in substance would seem to imply priority
in time too, but this does not seem to be the case for the individual.
Two distinct arguments, at least, are advanced, at l049b 18-29 and
1049b 29-lOSOa (cf. lOSOa 2-3), though the latter may in turn be
subdivided (see below); only the first argument, at 1049b 18-29, really
fits I049b 11-12.
I049b 18-29 19-23 illustrate d.pL8fl(j>
in 19; things the same
in species (l8) are not illustrated until 23 ff.
the eye of the embryo; the first actuality is already present
at birth (and for some time before).
6 6pWv in 20 is the second actua-
lity, the actual seeing; but also the eye itself when seeing - the
thing and the activity should not be distinguished too sharply.
need for a cause the same in species is stated twice, at 24-7 and 27-
9. Does it rule out the evolution of species?
Not if 'goat', e.g.,
1s a vague term, so that the problem will essentially be that of the
:writes; from 'the parent of any goat is a goat' it need not follow
that 'every ancestor of this goat was a goat'.
In Z 9 l034b 16-19
it is a peculiarity of substance that it has to be brought about by
another substance existing in actuality beforehand; cf. the reference
to in 28 here. Whiteness, on the other hand, can be produced
without the pre-existence of something else white as its cause. Health
in Z 7 1032b 23 ff. can be produced spontaneously because of heat which
is, or produces, a part of it.
Health is not a substance; what about
the spontaneous generation of grubs?
Z 7 unfortunately doesn't say.
i049b 29 - 1050a 1 A new argument, from examples. involving learning,
the ability is only produced by the activityi so in this case
actuality does precede potentiality in time, and not only 'in a way
but in a way not' (cf. 1049b 11-12). We might object Aristotle
does not - that in some sense the ability to housebuild must precede
the first act of housebuilding, though.)
in b 29 picks up not
the previous argument, but the general point that the actuality is
'!)49h 17
'lOTES 1JN 1 Ht::TA
pr 1.or in time; we now have a new po.tnt illustrating this, though the
Or ls it that 27-9 genera- qualification of 11-12 has been dropped.
lises the claim of 24-7, and 29 ff. does plc.k this up, the being
that the previous acts of .1.cc the in species but
.. ;umet i .al 1_y? This however doeq rnt fit the reference to prodt!Ction
from tn ?.7--9, or the analysis of coming-to-be 1..11 Z 7-9.
&o-.:et in 29; general princi!Jle of priority of the actual only
1lleged reason fur what follows, real 12-Xplanation in 31 ff.;
With of JS-6 compare
._,;__ 6 236b 32. Clearly not 'anyone who is building a house
must already have a house'; activity of housebuilding, rather
than of building (complete) houses? Or, rather, 35 ff. a}:lplles point
of VI. 6 not to housebuilding, but to : nlng to build. 29-
\4 .;J.ys you t. be a builder (capable of building) without having
built sumethlng; 35 ff. gtlpports this by applying theory of Physics
vr. 6 not to building, or to being a btlilder, but explicitly to learning
to build.
1050a l2-t4 l-lX'tciY'tE:' do not '3EWfH::tv except in some reduced sense:
the seems to make against the immediately pret:..eding l049b 29-50a 2,
J[Z. B (iii). May WOC mean that at any of acquiring knowledge
ot X pari:. of X is a1 ready atr ainP.d? The woLd does not typically
rcfeL back in Ar.; and this wrJuld cmoarrass ti1e IJresent r:laim, which
is that actuality is the intended of Lite process. The
Pliority argued for in B (ii) and B (iii) ts not in play here.
Suggestions: (a) on one reading on 4'JO 29-50a J tl1E:- mai.J. th;ust
nf the argument was to the effect that what precedr:s J.ctual possession
r)'[ the is a bit of the corresponding activity, not a bit of
l049b 35-SOa 3, which ..loes requiTe bit of the 66vrq.u.c
he 1lready present at any stage uf the process of uc..;.uiLLng the
was an aberrat1..or elit..itB:.t l.r; the .:Hr:histj_c ar'il;ument of 33-
l) Even so, even with 1049b 35-50a 2 set aside, \Ve face a
"icatlon Lo the earlier claim that r,e must be f!cquired. uv
ngag1ng in the activity. ':?n now it <m1; ,L6C th,__,. 1 ::arner
if this in turn implies he d .1;_r ,, the "p"ul,,-
hut t.hat tuo nnly in a reduv'ri '.-.'ay, then '.../f! (an "ffer Ar t!18 following
rccunciling gloss on the .vhule ,Jf tr}49b 29-50a 14: you have to have
icerns from geometry Hr;der control tu do ,;xercise on Pytndgoras'
'i'}Jeoremi :lour exercise 1s theretore tiE:wpe:t:v, Uut not as the master
t.;ould do it, and likewise
having svme items control' is havlnl2;
n tT;.; lJtLJ'tf1U11S but not in the "'a1 the l1.1.s the whole of it.
of the whole remains a condition for Bwptv
1. n the t ull sense. Consequently, and thi.s is the main point Ar has
tn the 8wp'Cv of ls not the counter-example
1 t might 3eem to be to the thes1s that 'Jt BEo;poOO'lY Lva Bt:wprruxflv
:E-xux::nv but the uther way round.
This :;nggestion relies ,-,n t:1e thought that the difference heu.;een
:1aving some and having the whole of a :>ubject is not just a difference
Ln how much is l.::n0vm but invr)lves 3 lii [terence in the manner of knowing,
:'he thought ls Aristnt_elian, hut it ls not explicitly announced
1,..,1"':;, wirh ':he tesult lh-3.t ?9-';0a 14 remains in a state of tension
the c-latr:l t.hat in the nf l1:;..;.rning actuality is tempo-
rellly prior (1CJ!+9b 2Q-50a 1) and t::e clalrn that it 1s temporally pos-
because prior in o6rrCa (1050a 4 .trld ff.).
Ross says that Alex.' s lemma
1us but not his but t.his is misleading: the quota-
r- i_r,n i.s embedded in the comment. '.1r they don't need to',
viz. because this :1'X..E-rdv is the mere mechanical practice of the expert?
the nearest tn this use lS lhe of speeches at Phdr.
228h. But this is irrelevant to the claim ..;hic.h 1::> being met, that
t::X.e:-rtOv-ce:c; theori.ze to get knowledge? Rc.>s \.laS sympathetic to Diels'
r_rnposal to treat thjs clattse ac; a hut a (JH Hhat?
none tl1at st:nnd 11p. (1) did not think that
'.lld :r.ean 't.h;:o,r do not c1ave to theorize',
-JS required for R0ss' secrmd and third nor that it
.,uuld lwlp if it could, the 1s 'lupposed to be one where
,_he'.' do 1 neorize, ,wd '>O jo Sxwcnv, but not in a way
; hat .,;ont radicts rhe l_hesis ,Jf lJ-l L \ ii) if, on the other
lrHi, :Jo (reallv) l,heorl.?.e LI-'C,tuse th0y dn not tr1, they
1 J ii) The '";ilme ohjection
l ft
'0 ';Oa 14 'Jnn:s ON THr.TA
applies if there are particular [upi.cs CJ about which they
do not need to theorize, and (iv) if there are 3Ubjects or parts
r)f subjects such that they can be acquired without the student needing
to practice in the activity. (v) Apelt suggested &:u'i\ wet, o6
6Uva.v'taL Aewpetv (sc. d7t:\Wc;). We also thought, tvithout enthusiasm,
of (vi) &:x.A'T; W&e Yl 0'tL. ob Bewpet'v (sc. at all) - intro-
riucing group of dunce ,ue).e'TWv,;ec; who oU Ber..opv\Xn.v 8EwprrnxTw
Exoxnv because they are not even up to doing the exercises. There
(vii) Ross' first suggestion: WOe, n obOEv
Oewpet'v, 'They are not theorizing, except in a quaified sense of that
word' (theorizing in the proper sense being for the sake nf theorizing
or for the sake of the truth), 'otherwise en -""' el 6 j...rr;, i.e. if they
are theorizing in the proper sense) they have no need to theorize'
(sc. t'va. Exux:nv, because in that case they must already
have it). For some reasons this interpretation was left out .in our
discussion.* And since we could not believe that 11 B'tL x'tA.. could
be even an unintelligent gloss on r, woe, or on anything else, and we
could not find a spot from which the words could have been displaced,
we remained in a state of dissatisfied &.1top a..
1050a 15 n .. eo, O:v: not 'may come' (Ross, Reale), nor 'might come'
(Apostle). For the optative as 'would normally except in odd cases',
cf. the treatment of tuche and at 196b 22, 197a 35,
l97b 32, l98a 6: 'what would usually be purposed', e.g. going to a
place and collecting, At 197b 26 the phrase is equated with TO
There was some dispute as to whether the parallels were
1arallcl, but the use of (implying no need of an external agent)
.1nd of -r6 confirms the reading here: the case is n0t that of e.g.
a table which could be painted green, or or any other colour,
b11t rather that of the apple will tn che Hormal course of events
turn red. There is a definite form which the U\11 is programmed to
acq11ire, so that one can say that a pntential F is ..;hat it is in a
,-ense much stronger than that in which one says Jf the table that it
'-,--1.n (might) be green. The sense is ..;hat is required for
the argument to show that actuality is prior obrrCa., not just A6Ytp.
*save for the credentials of
1050a 16-19 Ihe argument is extended to cases where the potentiality
is a potentiality for a certain x(vry:nc; instead of for having a certain
form. '(a.e in 17 should be explicative. The cases envisaged include
both that come from teaching (17-19) and the natural tendencies
for motion (19; cf. lOSOb 29-30) - both of which are needing
no external agent.
1050a 19-21 Inside/outside: is the contrast between the inside po-
tential and the oputside activity, or (Ross) between the absorbed know-
ledge inside and the still unattained knowledge without? The teacher
wants to show the pupils' knowledge in overt practice and thereby demon-
strate the telos: surely outside.
Objection: on the Ross reading el y&p straight-
forward: if the pupils cannot be seen in action, it is unclear whether
the knowledge you are teaching them has been successfully internalized.
But what do the words mean on the alternative understanding of the
inside/outside comparison? Perhaps this: if you do not distinguish
between inside and outside, 6Uva.j.JLc; and l.vlpyeLq. when giving the
of teaching, it will be unclear in which sense of l?tLO""r-fu.i.TJ t7tL<T"t"'fliJTl is
But the text sounds more concerned with a question of
fact than a question of meaning.
lOSOa 21 Epyov, covering both characteristic activity and, where appro-
priate (34), the end product. No attempt to connect the uses: cf.
'work' and 'job'. See, for a fuller version of EE 1219a 13-
18, which also uses x.piicnc;:; (popular in this use: cf. here 30,
lOSOa 28 Why f'il.l\l\ov? Rather than the activity, in such cases: ex-
plained by the following ydp: the lvpyELa. must be found in the pro-
duct, the house, so that the oxo06j.Jry:TLc; is (only) more the than
the is. But then how is the housebuilding simultaneous with
the house (29)? Ross says, 'that in which it (the most pro-
perly and directly resides is that which exactly answers to it, which
c,Jmes into being with it and exists simultaneously with it. This
is the house rather than the builder'. This does not face
10\lla 16
\.<ihen ,,,e have the house we 10 have tr,e r.uHk'-
A clue may be the present participles olxoOoiJ.OUfJlVq.J 29,
32, cf. 31, 33, 34: the building coincides with the house-that-is-
.. But what house is this? The troubles of Z 7-9 are
upon us, and seem to be ignored.
the subjects of the changes/ activities, e.g. the
seeing man; not therefore taking up the Oawv in 34 answering to that
in 30, 'in those actualizings of a 6Uva1-1"c;'.
105Gb 6 <iE ( 'eternal' according to Alexander, Ross, Reale, Apostle,
but better 'in each case'. See l049b 26 for the exact parallel.
This removes the supposedreference to the Prime Mover, whose activity
should in any case not be temporally prior.
lOSOb 2-4 We could not find a satisfactory answer to the question
why, in advance of the xvp!.w'tlpwc; argument of 6 ff., it is already
clear that actuality ts prior Tfl o6o-C9- to potentiality. No doubt,
as Ross says, what makes it clear is the whole section a 4-b 2. The
trouble is that the emphasis of the section is on ways in which form
as 'tl:\.oc; is prior in explanation, not prior ofx:rCq. in the sense at
; ssue 6 ff. y viz. that X must exist if Y is to exist but not vice-
versa (SOb i9, Ross ad loc.)
L050h 8-l.Q Is this out of style with Ar' s usual thesis that only ra-
tional are 'ti'Ov lva.v'tCwv, i.e. does it retract the distinction
between 2-track and 1-track &uvd.j.let.<:? An argument for thinking that
i.t r.ne.s not would be that refers to contradictories, not
1.nntrartes, and the point is simply that a possibility is by definition
;nm8thing that can be (realized) and also not be (realized), which
J.pplies to any potentiality, rational or irrational (cf. esp. ll-i2).
:hwqever, 30-4 both insists on the 1 track/2-track distinction and
:;haws how to reconcile it with the claim that every &6vnf..1Lt; is "ti')t; civ'tt.-
prio-e::wc;: irrational 6uv<4tet.c; produce the opposite of their proper effect
':JJ absence. Parallels to this point at 195a 11-14, Het .
. dl }b li 16 with the presence or absence of the pilot in place of the
;1resence of absence of the here. Cf. Ross ad l05la 3.
Elsewhere he
is alive to the a.nti-Eleatic advantages llf not 3llm.;oing locomotion
to count 1s but here he to admlt some not being for dC6La;
Lhe variant classification '3hnuld be seen dS tailored to present pur-
p0ses. l'he argument is that, in a universe of changeable things,
there must be some unchangeable ones, the universe would stop
( 23): the Aristotelian is the true unswer to the fears of
earlier physicists (22-4). But '.vhy? 1.19 is assertion, not argument.
:nthin the Aristotelian system the sun is needed for the genesis of
finite beings. but we are asking '..Jhy the system must be such as to
include some eternal things. One might argue: suppose there was
noLhing eternal, why should anything have come into being or why should
Lt not all stop? Rut equally, why shouldn't the eternal thing be
) 1tst the sequence of non-eternal ones? One could ngree with Ar that
event must have a sequel and yet deny that this entails the exis-
r0nce of any 8ternal, continuously thing.
l ()SOb in the properly Pldtonic sense that they ure and
are not 5r.p8a.p'ta. (immortal by ':onstantly changing into one another -
'if 337a 1-l), not .'iS an explanation of t:heir heha'liour, 'Nhich stems
fr11m their own nature (30).
lOSOb 34-Sla 2
applies the of tlte preceding argument to
The Piatonists do not 1aake the actuality-potentiality
distinction, so the arguments which are s1pposed to give them something
which is e.g. in the best possible way do not insist that
;. hE. txLrr'tlif..l0\.1 be exercising knowledge. The forms are thus no more
'han t'JVd.fJ.E!.c;, 1Vith the result that there ou)<',ht robe something more
than uU'tO '/iz. the activity.
i.J()h 15
by G.E.L. Owen
lOSla 5-10: dunamis ton enantiOn dgain. Tn l050b 30-34 Ar uses anti-
phasis not enantion to bring irrational dunameis under the generali-
zation: an irrational is responsible for the absence of some
effect by its own absence. At l95a 11-14 and 1'1"! I013b II-
16 he uses enantia for such cases, and in the second passage (repeating
the example of the steersman) adds that the presence and absence are
equally aitia h6s kinounta, which would need to be reconciled with
Ross's explanation that our present passage uses looser language because
in l046b 5 and l048a 8 'he was thinking of dunameis kat a ten kinesin,
positive powers or forces, while here he is thinking of mere potential!-
ties'. Certainly Ar' s examples here take no special notice of cases
,)f responsibility by absence: all are coveredby the poiein-paschein
of 1U48a 7, and note the positive oikodomein/katabalein of l05la 9-
10. Nor is he preoccupied with rational dunameis: eremein/kineisthai
are too wide, oikodomeisthai too indirect. In short our passage seems
to have no direct connection with the previous texts in 9 2, 5 and
8. It might be argued that it takes up the general argument of l050b
8-34, that what can be guaranteed to be always so and never not-s0
cannot be assigned the mere dunamis of
dunameis this depended on the argument
carried the weight. Not in our passage.
being so; but for irrational
from absence, and antiphasis
what is it that can either be built or be demolished (fall
To be sure, a house: but, on the first option (but not the
3econd), what house? Z 7-9 is 0n us ,1gain. 201a 29-34 taken
>'ith e.g. 188a 15-16 might suggest that what can be built up and knocked
down is t-he plinthoi, but one does not oikodomein plinthous, nor I
think kataballein. Are we to say that the built house, tode toionde,
0nce could he built though now it cannot? No: the past without a
!:>rev1ous present is bizarre, and in Z 8 l033b 22-26 \vhat is being and
st1ll can be built is not a tode toionde but a toionde. It is what
i pulled dmvn rhat is the t:o_inde.
not even congeners.
Building and pulling down
lOJla Ll
l05la 17-19: see notes of last session, dnd e,g, II
8, esp. 1 Q9a 33-b 7.
misplaced from e.g. l050b 2-3? or (as Ross
from l050a 3? or a fresh argument for l05la 4-5? If the last, \.,!hat
is the good dunamis involved? fhe capacity of being 'divided' in
mathematical constructions (21-24)? But mathematical objects are
not susceptible of change (DMA 698a 25-26, 19Jb 34, Met. 989b
32-33, etc.), though this is suggested by 23, 29-30, The potentiality
must rather be the inquirer's capacity for noesis (30), with the deri-
vative (kata sumbebekos) capacity in the objects for being discovered:
'.lith dunamei ant a in 29 supply no eta. The argument still seems to
be one for priority in ousia, not in value. On the question whether
being or coming to be understood requires a dunamis for poiein or
in the intellectual object recall (of course) Soph. 248a-49d.
the first geometrical example is strajghtforward, though
the moral is less clear. The angles about one point (24-25) are evi-
dently those on one side of a straight line, and this implies the ex-
tension of the triangle's base BCE:
the other addition is the line parallel
to the side (26), viz. CD oarallel

c E
i.s immediately clear to one
to AB. Then is 'seen' to be
equal to LncE and QAc to hence,
like their counterparts, LABC and bAG
who sees
with 'ACB make up two right-
- What is the 9ia ti which
the construction (26)? (a) A
proof employing a theorem about the angles formed by a straight line
:vhich intersects two parallel straight lines? Or (b) something more
intuitive, as perhaps with Thales' device for calculating the distance
of a ship at sea (pace Eudemus. who ascribed to T.
'must have used' to prove the construction valid)?
theorem that he
Alex. seems to
have it both ways (deixomen, 596. 15, 'It's a house' on seeing a house,
j96. 20); recent edd. opt for the first.
l.ne second 5eornctricat example is debatable. Bvnitz,
Heath, a11d Reale complicate needlessly by supposing that
:-,;t pr,,ves che theorem first for an isoc<2lt:!s triangle in the semi-
circle (viz. one whose apex is vertically ahove the centre of the circle)
rtnd then generalizes this to other triangles in the semi-circle by invo-
king a theorem (e.g. Eucl. iii. 21) to the effect that all,angles having
diameter as base in the semi-circle are equal (or. more generally
as in EucLid, all angles in the same segment). Aristotle makes no
mention of this further theorem; the ekeino in a 29 is surely the theo-
rem just mentioned, that the angles of a triangle are equal to two
right angles; and its introduction spoils the picture of simple recogni-
t ton at a 28-29. in a 28 does not mean 'perpendicular', as
Ross Reale and Apostle translate it (and as Heath supplies in brackets);
used of a line (not an angle) it means 'straight', and the whole phrase
,straight line erected from the centre' applies to any radius within
the semi-circle. Then:
l.et ACD be any triangle in a semi-circle
diameter as base and apex on the
circumference, and let BD be erected from
the centre of the circle to the apex.
BA, BC and BD are all radii and hence
equal. So the triangles ABD, CBD are
both isoceles. Hence ii:J>s is equal to
LoAB and LCDB is equal to j,pcB. Hence
in the trian$1;le ACD the total angle ADC
is equal to the sum of the remaining an-
p,les DAC, DCA. But the angles of a tri-
1ngle are equal to two right angles; hence
L .\DC is equal to a half of this, viz.
1ne right angle.
(iii 3l) extends AD to ADE and 3.rgues (from i 32)that the
_:.::t-=rna l .lm"';le CDE is equal to the sum 'lf two c-.pposite and internaJ
DAC. DCA. But. given the internal isoceles triangles DAB,
;1:B. the angle ADC is also equal to the sum of those angles. So LcDE
L to lAne j and since these lie on a straight line they amount
'l L', ) ungles. So ADC is une angle.
I.HAPTER 9 ln5ta 27
Biancano* also uses (as I construe him below) the same theorem about
external angles but in conformity with Aristotle's suggestion keeps
the construction internal to the semi-circle.
to our diagram.)
(1 adapt the lettering
In figura tres lineae sunt aequales,
duae nimirum in quas basis AC
dividitur, quae sunt CB, BA, et
tertia, quae ex medic basis eri-
gitur. estque BD, cum omnes sint
semidia metri eiusdem circuli.
Educta itaque linea BD de potentia
in actum, si cuipiam trium harum
linearum aequalitas innotescat,
continuo ei etiam manifestum erit
angulum ADC in semicirculo esse
rectum, quia statim apparent duo
isoscelia CBD, ABD, qorum anguli
ad bases CD, AD, sunt aequales
invicem; et anguli duo ad B sunt
dupli duorum angulorum ADB, BDC,
ex quibus conflatur totus angulus
ADC. Ergo duo anguli ad B sunt
dupli anguli ADC; sed duo anguli
ad B sunt aequales duobus rectis .
(How does he reach this? I suggest
arguing that the angle CBD external
to the triangle ABD is equal to the
two opposite internal angles and
therefore, given that the triangle
is isosceles, to twice either
of them; similarly with the angle
ABD and the triangle BDC.
he does not spell it out.)
Why do Euclid and (on my reading) Biancano base the proof on a theorem
about the external angle and not more directly on the theorem that
the angles of a triangle equal two right angles, as Ar's ekeino in
1051a 29 would suggest? Because, I take it, Euclid s proof of this
last theorem takes the theorem about the external angles as one step
which is established in the course of the proof, and the economy of
the system then makes him use the prior and more general theorem in
our case without drawing the particular unnecessary corollary about
the angles of a triangle. Ar. by contrast is following a narrower path.
*Loca mathematica Aristotelis (1615)
t05la 15 1') d:pa. The actuality is better because
the potentiality could be realized in either way. It is not that
a musician plays in actuality better than he is able to play paten-
tially. Rather, it is better that he should be playing actually than
t.hat he should just have the potentiality. Hence .. ,w""<lp<t ( Sla 4):
it is something more worth having. _!lli maintains that activity is
better than the corresponding so the possibility of exercising
a for the worse is not crucial.
!OS!a !7-!9 It is not clear "why the bad actuality is llo-cepov q>6a"<
As Ross, p. 268 points aut, all we have settled so
far is that it is less desirable than the latter. Perhaps we should
supply some thoughts about the need to understand the bad as a deviation
from the good. This would connect with the view we were inclined
to take of the claim that there is no evil xa.p& 'tO. 1tpd.YJ.ll.'fa. there
is an &yaabv 1:1t xp4y .. viz. that which things strive for, i.e.
the nature they are set to achieve, but there is not in the same way
something that things run from - they merely fail to strive success-
fully. We notes the absolute 'cosmic' (not kind-relative) use of
'good' in this passage, comparable to the remark in VI that there
are things better than man. Compare also Physics II. 8 esp. l99a

Suggestion (ii) where {i) uses
one set of alternate and one of
corresponding angles, (ii) uses
two sets of alternate angles.
Thus, according to Eudemus
Proculs, in Euc. I, p. 379, 2-15, the Pythagorean proof of the theorem.
(ii) squares with Ar's mentioning only one line to be drawn and with
his not using exterior angles in the second example either, in contrast
to Euc. III 3!. Heath (Euclid's Elements I, p. 320-l) following Hei-
berg, claims that our passage proves that (i) rather than (ii) is the
construction Ar has in mind in his many references to this theorem,
because &.viix"to (25) means the line is drawn up. It was also objected
I ' [ ! ; I{
to (ii) that one might have expected !tapO:, -thv rather than
Both objections could he met hy redrawing (ii) as (iii):
Text of !OS!a 26-7 Ross reads 6n 1:C. tv .. xa66\ou
6,a 1:C; tav Jaeger li<lt 1:C tv -l!.. ux\C'f' 6p61i xa86\ou; 6,6H
tttv... The asyndeton in Ross's version is dwkward; Ross objects to
that it makes the explanation of the theorem dependent on our
understanding of it (28): the theorem does not hold because it is
clear to one who knows i:xEtvo. But we thought this was pedantry:
Ar could well so express himself and the point comes across clearly
enough. We did not think there were any differences of substance
hetween the two texts. On the second geometrical example, see Owen's
The geometrical moral What is the 6 d which is immediately apparent
to one who sees the construction (26)? What is seen is at least some-
thing of the form 'E because 9.', not just the fact that E. This seems
clear from 24-6 and from the way the second example uses the theorem
of the first. And it must be seen from the construction
cp. on vo\3<; of the Eqa"tov in mathematics).
that every step of a Euclid-type proof is seen.
It would not follow
Thus it is further
grounds for rejecting Ross' story about the second example that his
two-stage procedure is too cumbrous for seeing'. There are 'seeings'
.:1nd 'graspings' of connections intermediate in articulateness between
a proof using a theorem and something more intuitive.
It is another question whether Ar thinks that what is seen in these
examples is the 6 "tC in the full APo. sense. The writing is compa-
tible with this, but he is surely not here Euclid's problem
of which thorem to bring in where in the finished organization of the
science. (And if not, neither is he correcting APo., pace Ross, p.
l 51
1)Ja J/
"! 9, 1051a 21 "JJ
\4r>'l1d draw attention to c_he published article
by ]r)rwthan Lear .-\pril 1982) )fl Aristotle's philosophyof mathe-
As against m11ch thdt l agree with, there is one point of
rietail on '..i'hich 9 9 to me an alternative im:.erpretation.
Lear's suggestion 1s that Aristotle's theory requires the existence
in the sensible world of at least one perfect specimen of each elemen-
tary geometrical shape. lOSla 21-33 suggests to me that all Aristotle
in connexion, say, with a sphere is that there should be an imper-
fectly spherical orange, dnd that the geometer should by a mental act
perfect sphericity in the orange. His mental act would
involve attending to the features of the orange relevant to sphericity,
,vhile ignoring the tmperfcction in a manner is described for
-'l different kind of case in De Hemoria 449b 30-450a 7 (q.v.).
l05la 29-33 and its relevance to Aristotle's philosophy of mathematics
The suggestion was advanced that our passage could be the basis
for an interpretation of Ar 's overall IJhilosophy of maths alternative
to that presented by Jonathan Lear in Phil. Rev. 1982. The Lear thesis
ts (I) that mathematics, in Ar's view, requires the existence of at
least one actual perfect sphere and one actual straight line (other
geometrical figures can be found by construction). To this might
be opposed the strong view (a) that there are no actual perfect spheres
}r straight Lines in the physical world for Ar, or the weaker view
(b) that there may be but the truth of mathematics does not depend
1n it. [(a) is disproved, as far as the spheres are concerned, by
ne example of the heavens; the case of straight lines is more per-
- the natural motions of the sublunary elements, perhaps? But
._ne these stra1.ght lines existing in actuality?] The role of our
1'ldssa.Q;e in support of (a) or (b) ,wuld be to suggest that what guaran-
tees the truth of geometrical theorems is not (I) the actual existence
:__n the physical Horld of the primitive geometrical elements but (II)
_heir .JOtential existence, actualizable by the mathematician's v6r,<rloC:.
scometer can take an imperfectly spherical orange and actualize
'eru:::cr. in it hy His mental act would involve
,1f tendin1.1; to r:he teatures .Jt' r:he Pran<J;e relevA-nt to sphericity while
ignoring the imperfection, in a manner which is rlescribed for a diffe-
rent kind of case in De Hem. 449b 30-450a 7. Further support for
(II) (b) might be found at Xet. M 3 1078a 29-31: the ov"a. stated by
geometry can be lv-re\e:xe:Cc; or b)..t.xWc:;.
Objections: (i) the meaning of this last passage could be that
bronze in itself does have the potential of a perfect material
sphere even if the bronze-smith cannot ensure its coming out just right
- a thesis about physical matter, not intelligible matter. (ii)
On (II) (b) it would be true to say that every book is a sphere, albeit
the sphere actualizable in it by v&-rp-c.c:; may be psychologically difficult
to realize. (iii) If Ar agreed that no physical object was an actual
perfect sphere or the like (Met. B 997b 35-8a 6, K 1 1059b 10-12,
commonly cited for this agreement, are statements of the Platonist
view that Ar is worrying over - Lear, p. 175-9), this would be an impor-
tant problem for him, and one would expect to find texts discussing
it and discussing the thinking away of imperfections in spherical oran-
ges. Ar talks often of thinking away the colour, motion, etc., of
the orange, i.e. properties other than its sphericity, but not of thin-
king away dents and imperfections .!. its sphericity.
account is of the former rather than the latter.
The De Mem.
Against (I) one objection is (i) that if the last physical straight
line were to perish, geometry would fall false. It seems, however,
quite Aristotelian to reply, reversing the objection, that since geome-
try is true there will always be straight lines in the world. (ii)
An alternative interpretation was put up of De An. 403a 13, cited by
Lear (p. 180-1) as proof that Ar did think there are physical objects
with perfectly straight edges capable of touching a hronze sphere at
a point. Namely, the claim could be that the straight line if en-
mattered touches the sphere at a point if we perform the abstraction,
but cannot touch the sphere at all if separated from its matter, on
the grounds that it cannot even be separated from the matter in the
first place. This, it was suggested, fitted better with the context
and the point about the soul; but it does require the reference to
ab5traction to be supplied in the first clause.
On the whole, however, we doubted that 9 9 ..Jas the place to decide
these larger issues. Our passage makes no suggestion that all mathe-
1051d 29
matical objects are actualized by v6ncn.c;. The potentiality under
discussion is a potentiality of something which is already a mathe-
matical object, and the point made here about the priority of
is special to constructions. Wherever it is best placed (see above),
the passage is what it purports to he, a footnote to the chapter which
proves that actuality is first.
Note by G.E.L. Owen
Structure After a proem ( lOSla 34-b2) the chapter divides broadly
into three, and it is a question whether the third part attaches more
directly to the first or the second.
(i) 1051b 2-17 (with back-ref. in (ii) at 1051b 34-35): truth invol-
ving combination and separation in the pragmata thought/spoken of.
Truth/falsehood are treated as properties of thought or speech, not
of their objective correlates: cf. E 4 1027b 25-33 and contrast 6
29 1024b 17-1025a 1. Combination/separation are treated as properties
of pragmata not explicitly of their spoken thought correlates: contrast
De Int_. 16a 9-18, De Anima 432a 11-12, Cat. 2a 4-10 (?-Here Ackrill's
tr. abandons his principle that 'the Cat. is not ... explicitly about
ndmes, but about the things that names signify'). The spoken/thought
correlates of combination and separation in pragmata seem to be respec-
tively true positive and true negative predications: ancestry in
253c 1-3, cf. Met. E 4 1027b 20-23. But the pragmata ground the truth/
falsehood, not vice-versa, 1051b 6-9, cf. Cat. 14b 13-22.
have being/not being (lOSlb 11-13), not truth/falsehood.
The first
The conclusion, that concerning things that cannot be otherwise
1n opinion/statement cannot be true at one time but not another (lOSlb
15-17), seems to lead naturally to (iii), maintaining that with things
that are understood to be unchangeable there cannot. be the error of
supposing a description of them true at one time but not at another.
The example seems to echo the type (i) truth of 1051b 18-
21. So apparently ps.-Alex. 601. 16 ff., Ross ('From the treatment
of asuntheta Ar now recurs to ct:rtain flut others have
read (iii) as an extension of (ii): S. Maurus apud Reale Li 94 ('Circa
simplicia cognita ut immobilia ... '), Rz., Reale, most recently Aubenque
(Etudes sur la Met., 86-87, 'leur cas est comparable a celui de 1a
logique . ').
(ii) lOSlb 17-1052a 2, with backward glances at (i) in 1051b 18-24,
34-35: being and truth with respect to incomposites.
.Jhat are these?
Specifically, do they include the definable essences whose unity was
an interest in Z and H? (Cf. on unity "' indivisibility Met. 1052b
15-16, 1041a 18-19.)
(a) H 3 1043b 28-32 seems to contrast the definable which is composite
with the indefinable incomposites which make it up; and at Z 1039a
14-23 Ar argues that if ousia is asunetheton it is indefinable - or
definable only in some way still to be explained. (l..Jhere?) Similarly
in 105lb 25-28 Ar seems to distinguish the ti esti from incomposite
substances, while still claiming something in common for them (see
((b) But at 105lb 30 the hoper einai ti is given as the subject of
the whole preceding argument about incomposites. In view of this,
what is claimed to be common to the ti estin and incomposite substances
at 1051b 25-30? That they cannot come to be and cease to be is not
enough, for this is true of the composites at 1051b 15-17. Concerning
incomposite substances, error is simply not possible ( 1051b 26-28);
concerning the ti estin, it is possible only kata sumbebekos (25-26).
This may suggest (using the familiar contrast from B 10) that in
producing a faulty definition one cannot be speaking of the intended
object but at best (erroneously) of a name. (Met. 1026b 13-14?) Still
there is a distinction. 'Incomposite substances' can hardly be those
identified by a definition by contrast with those introduced by acciden-
tal description, for do not the first fall under the suntheta of 1051b
(iii) 1052a 4-11, introducing a pate from ( ii) but otherwise apparently
developing (i) 1051b 9-16, unless it is assumed that (ii) has reduced
eternal verities to expressions of what is asuntheton and left synthesis
to accidental time-bound truths.
On simplicity-plus-definability cf. VI 4 141b 3-9 on priority
of points to lines and letters to syllables, the former not to be de-
fined via the latter but definable (Bz. 701b 25-29).
[ I \ ) L ..l J
( l05la
Why are rruth and apparently described here as )(UpLW'tcl.'ta. Ov
(dnd, presumably, r-1h 0v Contrast Z 1028a 13, E
It l.021b 31. Ross proposed deleting xupLtirta:ta. or transferring it to
a 34; he also mentioned the possibility of taking it wi!:h &:>..T)EI<; rather
than with Ov. t)ne objection to t-his is that no non-xupCwc; sense of
truth has been mentioned; however, lt was suggested that the contrast
could be with the application of truth to incomposites in (II) below
- rhougn it is not expressly stated there that that is an inferior
appl1cation of truth, and as will be seen there m1.ght he problems if
Lt were. It is indeed stated in ([I) that being deceived about the
-<: C L v can only occur ( 1051 b 26); but that needn
t imply
that truth too is only in such cases.
lf is taken with d.XT)Ob; the structure of the sentence
,.iLl lJe -rO Ov \l:':ye:'t/'l.C. (3) tO xupLUrta:ta. Ov d:XT)6c;, with the second
:)v being predicative. \Je felt that this was awkward but not impos-
sibJ.e t_and indeed the second Ov will have to he predicative if xupt.Urra.-ra.
Ls Jcleted or trdnsferred, too), but the fact that Ov is a key term
in the S"!ntence makes it very difficult to take xup Ll.lrta:ta. and A-x..,etc;
Jaeger postulates, and fills, a lacuna; it was also sug-
gested rhat xupt.W-ra.-ra. could be a marginal gloss :nisplaced from 34-
composites (104lb 2-17)
fruth and fa tsehood depend on the combination and separation of things
cf. Plato 253c l-2, Netaph. E 4 l027b 20-3. Truth
tnd falsehood are here located in spei'lking dnd thinking, rather than
<I I tte 1.11i ngs Lhought or spoken of 1 as in E 4 l027b 25-33; whereas
1n 1:::. 19 l'!L4b l7-1025a l they carried over to the things themselves.
rrtJth nnd fal.sehood depend on the 1tpdy).1a.'ta, not vice versa; l05lb 6-
cf. l4b 13-22. .;ombination and .'3eparat ion are treated as
ot the (lOSlb 11-13); Lhey are r10t 8Xplicitly trea-
' "d .Ls of the thought, by cnntTast ltJir::h [ l6a 9-
18, J. 8 432a tl-12, Cat. 4 2a 4-10.
J)ih 15-17 form a sort of ap!Jendix; for things that don't change,
't1e ;;unE' (tpiniuns are always either cr'.JE' or false. This pn,vides
a hridge to section II 1 where also there is no change; but there is
also a contrast lOSlb 17) - see further below.
II. Truth and incomposites (1051b 17 1052a 4)
l,.Jhat are these incomposites? Not just things that are always so;
these were described as composites in b 15-17, and N.B. the dia)?,onal
being incommensurable as an example of a composite in b 20-1. fhere
are explicit contrasts in this section with the composites dealt with
in (I): 1051b 18-24, 34-5.
Are the incomposites perfectly simple concepts, the irreducible
elements of definitions, or are they definable essences? fhe unity
of the latter was a theme of ZH (and on the link between unity and
indivisibility cf.
23. However, H
1 1052b 15-16 and Z 17 1041a 18-19); cf. H 6 1045b
1043b28-32 and Z 13 1039a 14-23 argued (dia1ecti-
cally?) that what is definable is composite; the latter st1ggesting
that if substance is incomposite it is indefinable, or definable only
in a way still to be explained (where?), One possible course would
be an appeal to the argument of Z 12, where the entire definition is
contained in the last differentia; but that did not remain Aristotle's
settled view of definition.
lOSlb 25-28 seems to distinguish essences from incomposites; but
30-31 seems to suggest that what has been said relates to whatever
The distinction is made that one can be mistaken
about the 't( per accidens , but about incornposites not at all
(b 25-6).
On one view put forward, the incomposites are a sub-class of things
defined (so that 6t in b 26-7 is restrictive); namely, forms
defined without reference to their matter, so that the riefinition is
dn identity statement (cf. De Anima 3. 6 430b 27 ff., intellect con-
cerned with the -rC l()"t'LV is true and does not say one thing xa'tO. an-
other; Sorabji in Language and Logos, 296-9). On this view a plurality
of elements in the definition does not imply that the object defined
is composite; and there is no need to appeal to the argument of Z 12.
The claim that one cannot be in error about a detinition of an
P.Ssence '..Jill rest on the fact that if one has the wrong definition
:::>ne simply fails to refer to the essence at all. If the -r(
as opposed to the incomposite 1 is the essence, the reference to bei.ng
,\!db 2
l05lb 17 /JTE3 DN tHETA
deceived per accidens (b 26) may suggest that one will refer to the
name, though not the essence (cf. 2. 10, Metaph. E 2 1026b
l3-14)i if however forms defined without reference to their matter
are incomposites, and those defined ..,rith such reference are not, the
point might be that in the case of the latter (only) one can at least
refer to the matter, even if one does not grasp the essence. If incom-
posites are the objects of definitions, it clearly cannot be the case
that the truth discussed in (II) 1-:i inferior to that relating to com-
posites in (I) (see above, on lOSla 34-b 2).
It was objected, against the interpretation of incomposites as
forms defined without reference to their matter, that there were rather
few plausible examples, and none that give the essences of natural
kinds. 'Righteousness' was suggested as an example, but it was pointed
out that that was not a substance. It was also suggested that, if
it was human souls that Aristotle had tn mind here, it was odd that
they were said not to come-to-be or pass away (b 28-9); but this could
be taken as a denial that they undergo of coming-to-be and
passing away, rather than as assertions of immortality. And Aristotle
does discuss forms considered apart from their matter at the end of
Another suggestion was that incomposites are predicate expressions
without their subjects, e.g. 'two-footed animal'. There would then
be no possibility of falsehood, as there could be no false combination,
tnere being no combination of subject and predicate at all. For the
description of a string of words not forming a sentence as a definition,
reference was made to De 5 17a 10. However, there were two objec-
!) How can such an incomposite be true? If the answer is 'because
there is an implicit reference in the context to the thing it is being
.:1dvanced as a definition of', it would seem that by the same token
it could be false as well. There are two possible solutions:
( i) an appeal to the suggestion above, that wrong definitions
are not so much false as simply failures to refer to
definiendum at all.
( ii) the point is not just that 'two-footed' animal is not
true in the context where it Ls a horse that is being
j_ 58
,.!!APTER lfl
defined, but rather that 'seventeen-footed animal'
is not true in any context. It ',.,laS noted, though,
that it is very tempting to analyse this point in terms
of composition by saying that 'seventeen-footed'
does not combine with 'animal', or, perhaps, that
venteen' does not combine with 'footed'.
2) If the basic point is that a simple term, as opposed to a complete
proposition, cannot be false, why is the point made as if it were a
special one about definitions, which obscures it? It was argued that
't( l<T'ti.V need not refer to definition; it could refer to classification
as well, and one MS, A , omits 't( in b 32; but N.B. o6crCa.<; in 27.
6t in 26-7 could indicate a restriction (as suggested above),
a generalising, or the introduction of a second, separate group of
cases; if the last, the reference to 1S1tep e!va.c. 'tL in b 30 is odd,
unless one reads lvfpyec.a.1. with the MSS in 31 and takes xa.i. not as
But is there, in fact, a contrast intended in lOSlb 25-28, so
that one can be deceived about the 't( lcrtc. per accidens but not be
deceived about the incomposites at all 1 How much force should be
attached to bj.i.oCw'? If being deceived per accidens is not really
being deceived at all, but failing to make contact with the subject,
Aristotle might not have felt it necessary to add the point that we
can be deceived per accidens in 26-8 as well as in 25-6.
Two suggestions had been put forward previously as to how one might
be deceived about something per accidens: ( 1) by using the name but
failing to refer to the thing, (2) by referring to the matter but not
to the essence. It was now suggested ( 3) that the point might be
that one could refer by an ace idental description ( c f. An Post. 1.
22 83a 6, 2. 8 93a 21 ff.)i for example, if we were talking about the
human soul yesterday, by saying 'the thing we were talking about yester-
day is a forked radish' - which doesn't completely fail to make contact
with the subject in the way the simple statement 'the soul is a forked
radish' would. It was pointed out that in An. Post. 2. 8 93a 21 a
contrast is drawn between cases where our knowledge of a thing is acci-
dental and those where we have some knowledge of the thing itself,
e.g. if we do know that thunder is a noise in the clouds, though not
,;hat causes it.
iJ5lh 17
'n51 b ! I
An application of ( 2) to a composite being would be our saying
'the oeing whose matter is f1esh and bone of a certain sort (i.e. man)
is a forked radish'. It was that where incomposites are
concerned - if we are now supposing that 26-28 does not rule out error
where these are concerned as well - a part analogous to
that played by the matter in the case of composites might here be played
by the genus. I.e., if we say 'the human soul has the faculties of
J lion's suul
, we are referring to soul, though our statement shows
that we are wrong if we think we are referring to human soul. That
tJne can talk about soul in general, rather than the souls of particular
species, is suggested by De Anima L. l 412a 3 ff.; on the other hand.
1. 1 402b 7 suggests that soul may not be a genus (cf. Alexander, Quaest.
l. lla -b).
If the incomposites are forms, as opposed to forms plus matter
- and 1t was pointed out that LU5Lb 28 supports this, .!...!_ it is taken
refer to coming into being without a process (see p. 158
- what are the wider class of things described as -rC lcrtL?
in categories other than that of substance, it was suggested.
In b 33. should we read tO J.Jc:; or We; -rO !he latter fits the
pattern of being in things corresponding to truth in propositions and
,-ice versa (cf. above on ()).
l05lb 33-l052a (Because the line numbers of different editions may
vary, it seems useful to set out this passage, numbering the clauses:)
10 w, (1) EV
:crnv, d (2) o'd +euoo, (3) to
Ov, o\hwc;: l<Y"tCv, (4) e: 6E. f-'"h oU'twt;, obx Ecrnv.
rhe quest.1on mark at the end of (4) in early editions of Ross is a
"lis print.
1 n part
( l) 1- ( l) c 1 early take up again the discuss ton of O""Oyxe::taBa.t.
of the chapter (lOSlb 2-17): tt1ey do not relate directly
'",J the discussion of a-Uv9e;'ta. and drr6v8e;--ra. .in part II (105lb 7-33i
_,E'low), except in so far as CT\.)"'("lte:t'cr13a.L is something that apply
t'J .u\composites (105lb 18-22).
,eem to he taken up by (3) + (4).
The &..cr-uvee-ra., on the t:..ther hand, do
takes ltv in (1) to be answered by -rO 6E in (J), each
't tn.ese (J"Lt:ki.ng out a different rype of heinJ?; - composite and incom-
R 1 .J I J :51 b jj
;Josite respectively - and each lti .3uhject of its clause and
of the following one. -r.? 6t in (2) ..lill then b.:: aJ.verbial, answering
a second f.-Lkv understood dfter in ( l). are not to understand
an in the introductory words at bJ3-4 ('being (i.s) like truth');
rather, the sense is 'being like truth .. ', this being the subject
subsequently divided between Ev and -rO 6: Or rather (see
below) the sense is 'that being (that is) like cruth'); this. it may
be noted. suits -rb Wt; -rO in b 34. and indeed one might have expected
a similar construction in b 33 as well. It was pointed out that,
with Ross's interpretation, one might have expected -rb l-rEpov rather
than in (3); Ross's parallel from Politics l285b 38-l286a
only has
Alternatively to Ross, it 'vas suggested that Ev in 34 could
be predicative, 'it is a single cruth' (cf. 105lb 12, in the first
part of the chapter). v will not of course then be carried over
to (2) as well as (1). But in this case, what is the subject
'' f (1 ) d nd ( 2 ) '! It could be 33-34, taken as referring or prima-
to composite being, Hith tO in (3) introducing composite being;
hut this is awkward. And, what are we then to tadke of (J) and (4)?
{f ev is predicative in (1), it is presumably so in (3) as well.
And indeed. 'the other sort of thing (--rQ 6!: ), if it is at all),
is one just like that' does make sense, though the reference to unity
doesn't seem particularly to the point here; but (4) '.Vill have to be
taken as 'but if (it is) not like this, it is not which is
not lhe sense required .
The general significance of lOSlb 33-l052a 4 and its place in the struc-
ture of the chapter as a
What do these lines add to what has preceded? They extend to
the force of the observations made in the earlier parts of the
chapter concerning t:_ruth. Being was referred to in 105Lb 23; even
if ,;.Ev in chat line is ilOt answered by 0 in b 13 (dnd we noted that
..Jith one reading in b 23 it is answered at once bY -tO Ot: +e:U6ot;). never-
b 22-23 does lead us to expect a subsequent ceference to being.
'J,1 the other hand, b 33 refers both the ccmposite and to incomposite
(see above), whereas b 23 refers only to Llle truth of incomposites;
on that of composites in (105lb 2-17) and that of the incomposites
05lb .lJ . !JJES 1/N THETA
in II (starting at lOSlb 17 and really ending at b 33). It was poin-
out that this makes it '!ven h<Hder t0 suppose that 1052a 4 ff.
takes up only (II). With regard to b 34-5, it was remarked that there
r:an al;so be truth in separation t1nd falsehood in combination, if one
truly asserts that things which are separated are separated; so b 34-
5 are a compressed statement which can only rf'ally be understood in
the light of what has gone before.
It was that since the object of discussion in 9 10, accor-
ding to the view that ZH9 carry out the progralllme of E, is concerned
with being in the sense of truth and falsity. lt is odd that it should
be felt necessary, after discussing truth in sections I and II, to
refer back to being again at b 33 ff. But it .vas pointed out that
'"hat is under rliscussion is the type of being in things (cf. 1051b
2; and the incomposites are things which are lvepye:CCf, 1051b 28; and
.:f. A 29) that corresponds to truth or falsity in statements; and it
is in place, after truth and falsehood have been discussed, to point
out how this type of being corresponds to them.
III. Deception and the changeless (1052a
This picks up the idea of heing deceived from (II), but otherwise
seems to be a development of the last part of (I) (10Slb 15-17),
(II) in t:he interimhas shown that timeless truths are incomposite.
That (III) picks up (I) is the view of ps. Alexander 601. 16 ff. and
Rossi that it continues (II) the vi.ew of S. Maurus Reale i.i. 94,
of Reale himself, Bonitz, .<Ind Aubenque ( R6-7).
a 7-11; quantification is possible in the case of a class, but not
l)f an i.ndividual.
'lur discussion centred around some nine questions, of varying
importance. treated here in roughly textual order:
( 1) D0es a 4 ff. belong with what immediately precedes ( "t"b lv
at b 35, introducing composite being), or wi.th the general problem
at the opening of the chapter? If the former, if must be
a different reason for absence ,,f deception than that given (so
f3r as one is at b 15-7 (';o.Thich in the composites
;f:'c t ion).
(2) Does &xCvn1'a. refer to the Unmoved Mover etc. or to mathematical
objects? It seems more intelligible to hesitate about the former,
but the ensuing examples are mathematical, and when Ar does express
hesitation about the UM at 1026a 29 he uses obo-Ca. of it. (Could
it refer to 'Platonic' intermediates? But the point of the sen-
tence doesn't seem to require such a metaphysical potulate, on
any interpretation.) Perhaps e:L uc; ... is simply to emphasise
that one won't then be tempted to attribute properties at one
time to what results and not at another.
(3} What is the force of xo"tb. 'tO 1to't'f (indefinite; not 1t6"t. Cf.
oe below)? Does it suggest there be decep-
tion in other respects? (See below). Could Jto-rf mean loosely
'in some cases' rather than strictly
at some times' (cf. &.el. at
II)? Only if we are willing to give a parallel interpretation
(4) Is "tb -rpCywv0\1 universal or particular? A particular triangle
would be the sort of thing that might change its properties, if
anything did, but Ar could mean the triangle as such, or any tri-
angle. The universal would be needed if meant 'vary'
or 'be different in different cases'; we would still get a false
belief as long as necessary properties were taken, but the contrast
with what follows would be spoilt.
( 5) How does "tt j.JkV... relate to Xo"tt ... ? Are we still talking
about triangles, the false belief being that some are of one kind
and some of another (or that the triangle in general appears now
in one and now in another, if 't'\ is predicate)? This would
make good sense, but makes very abrupt the change of subject in
the ensuing examples. More likely '<L f.LEv o'o6 is a vague
and general representation of the form an error might take, the
governing verb still being ol-f)ae-ra.L of course. But why is it
Lmmediately illustrated by an example where one doesn
t wrongly
think some are of one kind and some of another, but wrongly thinks
all are of one kind? We had no very clear answer, though Ar
main point seemed clear enough.
(6) Is -cl subject or predicate?
still be either.
On the vague interpretation it could
( 7) How does T, -cLvb.c: relate to the preceding Again we
had no clear answer. Could Ar mean (with Tr epexegetic) that
thinking no evens are prime is a 'Nay one could go wrong in speci-
fyl.ng one's belief that some numbers are prime and some not?
(8) Does tva. mean the number one, or any single number?
It seems easier to think of the number one as something one cannot
make mistakes about. But on the whole we preferred
any single
One could not there hold beliefs (and a fortiori not
false beliefs) of the form 'some are of one kind and some of an-
(9) Are 'tLvfl masculine singulars or neuter plurals?
seems to hang on this.
8 10 in 9 as a whole
Truth and falsity change where there is potentiality for change
in the i!:p6.y)..J.o.-ca. (l051b 10-15); incomposites, about which we cannot
be cleceived, exist in actuality only and not potentially (l05lb 28,
31). But it was felt that to try and link 9 10 as a whole to the
heme of potentiality and actuality was implausible, involving picking
->ut particular passages and phrases. Rather, it was noted that 9
l refers back to being in substance and the other categories, the theme
of ::H, before proceeding to being in potentiality and actuality; this
having been dealt with by the end of l3 9, 8 10 wi.llthen on to the
further sense of truth and falsity (cf. lOSla 34-b 2). Moreover E
2 l026a 33-l026b 2 lists these three uses (the categories, truth and
falsity, potentiality and actuality), together with the accidental;
: ;1e accidental is discussed in E 2-J, and E 4 then t.ntroduces being
as truth and falsehood, with a reference forward to the discussion
of simples and the 1:'( lO"'tt.Y in 9 10 (E 4 l027b 27; where, it may be
noted, the reference to -r<'i. 'X.o.\ 't'h -cC does not seem to suggest
that the former are a sub-class of the latter). The conclusion was
hard to resist that the l.Jhole of EZH9 form a systematically arranged
discussion of the senses of being - at whatever stage in their develop-
ment the arrangement took place.
General considerations on EZH9
Despite our previous discussions on it we remained not entirely
satisfied about how far EZHA formed a unified whole.
E 2 starts with
a list of senses of 'being', and after saying that 'being
is said
in many ways continues with a discussion of accidental being. E 4
takes up being as truth from the list 1 and after a brief discussion
says we must consider later those questions in this area which are
relevant ( l027b 28-9), but adds that since neither accidental being
nor being as truth is being in the proper sense (xupCwc;;: b 31) we must
dismiss them 1028a 3) and look at the causes and principles
of being qua being.
Ar does go on to do this of course. but we cannot
assume without futher ado that ZH9 forms a single connected whole car-
rying this programme out, let alone that Ar envisaged the whole of
t:H8 in detail when writing the end of E, which presumably no-one would
It may be true that ZH8 form a rough whole, however loose and
scrappy in places, but we still have the difficulties for the unifi-
catlon thesis that we noted earlier (see above, yp. l-3).
l in particular refers back copiously, but not apparently to z as we
noH have it.
It does not refer back on this topic of the matter,
and anyway the unity of Z itself is by no means unproblematic.
a 1-9 doesn It dO mUCh piCking Up from !Jlhat. precedes,
9 10 itself does pick up from E 4, but there are problems.
does it come where it does? There is no obviously better place for
it to come in e, but it does not seem to rely on the rest of 8, so
why didn
t Ar get it out of the way at the start in E, as he did with
being? What connexion has it with the preceding discussion
ot actuality and potentiality? ( l05lb 1) is embarrassing
on any view (cf. .(upCw; at 1027b 31, mentioned above), but perhaps
cc,uld be dealt with as or interpolated. On the other hand
some defence ..,;oas put up for tne unificati0n thesis as regards e 10.
If we assume (but should we?) that the programme sketched at 1028a
3 takes till 9 9 to complete, there would be no room for 9 10 earlier,
and if Ar had something to say about truth and falsity 810 seems as
good a place as any for him to sa7 it. reference to and
6Uvni-J.&.t; at lOSlb 28 might gi\e a positive reason for putting 9 10 after
the rest of 8; 10 does, in part, focus on truth concerning the
as compared with composites, which gi1es 11s a focussing that wouldn't
have arisen without the rest of A, especially 9 8. \.Je even briefly
flirted with the idea that 9 10 was written by someone warming to the
idea of being as truth presupposing being and actuality/potentiality,
and so up as xup\.<irta.'t'a. being - but we did not seriously think
9 10 was not by Aristotle, and ended by reaffirming that we couldn't
say definitely what 8 10 presupposed, and that the unification thesis
had not Oeen freed from all its difficulties.