Sei sulla pagina 1di 306


an i nternati onal j ournal
1 2007
Autorizzazione del Tribunale di Pisa n. 41 del 21/12/2007
Direttore responsabile: Fabrizio Serra
Sono rigorosamente vietati la riproduzione, la traduzione, ladattamento,
anche parziale o per estratti, per qualsiasi uso e con qualsiasi mezzo efettuati,
compresi la copia fotostatica, il microflm, la memorizzazione elettronica, ecc.,
senza la preventiva autorizzazione scritta della
Fabrizio Serra Editore

, Pisa Roma,
un marchio della Accademia editoriale

, Pisa Roma.
Ogni abuso sar perseguito a norma di legge.
Propriet riservata All rights reserved
Copyright 2008 by
Fabrizio Serra Editore

, Pisa Roma,
un marchio della Accademia editoriale

, Pisa Roma
Stampato in Italia Printed in Italy
issn 1973-5030
La Accademia Editoriale

, Pisa Roma, pubblica con il marchio

Fabrizio Serra Editore

, Pisa Roma, sia le proprie riviste precedentemente edite con

il marchio Istituti editoriali e poligrafci internazionali

, Pisa Roma, che i volumi

delle proprie collane precedentemente edite con i marchi Edizioni dellAteneo

, Roma,
Giardini editori e stampatori in Pisa

, Gruppo editoriale internazionale

, Pisa- Roma,
e Istituti editoriali e poligrafci internazionali

, Pisa- Roma.
Si invitano gli autori ad attenersi, nel predisporre i materiali da consegnare alla
Redazione e alla Casa editrice, alle norme specifcate nel volume Fabrizio Serra,
Regole editoriali, tipografche & redazionali, Pisa Roma, Istituti editoriali e
poligrafci internazionali, 2004 (ordini a: iepi
Il capitolo Norme redazionali, estratto dalle Regole, cit., consultabile Online alla
pagina Pubblicare con noi di
Forme di dibattito e di confutazione
degli avversari nel pensiero antico
Luca Castagnoli, Everything is true, everything is false. Self-refutation argu-
ments from Democritus to Augustine 11
Louis-Andr Dorion, Elenchos dialectique et elenchos rhtorique dans la dfense
de Socrate 75
Luc Brisson, Une rfutation contagieuse: Banquet (199c-201c et 201e-203a) 91
Giuseppe Cambiano, Come confutare un libro? Dal Fedro al Teeteto di Platone 99
Walter Cavini, Principia contradictionis. Sui principi aristotelici della contraddi-
zione ( 1-3) 123
Ermelinda Valentina Di Lascio, Solecisms on things. The arguments a e
in Aristotles Sophistical Refutations 171
Jean Levi, De la dispute sophiste au dialogue philosophique 205
Christopher Cullen, Actors, networks and disturbing spectacles in institutional
science: 2nd century Chinese debates on astronomy 237
Johannes Bronkhorst, Modes of debate and refutation of adversaries in classical
and medieval India: a preliminary investigation 269
Discussioni e ricerche
Luciana Repici, Aristotele, lanima e lincorruttibilit: note su De longitudine et
brevitate vitae, 1-3 283
Stefano Bacin, Kant, i flosof antichi e i limiti della loro recezione. A proposito di un
libro recente 307



Luca Castagnoli
Insignem continent Veritatis astum hae
demonstrationes, quo illa hostium suorum armis in
eorum perniciem pro se abutitur; sed imprimis
Dilemmate velut incantamento cogitur Veritatis
hostis de industria, pro Veritate tamquam pro aris
et focis dimicare.
Arnold Geulincx
1. Preliminaries
mong the theses which ancient philosophers charged with self-refutation (more pre-
cisely, with what modern readers have tended to identify with our1 self-refuta-
tion), we can single out a quite homogeneous class including Everything is true,
Everything is false, and other similar theses (Every appearance is true, To say the
false is impossible, Nothing is true, Truth could perish, etc.). The ancient argu-
ments denouncing such theses as incurring self-refutation will be the protagonists of
this article.2
Before starting our inspection some preliminary remarks are needed. It is worth
noticing that those theses have attracted very meagre attention in the modern studies
on self-refutation: as we shall see, only Nothing is true was taken into account by
Mackie in his infuential formal analysis, only to be neglected in the subsequent section
exploring the philosophical dividends of that analysis. The reason for this disinterest is
easy to diagnose: such theses are bound to sound to modern ears too blatantly absurd,
* This article is an abridged version of the frst part of my Cambridge Ph.D. dissertation The Logic of Ancient
Self-Refutation: From Democritus to Augustine (Castagnoli 2005), in which I provide a comprehensive analysis of the
history and logic of ancient self-refutation.
I wish to express my deepest gratitude to Nick Denyer, for the great skill and generosity with which he super-
vised my research, and to my examiners, Myles Burnyeat and David Sedley, for their invaluable comments and
suggestions and for the encouragement manifested on various occasions. In the last few years I have incurred
several other debts of gratitude towards many who, in various ways, contributed to the research of which this
article is an ofshoot. In particular, I wish to thank Walter Cavini, Valentina Di Lascio, Paolo Fait, Geofrey Lloyd,
Alex Long, Tony Long, Mauro Nasti De Vincentis and Robert Wardy. St. Johns College and Magdalene College,
Cambridge, ofered me the best conditions for pursuing my doctoral and post-doctoral research, and I am glad to
express here my gratitude. A fnal acknowledgement goes, again, to Myles Burnyeat: although on occasion I shall
criticise his views and try to improve on them, the extent to which my work remains indebted to his seminal
articles on ancient self-refutation cannot but exceed my numerous acknowledgements ad locum, and will not
escape the notice of any reader who is already remotely familiar with this topic.
1 For a disclaimer on the use of our here cf. p. 14.
2 By ancient philosophers and ancient arguments I refer here exclusively to ancient western philosophers
and their arguments. For analogous arguments in the eastern tradition cf. Graham 1989, pp. 183-186 and Harb-
smeier 1998, pp. 344-345.
and thereby hopelessly uninteresting, to be worthy even of scrupulous refutation. The
problem of the nature of falsehood and the puzzle of how it is possible to think and say
something false, on the contrary, were live issues for long time in antiquity, and among
the priorities on the philosophical agendas of no less thinkers than Plato and Aristotle.1
This is why radical views such as Nothing is false and Everything is true were not
only proposed by some, but also thought to require and deserve refutation by their op-
ponents. Also the contrary position, according to which nothing is true, found its eager
supporters, not only in the now obscure Xeniades,2 but also in the tradition, sceptical
and nihilistic, embodied (at least according to some ancient interpretations) by
renowned fgures such as Gorgias, Monimus, and Anaxarchus.
From this perspective, we should not confuse Everything is false and Everything
is true with the Liar and the Truth-Teller. The latter were arguments (often branded
as , sophisms, or , insoluble arguments), perceived by an-
cient logicians as a serious menace to the foundations of logic, whereas, however sur-
prising this might appear to us, the former were advanced as genuine philosophical
theses, never making their appearance in any ancient list of sophisms or insolubilia.
While the reconstruction of the ancient responses to the threat posed by the Liar and
the Truth-Teller is extremely controversial, and, given the scantiness of our evidence,
deeply conjectural,3 we are sufciently well informed on the ancient reactions to
Everything is false, Everything is true, and analogous theses to provide an analy-
sis of them which aspires to be both accurate and instructive on the nature of ancient
logic. What we do know for certain about the ancient Liar is that it gave ancient lo-
gicians no fewer headaches than its modern heir has aforded to their descendants:
Philetas of Cos is reported to have eventually fallen victim to the sleepless nights de-
voted to it,4 and, if without any such dramatic fnale, Chrysippus himself certainly
spent enormous eforts to solve it, if the catalogue of his writings seems to attest to
no less than twelve works in twenty-three books dedicated to the presentation and de-
fence of his own solution and criticism of others.5 We shall discover that, on the con-
trary, the self-refutation charges against Everything is false are very straightforward
in their logic, and sometimes depicted as almost trivial by their own proponents: if we
judge by the tone of our testimonies, whereas the Liar argument was felt as a real
challenge, the Everything is false thesis was regarded as an embarrassment only for
its nave supporters.6
One might protest that, after all, the latter cannot be less paradoxical than (and
should be treated along the same lines as) the Epimenides (the notorious claim of Epi-
menides the Cretan that All the Cretans say the false),7 which is not plainly false, but
1 For the history of the problem and its solution cf. Denyer 1991. 2 Cf. pp. 51 and 61.
3 Cf. e.g. Rstow 1910, Cavini 1993, Mignucci 1999, Crivelli 2004a. 4 Cf. Athenaeus 9.64, 34-35.
5 Cf. d.l. 7.196-197 and Barnes 1996.
6 For analogous reasons I shall not deal with convertible arguments ( ), i.e. argument
patterns which can be turned round against their proponent in such a way that both sides have equal force (cf. e.g.
the notorious Corax-Tisias dispute). The proponent of a convertible argument can be opposed by an argument
with similar structure but opposite conclusion, but he is not the clear-cut loser in the debate, which has no obvious
solution (convertible arguments were often classifed, like the Liar, among the ). For the diference between
convertible arguments and self-refutation cf. Burnyeat 1976a, pp. 67-69; for ample discussion of convertible
arguments cf. Nuchelmans 1991, pp. 13, 49, 64-75.
7 Cf. Paul. Tit. 1, 12-13: One of them [sc. the Cretans], a prophet of their own [sc. Epimenides], said: Cretans,
always liars (), evil beasts, lazy gluttons. This testimony is true [sic!].
12 luca castagnoli
either false (if some truth has ever been said by a Cretan) or Liar-paradoxical (if all the
other Cretan statements are false, or neither Epimenides himself nor any other Cretan
did ever say anything else).1 I shall suggest that the fact that ancient self-refutation ar-
guments appear totally innocent of this complexity does not betray any logical def-
ciency of their proposers, but comes as invaluable, albeit indirect, evidence that, unlike
the ancient refections on the Liar and many modern self-refutation arguments, they
did not aim at establishing the truth-value of certain propositions, but served a diferent
purpose. What this purpose was, how the ancients tried to achieve it, and the crucial
diference, in aim and structure, between ancient and modern self-refutation will
emerge progressively as we proceed.
2. Mackie on the absolute self-refutation
of Nothing is true
More than four decades after its frst publication in 1964, Mackies formal analysis of the
logic of self-refutation remains the best one on the market, on account both of its un-
deniable merits and of the scarcity of subsequent attempts.2 Since it has also become,
through Burnyeats partial adoption of it, the unchallenged benchmark in most subse-
quent studies on ancient self-refutation, I suggest we start from the end of our story,
and see what Mackie has to teach us on the self-refutation of Nothing is true.
After introducing pragmatic self-refutation, with which we shall not be concerned
here, Mackie analyses a second type, which he labels absolute self-refutation, distin-
guishing two varieties of it, based on two diferent properties of the main operators in-
volved: It is true that (T) has both properties, and thus is involved, in diferent ways,
in both varieties. To begin with, Mackie lists It is true that among the truth-entailing
operators (with I know that and It can be proved that), i.e. those operators ds for
which if dp is true, p itself must be true also (p. 194). On the basis of this law, Mackie
constructs the following argument:
(1) (p)(Tpp)3 T is truth-entailing
(2) T((p)Tp)(p)Tp From (1), by substitution
(3) T((p)Tp)(p)Tp Existential generalisation
(4) T((p)Tp) From (2) and (3), by destructive dilemma ((pq)(pq))p
Mackie clarifes what exactly an argument of this form is meant to prove:
With absolute self-refutation of this sort, an item that would be symbolized by d((p)dp), such
as my knowing that I know nothing [or being true that nothing is true], simply cannot occur.
Here we can say that each proposition of this form is self-refuting. It must be false; given that d is
truth-entailing, its form guarantees its falsehood. (p. 195)
1 Cf. e.g. Koyr 1946, Prior 1958. We do not know whether the Epimenides was treated together with the
Eubulides (Im saying the false) in the ancient analyses of the Liar. Analogously, Everything is true should be
either false or Truth-teller-paradoxical.
2 Cf. e.g. Johnstone 1964, Bonney 1966, Stroud 1968, Boyle 1972, Finnis 1977, Vanderveken 1980, Stack
1983, Champlin 1988, Johnstone 1989, White 1989, Page 1992, Herrnstein Smith 1996, Johansson 2003. Pass-
more 1961, the frst rigorous analysis of self-refutation, inspired Mackie and is still worth studying. Relevant ma-
terial can also be found in the vast literature devoted to Moores paradox and to the pragmatic paradoxes stem-
ming from it (cf. e.g. Moore 1942, Grant 1958, Hintikka 1962, Sorensen 1988, Green-Williams 2007).
3 I translate, here and hereafter, the Polish notation adopted by Mackie into a more easily readable notation.
self-refutation arguments from democritus to augustine 13
Mackies argument looks unimpeachable. Careful inspection, however, reveals that it is
afected by a potentially dangerous ambiguity. It is by no means clear how exactly Mack-
ie wants us to understand the conclusion (4), of which he presents at least two diferent
(4*) It is not the case (and it cannot be the case) that it is true that nothing is true;
(4**) The proposition It is true that nothing is true is (necessarily) false.
The former declares the impossibility of a certain state of afairs obtaining, the latter
the necessary falsehood of a proposition. (4*) and (4**) are, of course, strictly related:
ordinarily we would have no qualms about subscribing to their equivalence (an exem-
plifcation of Tarskis T-schema), but we shall discover shortly that their diference can
turn out to be signifcant in certain cases. Mackies wavering understanding of the main
negation is not the only ambiguity to be detected in his double paraphrase of (4). T is
taken sometimes as the sentential operator It is true that (equivalent, I presume, to It
is the case that) and sometimes as the truth-predicate is true (attached to propositions,
sentences, or whatever one might decide the truth-bearers are):1 It is true that nothing
is true, Mackies own explicit interpretation of T((p)Tp), is an odd hybrid of these
two understandings of T. Consistency would require either
(1) It is the case that nothing is the case
(2) (The proposition) Nothing (i.e. no proposition) is true is true
and, again, (1) and (2), although strictly related, do not express the same thesis (e.g. (2),
unlike (1), is committed to the existence of entities like propositions and truths). One
could protest that this indeterminacy is not, after all, very disturbing, since Mackies
conclusion appears to be sound (and soundly inferred) under all possible interpretations
(provided one interprets the premisses accordingly). It will shortly become clear why
this kind of relaxed attitude is not to be recommended.
But let us grant that the argument sketched above is acceptable as it stands: is it a
proof that T((p)Tp) is necessarily false (or cannot occur) by self-refutation? Mack-
ies argument shows that this formula entails both members of a contradiction: under
one of its possible interpretations, for example, the proposition Nothing is true is
true entails both Nothing is true (by the truth-entailing property of T) and Some-
thing is true (by existential generalisation: if Nothing is true is true, then certainly
there is something true, this very proposition). Is this a sufcient condition for self-refu-
tation? Since Mackie does not ofer us any general defnition of self-refutation, and, sur-
prisingly, there is no such defnition agreed upon in the literature,2 providing a clear-
cut answer is a fuzzy task (Socratically, one could despair that it is indeed an impossible
task). However, I suggest that such cases are best kept distinct from self-refutations, and
best catalogued as self-contradictions;3 Mackie himself comes close to admitting as
1 As we have seen, Mackies ofcial account treats T, like all the other ds, as an operator (p. 193).
2 The self-refutation jargon seems to be used (and sometimes abused) with a myriad of diferent senses and
nuances, and those who adopt it rarely make any efort to explain its exact import. Even the studies devoted to
the logic of self-refutation (cf. p. 13n2) cautiously eschew the task of providing us with defnitions encompassing
the various forms they identify and disentangle.
3 The way I intend the latter notion is itself hard to specify in satisfactory formal terms: I take self-contradic-
tion to include all those cases in which a single proposition, simple or compound, either entails or consists of a pair
14 luca castagnoli
much when he recognises that detecting absolute self-refutations with truth-entailing
operators is not philosophically fecund since we are merely avoiding logical contra-
dictions (p. 203).
It is time now to consider Mackies second and more interesting type of absolute self-
refutation involving the operator (predicate?) T. T is not only truth-entailing, but also a
member of the sub-class of ds which we may call prefxable, that is ones for which if p
itself is true, dp must also be true (p. 195). On T-prefxability Mackie erects the follow-
ing proof:
(1) (p)(pTp) T-prefxability
(2) (p)TpT((p)Tp) From (1), by substitution
(3) T((p)Tp)(p)Tp Existential generalisation
(4) (p)Tp((p)Tp) From (2) and (3), by transitivity and double negation
(5) ((p)Tp) From (4), by the logical law (pp)p1
How should we construe the conclusion (5)? Given the ambiguities I have pointed out
above, Mackie seems to be committed to four diferent interpretations, which he must
consider all sound:
(a) It cannot be the case that nothing is the case;
(b) It cannot be the case that nothing (i.e. no proposition) is true;
(c) The proposition Nothing is the case is necessarily false;
(d) The proposition Nothing (i.e. no proposition) is true is necessarily false.2
At least two of these immediately strike me as dubious. How could Mackie prove any-
thing like (b)? Suppose, someone as sophisticated as the mediaeval philosopher John
Buridan might protest, that God had annihilated all true propositions:3 doubtless no
proposition would be true, therefore that nothing is true, although not possibly-true,
seems to be possible, i.e. something which can be the case.4 Mackies conclusion asks us
to accept that, somehow, such possibility is logically barred: but it is difcult to see why
this should be the case (did Mackie establish, as a remarkable by-product of his argu-
ment, that necessarily either God does not exist or is not omnipotent?). Reading (d) is
no less problematic. Certainly the proposition No proposition is true cannot be true,
of contradictory propositions. This broad category would include instances both of formal self-contradictions, ei-
ther explicit (e.g. pp, It is raining and it is not raining) or implicit (e.g. (pq)pq, If it is day, it is light,
and it is day, and it is not light), and of analytic self-contradictions (e.g. This triangle has four sides). Self-con-
tradictions are also, intuitively, necessary falsehoods (and typically falsifable through reductio ad absurdum), and
are treated as such in most logical systems. One might argue that self-refutation must be a subclass of self-con-
tradiction: on some analyses, a proposition refutes itself when it entails its own contradictory, and since anything
seems to entail itself, any self-refuting p would always entail the contradiction pp. I shall not assess this view at
this stage, but I hope it will become clear later why this classifcation could be problematic.
I suggest that self-refutation should also be carefully distinguished from inconsistency, both semantic and prag-
matic, and, as I have mentioned on pp. 12-13, from the semantic paradoxes (cf. Castagnoli 2005).
1 We shall become very well acquainted with this law of classical logic (a form of the so-called Consequentia
Mirabilis) in what follows (cf. sect. 6. 1).
2 Mackies own unique paraphrase of (5) is There are no truths is absolutely self-refuting and There are
some truths is necessarily true (p. 197), which seems to be equivalent to (d).
3 Adopting modern jargon, for Buridan a propositio is a meaningful sentence token (i.e. a particular utterance
or inscription), spoken or written with assertive intent (Hughes 1982, p. 5). This need not be Mackies own con-
ception of proposition, which unfortunately he fails to clarify, but the use of statement (p. 194) and the claim
that T-prefxability is a condition of discourse (p. 202) might suggest that Mackies propositions are quite con-
crete linguistic items, not unlike Buridans propositiones.
4 Borrowing an important distinction which Prior 1969 extracted from Buridans remarks in Sophismata 8.
self-refutation arguments from democritus to augustine 15
but suppose, again, that no other proposition existed, or that only false propositions ex-
isted (something that Mackies self-refutation argument cannot exclude): No proposi-
tion is true would be paradoxical, for the very same reasons which make the Liar as-
sertion Im saying the false paradoxical, and thus Mackies conclusion that it is
necessarily false, by self-refutation, appears too hasty.1
The conclusion (5) has turned out to be much less palatable and to require much
more cautious refection than we might have thought at frst glance; but how was it
reached? The key step of Mackies argument, (2), is an exemplifcation of T-prefxabil-
ity: on one of its possible interpretations the one eventually leading to the two read-
ings of the conclusion just discussed (2) can be paraphrased as If nothing is true, then
Nothing is true is true. Is this kernel of the self-refutation argument acceptable? As
an instance of T-prefxability, (2) seems to be perfectly sound, on a par with all the oth-
er instances of semantic ascent: if snow is white then Snow is white is true, if 2 + 2
= 5 then 2 + 2 = 5 is true, etc. My contention, however, is that T-prefxability should
not be light-heartedly assumed here. Since T-prefxability takes for granted the existence
of truth, to employ it to disprove that Nothing is true appears question-begging.
Nothing is true is obviously inconsistent with the law of T-prefxability, and it comes
as no surprise that by assuming the latter Mackie can contradict the former. However,
although we lack a proper defnition of self-refutation to which we can make appeal, it
is not idiosyncratic to suggest, minimally, that a self-refutation argument should show
how a certain thesis is refuted (whatever this refutation amounts to) by itself alone or,
at most, with the help of other assumptions which have been granted, or would need
to be granted, by its proponent in virtue of his commitment to that thesis. But no cautious
supporter of Nothing is true would grant T-prefxability, and hence step (2) in Mack-
ies argument: by advancing his extraordinary thesis he must be at the same time im-
plicitly asking us to revise many of our basic assumptions about truth, and T-prefxa-
bility is no doubt among these.
Assessing a revolutionary thesis against an extraneous and hostile conservative set-
ting produces a refutation that is suspiciously too easy. This is not to deny that such a
strategy can be very instructive: the boundary between begging the question by tacit-
ly foisting upon you admissions you would never grant, on the one hand, and changing
the subject by stubbornly refusing to grant me anything whatsoever which is commonly
recognised as a defning feature of the subject (making your position unintelligible), on
the other, is often quite indeterminate. Perhaps, however, this kind of strategy should
not be misleadingly presented as a self-refutation argument (and certainly not as an un-
controversial and paradigmatic one). But what seems especially difcult to swallow is
Mackies idea that with absolute self-refutation, unlike pragmatic and operational self-
refutation, it is the self-refuting propositional content that falsifes itself, all by itself: not
only is a supplementary assumption (T-prefxability) required, but the substantial bur-
den of the refutation is carried by it, and not by the alleged self-refuting proposition.2
Furthermore, we have seen that Mackies tactic delivers, on this occasion, question-
able conclusions. Starting from the next section, we shall begin appreciating the difer-
ence, in structure and purpose, between his absolute self-refutation and various ancient
self-refutation charges against theses like Nothing is true and Everything is false.
1 For discussion of the kind of problems raised by Mackies argument cf. also sect. 7.
2 For perplexities on Mackies accounts of pragmatic and operational self-refutation cf. Castagnoli 2005.
16 luca castagnoli
3. Setting the stage:
Dissoi logoi 4. 6
Democritus might be the frst fgure whom our sources credit with deploying a clear
self-refutation charge against a thesis belonging to the family we are interested in (Pro-
tagoras Every is true). It has been plausibly remarked, however, that our
late source, Sextus Empiricus (m 7.389-390), employs technical jargon and an argumen-
tative structure which bespeak a more sophisticated consciousness of logical form
than we may suppose was to be found several centuries earlier1 and represent a lega-
cy of later (in particular Hellenistic) refections and developments. For this reason I shall
consider Sextus testimony on Democritus anti-Protagorean argument together with
the other Sextan evidence in section 6, cautiously postponing its scrutiny to a more ad-
vanced phase of our inquiry.
With Democritus temporarily sidelined, the earliest argument relevant for us could
be one contained in the fourth chapter of the untitled anonymous treatise usually re-
ferred to as d (Twofold arguments) from its opening words.2 This sophistic-
style collection of arguments for and against various theses was included by Diels in Die
Fragmente der Vorsokratiker and is standardly dated around 400 bc;3 this dating, however,
was more recently (and compellingly) questioned as wholly speculative by Conley and
Burnyeat, who argued that, as far as our poor evidence can show, the Dissoi logoi could
have been written centuries after the 404 bc (the likely terminus a quo).4 While granting
this cautionary view on the possibility of dating the Dissoi logoi with any confdence and
precision, I hope it will not appear too conjectural to assume that this work draws ulti-
mately (if only indirectly) on sources belonging to the sophistic milieu of the late 5th-
early 4th century bc, or at least represents a quite successful later attempt to mimic them
as faithfully as possible. Even if the author should be much later, no evidence suggests
that in the short passage in which we shall be interested he might be contaminating the
material he is working on with anachronistic insertions.5 Therefore, I shall begin my
analysis from the Dissoi logoi, with no commitments about its actual date or authorship.
We are in the middle of the fourth chapter, On truth and falsehood: the author has
just presented some arguments in support of the thesis that the true and the false
are the same thing (henceforth, Identity Thesis, it), and now is ready to ofer a
series of arguments for the opposite thesis that the false and the true are
diferent things, difering by name and also in reality. Here is how the frst of these ar-
guments runs:
T1 For if one were to ask () to those who say that the same is false and true which
of the two their own is, if <their reply were> false, it is clear that <the false
and the true > would be two things, while if they were to answer ()6 true,
then this very <> would be also false. (dk90b4, 6)
1 Burnyeat 1976a, p. 47.
2 The phrase occurs also in the opening sentence of the next three chapters.
3 For extensive discussion about the date of the Dissoi logoi cf. Robinson (1979, pp. 25-41), according to whom
the . . was written some time around 403-395 (the date accepted by most scholars).
4 Cf. Conley 1985, Burnyeat 1998.
5 Of course this diagnosis is largely based on my overall assessment of ancient self-refutation, and cannot be
vindicated at this stage.
6 As Robinson (1979, p. 194) notices, this disconcerting example of a change from plural to singular is not a
hapax in the Dissoi Logoi.
self-refutation arguments from democritus to augustine 17
This argument rephrases it as the same is false and true, and the structure of
the refutation makes it clear that this must be in turn understood as every is (un-
qualifedly) both false and true, equivalent to the conjunction of the contraries Every
is true and Every is false:
(IT) (p)(TpFp).
But the original thesis argued for in the frst part of chapter 4 had a quite diferent shape,
at least if we judge by the arguments collected in its support: the true and false
are the same because
the same can be true, if the event it describes has taken place, or false, if that
event has not taken place (2-3);
the same can be true, if uttered by a certain person, and false, if uttered by another
the same can be true now, and false tomorrow (5).1
Burnyeats general qualm that in the Dissoi logoi many of the arguments for and
against do not even manage to contradict each other fts our case nicely: the argument
in T1 seems liable to the charge of ignoratio elenchi, since the thesis it attacks (it) is not
the same as the one established by the previous set of arguments (that the same 2
can turn out to be true or false depending on diferent circumstances). To be precise,
T1s argument does not manage to contradict any thesis which the previous arguments
succeeded in establishing: from the opposite perspective, one could hypothesise that the
frst set of arguments was aimed at proving, unsuccessfully, the outlandish thesis it
which T1 targets.
Two fundamental features of T1s argument immediately leap to the eye: its dilem-
matic form and its dialectical context. We shall encounter both of them repeatedly: the
presence of some kind of dialectical context, in particular, seems to underlie all ancient
self-refutation arguments, with only very few possible exceptions (hereafter by dialec-
tical context I shall intend, loosely, any dialogical situation in which two opposing par-
ties either individuals or groups, not necessarily facing each other in fesh and blood
advance and support incompatible views and agree to try to settle their dispute
through arguments which are structured in the form of question and answer and re-
spond to some shared rational standards or rules, with the purpose of establishing the
truth, or at least the relative merits and plausibility of their clashing positions, and not
merely of winning the debate at any cost, unlike the case of eristics).3 This should come
1 For the same pattern cf., e.g., ch. 1 of the Dissoi logoi:
(1) Illness is bad for the sick, good for the doctors; victory is good for the winner, bad for the loser; etc.;
(2) <therefore the same thing (e.g. illness, victory) is good and bad>;
(3) therefore the good and the bad are the same thing.
The inference from (2) to (3) would sound less problematic to Greek ears than to ours because of the well-
known fact that in Greek the X, where X is a neuter singular adjective, can function not only as an abstract
(X-ness) but also as a collective (the Xs, the class of the things which are X), much as in the English phrase the
poor: e can therefore be taken to mean both goodness and the things which are good. For a very
diferent conclusion from premisses of the same kind as (1) cf. Pl. Resp. v, 479a-d.
2 It is difcult to decide whether the same is best understood as a single sentence-token (as suggested
by the argument of sects. 2-3) or as a diferent token of the same sentence-type (as required by the argument of
sect. 4). The argument of sect. 5 seems to be compatible with both options.
3 Of course such a loose working notion would require to be clarifed and narrowed case by case.
18 luca castagnoli
as no surprise, since ancient logic never lost its well-known original connection with
the concrete practice of dialectic and disputation: nevertheless, I think that some con-
fusion has arisen in the literature from disregard for, or underestimation of, the full im-
port of this datum.
Let us reconstruct the details of T1s dialectical exchange. The proponent of it is
faced by his opponent with a dilemma: does he believe that the expressing it is
false or that is true? The reasoning underlying the frst horn is easy to understand: if
the supporter of it answers that his is false, then he is conceding that the con-
tradictory of his it is the case (as long as he endorses the platitude Fpp), i.e. that,
to borrow the opaque but now familiar jargon of the author of the Dissoi logoi, the
false and the true are two diferent things. On the other hand, if the pro-
ponent of it grasps the second horn (My is true), he is thereby confrming that
he takes it to be the case; but if he accepts that every is both false and true, he
must thereby admit that also the expressing his own thesis it is (also) false. Here
the argument suddenly stops, with no further comment or clarifcation: but what has
it achieved exactly? Under both the horns of his opponents dilemma, the proponent
of it has been forced into undesirable admissions: in the frst case the straightforward
denial of his own thesis, in the second the concession that the expressing it is it-
self (also) false, by self-application. It is not difcult to see why both outcomes are
deeply embarrassing, and can be interpreted as amounting to a ruinous dialectical de-
feat, which anyone should be extremely careful to prevent, by refraining, at the outset,
from endorsing it itself.1
it turned out to be a loser in debate; does this mean that what it expresses its
propositional content, we might say has been proved not to be the case? I suggest
that the answer is no, and what is more important that the author of the Dissoi logoi
shows no interest in establishing the latter, diferent point. Unlike Mackies absolute
self-refutation, which is supposed to prove the necessary falsehood of certain proposi-
tions (or the impossibility of certain states of afairs obtaining), T1s argument simply
seems to aim at showing the untenability of it under dialectical scrutiny. Moreover,
the structure of the charge embedded in the second horn is signifcantly diferent from
that of Mackies argument: if the every is false (and true) is true, then
every is false (and true), and therefore the every is false (and true)
must itself be (also) false. This kind of self-application is what one would expect to fnd
as a prominent trait of self-refutation, and we shall discover that this expectation is met
by various ancient arguments. However, we have seen Mackie follow a diferent (and
indeed opposite) route: the fundamental step of his absolute self-refutation was not If
nothing is true, then Nothing is true is not true either, but If nothing is true, then
Nothing is true is true.2
1 I use can because a full-blown supporter of it could be prepared to subscribe to the idea that his own
is itself, like every other , both false and true. Such a hardcore position, however, could still be attacked on
dialectical grounds: why has the supporter of it advanced his thesis, if he believes that it is the case no more than
it is not? Why has he answered true to the dilemma, when he believes that his is both true and false, in-
stead of immediately asking to reformulate the question more properly, as a trilemma? [On the basis of the letter
of the Greek text, I am assuming that both horns are explicit: Is your true or false?, and not Is your
true (or not)?].
2 Robinsons (1979, pp. 193-194) analogy between it and the Liar is also ungrounded. Levis suggestion that
T1s argument resembles the of s.e. m 8.389 (cf. T20 in sect. 6. 1) is more to the point, but his assump-
tion that therefore it also derives from Democritus (1940, p. 298) appears wholly speculative.
self-refutation arguments from democritus to augustine 19
My reconstruction of T1 as a dialectical silencer of its proponent could be chal-
lenged by observing that the text does not make it explicitly clear that the consequents
of the two conditionals describe what must be granted by anyone who has subscribed
to the corresponding antecedents. This might encourage a diferent reconstruction of
our argument, not as a dialectical dilemma, but as a proof by cases of not-it:
(1) T(it)F(it) Bivalence
(2) F(it)it Semantic descent (Fpp)
<3> T(it)it Semantic descent (Tpp)
<4> it(T(it)F(it)) By substitution (self-reference)
<5> (T(it)F(it))F(it) -elimination
(6) T(it)F(it) From <3>, <4> and <5>, by transitivity
<7> T(it)it From (6) and <2>, by transitivity
<8> it From <7>, (2), and (1), by simple constructive dilemma
I recommend that the temptation to read T1 along these lines must be resisted, for at
least two reasons. On the one hand, this reconstruction forces us to supplement nu-
merous additional steps of which no trace can be found in the text (in particular, the
crucial <7> and <8>); on the other, the two key premisses (2) and (6) patently distort
the literal sense of T1. The protases of the two conditional sentences are, respective-
ly, If <they answer1 that their own is> false and If they answer <that their
own is> true, and not If their own is false and If their own
is true. This strongly invites us to interpret the apodoses accordingly, since
the conditional If they answer that their own is false, then it is clear that the
false and the true would be two things, if taken verbatim, is a sheer non
sequitur: obviously it is not sufcient to say that p is false for p not being the case (the
same holds, mutatis mutandis, for the second conditional). The most plausible way of
making any sense of these conditionals, as charity suggests, is interpreting their
apodoses as elliptical, in the way I have proposed above when frst reconstructing the
argument: If <they answer that their own is> false, then it is clear that <they
are conceding that the false and the true > are two things; if they answer
<that their own is> true, then <they must admit that this very > is
(also) false.
We have just come across another trait which characterises several ancient self-refu-
tation arguments: a tendency to elliptical formulations, in which it is not always
immediately transparent whether what is on the table is the truth-values and logical
consequences of certain propositions or, rather, the tenability and unavoidable com-
mitments of certain positions in dialectical exchanges.
4. Plato: dialectical self-refutations
Although in the second part of Dissoi logoi 4 the position under attack had taken the
shape of a conjunction of both the theses we are interested in (Every is true and
every is false), T1s self-refutation charge exploited the self-applicability of the
second conjunct only, while the frst one remained, as it were, logically inert. I shall ex-
1 Some verb such as is clearly presupposed also by the frst conditional.
20 luca castagnoli
amine now two Platonic passages in which self-refutation charges are levelled against
variants of that frst conjunct.
I have emphasised above some crucial diferences between Mackies absolute self-
refutation of Nothing is true and the Dissoi logoi dilemma against it: no similar com-
parison will be possible in this section, since in his taxonomy Mackie left no place for
Everything is true, Nothing is false, and analogous theses. I suppose he did not see
them as self-refuting at all, since, as we shall discover, their refutation requires one to
consider some external conficting proposition which either should be accepted exactly
in virtue of the supposed truth of those theses (section 4. 2) or must be eventually grant-
ed because of some broader kind of dialectical necessity (section 4. 1).1
4. 1. Dionysodorus downfall
Let us begin with a passage from the Euthydemus. The sophist Dionysodorus has just ar-
gued, very deftly, for the surprising thesis that contradicting () someone else
is impossible,2 stunning and silencing his interlocutor Ctesippus (285e-286b). Socrates
takes over the conversation:
T2 even though I have heard this particular [sc. that it is impossible to contradict (
)] from many and at many times, Im always amazed (d ). Protagoras
and those like him made considerable use of it, and also some still earlier: but it always seems
to me that its something amazing, and that it overthrows not just the other , but itself
as well (d b d r d d e ).
[] The amounts to claiming that it is not possible to say what is false (
), doesnt it? And when speaking either one says the truth or else doesnt say any-
thing at all? (286c1-8)
Why does Socrates believe that the that has the amazing pe-
culiarity3 of refuting itself ? This is surprising:4 after all, it proclaims that nothing can be
refuted, since it is equivalent to, or at least entails (), that it is impossible to say
what is false.5 Socrates immediately infers that, according to the Protagorean6 ,
false judging (), false judgement (), ignorance (), and ignorant peo-
ple () will not exist either, obtaining Dionysodorus eager assent to the whole
lot (286d). Socrates protests that Dionysodorus must be speaking only for the sake of
argument, but his opponents reply is dry: But do refute () me, then (286e1).
Socrates complains that there cannot be such a thing as refutation if one accepts, with
Dionysodorus, that nobody speaks falsely (286e2-3). The underlying charge is devastat-
ing: what Dionysodorus has just done (challenging Socrates to refute him) is de-
nounced as inconsistent with what he says, since by implying that refutation is possible
1 I have argued in sect. 2 against Mackies pretension that in his absolute self-refutation it is a single proposi-
tion, all alone, which refutes itself. It is true, however, that the fundamental extra assumption required (T-prefx-
ability) is at least supposed to describe a basic, non-contingent trait of the grammar of a predicate (true) included
in the allegedly self-refuting proposition.
2 Dionysodorus argument at 285d-286b need not concern us here. For an interpretation of it and parallel pas-
sages cf. Denyer 1991, Burnyeat 2002. 3 I interpret the frst of 286c4 as epexegetic.
4 Pace Rankin, according to whom this is a semantically self-evident point (1981, p. 25).
5 A thesis previously defended by the two brothers (283e-284c).
6 The was better known as an Antisthenic warhorse (cf. e.g. Arist. metaph. 29, 1024b32-
34, top. i 11, 104b20-21 and p. 23n5). Notice the air of paradox of attributing to Protagoras, the author of \A,
the view that is impossible.
self-refutation arguments from democritus to augustine 21
it seems to commit Dionysodorus to the idea that falsehood and contradiction are also
possible after all (a form of pragmatic inconsistency). Euthydemus comes to rescue of
his companion, helping him to avoid tackling Socrates criticism by picking up and em-
bracing Socrates own mocking suggestion that since falsehood is indeed impossible,
refutation is also impossible, and thus Dionysodorus cannot have challenged Socrates
to refute him, despite all appearances (286e4-7):1 since no one is capable of doing what
is not (284c), you cannot order what is not, just as you cannot say what is not. Socrates
does not lose his patience, and immediately launches a second attack: if ignorance and
error (in action, speech and thought) do not exist, what in heavens name do you two
come here to teach? (287a1-b1). The sophists previous claim to be teachers (cf. e.g.
274a) is charged with blatant inconsistency with the thesis they are
now defending, and all its corollaries: teaching certainly involves, among other things,
contradicting and purging the false beliefs of the learner and replacing them with true
ones, elevating him from mistake and ignorance to knowledge.2 But again the two
sophists refuse to tackle Socrates challenge, accusing him of bringing back into the
discussion something said at the very beginning of their exchange (their boasts as
teachers) only because he is unable to deal with what is being said presently (287b2-
5). With this move (a shameless refusal to consider diachronic inconsistency a dialecti-
cal sin),3 however, they give Socrates a chance to revive his previous charge: what
could the sense of the phrase unable to deal with what is being said presently be if
not unable to refute the present argument (287b6-c1)? Dionysodorus challenge is
meaningless, if refutation is impossible. Once again Dionysodorus refuses to answer
and wants to restore his preferred role of questioner, but Socrates quickly spots anoth-
er dangerous inconsistency displayed by this very behaviour: the principle on which
Dionysodorus refuses to answer must be that he is most skilled in discussion and
knows, unlike the ignorant Socrates, when an answer is to be given and when not
(287c9-d2). Dionysodorus either does not grasp Socrates allusion or, more likely, as his
reaction betrays (You are just babbling), pretends not to grasp it, and goes on with
questioning: if only animate beings have sense (), and phrases are inanimate, as
Socrates admits, why has Socrates asked the sense () of Dionysodorus phrase un-
able to deal with what is being said presently (287d7-e1)? Dionysodorus argument is
yet another piece of sophistry, based on a blatantly homonymous use of the verb
,4 but Socrates, instead of denouncing its fallaciousness, once again suddenly
brings it to bear against its proponent:
T3 Are you saying that I made a mistake or not? Because if I did not make a mistake you will
not refute me, no matter how wise you are, and it is you who are unable to deal with the ar-
gument. But if I did make a mistake not even then are you right to claim that it is impossi-
1 Both Socrates charge and Euthydemus defence of his companion seem to overlook the possibility of inter-
preting Dionysodorus imperative not as committing him to the existence of contradiction and falsehood in pro-
pria persona, but only as challenging Socrates to do, if he is capable, that very thing which he insists is possible and
which Dionysodorus denies. 2 Cf. Tht. 161d8-e1.
3 One might object that this refusal appears shameless only because of our failure to take into proper account
the real nature of the eristic display of the two brothers, who are not bound to consistency between diferent
episodes or rounds of dialectic. In Socrates eyes, however, such a rejoinder would be tantamount to confrma-
tion of his initial suspicion that the brothers speak only for the sake of argument.
4 With the meaning of to mean, in one case, and to think, in the other one. My English translation is not
completely faithful to the Greek original, but at least it manages to reproduce the same kind of homonymy
22 luca castagnoli
ble to make mistakes. And Im not talking about things you said last year. So, Dionysodorus
and Euthydemus, it looks as if this has made no progress and still, as in the past, throw-
ing down <the others> falls down itself (g ). (287e4-288a4)
The loop is now complete. Dionysodorus last sophistry has provided Socrates with
confrmation of his initial impression that Dionysodorus throws itself down:
g 1 is clearly meant to be equivalent to d
e of T2 (286c4). Both phrases have a nice pictorial force: the verbs
, , and were, most probably, all borrowed from wrestling
jargon.2 This introduces us to another typical feature of ancient self-refutation charges:
the large use of metaphors and similes to express and illustrate them.3
Dionysodorus has criticised Socrates for speaking as if phrases had sense: but what
can the point of this criticism be for one who takes Dionysodorus stance on contra-
diction and falsehood? He is faced with a dilemma: if Socrates did not make any mis-
take by speaking in that way, then Dionysodorus must admit that his censure has been
pointless (a dialectical error), and he cannot dismiss Socrates previous inconsistency
charges (he still owes him an answer, and thus he appears to be unable to deal with the
argument); if Socrates did make a mistake, then perhaps Dionysodorus won the penul-
timate round of the dialectical contest, but the thesis and all its
companion theses are automatically admitted by him to be false, and he looses the
whole match. Either way, Dionysodorus is defeated.4
Socrates argument looks like a knock-out blow: but why should we classify it as a
self-refutation argument, as Socrates metaphors suggest? Spragues (1962, 19) solution
is unconvincing: to endorse theses which make refutation impossible, and then go on
in ones daily business of trying to refute everyone certainly is not a commendable po-
sition, but what it can be charged with is only pragmatic inconsistency, and not self-
refutation. A diferent hypothesis which could explain Socrates self-refutation jargon
here follows the guidelines of a solution Sprague herself discards: the argument pur-
porting to prove that it is impossible to contradict (or, perhaps, this very thesis) contra-
dicts (or, more precisely, is meant to contradict) the commonsense view that contradic-
tion is possible, and thus commits its proponent to the existence of contradiction,
refuting itself.5 The difculty with this proposal6 is, trivially, that T3s dilemma, which
1 K could be a polemic allusion to Protagoras work \A under its alternative title O
(The downthrowers; cf. s.e. m 7.60 and sect. 4. 2), but an early currency of this title is far from cer-
tain. On the use of the verb as a metaphor for refuting someone in the agonistic context of debate
and on further ancient references (e.g. Eurip. Bacc. 202; Plut. Peric. 8; Stob. ecl. ii, 23.15) cf. Lee 2005, p. 24n31.
2 Cf. also Euthd. 277d1-2; Phdr. 256b; Pol. 583b; Hawtrey 1981, p. 70. Some translators take also e
at 288a4 as referring to some proverbial expression (as the old saying goes) or to some piece of wrestling jargon
(in the old phrase of the wrestling school), but both proposals seem unwarranted.
3 For an early occurrence of the same wrestling jargon (, , ) for self-refutation
cf. Democritus fr. 125. Notice that Protagoras is credited with a d (On wrestling, d.l. 9.55). For the am-
ple use of similes in ancient self-refutation cf. Castagnoli 2000 and Castagnoli 2005.
4 As Chance notices, Socrates refutation is also a perfect illustration of a perennial feature of all comic ac-
tion: comic inversion. Just as the comic playwright presents, for example, a robber robbed or a mugger mugged,
so too Plato has presented the refuters refuted (1992, p. 108).
5 I believe that the proposed line of reasoning underlies this compressed passage in Diogenes Laertius (3.35):
They also say that Antisthenes, being about to read publicly something that he had composed, invited him [sc.
Plato] to be present. And on his inquiring what he was about to read, Antisthenes replied that it was On the im-
possibility of contradicting. How then, said Plato, can you write on this subject?, thus teaching that it incurs
reversal (). For the meaning of cf. sect. 6.
6 Accepted by Hawtrey (1981, p. 108).
self-refutation arguments from democritus to augustine 23
appears to be meant by Socrates to back his fnal self-refutation charge, does not ap-
pear to work along these lines. I suggest we can preserve the core of this solution only
if we understand the term as referring here not only to the thesis that contradic-
tion is impossible (and its corollaries), but also to all the arguments advanced in its sup-
port, both to establish it and to undermine potential counter-arguments (the whole
theory or philosophical outlook, we might gloss). By grasping the second horn of
Socrates dilemma in T3 and thus confrming his allegation that Socrates was mistaken
in speaking as if phrases had sense, Dionysodorus would be trying to defuse Socrates
accusation of inconsistency at 287b6-c1; but by overthrowing (or at least trying to over-
throw)1 that accusation, he would be at the same time overthrowing his own position
too, by unwittingly conceding that mistake and contradiction do exist after all. One
could still feel uncomfortable with the idea that in this way Dionysodorus makes him-
self liable to the diferent charge of self-refutation, rather than to a second, renewed
charge of inconsistency: strictly speaking, it is not the thesis, tak-
en in isolation or together with some other assumption,2 that overthrows itself here,
but Dionysodorus who contradicts himself by counter-arguing in its defence. However,
Dionysodorus sophistic refutation of Socrates, which clashes with his endorsement of
that thesis, is not just any old refutation; it somehow stems from that endorsement for
a sort of dialectical necessity.3 In a dialectical context like the one depicted in the Euthy-
demus it is necessary that you try to undermine the arguments that your opponent ad-
vances against the position you are advocating, unless you prefer to give it up and ad-
mit defeat: but if your position is that it is impossible to contradict then any such
attempt (whatever its precise content and force) will speak against that position, rather
than support it, and will transform you into the best, albeit reluctant, ally of your op-
ponent. Although in T2 Socrates could not possibly have foreseen to what specifc ar-
gument of Dionysodorus he would later apply the fatal dilemma of T3, he could be
fully confdent, from the very beginning, that the sophist, when challenged, in order to
defend his would have to attempt some argument against him to which the dilem-
ma could be easily applied.4
This makes Platos idea of self-refutation here perhaps looser than we might have
expected, and the aim of Socrates argument certainly diferent from that of Mackies
absolute self-refutation. Socrates has shown why is bound to be a
dialectical loser: every attempt to defend it from attacks would amount to an involun-
tary admission of its falsehood. That thesis is in fact incompatible with dialectic ( just as
it is incompatible with teaching), because by denying the possibility of falsehood and
1 Neither nor in Socrates formulations of the self-refutation charge should be intend-
ed as success verbs. I suggest they must have a conative nuance: Socrates cannot be saying that the arguments in
favour of the thesis do manage to overthrow the opposite thesis and the arguments in its de-
fence (and then overturn themselves too), but only that they purport to overthrow them (cf. p. 41).
2 Contra Burnyeat (2002, p. 41): if it is true that there is no false judgement, but Socrates thinks it is false, then
it is false that there is no false judgement. Contra also Narcy (1989, p. 80), Fine (1998, p. 201n2), and Kahn (2000,
p. 91), who compare this passage to the peritrop or self-refutation of Protagoras in the Theaetetus (cf. sect. 4. 2).
3 McCabe (1998, p. 155) notices that the claim that falsehood is impossibile does not directly imply its own false-
hood []; it needs, instead, a more complex dialectical context to be overthrown, but she fails to clarify what this
context is in our passage.
4 Socrates remark I am not talking of things you said last year can be interpreted not only chronologically
(he is applying his dilemma to the argument Dionysodorus has just proposed) but also logically (that argument is
integral part of the current dialectical round, unlike Socrates previous reference to the two sophists boast as
24 luca castagnoli
error it destroys the rationale of debating (proving your own position correct and your
adversarys wrong). All you can aford with that thesis is the undialectical behaviour of
stating it and then remaining silent. And this too only with the crucial proviso that yours
is no ordinary assertion: without such a proviso, the bare statement of it could already
be charged with overthrowing itself , since normally when you assert that p you can
be taken to be committing yourself to the idea that p is true and not-p is thereby false.
Needless to say, this is a most unpalatable result for any philosophical position. But
has the proposition itself It is impossible to contradict thereby been proved to be
false? I suggest that the correct answer is, again, no. The fact that Dionysodorus could
not help committing himself to the existence of error and contradiction as soon as he
decided to enter the public dialectical arena which is built around those notions does
not imply, in point of logic, that error and contradiction do really exist.1 For, although
the sophists downfall was not the result of any preventable error on his part, but re-
fected an objective indefensibility of his thesis in that setting,2 one could protest that
that arena hosts worthless games governed by rules which have no correspondence
with reality itself: contradiction and falsehood are not possible, although we foolishly
behave as if they were (but, then, are those who engage in this sterile dialectical game
ignorant and mistaken?). I am not suggesting that Plato believed he could not properly
establish the stronger point of the absolute falsehood of Dionysodorus , but
only that his purpose in the Euthydemus section we have just analysed was diferent and
more modest.3
4. 2. Protagoras most clever self-refutation4
In the last three decades or so very few Platonic dialogues have attracted the same
amount of scholarly interest as the Theaetetus; no single passage in the Theaetetus has
managed to excite the same lively debate as Socrates so-called most clever refutation
of Protagoras Measure Doctrine (hereafter, md) at 171a-c. Since md, however one in-
terprets it, bears an obvious resemblance to the thesis that Everything is true and
Socrates argument is usually treated as a paradigmatic example of ancient self-refuta-
tion,5 I shall dare to ofer my own interpretation of the logic of that argument.
4. 2. 1. The Measure Doctrine
and the context of Socrates argument
Immediately after Theaetetus formulation of his frst admissible defnition of knowl-
edge, Knowledge is perception (hereafter, kp), Socrates remarks that Protagoras used
to say the same, although in a diferent way:
1 Just as the mere fact that one says that it is false does not prove that the false and the true are
diferent (cf. sect. 3).
2 In Socrates own terms, an old defect that no one has ever found a way to overcome.
3 Whether he thought that dialectical indefensibility is a telling sign of something more fundamental about
the truth-value of the proposition involved is a question which I cannot address in this article (cf. Castagnoli
4 This section is an abridged and slightly revised version of Castagnoli 2004a, to which I refer for all the details
which I could not discuss here, and in particular for analysis of other interpretations.
5 Some alternative labels: recoil argument (Newman 1982), turning the table argument (Ketchum 1992),
and, especially, peritrope argument (e.g. Lee 1973, Burnyeat 1976b, Chappell 2006).
self-refutation arguments from democritus to augustine 25
T4 Man is the measure of all things, of those which are, that they are, of those which are not,
that they are not. (152a2-4)
Protagoras md is paraphrased, in turn, as follows:
T5 As each thing appears () to me, so it is for me (), as it appears to you, so it is for
you (). (152a6-8)
Since, as Socrates goes on to explain, appears and is perceived amount to the same
in the case of perceptual objects (152b12-c2), md implies that things are for each person
() such as he perceives them (152c2-3) and thus that perception is always of
what is, and cannot be false, as befts knowledge (152c5-6). Therefore, Protagoras md
provides the sufcient supporting ground for Theaetetus kp: in a world in which a Pro-
tagorean epistemology holds good, perception is always of what is the case (and thus
always true) for the perceiver. In a case in which, when the same wind is blowing, one
of us feels cold, and the other not (152b2-3), we must say that the wind is cold for the
one who feels cold, and not for the other (152b8). This move allows Protagoras to in-
sist that neither perception is false and, at the same time, to avoid accepting the sheer
contradiction The wind is both cold and not cold (thus preserving, we would say, the
principle of non-contradiction hereafter, pnc). But what on earth does The wind is
cold for Socrates mean? This is explained when Socrates introduces, and progressively
unfolds in more and more detail, the Secret Doctrine (hereafter, sd) which Protago-
ras allegedly revealed to his pupils (152c10), meant to provide an ontological setting for
the epistemological md. sd is frst introduced as the thesis that there is nothing which
is just one thing by itself (152d2-3), to become very soon the quite diferent thesis that
nothing is one, either one thing or qualifed in one way (152d6):
T6 all the things of which we incorrectly say that they are (r) are in the process of coming
to be (), as the result of movement and change and blending with one another.
After a preliminary sketch of an sd-based theory of perception (153d7-154a5), sd appears
in the new form everything is change () and there is nothing else besides
change (156a5). A long description of what the world is like according to this theory
follows (156a5-157b1): since each perception () and its twinned perceived thing
(), i.e. quality, are generated together only on the occasion of the interaction
of their two parents (the perceiver and the perceived object), they are relative to both
parents. Each perception is of a perceived object no less than it is of (belongs to) a per-
ceiver; more surprisingly, each perceived quality is for a perceiver no less than it is (a qual-
ity) of a perceived object. There cannot be any perception to which no perceived qual-
ity corresponds (every perception is true: perception is infallible), and there cannot be
any unperceived quality (every perceptual quality is perceived: perception is omni-
scient). A and are like twins, and this should ensure that perception
is unerring, as befts knowledge.1 Now we can understand what The wind is cold for
Socrates means. Socrates and the wind generate in their encounter twin ofspring,
coldness and a perception of cold. Coldness quickly moves towards the wind and qual-
ifes it (relatively to Socrates), so that the wind becomes, for Socrates, cold; the percep-
1 They must be identical twins, then.
26 luca castagnoli
tion of cold quickly moves towards Socrates and qualifes him (relatively to this wind),
so that he becomes a feeling-cold Socrates (in respect to this wind). Is the wind warm
or cold in some absolute sense? No, nothing (i.e. no parent) is qualifed in any way by
itself: each thing becomes whatever it is in relation to something.
The import of the qualifers is not the only aspect of md in need of clarifcation. md
is a complex generalisation of the form
xyF (if y appears F to x, then y is F for x),
but whereas the domain of the xs is obvious from the beginning (men are measures),
the domains of the ys and of the Fs appear less defnite, and progressively expand from
the narrow perceptual objects and qualities we have seen so far to include at the end all
possible objects of judgement, even in the non-perceptual sphere.1
At 160e the depiction of kp, md, sd, and their correlations is complete. Socrates pres-
ents a series of objections to md (161c-164b), but also voices a possible disdainful apol-
ogy on behalf of Protagoras, who protests that those objections have been unfair and
based on mere plausibility and verbal traps, hints at how he would reply to them, and
challenges Socrates to attack and refute what he actually says, if he is able to (162d-e;
166a-c).2 Socrates can present his objections in a continuous speech, or, if he prefers,
use his favoured method of question and answer (\ ), provided he is fair in
his questioning (167d4-e1).3 Subsequently, Socrates persuades a reluctant Theodorus to
participate, in place of young Theaetetus, in a more mature examination of Protago-
ras doctrine, and in particular of the issue whether it was correct on Socrates part to
have Protagoras concede in his apology (166d-167d) that wisdom does exist, but that
the wise are superior to others not on the question of what is true or false (everyone is
an infallible measure), but on that of what is better/benefcial or worse/harmful
(169d3-8): the wise are those who can change the appearances, who, when things appear
(and therefore are) bad for someone, can produce a better state by making things ap-
pear (and therefore be) good for him (166d6-7). Socrates plan is to obtain Protagoras
agreement () in the quickest and safest possible way, starting from his own
4. 2. 2. Socrates dilemma
Socrates begins by recalling md:
(1) what seems to each one also is for him. (170a3-4)
I have explained the scope md has reached at this point of the dialogue: what seems (e
) to each one is whatever is judged by each man. Socrates obtains then Theodor-
us concession that
(2) everyone agrees that all men believe in the existence of both wisdom () and ignorance
(). (170b6-7)
1 For detailed analysis of the various stages of this expansion and discussion of diferent interpretations of md,
according to which Protagoras is not a relativist about truth, but of fact (e.g. Waterlow 1977), or is not a rel-
ativist at all, but an infallibilist or subjectivist committed to absolute truths about private perceptual objects
(e.g. Fine 1998) cf. Castagnoli 2004a, pp. 5-9.
2 For this rejoinder and its intrinsic difculties cf. Euthd. 286e1-3 (p. 21) and sect. 4. 2. 6.
3 On Protagoras own mastery of both modalities of speech cf. e.g. Pl. Prt. 329b1-5 and 334e4-335a1.
self-refutation arguments from democritus to augustine 27
As evidence for this universal consensus Socrates does not bring the explicit declara-
tions of people, but their behaviour: men look for (or, alternatively, propose them-
selves as) experts, teachers, and leaders, on the evident assumption that experts, teach-
ers, and leaders are wise in those very spheres in which the laymen are ignorant, and
that wisdom is something valuable, sometimes even vital (170a6-b7). Theodorus also
grants that
(3) men believe that wisdom is true thinking, ignorance false judgement. (170b9-10)
This ordinary conception of and is not the one we have seen Socrates
attribute to Protagoras in his apology: therefore, men should be intended only as
most men (at this stage, fairness requires that Protagoras and his followers are ex-
cluded). It is easy to see how (2) and (3) jointly imply that
(4) everyone agrees that (most) men believe that there are false judgements,
from which, as we shall discover shortly, the second premiss of Socrates argument is
easily secured.
Let us now proceed in our analysis:
T7 Socr.: What then, Protagoras, are we to make of your ? Are we to say that men always
judge what is true, or that they judge sometimes what is true and sometimes what is false?
( d f , j b b , b b ;) For,
I suppose, from both alternatives it follows that men do not always judge what is true, but
both <what is true and what is false>. For think, Theodorus, would anyone of Protagoras
followers, or you yourself, contend that no one ever thinks that anyone else is ignorant and
judges the false?
Theod.: Thats not a thing one could believe, Socrates.
Socr.: And yet it is to this that the saying that man is the measure of all things is nec-
essarily driven.
Theod.: How is that? ( ;) (170c2-d3)
Protagoras , which we had found in its usual relativised form only a few lines above
in (1), and which we fnd at the end of T7 in its ofcial Protagorean formulation (Man
is the measure of all things), appears diferent at the beginning of the passage (170c3):
(5) men always judge what is true.1
The absence of the qualifer for them is puzzling, since it seems to make of (5) an in-
fallibilist or subjectivist thesis, and not that formulation of relativism which we
would have expected. Before considering some possible explanations for this absence,
let us see what role (5) plays in Socrates argument:
(a) If (5) men always judge what is true, then men sometimes judge what is false. pp
(b) If men sometimes judge what is false, then men sometimes judge what is false. pp
(c) Either men always judge what is true or men sometimes judge what is false. pp
(d) Therefore, men sometimes judge what is false. p
1 For the reasons why I believe that (5) is meant to be a formulation of md and discussion of diferent inter-
pretations cf. Castagnoli 2004a, p. 12.
28 luca castagnoli
Whereas both the validity of this constructive dilemma and the truth of premisses (b)
and (c) are apparent, the conditional (a), the core of the argument, requires some
explanation: that men sometimes judge what is false follows from (5) and the further,
external assumption, already guaranteed by (4), that people believe that there are false
judgements. For if every judgement is true, then also the judgement according to which
there are false judgements must be true, and therefore there must be false judgements.
If one wanted to block this line of reasoning and deny that false judgements exist, one
should deny what everyone agrees upon, that people believe in the existence of
ignorance and false judgement: but, as T7 reminds us, not even Protagoras or the Pro-
tagoreans can be ready to reject this undeniable datum.1
Socrates argument, as it stands, is clear, elegant, and unimpeachable: however, as a
refutation of Protagoras it is vitiated by the fact that the unrelativised (5) does not look
like a fair depiction of md. One might suggest that Socrates is unwittingly guilty of ig-
noratio elenchi: surely, however, this is to be ruled out, given that the qualifers are frm
in their place both a few lines above and, as we shall see, a few lines below T7, and it
would be impossible for any lucid writer to commit such a mistake, let alone for a Pla-
to.2 Might Socrates be dropping the qualifers purposely to get an easy win over Pro-
tagoras? One could object, with Burnyeat, that after Protagoras request to be given a
fairer treatment in the discussion of his doctrine, this would be nothing less than per-
verse dishonesty, and perverse dishonesty is not a charge to be leveled lightly against
a philosopher of Platos stature and integrity (1976b, p. 177).3
How must we interpret T7s argument then? We could understand (5) as an elliptical
formulation of
(5*) men always judge what is true for them,
supplying the missing qualifer as implicitly meant. This proposal is not as arbitrary as it
could appear: there are a few other instances in the Theaetetus in which true, although
used within the framework of md, is not explicitly relativised (161d7, 167a8, 167c2, 172b6),
but it is evident that the reader is asked to supply the qualifers in thought.4 If we
understand (5) as (5*), Socrates argument can be reconstructed as follows:
(a*) If (5*) men always judge what is true for them, then men sometimes judge what is false.
(b) If men sometimes judge what is false, then men sometimes judge what is false.
(c*) Either men always judge what is true for them or men sometimes judge what is false.
(d) Therefore, men sometimes judge what is false.5
1 In any case, by denying what everyone believes, despite acknowledging that people do believe it, they would
at the same time be involuntarily admitting the existence of false beliefs.
2 This appears even more clear when we consider that the argument is presented after eight Stephanus pages
worth (160e-168c) of close study of arguments [] all of which Plato evidently takes to be fallacious precisely
because [] these arguments are careless about qualifers in various ways (Chappell 2006, p. 112).
3 I shall argue later, however, that this diagnosis needs to be qualifed.
4 The unrelativised occurrence at 161d7-8 (e a , b a d
) is particularly enlightening, since it comes only a few lines after a relativised occurrence, at 161d2-3 (
b n i \ ) and the two sentences seem to be meant as interchangeable formula-
tions of md.
5 The reader will wonder why I have relativised only true, and not also false, which makes this new ver-
sion of the argument rather asymmetrical. It is reasonable to suppose that Socrates argument is aimed at refut-
self-refutation arguments from democritus to augustine 29
Is this new argument sound? Whereas its validity and the truth of premisses (b) and (c*)
are, again, unproblematic,1 it is hard to imagine any rationale for (a*). The structure of
the passage suggests that its consequent (the existence of false judgements) should fol-
low from its antecedent (5*), i.e. md, and (4), the universally agreed fact that (most) men
believe that there are false judgements, something on which Protagoras too must agree.
But what can be inferred from these premisses is the relativised conclusion that it is true
for those who judge so that men sometimes judge what is false, where those who judge
so should include everyone except Protagoras and his acolytes. Consequently, it be-
comes mysterious how Socrates argument is supposed to work as a whole, and how it
can be meant to establish the unrelativised conclusion (d).
One could infer that the attempt at reading (5) as implicitly relativised has turned out
to be a dead end: we have avoided the Scylla of a sound but irrelevant or question-beg-
ging argument only at the high price of falling into the Charybdis of sheer unsound-
ness. There is a textual clue, however, suggesting that we might be on the right route
after all. Our uneasiness is Theodorus uneasiness; we feel in need for clarifcation, and
Theodorus asks for clarifcation too (How is that?). If T7s argument were to be in-
terpreted at face value, without adding the qualifers, it would be quite easy to see how
it works, even for a character, like Theodorus, not particularly philosophically-minded.
Since Theodorus is asking for clarifcation, it seems reasonable to suppose that Socrates
will proceed by providing it. This is, at any rate, what happens in the Theaetetus in all
the six other circumstances in which Socrates interlocutor demands some explanation
by uttering ; (152d1, 154b10, 164c7, 172c7, 199c12, 201a6);2 this, or at least some dis-
cernible signal of the fact that one is not going to satisfy ones interlocutors request, is
what the possession of a decent amount of dialectical politeness should guarantee, I
suppose, in Socrates time just as today. Most modern commentators do not seem to
agree: they suggest that in what follows we will fnd a brand new argument (or even
two new arguments) against md.3
4. 2. 3. How to relativise truth and falsehood:
the transitional passage
T8 Socr.: When you have decided something in your mind, and express a judgement about it
to me, lets grant that, in accordance with his [sc. Protagoras] , this thing is true for
you; but isnt it possible for the rest of us to decide about your decision? Or do we always de-
ing relativism tout court, and thus at establishing the existence of judgements which are false simpliciter, and not
merely false relatively to the judger (which would mean remaining in a broad relativist framework). Therefore,
false must not be relativised in (d) and in the consequents of (a*) and (b); but then, since it is logical to suppose
that (b) is true simply in virtue of its being a duplicated conditional of the form pp, false in the antecedent of
(b), and consequently false in the second disjunct of (c*), must not be relativised either. One might even con-
jecture that Plato left out all the qualifers in T7 to avoid making this asymmetry distractingly obvious.
1 (c*) lists only the two alternatives relevant to the present discussion: Protagoras relativism and the laymens
(and philosophers) ordinary view.
2 It is noteworthy that in all these cases the need for clarifcation does not betray any defciency of Socrates
interlocutor (Theaetetus in fve cases, Theodorus in one), but is caused by the elliptical and obscure way in which
Socrates has expressed his thought, alluding to something he is still to explain properly.
3 McDowell (1973, pp. 169-170) is one notable exception. Sedleys (2004, p. 57) interpretation of the structure
of the passage is not liable to my objection, since it locates the beginning of the second argument before Theodor-
us request, at 170d1-2.
30 luca castagnoli
cide that you judge the truth? Or is it rather the case that on every occasion there are count-
less people who are in confict with you, judging the opposite (), and believe
that you decide and think the false?
Theod.: Good heavens, yes, Socrates, countless thousands, as Homer says, who give me all
the trouble humanly possible!
Socr.: What then? Do you want us to say that what you judge on those occasions is true for
you (), but false for those countless people ( b )?
Theod.: It looks as if it is necessary, according to the , at any rate. (170d4-e6)
According to Protagoras md (a e , ), we should say
that since Theodorus judges that p then p is true for Theodorus, and since his opponents
judge that p is false then p is false for them. Of course there is nothing excitingly new in
this: this is exactly what one should have expected given the way md has been shaped
throughout the dialogue. But since only a few lines before, in T7, the Protagorean
had apparently altered into a diferent, non-relativistic thesis, this transitional passage
turns out to be important: it supports my hypothesis that the qualifers should be sup-
plied also a few lines before, in Socrates dilemmatic argument, unless one prefers to at-
tribute to Plato a quite schizophrenic attitude towards md (and much more so, if one
accepts my contention that T8 is supposed to begin an explanation of how Socrates
dilemma in T7 works).
Having proposed a model treatment of qualifers in case of conficting judgements,
Socrates applies it to Protagoras case. He starts from what he takes to be a datum: the
vast majority of people do not believe that man is the measure. If Protagoras himself
does not believe1 his own doctrine either, then no one believes it,2 and thus, according
to md itself, it is true for nobody (170e7-171a1).3 This might look prima facie like a counter-
factual hypothesis: Socrates will aim at showing that it is not as notional as it could ap-
pear. If we suppose, on the contrary, that at least Protagoras does believe his md, a frst
consequence is that md is still false for many more people than it is true for: just as in the
previous example about Theodorus, we might say that md is true only for Protagoras but
false for countless thousands, and thus, in this sense, more false than true (171a1-5).
4. 2. 4. The mds most clever feature
There is a more unexpected consequence coming next:
T9 Socr.: Secondly, it has this most clever feature (\ ): on the one hand, he
[sc. Protagoras] concedes (), in some way, that regarding his own opinion the opin-
ion of those who judge the opposite ( ) (by which they think that he says
the false) is true, since he agrees () that all men judge what is the case (a
Theod.: Undoubtedly.
Socr.: And then, if he admits () the truth of the opinion of those who think that
he says the false (e ), he is conceding () the falsehood of his own
1 More precisely, did not believe: at the dramatic date of the dialogue Protagoras is dead (cf. sect. 4. 2. 6).
2 This sounds quite surprising: one would think there must be other Protagoreans around endorsing, or at
least pretending to endorse, md. Perhaps one could intend Protagoras as Protagoras and his faction. However,
this slight inaccuracy is not too damaging: the same argument which Socrates will use to show that Protagoras
can be forced to join the anti-md consensus could be used against other Protagoreans.
3 The truth that he wrote is a pun, referring to md but also to the work, Truth (\A), beginning with it.
It also echoes Protagoras own words at 166c9-d1: For I do say that the truth is as I have written it.
self-refutation arguments from democritus to augustine 31
Theod.: Yes, necessarily.
Socr.: On the other hand, the others dont concede ( ) that they say some-
thing false (f )?1
Theod.: No indeed.
Socr.: But he, again, admits () that also this belief is true, according to what he
Theod.: So it appears.
Socr.: It will be disputed (), therefore, by everyone, beginning with Pro-
tagorasor rather, it will be admitted () by him, when he concedes ()
to the person who contradicts him that he believes the truthwhen he does that, even Pro-
tagoras himself will be conceding () that neither a dog nor just any human be-
ing is the measure of anything at all which he hasnt learnt. Isnt that so?2
Theod.: It is. (171a6-c4)
We have fnally arrived at Socrates most clever self-refutation argument. I have
adopted this label so far noncommittally, simply because it or some equivalent label
suggesting that the argument is particularly ingenious, exquisite, or subtle are
used by most commentators,3 with very few exceptions. However, I abandon this es-
tablished use here, since it is based, I maintain, on a misreading of 171a6: the unstated
subject in \ must be the that he [sc. Protago-
ras] wrote (170e9-171a1), i.e. md itself. It is not Socrates forthcoming self-refutation ar-
gument against md that is singled out as most clever at the beginning of T9; it is md
that has this most clever feature,4 where this refers forward to the surprising fact
that also its inventor, Protagoras, can be forced into rejecting it through a self-refuta-
tion argument. Is being necessarily repudiated by its own creator a sign of particular
cleverness for a philosophical thesis? Since the answer must be a defnite no, we should
understand Socrates remark as mocking:5 after all, also the two other occurrences of
the adjective in the Theaetetus (156a3,6 202d10)7 do not seem to express Socrates
unmixed praise.
Having bracketed the prejudice that T9s argument must be particularly clever, let us
examine it. Its most puzzling feature is, again, the absence of qualifers, beginning with
the unrelativised formulation of md as all men judge what is the case at 171a9: we have
appreciated how crucial the qualifers are in Protagoras and we have found them
in the statement of it introducing the whole section (170a3-4) and again, repeatedly, on-
ly a few lines before T9 (170d5, 170e4-5, 170e9). One possibility, powerfully advocated by
1 I shall argue on pp. 34-35 that this line must be read and interpreted diferently.
2 For a diferent construal of this convoluted sentence based on diferent punctuation cf. Sedley 2004, p. 60.
3 E.g. especially clever refutation (Fine 1998, pp. 208-209), the cleverer second phase [of the argument]
(McCabe 1994, p. 278), very subtle argument (Bostock 1988, p. 89), the subtlest argument (Polansky 1992, p.
130), most subtle argument (Sedley 2004, p. 60), really exquisite argument (Burnyeat 1976b, p. 177), most in-
genious argument (Waterlow 1977, p. 19), argument [] ingenious (Chappell 2006, p. 135), exquisite argu-
ment (e.g. Lee 1973, p. 243; Long 2004, p. 35, Wedin 2005, p. 171).
4 For the same construction cf. Phdr. 275d4, Chrm. 172b1. Gottlieb (1992, p. 189) seems to construe the text sim-
5 On this interpretation Socrates remark echoes his claim in the Euthydemus that the (Protagorean) ac-
cording to which it is impossible to contradict appears always (amazing) to him, because it over-
throws not just the other , but itself as well (286c3-4; cf. T2 on p. 21). Cf. also Euthd. 303d-e.
6 The proponents of sd are much more subtle (f ) than those uninitiates who do not admit
the existence of anything except what they can touch and see.
7 What seems the most subtle point () in Socrates dream (201d-202d) is that the elements are
unknowable but the complexes knowable.
32 luca castagnoli
Burnyeat, is that Plato wanted his readers to realise by themselves that they should re-
store the missing qualifers and to understand by themselves how the argument would
then work. In sect. 4.2.2 I provisionally adopted the same conjecture as the most plau-
sible in interpreting T7s dilemmatic argument; however, I had to admit that it is mys-
terious for me ( just as for Theodorus) how that argument can remain sound once we
have supplied the qualifers. Prima facie, adding the qualifers does not help us with T9
either: since Protagoras believes that (1) all men judge what is the case <for them>, he
must admit that (2) also the judgement of those who believe that md is false is true <for
them>, and thus that (3) md is false <for them>. One might argue that (3) is already a dan-
gerous admission for Protagoras to grant, for although it does not imply that md is ab-
solutely false, it does imply that md is not absolutely true. This is not surprising: md can-
not be an absolute truth if it falls into its own scope (something which, interestingly, is
never called into question in the Theaetetus)1 and thus each man must be the measure
also of his own being a measure and of other mens being measures. More surprising
is the fact that in the Theaetetus Protagoras does seem to consistently assert his md as an
absolute, unqualifed truth:2 Man is the measure of all things (152a2-3), I do say that
the truth is as I have written it (166c9-d1), You have to put up with being a measure,
whether you like it or not (167d3-4).3 Protagoras inescapable admission of (3) is sufcient
to compel him to abandon this blunt and inappropriate way of presenting his theory;
nonetheless, by granting that md is false for his opponents, albeit true for him,4 Pro-
tagoras does not seem committed yet to admitting in propria persona that md is false sim-
pliciter, since the idea that md must be true for everyone, or true absolutely, does not
seem to be part of md itself.5
1 Contra Sedley 2004, p. 48.
2 Thus violating the rule expressed at 160b8-c1. Analogously, sd is presented as an unqualifedly true ontology
(cf. Sedley 2004, p. 48), but this clashes with the relativism such an ontology is supposed to back and justify.
3 Protagoras book was entlitled Truth, without qualifcations.
4 Several interpreters stress the point that this concession is already sufcient to make Protagoras position ut-
terly uninteresting and solipsistic, and that this is the strongest lesson of Platos refutation of Protagoras, even if
Socrates argument fails to establish that Protagoras himself must admit the absolute falsehood of md in propria
persona (cf. e.g. Lee 1973; Waterfield 1973, p. 176; McDowell 1973, p. 71; Waterlow1977, pp. 35-36; Bostock 1988,
p. 95; McCabe 1994, p. 279; Chappell 2005, p. 114).
5 Burnyeat

s (1976b) contention is that this impression is misguided: once we have understood the import of
the admission that md is false for those who judge it false, we realise that Protagoras cannot concede this and at the
same time refuse to grant that md is false simpliciter. I cannot present and discuss Burnyeats infuential proposal
here, for which I refer to Castagnoli 2004a, pp. 15-18 (for recent criticism of Burnyeats reading along similar lines
cf. now Wedin 2005, pp. 174-178). One of my perplexities concerned Burnyeats use of the metaphor of private
worlds (something is true for x if and only if it is true in xs world), but I feel now that I failed to emphasise the re-
al crux of that use. By excluding the possibility that a world might be incorporated into another as incoherent,
solipsistic, and almost nonsense (1976b, p. 191), Burnyeat is denying the possibility to say that it is true for Protago-
ras that what Socrates believes is true for Socrates. In other terms, denying the possibility of private worlds embed-
ded into other private worlds is the same as denying the possibility of repeatable qualifers (the idea that Pro-
tagorean qualifers must be unrepeatable is argued for by Denyer 1991, pp. 90-94). This single-relativisation
assumption (no truth is or could be hierarchically relativised to two or more subjects) is used by Sedley (2004, pp.
57-62) to vindicate the soundness of Socrates argument: When Protagoras is forced to agree that his opponents
view is correct, the reason why this is not qualifed as correct for them is that his responses are establishing what
is true in his own world (2004, p. 61), since T9s argument is implicitly governed by What is the case for Protagoras
himself ? at 170e7 and double-relativisations are barred. (Sedley adopts the same strategy to explain the apparent
lack of qualifers in T7: he reads 0 at 170c2 with mss b and d, and takes the whole following dilem-
matic argument to be governed by this qualifer [How then shall we run the argument for Protagoras?], which bars
the addition of further qualifers.) I am not sure that in the absence of any explicit evidence for this single-relativi-
sation assumption in the Theaetetus we are entitled to supply it and to make it the core of Protagoras refutation
self-refutation arguments from democritus to augustine 33
The frst step towards the solution to our puzzle will consist in reinterpreting the cen-
tral lines of Socrates argument in T9 (171b4-9: On the other hand [] so it appears):
their exact function has been left unexplained by commentators until Emilsson, in a
thought-provoking article published in 1994, frst highlighted their crucial role. Since
these lines come immediately before the conclusion, which is introduced by (there-
fore), one would expect them to make some major contribution towards it. Emilsson
notices that it would be unnatural to describe Protagoras opponents spontaneous
assertion that they are not wrong in maintaining that md is false as the refusal to concede
() something (lines b4-5): in a dialogue context, as here, dont admit in-
deed suggests that Protagoras is supposed to have said something to which the opponents
refuse to give their assent, something which Plato fails to record (1994, p. 140). But what
might Protagoras have said? Certainly not that his opponents are mistaken (this would
amount to a disastrous admission that false judgements do exist);1 what Protagoras is
allowed to retort is that his opponents view on md is false for him(i.e. that md is true for
him). But his opponents do not concede this, do not concede that they are saying some-
thing which, although true for them, is also false for Protagoras: since they do not believe
md, the very concepts of relative truth and falsehood are unacceptable for them. And
Protagoras himself cannot help admitting (), in accordance with his own theo-
ry, that also this judgement is true (lines b7-8). True for whom? Surely only for them? No,
since according to Emilsson Protagoras has now been disarmed of his qualifers: the
objection that Platos argument depends on ignoring the qualifers is misplaced, since
the function of lines b4-5 is precisely to show that Protagoras opponents will not accept
the answer which the critics have thought available to him (1994, pp. 142-143).2
I shall try to improve on Emilssons idea by presenting a more straightforward read-
ing of lines b4-8 and by elucidating how exactly they are meant disarm Protagoras of
the qualifers. My interpretation relies on the adoption, at line b4, of the lectio
(mss b, d, and t) in lieu of (ms w), which we fnd printed in the most recent
edition of the Theaetetus3 and was accepted by Emilsson. This apparently minor choice
on the basis of an argument ex silentio (no double-relativisation occurs in the Theaetetus). On such an assumption,
any confict of judgments of which one is aware would commit md to blatant contradiction: if Protagoras judges
that Socrates is pale and also that Socrates believes of himself that he is tanned, then Socrates must be at the same
time pale and not pale (tanned) for Protagoras (i.e. in Protagoras world), since Protagoras cannot say that it is true
for him that it is true for Socrates that Socrates is tanned. Moreover, one might wonder why Plato should have failed
to emphasise the crucial point that double relativisations would be senseless. (More specifcally, Sedleys proposal
to read 0 at 170e7 as a Protagorean relativiser appears difcult, especially because Sedleys inter-
pretation requires us to believe that this relativiser does not govern the immediately following lines 170e7-171a1,
where other qualifers occur, begins governing the arguments from 171a1, including T9, and ceases governing the
argument again at 171c5 [T10 on p. 37], all of this without any signal in the text.) For recent and insightful discussion
and criticism of the single-relativisation assumption cf. also Chappell 2006, pp. 113-120. Chappels own positive
proposal is that Socrates is warranted in dropping the qualifers because Protagoras aim is to reduce truth to rela-
tive truth: the supposition that the properties of the analysandumtruthtransfer across to the analysanstruth for
leads Protagoras into the contradictory position of accepting that his philosophical opponents views may just be
described not just as true for them, but as true simpliciter (2006, p. 130). However, to refute Protagoras by saddling
him with such an estreme brand of reductionism seems to me no less question-begging than treating him as a sub-
jectivist while his position is a relativist one.
1 Cf. 161e2-3: we who are ourselves each the measure of his [sc. Protagoras] own wisdom.
2 Emilsson 1994, pp. 142-143.
3 Hickens edition in Duke, Hicken, Nicoll 1995 (cf. also Campbell 1861, McDowell 1973, Narcy 1995).
Wohlrab 1891, Burnet 1900, and Dis 1924 read (but Dis does not translate it). Among the interpreters,
only Polansky (1992) and Bemelmans (2002) adopt and read it as a Protagorean qualifer.
34 luca castagnoli
will allow us to reinterpret this portion of the exchange between Protagoras and his op-
ponents in a way resembling Emilssons without any need for supplying implicit steps
in the argument (notably, nothing in Platos Greek corresponds to Protagoras retort
that his opponents view on md is false for him).
By admitting that the opinion of his opponents about md is true (for them), Pro-
tagoras is conceding that md is false: false for them, of course (I am supplying here the
missing qualifers at 171a6-b3 like Burnyeat and Emilsson). But his opponents are not
content with this concession: they are not ready to grant this qualifcation, that he says
something which is false for them( ). My reading takes
(Protagoras) as the unstated subject of (cf., similarly, e
immediately before, at line b2): since, if there is no specifc subject of the infnitive then
the indefnite accusative idea takes over [] but such an indefnite or generic turn of
thought is often used when there is in fact a specifc reference within the context,1 the
most accurate translation would be the others do not concede that one says something
false for them,2 where one alludes to Protagoras, referred to immediately before. Al-
ternatively, it is not wild speculation that Plato might have originally written
e but was inadvertently dropped out of our manuscript tradition
at some early stage. The alternative lectio is suspect also on purely linguistic
grounds: if the subject of the infnitive is the same as the subject of the leading verb,
then, the proper accusatival subject is usually displaced by the nominative of the orig-
inal expression of the idea.3
But why are Protagoras opponents unwilling to accept Protagoras qualifcation that
it is for them that md is false? What they believe is that Protagoras, by advancing his md,
is saying something false simpliciter, and not only for them (or indeed for anyone else).4
On my interpretation at lines b4-5 the qualifer () fnally makes its appearance in
the text: unlike Emilsson, who makes qualifers a major issue in T9 in spite of their com-
plete absence, I ask the reader to intend them as implicitly meant only in its frst part (at
171a8, a9, b1, b2), where it is not so difcult to accept the integration given that they are
in their place both immediately above (170e9: ) and, on my reading of the text,
below (171b4: ).
What can Protagoras reply to his opponents refusal to qualify their denial of md? He
cannot protest that they are mistaken: according to md, he cannot help saying that their
1 Cooper 1998, p. 774.
2 Although in connection with Protagorean qualifers (to say something false for) is not used
anywhere else in the Theaetetus, we have encountered two sufciently close parallels at 170e4-5 (
and <>).
3 Cooper 1998, p. 771. Unfortunately, however, this cannot settle the question, since there are exceptions to the
general rule, some of which can be found in Plato.
As we have seen, Emilsson, who reads at line b4, must supply before it Protagoras implicit reply But
you must admit that your view is false for me. Emilsson examines the reading , but rejects it on the grounds
that not even in Protagorean language does there seem to be anything describable as being wrong for oneself
(1994, p. 139n8). This is correct, but Emilsson fails to see the possibility, which I defend here, of taking (Pro-
tagoras), and not (Protagoras opponents), as the subject of .
Bemelmans, who reads , translates lines b4-5 as follows: But the others dont concede that it is (true)
for themselves, that what he thinks is false (2002, p. 80). This is consistent with my reading, but I fnd the addition
of it is true unnecessary. Polanskys paraphrase is very similar to Bemelmans translation (1992, p. 131), but he
adds that Socrates withholding of the qualifying labels to him and to them is playfully unfair.
4 Textual support for this reading might come from 179b7-9, at which Theodorus, referring back to T9, says
that Protagoras is refuted also when it makes other peoples judgements authoritative, but they clearly
think that his theories are in no way () true.
self-refutation arguments from democritus to augustine 35
belief that md is not false for them, but simply false, is true (171b7-8). We would have ex-
pected a qualifed admission that this second-order belief too is true for them only; sure-
ly we must supply again the missing relativiser? If we did so we would end up with our
original problem unsolved: how could Socrates shift in the lines which immediately fol-
low (b10-c3) from Protagoras admission that it is true for his opponents that md is false
(and not only for them) to the devastating admission that md is unqualifedly false?
Emilssons insight must be correct: at lines b4-8 Protagoras is being disarmed of the
qualifers; the problem is to understand how. Lines b4-5 suggest that Protagoras oppo-
nents are not ready to accept any relativisation of the truth of their judgements; how-
ever, prima facie this does not seem to require that Protagoras himself gets rid of the
qualifers. For, apparently, he might continue relativising the truth and falsehood of his
opponents claims, without contesting their refusals to accept such relativisations: in
this way Protagoras would be giving up any hope to persuade his opponents, but at least
would steal a draw.1 Emilsson suggests that, since Protagoras never objects to his op-
ponents refusals to relativise the truth-value of their claims, his opponents could sim-
ply say Since you have no objection at all, Protagoras, we understand that you have
given your admission to our statement (1994, p. 145), i.e. pretend that Protagoras him-
self has admitted the unqualifed falsehood of md. However, this would be a rather un-
fair treatment of Protagoras dialectical behaviour: as long as he continues relativising
the truth of his opponents claims (something from which he has not yet been proved
to be barred), he should not be charged with endorsing them on the mere grounds that
he does not charge them with absolute falsehood (to require him to do this would be
only another way of begging the question against md).2
I suggest that, despite appearances, after his opponents refusal to accept relativisa-
tions described at lines b4-5 Protagoras cannot continue adding his qualifers. For let us
reconstruct in direct form how the entire dialectical exchange between Protagoras and
his opponents would unfold if Protagoras clinged to his qualifers:
Opponents: By advancing md you say something false.
Protagoras: I concede that this view of yours is true for you, since all men believe what is the
case for them.
Opponents: So your md is false.
Protagoras: I concede that its false for you.
Opponents: But we dont concede this to you. We dont believe that md is false for us: its false
Protagoras: I admit that also this belief of yours is true, for you. Its true for you that md is not
false for you, but false simpliciter.
Protagoras fnal remark is only apparently one more harmless admission of the rela-
tive truth of his opponents claim. Since It is true for you that md is false seems to
amount to the same as md is false for you, Protagoras attempt to remain faithful to
his md by conceding, once again, the relative truth of his opponents claim would turn
1 But in this way he would fail to qualify for wisdom in the sense of the superior skill to change others ap-
pearances (cf. p. 27). According to Bemelmans, continuing qualifying the truth of his opponents claims is not a
viable option because this would lead to an infnite regress (2002, p. 82), but he fails to explain why this regress
should be dangerous for Protagoras, or at any rate attributable to Protagoras more than to his opponents (they
stubbornly continue denying Protagoras qualifcations just as Protagoras continues qualifying their denials).
2 Waterlow (1977, p. 31) is liable to a similar objection.
36 luca castagnoli
out to be a straightforward and unqualifed denial of that claim, oddly presented as an
agreement (I admit):
Protagoras: I concede that md is false for you.
Opponents: But we dont concede this to you. We dont believe that md is false for us: its false.
Protagoras: I admit this too: md is false, for you.
On this occasion relativising would not be, ultimately, a way of granting a qualifed ac-
ceptance to someone elses belief, without endorsing it in propria persona, but a way of
openly contradicting it, thus denying md itself.1 Moreover, Protagoras would be repeat-
ing, in slightly diferent terms, exactly the same thing which he has already said just a
few seconds before and which he knows has not (and will not) be accepted by his op-
ponents. In other words, he would be merely babbling.2
It is for these reasons that after lines b4-8 Protagoras must give up his qualifers, willy
nilly, and admit that the belief of his opponents, according to which md is false, is true.3
This admission is taken by Socrates and Theodorus as evidence that everyone, includ-
ing Protagoras, disputes md (171b10-c4): the anti-md consensus is now universal. What
at 170e7-171a1 had appeared as a merely counterfactual possibility (if not even he him-
self thought that man is the measure) has turned out to be a necessary dialectical
outcome of the clash between Protagoras and his opponents. Even Protagoras will be
compelled to reject md when faced with disagreement. It is crucial to emphasise that
the result of the dialectical manoeuvre described in T9 is not (and cannot be) the
demonstration of the necessary (logical) falsehood of md, but Protagoras admission
of mds falsehood and consequent defeat by reversal.4 Grasping this point is funda-
mental to understanding the rationale of the next and fnal step of Socrates argument:
T10 Then, since it is disputed () by everyone, Protagoras truth is not true for any-
one, neither for anyone else, nor for himself. (171c5-7)
md is still assumed here, along with T9s conclusion that everyone disputes md, to draw
the further conclusion that md is not true for anyone (Protagoras included): if his pre-
vious argument had been recognised by Socrates as a sufcient proof of the mds ab-
solute falsehood, this fnal step would be unwarranted or at least redundant, and even
more so given that the conclusion that md is false for everyone sounds like a weaker
(and perhaps incompatible) one.
1 As McCabe notices, disagreement with everyone else is inaccessibile to him [sc. Protagoras]; all he can ever
do is agree (2000, p. 43); but in this case even qualifed agreement would be inaccessibile, since it would amount
to sheer disagreement, dooming Protagoras to inconsistency.
2 I adopt this term in the narrow sense in which Aristotle uses it at Top. viii 2, 158a25-28: Whoever keeps on
asking one thing for a long time is a bad inquirer. For if he does so though the person questioned keeps on an-
swering the question, clearly he asks a large number of questions, or else asks the same question a large number of
times: in the latter case he merely babbles (), in the former he fails to deduce. Protagoras would not be
asking the same question many times, but would be proposing the same relativised claim many times, thus
implicitly asking his opponents to concede it.
3 Could Protagoras say that their belief is false? Of course he could, but he would be thus unwittingly denying
md (cf. Dionysodorus downfall in sect. 4. 1).
4 I agree with Waterlow (1977, p. 27) that 171a6-c7 is not a proof of inconsistency. Some interpreters con-
strue T9s argument as a logical proof by Consequentia Mirabilis (cf. sect. 6. 1): if md is true, then it is false; there-
fore md is false (cf. e.g. Vailati 1904; Kneale 1957, p. 63), overlooking T9s dialectical context.
self-refutation arguments from democritus to augustine 37
4. 2. 5. One argument, two formulations
At the end of section 4.2.2 I suggested that the overall structure of the section 170a-171c
has been generally misunderstood by commentators: it is time to illustrate more clear-
ly my contention that T9s self-refutation argument must be interpreted as a clarifca-
tion of T7s dilemma. I argued that T7s dilemma needed to be supplemented with miss-
ing qualifers: although the addition was not too problematic in itself, it left us with an
argument whose frst premiss, (a*) (If men always judge what is true for them, then
men sometimes judge what is false), sounded unacceptable. T9 is designed to show
Theodorus and us why (a*) is justifed after all, by disclosing the most clever facet of
Protagoras md:
170c3-5 (a*) 171a6-c4
md men always judge what is true all men judge what is the case
<for them> (170c3) <for them> (171a9)
Anti-md (most) men believe that there are false (some) men (Protagoras
consensus judgements (since men believe in the opponents) believe that md is false
existence of ignorance [170a6-b8] and (171a7-8)
(most) men believe that ignorance is
false judgement [170b9-c1])
Inference left unexplained Inference explained at 171a6-b9
md is men sometimes judge what is false man is not the measure of anything
false (170c5) at all which he has not learnt (171c1-4)
The antecedent and the consequent of (a*) (md and its contradictory, respectively)
match, in content if not in exact wording, the initial premiss and the conclusion of T9s
argument. What might lead us to suppose that the mysterious reasoning underlying
(a*) cannot be the same as the one we fnd in T9 is that their additional premisses seem
to difer: (a*) relies on the widespread general belief in the existence of false judgement,
whereas T9s argument seems to appeal to the anti-Protagoreans specifc judgement
that md is false. However, this surface diference fades and almost disappears altogeth-
er as soon as one considers more carefully the nature of the anti-md consensus in T9:
although the opinion of those who judge the opposite (by which they think that he
[sc. Protagoras] says the false) suggests an explicitly verbalised and specifcally direct-
ed dissension against Protagoras, important clues show that the anti-md consensus
need not assume such a defnite form. At 170e8-9 we are told that the masses do not be-
lieve md, where the context makes it quite clear that this is intended as meaning that
they believe md to be false, and at 171b10-11 we fnd the bold conclusion that, since Pro-
tagoras is compelled to admit the falsehood of his md, everyone disputes it. Since it is
plausible that in the Athens of the beginning of the fourth century b.c. the vast ma-
jority of people never heard of Protagorean relativism, and, a fortiori, so much the less
bothered disputing it actively, Socrates must be adopting a generous notion of belief
here: you can be said to disbelieve, and even dispute, md even if you have never heard
of it, as long as you believe in the existence of false beliefs, which is inconsistent with
md.1 Furthermore, this belief in the existence of false beliefs does not need to be ex-
1 Pace Newman 1982, p. 49.
38 luca castagnoli
plicit itself: you can be said to have it even if you have never expressed it,1 or you do
not have any full-blown concept of belief at all, provided that your behaviour shows
this belief of yours. As we have seen above (p. 28), Socrates own reasons for attribut-
ing to people the belief in the existence of ignorance and false judgement are based on
the observation of their behaviour (in particular of their search for experts and lead-
ers). Since the belief that md is false is a corollary of the belief that false beliefs exist,
T9s argument is nothing less than the expected elucidation of why (a*) is true and thus
Socrates initial dilemma is sound: it is meant to satisfy Theodorus request for clarif-
cation at 170d3.2
A notable consequence follows from this analysis. T9s self-refutation argument is di-
alectical: it does not aim at proving that md is false, but at showing why even Protago-
ras is bound to admit the falsehood of his md when faced by someone who disagrees
(or would disagree if asked), and so why no one can really uphold md. Accordingly, T7s
dilemma must also be interpreted as a dialectical challenge to Protagoras (What then,
Protagoras, are we to make of your ?): whether he says ( [] ) that
md is true, or that it is false, he will be fnally compelled to admit that it is false, that
men do not always judge what is true <for them>. Socrates dilemma is not a proof
by cases of the necessary falsehood of md: if this should appear an unwelcome result
of my interpretation, it will be helpful to recall that Socrates declared aim was to ob-
tain Protagoras own agreement starting from his own (169e8e-170a1), and not to
demonstrate the falsehood of md.
4. 2. 6. Protagoras return
Faced with T10s unfortunate conclusion, Theodorus, who was supposed to defend his
dead friend (168e7-169a1) but could not help conceding all the steps which led to Pro-
tagoras dialectical rout, protests (171c8-9). Socrates reply is worth examining:
T11 Its likely, then, that he [sc. Protagoras], being older, is wiser than us; and if he suddenly
popped up here from below, as far as the neck, he would probably accuse () me of
talking a great deal of nonsense, and you of agreeing with it, and then he would duck down
again rushing of (f i ). But I think we have to take ourselves as
we are, and always say what seems to us to be the case (a d ).
This vivid image has attracted some attention in the literature, concerning not only its
pictorial details,3 but also its philosophical signifcance. What would Protagoras say
should he come back from Hades? And why would he rush of immediately after-
wards? According to Burnyeat, Protagoras might try to defend himself by insisting that
he presented md not as an absolute truth, but as something true for himself alone. By
this solipsistic move Protagoras would be refusing to enter fully into a common world
with his opponents for genuine discussion, and Platos image would represent this at-
titude: coming from and retreating to another world from ours, he [sc. Protagoras]
does not really leave the underworld (1976b, p. 193n23). However, this would look more
like a way of defending md by clarifying and narrowing it than a way of accusing Socrates
1 Cf. Ketchum 1992, p. 78.
2 Theodorus seems to understand this new formulation of the argument without difculty.
3 Cf. Burnyeat 1976b, p. 192n23; Ford 1994.
self-refutation arguments from democritus to augustine 39
of talking nonsense. In a similar vein, Lee conjectures that Protagoras would try to es-
cape the exquisite argument by restoring the qualifers that Socrates deliberately and
unjustly left out, but at the high price of showing us that he himself is not asserting
anything we can or should take seriously (1973, p. 248),1 and thus he would be no
better than a vegetable sprouting up from the ground (p. 252). This sounds implausi-
ble: frst, a plant could hardly be depicted as running away; second, like the imaginary
opponents of Aristotle in Metaphysics and unlike the returned Protagoras, plants do
not speak at all.2
I suggest a more literal interpretation of T11: by attacking and trying to refute
() Socrates, presumably protesting that his argument was unsound,3 Protago-
ras is at the same time refuting himself, because his md is inconsistent with the possi-
bility of someone being mistaken and someone else proving that he says something
false.4 Protagoras is trying to do (it does not matter whether successfully) something
which, according to his own doctrine, is impossible, thus betraying the fact that he him-
self, at the end of the day, does not really believe and live his md. Protagoras does not
choose to rush of, but must rush of: his ducking down can be interpreted as the
metaphorical counterpart of the further reversal which he involuntarily incurs by at-
tacking Socrates.
As Socrates had previously remarked, md is utterly incompatible with dialectic and
refutation5 ( just as it is incompatible with teaching, at least as ordinarily intended):6
T12 I say nothing about my own case and my art of midwifery, and how much ridicule we in-
cur; and I think the same goes for the whole business of dialectic (). For must-
nt it be a long and enormous nonsense to examine and try to refute () one anoth-
1 For similar views of what qualifying md as true only for Protagoras would amount to cf. the literature list-
ed on p. 33n4.
2 For analysis of other interpretations of T11 cf. Castagnoli 2004a, p. 30n146.
Cf. also Bemelmans 2002, p. 83: Socrates alludes in this passage to the option he did not choose for Protago-
ras in the self-refutation argument []. The self-refutation would not have succeeded, if Socrates had let Pro-
tagoras hold on to qualifcation. With this choice, however, Protagoras would have deprived the others of their
status of measure.
According to McCabe, the image is meant instead to express the doubt whether Protagoras who has Hera-
clitean leanings could persist long enough, or with enough concreteness, to engage in conversation and defend
himself (2000, p. 47). I do not see any generic question of personal identity at stake here; there seems to exist a di-
rect link between what the returned Protagoras says and his sudden ducking down (Protagoras does not dissolve
while rushing of).
Ford argues that Protagoras only possibile reply is simply to restate his kephalaion, the head statement of his
book: Protagoras head [] runs of because it will still survive as a kind of pheme (saying). Dismissed but not
obliterated, the dead thinkers saying is so well known, even notorious, that it is likely to pop up elsewhere among
others interested in philosophy, and vex them with the same difcult words (1994, pp. 204-205). This reading also
ignores the point that Protagoras head is supposed to try to refute Socrates and Theodorus, and not simply to re-
state his position.
Other commentators do not try to conjecture what Protagoras rejoinder might have been, but suppose Pro-
tagoras rushes of because he has no good point to make (cf. e.g. Cornford 1935, p. 80; Polansky 1992, p. 132;
Emilsson 1994, p. 144n14; Lee 2005, p. 56, who speaks of sterile stubborness). McDowell (1973, p. 171) believes
that the image of Protagoras return could indicate that Plato is not content with his argument (possibly because
of the lack of qualifers). Narcy (1986, p. 81) too believes that the image indicates Platos dissatisfaction with his
argument, but for diferent reasons (cf. p. 41n1).
3 The same kind of complaint had already been voiced, not without irony, at 166a-c, where Protagoras had
lamented Socrates use of unfair verbal traps. Protagoras return had been foreshadowed at 169c8-d1.
4 Cf. Waterlow 1977, pp. 28-29; Gottlieb 1992, p. 190. A similar interpretation is proposed but discarded by
Burnyeat (1976b, p. 191). 5 This important point is nicely stressed by Long 2004.
6 Cf. Tht. 161d8-e1.
40 luca castagnoli
ers appearances and judgements, when everyones are correct if Protagoras truth is true
[]?1 (161e4-162a2)
It should come as no surprise then that md is dialectically untenable: any attempt to de-
fend it inevitably results in an involuntary admission of its falsehood. This makes T11,
rather than T9,2 very similar to the Euthydemus passage of section 4.1: the
thesis was considered self-refuting in a dialectical context because its defend-
ers were obliged, by the nature itself of that context, to try to throw their opponents
arguments down, thereby overthrowing their own position too, by implicitly conceding
that false judgement and contradiction do exist. Both in the Theaetetus and in the Euthy-
demus it is absolutely immaterial whether Socrates attacks are backed by sound argu-
ments or poor reasoning:3 as Socrates mockingly reminds us in T11, it is Protagoras him-
self who guarantees his adversaries the right of always saying whatever passes through
their mind. In the light of these considerations, Burnyeats motivation for discarding the
hypothesis that Socrates purposely begs the question against Protagoras by omitting the
qualifers4 weakens: by such a move Socrates would not be displaying perverse dishon-
esty, but showing how hopeless Protagoras md turns out to be as soon as submitted to
dialectical scrutiny. Either Protagoras remains silent, or he angrily protests that Socrates
is misrepresenting his doctrine and proposing an incorrect refutation (thus depriving
Socrates of the status of measure as far as mds exegesis and logical soundness are con-
cerned), thus contradicting md itself:5 in either case, Protagoras is the loser, because he
cannot downthrow adversaries and arguments which anyone else could have charged
with perverse dishonesty.6 Even an apparently shameless move would establish a pow-
erful philosophical point: although I have provided a reconstruction of Socrates argu-
ment which does not appeal to the absence of qualifers, one might suppose that Platos
undeniable ambiguity, in particular in the elliptical T7, is deliberate.7 Also an irrelevant
charge can be lethal for one endorsing Protagoras position.8
1 Socrates fails to consider the possibility that Protagoras chooses a radically diferent form of dialectic, a ther-
apeutic dialectic in which the sophist argues to change his interlocutors beliefs not from false to true, but from
harmful to benefcial (cf. p. 27). According to Narcy (1986, pp. 80-81; 1995, pp. 93-100) the self-refutation argument
is weak and eristic exactly because it fails to consider this possibility. Notice, however, that this line of defence
could be countered by Socrates later argument for the non-relativity of expert predictions about the future and
what is benefcial (177d-179b). 2 Cf. p. 24n2.
3 Just as it is immaterial whether Protagoras and Euthydemus responses, respectively, really undermine
Socrates arguments or only purport to undermine them. 4 Cf. p. 29.
5 Cf. the dialectical manoeuvre described by Aristotle at se 15, 174a20-23, in a chapter in which he lists various
methods which help one to refute ones adversary more easily: Moreover, there are anger and contentiousness,
for when agitated everybody is less able to be on his guard; elementary rules for producing anger are to make it
clear that one wishes to be unfair and to be completely shameless.
6 It is then quite ironic that Protagoras Truth came to be known in antiquity also under the title The down-
throwers (cf. p. 23n1). The use of the wrestling metaphor for dialectical argumentation is widespread also in the
Theaetetus (cf. e.g. 162b, 166b1, 167e6, 169a-b).
7 Along these lines, one might argue that Socrates insisted that Theodorus participated, in place of Theaete-
tus, to the examination of md (168c-169c) because of his awareness that Theodorus, unlike the philosophically more
gifted Theaetetus, would fail to notice the fallacious drop of the qualifers, thus compelling Protagoras to return
and defend himself in propria persona, with the disastrous consequences just explained. Notice that Theaetetus had
already been made aware of the importance of dealing carefully with qualifers in his discussion of previous ob-
jections to Protagoras (cf. p. 29n2), especially at 165b-c. For the related but diferent idea that the relativisers can be
dropped in the self-refutation argument because Theodorus is Socrates interlocutor, and a sober mathematician
would be the last person we would expect to take exception to the use of unrelativised terms cf. Long 2004, p. 36.
8 In the light of what we have seen in this section, I cannot agree with Barnes comment that in Plato the di-
alogue form is extrinsic in this sense: Platos arguments can all be turned into monologues without any logical or
philosophical loss (Barnes 2003, p. 28, italics mine).
self-refutation arguments from democritus to augustine 41
5. Aristotle: Speaking to Antiphasis
The most conspicuous feature of Aristotles attitude towards self-refutation is probably
his apparently scarce interest: in nearly 1,500 Bekker pages of Greek prose often stufed
with dense argumentation, only a bunch of serious candidates to the role of self-refu-
tation arguments leap to the eye. Nevertheless, since most of them target theses be-
longing to the broad family we are currently investigating, our inquiry will not lack an
Aristotelian section.
5. 1. Metaph. 4, 1008a27-30
In the war he wages in Metaphysics against whoever denies pnc (hereafter, also An-
tiphasis),1 starting from 1007b18 Aristotle focuses on the thesis that the contradicto-
ries are all simultaneously true of the same thing, i.e. anything afrmed may also be
denied and anything denied may equally be afrmed (1008a12-13):
(1) (x)(P)(Px Px).2
Having argued that this thesis commits its supporters to the view that everything is
one (1007b18-1008a2) and to a breakdown of the principle of excluded middle (1008a2-
7), Aristotle considers the issue whether, according to it, one could truly state the two
members of the contradictory pair separately or only their conjunction, and shows that
in either case unacceptable consequences follow. Aristotle depicts the unwelcome con-
sequences of the frst option as follows:
T13 Equally, even if it is possible to say the truth <in stating afrmations and denials> sepa-
rately it follows [] that (a) everyone will say the truth and everyone will say the false, and
(b) he [sc. Antiphasis] himself admits () that he is saying something false. ( 4,
Let us consider frst consequence (a). Why if, for any predicate P and any thing x, one
could truly afrm Px and truly deny it (i.e. assert Px), would everyone say the truth
and at the same time say the false? Would not everyone simply say the truth? Aristotle
must be tacitly applying here the basic semantic principle according to which whenev-
er an afrmation is true the corresponding denial must be false, and vice versa:
(2) (x)(P)(TPx FPx).
Suppose that both Pn and Pn are true; Pn will be true, by hypothesis, and at the same
time false, because of the truth of its contradictory, Pn (for (2) and double negation).
On the basis of this simple reasoning, it is easy to see that if the contradictories are all
simultaneously true of the same thing whatever one can say will be both true and false,
and thus everyone will be both right and wrong about everything. This entire train of
reasoning is crystal-clear: is it also sound? (2), which is the real, albeit implicit, pivot of
Aristotles argument in T13(a), looks unimpeachable: far from being a contentious prin-
1 Cf. Dancy 1975. For in-depth discussion of Aristotles views on contradiction, pnc and related principles cf.
Cavini 2007 and 2008.
2 (1), together with the principle of excluded middle, entails (x)(P)(PxPx), i.e. the contrary (and not simply
the contradictory) of pnc. For discussion of this curious turn in the argument of cf. Wedin 2003.
42 luca castagnoli
ciple, it can be described as a corollary of the classical defnition of truth and falsehood
which Aristotle himself will provide in 7:
T14 For to say that that which is is not or that which is not is, is false; and to say that that which
is is and that which is not is not, is true. (1011b26-27)
If Pn is true, it must be so because it says that n (which is P) is P; but then Pn must be
false, since it says that n (which is P) is not P.1
However unproblematic (2) might be in itself, I suggest that Aristotles use of it could
be questioned in this context. One who supposes that all contradictories are simulta-
neously true, like Antiphasis here, is thereby likely to be asking us, if implicitly, to brack-
et the validity of a principle like (2). Otherwise, it would have been reasonable for him
to claim that all afrmations and denials are at the same time true and false, and chari-
ty demands that we attribute him a position which is at least prima facie internally con-
sistent. Given what I have explained about the intimate link between (2) and Aristotles
defnition of true and false in T14, we must suppose that Antiphasis is also asking us to
revise radically our (for him) misguided conception of these notions. One might then
argue that (2) cannot be tacitly presupposed in any argument against him, insofar as such
an argument aims at being dialectical (as T13s fnal clause e e
suggests), and thus the opponents defeat should be a consequence of his own thesis
alone, or at least of views upon which he has agreed or would certainly agree. Aristot-
le might be accused of begging the question and allowing his petitio principii to sneak
unnoticed into his argument in T13(a) in the form of (2).
That Aristotle could hardly have failed to be aware of such a possible rejoinder is tes-
tifed by a passage occurring only a few lines below T13:
T15 if whenever the afrmation is true the denial is false, and when the latter is true the afr-
mation is false, there can be no such thing as simultaneously afrming and denying the
same thing truly. However, they would probably assert that this is the issue originally posed.
( 4, 1008a34-b2)
Here Aristotle formulates the principle which I have labelled (2), and claims that pnc
can be inferred from it.2 However, he comments that the deniers of pnc would proba-
bly reject his argument and complain that it begs the question, because (2), in a sense,
is nothing else than what Aristotle posed as the thesis he wanted to defend, pnc itself.
1 As Walter Cavini suggested me, one might object that, if taken literally, T14s defnition actually expresses
only sufcient conditions for (saying the) true and (saying the) false, i.e. the rules of semantic ascent
Pn TPn Pn TPn
Pn FPn Pn FPn
and not necessary and sufcient conditions (i.e. equivalences), and that the corresponding rules of semantic
TPn Pn TPn Pn
FPn Pn FPn Pn
and thereby the equivalences can be extracted only from Cat. 12, 14b15-20. It should be noticed, however, that at
Metaph. 8, 1012b5-11 Aristotle himself seems to treat a variant of (2) as fully equivalent to his previous account
of what false and true signify. For a recent analysis of T14 with extensive bibliographical references cf. Criv-
elli 2004b, pp. 132-136.
2 I do not see any reason for interpreting T15 as presenting an indirect argument, along Wedins proposal (2000,
pp. 160-161): if pnc were false, then (2) would be false, but this is absurd, and thus the denial of pnc is absurd too.
I see even less reasons for describing the passage as arguing that the argument may be self-refuting (p. 159) or
that the argument may be self-defeating (p. 162) (where the argument is, I suppose, Antiphasis denial of pnc).
self-refutation arguments from democritus to augustine 43
To be precise, Aristotle does not admit that his opponents would be justifed in com-
plaining, but his hidden reliance on (2) in T13 could seem liable to the same charge as
his explicit assumption of it in T15.1 The fact that, to my knowledge, no commentator,
apart from Alexander of Aphrodisias, noticed Aristotles tacit, and possibly question-
begging, manoeuvre in T132 shows how difcult it is to identify certain unstated as-
sumptions when these are so deeply rooted in our ordinary practice of inference, and
actually constitutive of it. These are assumptions which usually we do not need to make
explicit and we do not challenge, but in Metaphysics it is some such assumptions and
their consequences that face Antiphasis challenge, and so what is not question-begging
in other contexts risks becoming so here.
Also T13(b), i.e. Antiphasis self-refutation by the admission that his own thesis is
false, crucially relies on the application of (2). First, the supporter of the thesis that
whenever the afrmation is true the denial is also true will be forced to agree that the
denial of his thesis is true as well.3 But then, as Alexander remarks,
T16 by his own admission that the denial of the afrmation which he himself posited [] is
true, he admits that he himself says what is false (e ). (in Metaph.
296, 19-21)
The fnal step of the argument is not the one based on the self-application of Antipha-
sis thesis; it relies, once again, on the further application of (2) to the result of that self-
application. In the absence of this extra step, Antiphasis, who defends the contradicto-
ry4 of pnc, could be easily led to admit the truth of pnc too: he could thus be accused
of having omitted some relevant truth, but not of having afrmed something he him-
self must admit to be false.5 Emphasising that Antiphasis, the champion of contradic-
tion, is committed to contradictory positions also about pnc is not necessarily a knock-
out blow. Lear noticed that Aristotles arguments in Metaphysics are
constructed so as to reveal to us that Aristotles opponent is in a contradictory position. [] Ar-
istotle is not trying to persuade him [sc. Antiphasis]: the argument is for our sake, not for his.
(Lear 1980, p. 113)
While agreeing on the essence of Lears comment, I signal that it overlooks two im-
portant details. First, Aristotle believes that those who deny pnc as a result of philo-
sophical confusion (caused, for example, by the fux of perceptibles or conficting ap-
1 Subsequently Aristotle shows no qualms about arguing, in a similar vein, for the principle of excluded mid-
dle starting from T14s defnition of truth and falsehood ( 7, 1011b25-29) and against the thesis that everything is
false starting from a variant of (2) ( 8, 1012b5-11). These diferent attitudes might depend on the diferent degrees
of similarity between the disguised question-begging premiss and the conclusion, on the presence and number of
additional premisses involved, and on the number of deductive steps needed to remove the disguise (cf. Dancy
1975, p. 20). For Aristotles views on begging the question cf. top. viii 13; se 5, 6, 27; APr ii 16, 64b33-38; see Schreiber
2003, pp. 98-106. The concern to avoid begging the question against Antiphasis (or at least to avoid appearing to
do so) by choosing a strategy, the elenctic proof , immune to this charge, was a priority in Aristotles agenda from
the very beginning of 4.
2 Alexander, commenting on T15, remarks that Aristotle has implicitly used this argument already, when he
said that all were in error (297, 11), clearly referring to T13.
3 Aristotle does not take into account here the possibility that Antiphasis might want to exempt his higher or-
der generalisation from self-application (whenever the afrmation is true the denial is also true, except for this
very afrmation, whose denial is not true). For Aristotles awareness of the possibility of such a move cf. the end
of sect. 5. 2. 4 Or even the contrary of pnc, depending on the diferent occasions (cf. p. 42n2).
5 Moreover, there would be an answer available to him: I omitted to state and defend pnc because, unlike its
denial, it is something which is already so widely accepted, as you yourself suggest, that it does not need to be
advocated by me or anyone else.
44 luca castagnoli
pearances) can and must be convinced, and doubtless a way of achieving this purpose
consists in fully spelling out all the bizarre consequences of their denial ( 5, 1009a17-
20). Perhaps the double consequence of T13 is not the most impressive within Aristo-
tles overall enterprise of Metaphysics 4-8, but it could still contribute to this peda-
gogical aim. Second, Lear undervalues the specifc force of the admission that ones
position is false. The fact that Antiphasis should be ready to accept contradiction does
not automatically imply that he will be happy to admit that his view is false. The extra
step taken by Aristotle in T13 thanks to his tacit application of (2) is not unimportant:
ones admission that one is in error is much more embarrassing than ones admission
that ones adversary is right too, and so much more so for a character like Antiphasis,
whose position has already been assimilated to Protagoras at the beginning of 5 and
whose denial of pnc could therefore be read as a companion to the now familiar theses
that everything is true and falsehood and error are impossible.
This is why Aristotles strategy in T13 is, at the end of the day, less airtight than it
might appear: we are examining a peculiar case in which forcing ones adversary to con-
cede the contradictory of his own thesis is not a sufciently clear-cut victory, and An-
tiphasis stronger confession that he is in error can be obtained only by relying, surrep-
titiously, on an assumption which he might want to reject. But at what cost? Could he
really be so bold to complain that Aristotle is guilty of petitio principii and refuse to grant
(2)? So far I have assumed he could, and I confrm that this is theoretically correct. But,
as I have already noticed, by rejecting (2) Antiphasis would be rejecting, ultimately, that
commonsense notion of truth and falsehood which is conveyed, for example, in T14. If,
on the one hand, Aristotle risks begging the question in his struggle with Antiphasis,
Antiphasis, on the other, seems to be liable to the charge of changing the subject if he
stubbornly refuses to grant at least some minimal features of our (and Aristotles)
semantic notions. Of course he is free to provide an alternative account for truth and
falsehood; however, it cannot be so diferent from ours as to obscure the fact that he is
trying to give an extremely revisionary account for those very things which we call
truth and falsehood.
Does Aristotle tacitly avail himself of (2) in T13 because he is confdent that his op-
ponent could hardly dare reject it? Given Aristotles caution in T15, we cannot exclude
that he simply hoped his move would pass unnoticed (what indeed happened).1 How-
ever, this does not mean that Antiphasis emerges in very good shape from T13s twofold
attack: his only possible line of defence would make his thesis even more unpalatable,
by requiring him to uncover and spell out certain consequences of his position on truth
and falsehood which make it almost unintelligible.
5. 2. Metaph. 8, 1012b13-22
Aristotles most manifest adoption of the self-refutation charge against Antiphasis
occurs almost at the end of Metaphysics :
T17 Indeed all such theses2 are exposed to the stock objection (e ) that they elim-
inate themselves (f f ). For anyone who says that everything is true al-
1 Cf. however p. 44n1.
2 Ross (1924) translation of as arguments is incorrect: as Aristotles subsequent explanation clarifes,
it is the statements Everything is true and Everything is false that eliminate themselves (contra also Cassin,
Narcy 1989: argumentations; DAgostini 2002, p. 17: ragionamento).
self-refutation arguments from democritus to augustine 45
so makes the thesis contrary to his own true, so that his own is not true (for the contrary
thesis denies that his is true), and anyone who says that everything is false <makes> him-
self also <false>. And if they make exceptions the former that only the contrary <of his
own thesis> is not true, the latter that only his own is not false they will end up none the
less begging an infnite number of theses, true and false; for he who says that the true the-
sis is true says something true, and this will go on to infnity. ( 8, 1012b13-22)
Whereas T13s consequence (b), which I have reconstructed as a self-refutation charge,
was not highlighted as such, Aristotle refers here to the strategy employed as
(expressed over and over), thus hinting at its previous history, and de-
scribes its thrust in wholly general terms: showing that certain eliminate
themselves (or, more tragically, kill themselves, are suicidal). Aristotles use of
with reference to this charge captured Narcys attention:
Platon, corrlativement, nest pas nomm: Aristote nestime pas que la mention de cet argument
mrite plus que lanonymat du lieu commun. Le terme dont il le dsigne, to thruloumenon (1012
b 14), sinscrit [] dans le registre du bavardage []. Cette faon de mentionner largument pla-
tonicien nest pas proprement parler invalidante: dun tel lieu commun il peut loccasion tre
commode se servir [], et cest pourquoi Aristote lajoute ici son arsenal; mais elle est coup
sr dprciative. Largument platonicien nest rappel que pour mmoire, en quelque sorte,
moyen vulgaire de rfutation. (Narcy 1989, p. 81)
I agree that Aristotle shows no special enthusiasm for the two self-refutation argu-
ments of T17.1 Certainly their position and role within the overall architecture of Meta-
physics appear to be peripheral when compared with other arguments, in particular
with the elenctic proof of pnc in 4. The self-refutation argument indisputably plays
a more crucial part in Platos Theaetetus, but there too it does not occur in a climactic
position and I have argued that Plato himself, contrary to what is usually supposed,
does nothing to highlight it as especially clever. I suggest that, pace Narcy, Aristotles
way of introducing T17s self-refutation arguments does not necessarily refect a veiled
criticism of Platos strategy.2 Even if we understood in its most pejorative
nuance (babbled over and over), which is by no means necessary given Aristotles own
usage of the term,3 the fact that at the time in which Aristotle wrote 8 the self-refu-
tation objection seemed quite trite does not imply that he considered the achievement
of those who frst devised and used it trivial, or the force of the objection itself ques-
tionable. More importantly, calling an argument commonplace and then going on to
deploy it successfully and show that it thoroughly defeats ones opponent is not so
much a way of devaluing the argument itself as a way of increasing the humiliation of
that opponent. If your thesis is so hopeless that even a hackneyed argument is suf-
cient to subvert it, then your defeat is certainly less honourable than it would be if an
ingenious, and possibly controversial, refutation were needed to beat you: At least
since the time of Democritus and Plato everyone knows that certain theses are suici-
dal,4 and you still wander around trying to sell them to us? From a rhetorical point of
view, Aristotles use of is no less efective than the argument itself which
he thus labels.
1 Cf. also Wedin (2005, p. 172), who speaks of Aristotles less sanguine attitude [sc. than Platos] about the
prospects for self-refutation arguments. 2 Contra also Cassin 1993, p. 535.
3 Cf. Ath. 16.7.4; ha ix 13, 615b24; ix 37, 620b11; ga iii 5, 756b6; Metaph. M 1, 1076a28; Rhet. ii 21, 1395a10; iii 7,
1408b2; iii 14, 1415a2. 4 Cf. T20 in sect. 6. 1, p. 50.
46 luca castagnoli
So far I have discussed some clues to the Aristotelian attitude towards self-refutation
emerging from T17; I shall focus now on what T17 teaches us about the logic of self-
refutation. Once again, what is subject to self-elimination is not abstract propositions,
but statements along with their proponents (, ). Aristotle is not saying that
if everything were true then also It is false that everything is true would be true and,
therefore, the proposition Everything is true is necessarily false: the fnal, crucial infer-
ence is missing.1 Aristotles reference to the asserter of this thesis would be irrelevant
if his aim were to prove something about its propositional content. What Aristotle
claims is that whoever says that everything is true makes the contrary of his thesis true
too: since asserting something is not a sufcient condition for making it true (except for
a few special cases which are not at stake presently), Aristotle must mean that whoev-
er says that everything is true is thereby committing himself also to the truth of the con-
trary of his own thesis,2 and thus can be forced to admit the falsehood of his own (so
that his own is not true). The nature of the argument as a purely dialectical silencer of
ones opponent is confrmed by the case of Everything is false: he who says that every-
thing is false makes himself also false, i.e. unwittingly concedes, by self-application,3
that what he is saying must be false too (again, the further conclusion Therefore it is
false that everything is false is missing). The proposal that nothing is true is thus self-
defeating and suicidal: this is all that Aristotle seems to be interested in establishing
through the self-elimination stock objection. This is not to say that Aristotle did not
believe that the propositions Everything is true and Everything is false are false, or
that he thought he could not prove (at least in some weak sense of proving) their false-
hood, but only that this is not what T17 purports to establish. This is a welcome dis-
covery, since, as we have learnt, without frst considering what the truth-bearers are and
what their truth-value is,4 all one might prove is that Everything is false is either false
or liar-paradoxical.
One fnal noteworthy aspect of T17 is Aristotles mention and criticism of the at-
tempt to elude the self-elimination charge by making some exceptions, i.e. by limiting
the scope of the universal everything. To my knowledge, this is the frst discussion of
this strategy to be found in the ancient texts: what was strikingly absent from the
Theaetetus was precisely the suggestion that Protagoras might try to defend himself by
claiming that man is the measure of all things, apart from mans being or not being a
1 Pace DAgostini 2002, p. 17. This extra inference would be required to have what in logical jargon is often
called Consequentia Mirabilis. I shall argue in sect. 6. 1 that Consequentia Mirabilis might have been unavailable to
2 It is not clear whether the contrary () of his own thesis is supposed to be Something is false, Every-
thing is false, or, more probably, given Aristotles own wording, It is not true that everything is true.
3 From this point of view, the present argument difers from Mackies proof based on T-prefxability, pace De
Praetere (1993, p. 356).
4 For detailed discussion of Aristotles complex position on the bearers of truth and falsehood (including,
among the others, sentence-tokens, thought-tokens, and perhaps states of afairs and sentence-types) cf. Nuchel-
mans 1973, pp. 23-44 and Crivelli 2004b, pp. 45-76.
5 Contra Sedley 2004, p. 48. Aristotles criticism of this manoeuvre is based on the infnite repeatability of the
true and false predicates, which would commit the deniers of truth and falsehood to conceding anyway infnitely
many truths and falsehoods respectively. For the analysis of the exception move in two later self-refutation
contexts cf. Burnyeat 1997, Castagnoli 2000, pp. 303-306.
self-refutation arguments from democritus to augustine 47
5. 3. Metaph. K 5, 1062a36-b11
Immediately before the passage we shall analyse in this section, the author of K1 re-
marks that although there can be no proper proof () of pnc and similar prin-
ciples, an ad hominem (e ) proof against those who deny them is possible
(1062a30-31): for example, if one had questioned () Heraclitus in the right way,
one might have forced him to admit () the truth of pnc, since he had adopt-
ed his opinion without really understanding what he was saying (1062a31-35).2 We fnd
then two arguments in a sequence. Let us begin by examining the frst one:
T18 And in general if what is said by him [sc. Heraclitus] is true, not even this itself will be true,
i.e. that the same thing can at one and the same time both be and not be. For as, when they
are separated, the afrmation is no more true than the denial, in the same way their com-
bination and conjunction being like a single afrmation the whole thing put forward as
an afrmation will be no more true than its denial. (K 5, 1062a36-b7)
Unlike the other cases we have examined so far, no dialectical context is obviously pre-
supposed here: what T18 does say is that if Heraclitus thesis is true, then it is not true,
and not that if one endorses the truth of Heraclitus thesis, then one must admit it to
be false. Surely we have found, fnally, an unambiguous ancient example of a proof, by
self-refutation, of the logical falsehood of a proposition? Such a diagnosis would be
hasty. We should not overlook T18s broader context, and in particular Aristotles ob-
servation, only a few lines above, that if one had questioned Heraclitus in the right way
one might have compelled him to accept pnc. Must the conditional if what is said by
him is true, not even this itself will be true be understood as the pivotal premiss of an
implicit constructive dilemma argument along the following lines
(1) If Heraclitus thesis is true, it is not true; pp
<2> <If Heraclitus thesis is not true, it is not true;> pp
<3> <Either Heraclitus thesis is true or it is not true;> pp
<4> <Therefore, Heraclitus thesis is not true.> p
or, once again, as an elliptical reminder of the dialectically self-defeating nature of Her-
aclitus position? Undeniably, neither the unproblematic premisses <2> and <3> nor,
more importantly, the alleged conclusion <4> appear in the text.3
T18 is followed by what sounds like a second self-refutation argument:
T19 Furthermore, if it is not possible to afrm anything truly, this itself will be false, the asser-
tion that there is no true afrmation. But if some true afrmation exists, this would refute
what is said by those who raise such objections and utterly destroy dialectic (e
). (1062b7-11)
1 There is some doubt as to the authorship and chronology of Metaphysics K(cf. e.g. Jaeger 1934, pp. 208-219;
Aubenque 1983). I shall analyse some Metaphysics Karguments without any commitment to their authorship and
2 This is not in accordance with the typical current usage of informal fallacies theories, in which an ad hominem
argument is an attempt to impugn ones opponents views or arguments by somehow denigrating his person,
character, and motivations (abusive ad hominem).
3 The latter fact prevents us also from construing T18s argument as a Consequentia Mirabilis of the form
(pp)p. I shall argue in sect. 6. 1 that Consequentia Mirabilis might have been unavailable to Aristotle.
48 luca castagnoli
The comments just made on T18 hold good, mutatis mutandis, here. It is hard to decide
whether the argument is best interpreted as a dialectical silencer of Antiphasis or as a
proof of the necessary falsehood of the propositional content of Antiphasis assertion
that no afrmation is true. This sort of indeterminacy often occurs when one tries to
cast arguments formulated in a natural language into some precise logical form. How-
ever, the context and especially T19s fnal remark about the incompatibility of An-
tiphasis objection with the practice of dialectic make me inclined to favour, again, the
frst interpretative line. This tentative approval might also be backed by familiar logical
considerations: it would be incorrect to conclude that the afrmation that there is no
true afrmation must be false, solely on the basis of its refexivity, for supposing that in
fact no other afrmation were true, No afrmation is true would be liar-paradoxical,
in a way similar to Epimenides All Cretans say the false, and thus it would be hasty
to call it unreservedly false. The only conclusion one can safely draw about the truth-
value of No afrmation is true without inspecting what afrmations there are and what
their truth-value is is, once again, that No afrmation is true is either false or liar-para-
doxical. Although we are not sufciently informed about the Aristotelian attitude
towards the Liar,1 charity invites us to read T19 in a way which does not commit its
author to an incorrect, or at best superfcial, approach to the thorny issues raised by this
paradox and its cognates.2
6. Introducing : Sextus Empiricus
In the previous sections we have analysed several ancient arguments denouncing cer-
tain extremist views on truth and falsehood as hopelessly self-defeating, and detected
numerous interesting analogies in their underlying logic. The passages we shall con-
sider in this section, all from Sextus Empiricus corpus, both testify to the continuous
survival, and indeed fourishing, of the same argumentative pattern over the fve cen-
turies which separate Aristotle and Sextus and refect a more precise awareness of its
distinctiveness, starting from the frequent adoption of a semi-technical vocabulary to
label it: the verb and its substantival form .
1 Crivelli argues that Aristotle does refer to the Liar at se 25, 180a34-b7 and attempts to solve it by assuming
that someone uttering I am speaking falsely (or whatever sentence-type the paradox turns on) is neither speak-
ing truly nor speaking falsely absolutely (2004a, p. 61; 2004b, p. 31). If this were correct (but cf. now Cavini 2007,
pp. 127-128), my argument for not interpreting T19 as a logical proof would fnd some further corroboration
(provided, of course, that the author of K is Aristotle or someone who shares the same views on the Liar).
2 There are at least two other arguments in defence of pnc often referred to in the literature as Aristotelian
self-refutations: the short argument at K6, 1063b30-35 and the elenctic proof of pnc in 4. I shall not deal with
them here because, as I have argued at length elsewhere (Castagnoli 2005), these are not genuine examples of
self-refutation (at least on a reasonably precise and narrow conception of what counts as self-refutation). At K6,
1063b30-35 All statements are true and All statements are false are rejected because absurdly incompatible with
the defnitions of true and false of T14, and not because of self-refutation (pace [Alex.] in metaph. 658, 20-26;
Cherniss 1935, p. 87n364; Burnyeat 1976a, p. 44; Reale 1993, vol. iii, p. 531). In 4, Aristotles elenctic proof ap-
pears to have the structure of a direct refutation of not-pnc, where, importantly, Antiphasis thesis, not-pnc, does
not fgure among the premisses of its own refutation (cf. e.g. Wedin 2000, p. 129), so not-pnc is not disproved by
self-refutation (pace Irwin 1988, Cassin, Narcy 1989, Baltzly 1999, Politis 2004, pp. 147-148). In a nutshell, the
elenctic proof of pnc is, unsurprisingly, an elenchus of Antiphasis not-pnc, based on certain other admissions
Antiphasis cannot help granting if he wants to participate to any dialectical exchange (and not, narrowly, if he
wants to state or defend his own denial of pnc).
self-refutation arguments from democritus to augustine 49
6. 1. Every appearance is true:
dialectical reversal or Consequentia Mirabilis?
Let us begin with one of the best known and most discussed self-refutation arguments
reported by Sextus:
T20 One could not say that every appearance () is true, because of , as Dem-
ocritus and Plato taught contradicting Protagoras; for if every appearance is true, it will al-
so be true, being based on an appearance, that not every appearance is true, and thus it will
become false that every appearance is true. And even apart from of this kind
The noun (reversal, about-turn) is used twice here to brand the kind of
argument presented in the middle of the passage: what is this label supposed to mean
and how does that argument work? In his frst seminal article on ancient self-refutation,
Burnyeat suggested that although any refutation, of course, establishes the contrary
of what it refutes, and tended particularly to be used of the
special case where the thesis to be refuted itself serves as a premise for its own refuta-
tion, where starting out with p we deduce not-p and so conclude that the original
premise was false (1976a, p. 48).1 In Hellenistic and post-Hellenistic philosophical
jargon would indicate not simply a reversal or refutation of a thesis into its
contradictory, but a self-reversal or self-refutation having that thesis as its premiss.
Let us investigate how this broad outline fts the logic of T20s argument. If we take
T20 at face value, we might infer that Democritus and Plato presented the following an-
ti-Protagorean argument:
1 (1) Every appearance is true Assumption (Protagoras thesis)
2 (2) It is an appearance that not every appearance is true Assumption
1,2 (3) It is true that not every appearance is true from (1) and (2)
? (4) It is false that every appearance is true ?2
I shall not deal with the question of how Protagoras relativistic outlook could turn, in
most of the following philosophical and doxographical tradition (starting as soon as
Aristotle), into an infallibilist thesis to the efect that every appearance is (unqualifed-
ly) true. The Theaetetus passages we have examined in section 4. 2 could have con-
tributed to the process, giving some readers the false impression that the Protagorean
thesis under fre was an unqualifed assertion that whatever appears is true, and Plato
complex argument might have been thereby read along the simplifed lines recorded in
T20. As for Democritus, although we do not possess any independent evidence for his
employing (and possibly devising)3 any such argument against Protagoras, I see no seri-
ous reasons for doubting the reliability of Sextus attribution (Plutarch informs us that
Democritus wrote many and persuasive things against Protagoras).4 One could even
1 I shall suggest that the fnal clause is questionable. It should be noticed that actually the fact that the self-
refuting thesis serves as a premiss for its own refutation is not distinctive of self-refutation: any indirect proof,
e.g. reductio ad absurdum, can be seen as a refutation of the contradictory of the conclusion having it as a premiss.
2 I shall suggest that there are two alternative interpretations for this fnal step to be considered.
3 Sextus order might indicate that Democritus used the argument before Plato, but cannot settle the issue. That
Democritus used the argument before Plato is taken for granted, without argument, by Burnyeat (cf. p. 51n3). For
Democritus uncertain dates cf. Lee 2005, pp. 182-183. 4 Adv. Colot. 1109A4.
50 luca castagnoli
fnd a thin clue in favour of Sextus accuracy in another Sextan passage, in which we
are informed that Xeniades of Corinth, who was mentioned by Democritus too, as-
serted that everything is false, and that every appearance and judgement are false (m
7.53). If Democritus had something to say about such an obscure fgure, we can per-
haps suppose that it was to criticise the theses which Sextus attributes to Xeniades.1
One might thus conjecture that Democritus attacked Xeniades Everything is false
by some self-refutation argument analogous to that which he employed against Pro-
tagoras contrary position.2
What we can safely afrm is that, since the noun certainly did not appear
in Plato, Sextus own phrasing does not authorise us to believe that it was used by Dem-
ocritus either: our source might be applying a later, Hellenistic label to an argument
otherwise dubbed in both the authors he mentions. However, we should not even be
too hasty in discarding the possibility that Democritus, unlike Plato, did use ,
or , and perhaps even prompted the philosophical career of these terms.3
But let us return now to the argument itself. It is clear that premiss (2) (It is an ap-
pearance that not every appearance is true) is necessary to infer, in conjunction with
(1), the conclusion (3), that it is true that not every appearance is true. Burnyeat found
this puzzling: what is the rationale of saying that (1) is self-refuting, if its reversal can be
arrived at only with the aid of (2), which seems to be only contingently, albeit indis-
putably, true?4 One might protest that this perplexity arises from the ungrounded pre-
supposition that in ancient , just like in modern absolute self-refutation, a sin-
gle proposition must be involved: it is worth stressing that in fact in T20 not only we are
not told that (1) is reversed by itself alone, but actually also the by itself is missing.
Nonetheless, Burnyeats solution deserves careful scrutiny:
We are to imagine Protagoras putting forward a subjectivist doctrine, according to which what-
ever appears to anyone to be so is so in fact, (1). He is opposed by someone saying that to him it
appears, on the contrary, that not everything that appears to someone to be so is so in fact, (2).
1 One could argue that actually Democritus might have had some sympathy for Xeniades second thesis, but
only if appearance is interpreted in a narrow perceptual sense (cf. Morel 1996, pp. 430-431). Cf. Metaph. 5,
1009b11-12, where Aristotle claims that Democritus asserts that there is no truth, or at least to us it is unclear (but
the context indicates that this report is meant to refer to sense-perception only) and Cic. Luc. 73, where Dem-
ocritus is reported to have fatly denied that truth exists.
2 Cf. T30 and T31 below. Democritus devised at least another famous self-refutation charge (fr. 125), with which
we shall not be concerned here (cf. Castagnoli 2005).
3 Contra Burnyeat 1976a, p. 66: since philosophical writing before Epicurus has plenty of occasion to speak
of self-refutation, but the varied vocabulary used for the purpose makes no mention of reversal, [] the idea of
reversal can be traced to the frst decade of the third century b.c. when the Hellenistic philosophies were taking
shape. Burnyeat is perhaps too cautious by presupposing that it is unlikely that the vocabulary had
originated well before Hellenistic times, with Democritus, on the sole basis that it is not to be found in Plato and
Aristotle (after all, Democritus name is conspicuously absent from Platos writings, and as we have seen in sect.
5 Aristotle did not rely on self-refutation charges very frequently). One might suppose that Epicurus borrowed the
jargon from Democritus and revived it.
Burnyeat seems to suggest that is, just like , a later technical concept, extraneous to the
original formulation of the argument (1976a, p. 47n5). This is not obvious: is used twice in the Theaete-
tus, at 152c1 and 161e8, in the exposition of Protagoras doctrine, both times with the relevant meaning of what
. Moreover, to insist that is intended in its later technical sense in T20 could be dangerous for
Burnyeats own interpretation (see below): from a Stoic perspective, any mind content, even when not assented
to but simply entertained, is a , so the presence of a dialectical opponent risks becoming redundant,
because the proponent of the thesis according to which every is true must presumably entertain also
the contradictory thought that not every is true. 4 Cf. Burnyeat 1976a, p. 49.
self-refutation arguments from democritus to augustine 51
But Protagoras has only to be opposed like this and he will be forced to deny his own thesis and
admit defeat, i.e. that not every appearance is true. His subjectivism is a non-starter, bound to lose
him the debate before it has a chance to get going. It is necessarily a loser because in a dialectical
context (2), contingent though it is, is in a sense guaranteed to hold; for there is no debate with-
out disagreement and clash of views. [] We began with the simple and, if you like, strictly self-
refuting case of a thesis falsifed by its own content. Then came falsifcation by the way a propo-
sition is presented [] Now it is the act of submitting a thesis for debate or maintaining it in the
face of disagreement that causes its reversal and shows it up as false. One might call this dialecti-
cal self-refutation, and say that a thesis so falsifed is dialectically self-refuting. (1976a, p. 59)
To have underscored the necessity of placing T20s in its dialectical back-
ground to do full justice to its logic was a fundamental contribution to our under-
standing of the argument; notice how similar the skeleton of the exchange recon-
structed by Burnyeat is to that of Platos own argument in the Theaetetus on the
interpretation I defended in section 4.2. I have argued that all the ancient self-refutation
arguments we have analysed so far are best understood as presupposing dialectical con-
texts, even when such contexts are not explicitly mentioned; Burnyeat suggests that the
same approach might be required for T20 too. I can add here another clue to the exis-
tence of an implicit dialectical background in T20. According to Sextus, the
argument shows why one could not say ( i ), with Protagoras, that every
appearance is true: while this could be the case in virtue of the fact that the self-refuta-
tion argument supposedly proves the falsehood of that thesis (and one should avoid say-
ing something demonstrably false), it seems more natural to suppose that the reason is,
simply, that if you do say that every appearance is false you will be easily defeated by
your dialectical opponent in the way described.
This brings us to my frst doubt about Burnyeats conception of dialectical self-refu-
tation, according to which dialectical self-refutation would show up a thesis as false,
or falsify it. If this were an alternative way of conveying the idea that the proponent
of a dialectically self-refuting thesis is forced to deny it, or admit its falsehood, I would
have no qualms about accepting such jargon. But since Burnyeat adopts the same vo-
cabulary of falsifcation also for two other species of self-refutation (absolute and prag-
matic), in which the falsifcation of the thesis involved is supposed to amount to an ob-
jective, non-dialectical proof of its falsehood, the notion of falsifcation is not
innocuous here. The falsehood of the reversed thesis was indicated, in fact, as the fnal
deductive step of any argument in Burnyeats general account quoted above
on p. 50 (and so conclude that the original premise was false): it is not clear whether
Burnyeat fails to distinguish falsifcation from dialectical reversal or supposes that the
dialectical manoeuvre he describes is, or brings with itself, a falsifcation of the thesis
involved as well. If we come back to T20s argument, however, we notice that I left its
last step, (4), unexplained. From the Protagorean thesis, (1), and Protagoras opponents
dissent, (2), Protagoras unavoidable confession follows that (3) It is true that not every
appearance is true; but then (4), It is false that every appearance is true, is added as
a further and fnal consequence. I suggest that the most natural way of reading this in-
ference is to see (4) as a straightforward consequence of (3) (TpFp), an extra turn of
the screw aimed at making Protagoras defeat more glaring by having him admit ex-
plicitly the falsehood of his own thesis (and not only the truth of its contradictory). But,
once again, Protagoras own confession that (1) is false is not the same as a proof that
(1) is false, nor does it imply it.
52 luca castagnoli
One might be tempted to propose a diferent interpretation of T20s argument as a
non-dialectical proof of the logical falsehood of proposition (1), along the following
1 (1) Every appearance is true Ass. (Protagoras thesis)
2 (2) It is an appearance that not every appearance is true Assumption
1,2 (3) It is true that not every appearance is true from (1) and (2)
1,2 <3.1> Not every appearance is true from (3), by Tpp
2 <3.2> If every appearance is true, not every appearance is true from (1) and <3.1>
3.3 <3.3> If something implies its own contradictory, it is false Law of classical calculus
2, 3.3 (4) It is false that every appearance is true from <3.2> and <3.3>
I will not stress further the problem that, on such a reconstruction, (2) would be the as-
sumption of a contingent extra-logical truth, which would remain undischarged at the
end.1 It is the status of another assumption that appears to me no less problematic here:
the conclusion, (4), would be a consequence (d ) not of (3), the apodosis of the
preceding conditional sentence, but of the whole conditional, <3.2>, together with the
implicit <3.3>. Now, <3.3> is a tautology of the classical propositional calculus
(cm) (pp)p
often referred to by the specialists as Consequentia Mirabilis:2 any proposition implying
its own contradictory is false3 (precisely, necessarily false). cm has been lurking more
than once in the footnotes in the previous sections,4 but I postponed the discussion of
it to this stage of our inquiry for expository convenience.
I suggest there are various reasons for doubting the correctness of any reconstruc-
tion of T20s argument relying on cm. To begin with, and most obviously, no trace of
cm appears in our text. To this objection, one might reply that cm does not need to be
stated explicitly, because it is not a premiss, but an inferential schema (ppp), the
1 See the left column in the derivation, where the assumptions on which each step depends are listed. Given
the assumption (2), the argument does not sufer the same weakness as we encountered in previous logical
reconstructions of ancient self-refutation arguments: if it appears to someone that (p) not every appearance is
true, then it is in fact impossible that every appearance is true. For even if all the other appearances should be true,
at least the appearance p will not be true, but Liar-paradoxical.
2 Often the name Consequentia Mirabilis (attested for the frst time in the 17th century among Polish Jesuit
scholars: cf. ukasiewcz 1970, p. 168n19) is primarily attributed to the formula
dubbed also Lex Clavii, and only consequently to cm above, which follows from it by a simple substitution of the
variables and the law of double negation (for the fortuna of this label cf. Bellissima, Pagli 1996, pp. 205-212). I
shall use cm both for (pp)p and for (pp)p (notice, however, that the two formulas are not equiva-
lent in all systems: in intuitionistic logic, for example, only (pp)p is valid, whereas (pp)p is not). For
a comprehensive history of Consequentia Mirabilis cf. Nuchelmans 1991, pp. 124-137; Bellissima, Pagli 1996.
3 In this paraphrase of cm is false is used only as a convenient method of generalising about negation. How-
ever, if one accepts the basic semantic principle pFp, cm can be easily reformulated metalinguistically with a
falsehood predicate: (pp)Fp. To be precise, this reformulation seems to be the one required in the argument
above to infer the conclusion (4):
1 (1) (p)ApTp Ass. (Protagoras thesis)
2 (2) A((p)ApTp) Assumption
1,2 (3) T((p)ApTp) from (1) and (2)
1,2 <3.1> (p)ApTp from (3), by Tpp
2 <3.2> ((p)ApTp)((p)ApTp) from (1) and <3.1>
3.3 <3.3> (pp)Fp metalinguistic cm
2, 3.3 (4) F((p)ApTp) from <3.2> and <3.3>
4 Cf. nn. 1 on p. 15, 4 on p. 37, 1 on p. 47, 3 on p. 48.
self-refutation arguments from democritus to augustine 53
obvious validity of which would be recognised by any attentive reader as what justifes
the inference from a conditional formed by contradictories, <3.2>, to its consequent,
(4). However, not only is cm never presented as a logical law by our sources for Hel-
lenistic logic1 or, more broadly, for ancient logic, but no argument is attested in which
it is unequivocally employed (our T20 has been seen as one of the few and most prom-
ising texts for an attempt at tracing back the use, if not the theoretical formulation, of
Consequentia Mirabilis to antiquity).2 Although this might be partly imputed to an ob-
jective difculty in univocally translating into logical form arguments expressed in a
natural language, to say confdently that an ancient reader would have easily supplied
the unstated cm may mean providing him anachronistically with a tool he would have
been unable to handle. Moreover, the suggestion that cm is the underlying inferential
schema which allows us to infer (4) from <3.2> is liable to another objection: deduc-
tions from a single premiss did not meet particular enthusiasm to put it mildly in
antiquity, at least as far as we can judge from the two most infuential ancient logical
constructions, Aristotles syllogistic and Chrysippus dialectic.3
To avoid these difculties, we could attempt a slightly diferent reconstruction, not
involving cm, but a distinct, albeit strictly related, inferential schema. On this interpre-
tation, after <3.1> the argument would proceed as follows:
2 <3.2> If every appearance is true, not every appearance is true from (1) and <3.1>
3.3 <3.3> If not every appearance is true, not every appearance Duplicated conditional
is true
3.4 <3.4> Either every appearance is true or not every appearance Excluded middle
is true
2 <3.5> Not every appearance is true from <3.2>,<3.3>,<3.4>,
by constructive dilemma
2 (4) It is false that every appearance is true from <3.5>, by pFp
Although such a reconstruction would require us to supply even more implicit reason-
ing than before, which is sufcient to make it rather questionable as an exegesis of the
text, the inferential pattern to the conclusion would be in accordance with a schema
which, apparently, was recognised as sound and used in antiquity and which occurs else-
where in Sextus corpus (cf. p. 57n6):
1 Bobzien puts the Mirabilis in a list of sequents that have such a form that no compounds of propositions of
that form would be syllogisms in the Stoic system, although all of them are correct sequents in pc [sc. classical
propositional calculus]. She claims also not to have found any documentation in the sources that the Stoics ac-
cepted either all corresponding conditionals of a form as true, or a metalogical principle that in some way corre-
sponds to the sequent (Bobzien 1996, pp. 183-184). These sequents (including the Mirabilis) cannot be analysed in-
to indemonstrables either.
2 Commenting on T20, Bellissima and Pagli write: Non c dubbio che lo schema di ragionamento, efettiva-
mente parallelo a quello del Teeteto platonico [], ripercorra le linee della Consequentia Mirabilis nella forma
(AA)A (1996, p. 178). It should be clear from my reconstruction of the Theaetetus argument that I do not
agree that it relies on cm either (cf. p. 37n4).
In sect. 5. 2 I have reconstructed Aristotles treatment of the self-eliminating statements Everything is true
and Everything is false at Metaph. 8, 1012b13-22 in a straightforward dialectical way which does not presuppose
any unstated application of cm.
3 Aristotles phrasing of the defnition of (cf. e.g. APr. i 1, 24b18-21: A is a in
which, certain things being stated, something other than what is stated follows of necessity from their being so)
seems to exclude single-premiss deductions (cf. Alex. Aphr. in APr. 17,10-18,8); the Stoic Antipater is reported to
have endorsed a very unorthodox (and much criticised) view by allowing single-premiss arguments (
54 luca castagnoli
(a) pp <3.2>
(b) pp <3.3>
(c) pp <3.4>
(d) p <3.5>
This form of constructive dilemma is strictly related to cm: it might be seen as an ex-
tended version of it, in which the two tautological extra premisses (b) and (c) are spelled
There is a fnal obstacle which such an exegetical approach to T20s argument must
overcome: could the key premiss If every appearance is true, not every appearance is
true, if taken at face value,1 ever be true? More generally, is it possible for any conditional
of the form pp to be true? If one accepts the truth-conditions for material impli-
cation typical of (modern) classical calculi, the answer is, of course, yes: pp is true
whenever p is false. Those truth-conditions were accepted in antiquity by the dialecti-
cian Philo2 and, most probably, by some Stoics.3 A diferent analysis, proposed by
Diodorus Cronus,4 required the present impossibility of p (i.e., in Diodorean terms, the
falsehood of p now and at any time from now on) for pp to be true. On both con-
ceptions nothing would prevent <3.2> from being accepted as true. However, there is
strong circumstantial evidence to make the case that two other particularly infuential
ancient conceptions of sound implication and conditional would have barred that pos-
sibility. In modern discussions of non-classical logics one radically non-classical thesis is
sometimes singled out as distinctive:
(ta) (pp) (pp)
McCall, who propounded one of the frst systems of connexive logic, characterised
by a brand of connexive implication such that no proposition connexively implies or
is implied by its own negation, baptised this property Aristotles thesis (hereafter, also
ta).5 This choice was not random, since according to McCall the following passage
testifes Aristotles endorsement of ta:
T21 But it is impossible that the same thing should be necessitated by the being and by the not-
being of the same thing. I mean, for example, <that it is impossible> that B should neces-
sarily be large if A is white and that B should necessarily be large if A is not white. [] If then
B is not large, A cannot be white [from If A is white, necessarily B is large, by contraposi-
tion]. But if, if A is not white, it is necessary that B should be large, it necessarily results [by
transitivity] that if B is not large, B itself is large. But this is impossible. (APr. ii 4, 57b3-14)6
ta seems to be accepted at the very end of the passage and used by Aristotle to prove
the incompossibility of two implications of the form pq and pq (baptised by
McCall Boethius thesis, hereafter also tb):
1 I.e. not as shorthand for If every appearance is true and it appears that not every appearance is true, then not
every appearance is true, which is a conditional of the diferent form (pq)p.
2 Cf. e.g. s.e. ph 2.110. A Philonian conditional, like a material implication, is false only when its antecedent is
true and its consequent is false.
3 The adoption of Philos truth-conditions by some Stoics emerges sometimes in our sources (cf. e.g. s.e. ph
2.104); the problem is to establish whether by pre- or post-Chrysippean Stoics, and how reliable these testimonies
are. 4 Cf. e.g. s.e. ph 2.110-111.
5 McCall 1966, p. 415. 6 On this passage cf. Patzig 1959; Mignucci 1969, pp. 610-614.
self-refutation arguments from democritus to augustine 55
Suppose that both (a) pq and (b) pq
If (a) pq, then (c) qp by contraposition (57b9-11)
If (c) qp and (b) pq, then (d) qq by transitivity (57b6-9)
But (d) qq is impossible ta
Therefore, (both pq and pq) tb
As William Kneale remarked, if Aristotle was right in asserting this, there could never
be any valid argument in the pattern of the consequentia mirabilis (Kneales bold con-
clusion was that Aristotle was certainly wrong, and Aristotles thesis was indeed Aris-
totles error).1 No constructive dilemma in the form presented on p. 55 could be sound
either, since its frst premiss would be bound to be false.2 More generally, any attempt
at proving the necessary falsehood of a proposition (e.g. Protagoras thesis) starting
from the fact that it implies its own contradictory would be a non-starter.
One could object that T21 alone ofers insufcient grounds for establishing Aristotles
real commitment to ta and tb;3 moreover, even conceding that T20s anti-Protagorean
argument, in its non-dialectical reconstructions, perhaps would not appeal to Aris-
totelian devotees, it still could be convincing for readers with diferent logical tastes. In
response to this, I suggest that the non-classical attitude towards implication which ap-
parently emerges from T21 was not isolated in the ancient logical landscape. There are
compelling reasons to believe that the (most likely Chrysippean) conception of condi-
tional () dubbed (connectedness) by some of our sources,
which seems to have been Stoic orthodoxy for some time,4 also involved the truth of
ta. Here is the defnition of reported by Sextus:
T22 Those who introduce say that a conditional is sound when the contradictory of
its consequent is in confict () with its antecedent; according to them, the condi-
tionals mentioned above [sc. If it is day, I converse5 and If there are not indivisible ele-
ments of the things, there are indivisible elements of the things]6 will be unsound, while
If it is day, it is day will be true. (ph 2.111)
Admittedly, the attested falsehood of the conditional If there are not indivisible
elements of the things, there are indivisible elements of the things on the
conception is insufcient to establish that every conditional of the same form pp
must be false on that conception. However, another Sextan passage suggests the gen-
eral validity of ta for :
1 On this occasion Aristotle wrote more than was needed and fell into error (Kneale 1957, p. 66). ukasiewicz
(1957, pp. 48-51) and Patzig (1959, p. 191) share the same opinion. For a more sympathetic approach, based on the
request that the antecedent is explanatory of the truth of the consequent, cf. Smith 1989, pp. 190-191.
2 (qq) entails (qq) provided one accepts contraposition and double negation.
3 One could object that the second premiss of Aristotles famous argument in the Protrepticus
(a) pp If one must philosophise, one must philosophise
(b) pp If one must not philosophise, one must philosophise
(c) p Therefore, in any case one must philosophise
is obviously incompatible with ta and that this suggests that we should not attribute ta to Aristotle. I have argued
elsewhere at length (Castagnoli 2005) that the argument in this form attested by our late sources should not be
attributed to Aristotle himself, and that the original argument in the Protrepticus had a dialectical structure not re-
ducible to cm and compatible with ta.
4 Most notably, the truth-conditions for conditionals are the only ones presented by Diogenes Laer-
tius in his testimony on Stoic logic (7.73).
5 True for Philo (ph 2.110). 6 True for Diodorus (ph 2.111).
56 luca castagnoli
T23 But it is impossible, according to what they1 say, that a sound conditional is composed of
conficting propositions. (ph 2.189)
Since contradictory propositions are always in mutual confict () too, the thesis ex-
pressed in T23, which I shall call Chrysippus thesis (hereafter, tc), sounds like strong
evidence that ta conveys a genuine property of . The reliability of this tes-
timony, the importance of which was frst underscored by Nasti,2 has been questioned,3
but seems to have now found sufcient corroboration in other sources which can be
thought to provide trustworthy, albeit indirect, information about Stoic logic. Although
I cannot enter the details of this debate, which I have discussed at length elsewhere,4 let
us consider at least a couple of passages:
T24 Where one can use the disjunctive connective ( ), one cannot use the condi-
tional one ( ); and where one can use the conditional one, one cannot use the
disjunctive one. And it is clear from what has been said that what the conditional <con-
nectives> [] announce is in confict with what is announced by the disjunctive <ones>
[] between the disjuncts there is no consequence. (Apoll. Dysc. de conj. 218, 11-15)
T25 Moreover, the antecedents and the consequents cannot leave one another, nor can con-
ficting propositions (repugnantia) be connected (cohaerere) one to another; the former are
necessarily connected one to another, the latter disconnected. (Boeth. in Cic. top. 349, 40-42)
Both Apollonius Dyscolus and Boethius report that a true conditional cannot be formed
of conficting clauses (tc), or, a fortiori, contradictory clauses (ta).5
There seems to be credible, albeit circumstantial, evidence, therefore, for attributing
ta to at least some Stoics (Chrysippus and those endorsing his notion of ).
If this attribution is correct,6 any reconstruction of T20s argument along the lines of
1 They are the dogmatists (ph 2.111), but the context indicates that the reference is to philosophers who
adopt the truth-conditions. 2 Cf. Nasti 1981.
3 Cf. Stopper 1983. 4 Cf. Castagnoli 2004b, 2004c, 2008. Cf. also Nasti 2002.
5 For further details about these passages and their signifcance cf. Nasti 2002; Castagnoli 2004c, 2008;
Barnes 2005. One might question the reliability of Apollonius Dyscolus and Boethius as sources for Stoic logic.
Without tackling this objection directly here, I shall just reply that it need not be damaging for my present pur-
poses: if ta appears in their reports but did not necessarily derive from Stoic logic, ta could have been endorsed
even more widely. If ta holds good for , given the ascending order of strength of the conceptions of
conditional presented by Sextus at ph 2.110-112, the fourth one, (manifestation), should have been gov-
erned by ta as well.
6 Before we can accept ta as a genuine Stoic thesis, however, it is necessary to show how this can be reconciled
with some attested dilemmatic arguments apparently incompatible with ta but usually attributed to the Stoics
(cf. ph 2.131, 186; m 8.281, 466):
(a) pp If cause (sign, proof ) exists, cause (sign, proof ) exists
(b) pp If cause (sign, proof ) does not exist, cause (sign, proof ) exists
(c) pp Either cause (sign, proof ) exists or cause (sign, proof ) does not exist
(d) p Therefore, cause (sign, proof ) exists
I have argued at length elsewhere (Castagnoli 2005 and 2008) that various textual clues allow us to reinterpret
those dilemmas as elliptical and enthymematic formulations of dialectical arguments perfectly compatible with ta
(and tc):
(a) qp If <you answer that> cause (sign, proof ) exists, then <you yourself
admit that> cause (sign, proof ) exists
(b) r<s s>p If <you answer that>cause (sign, proof ) does not exist, then <you must
present a cause (sign, proof ) of what you assert, in order to be credible,
but thus you yourself admit that> cause (sign, proof ) exists
self-refutation arguments from democritus to augustine 57
cm or cognate schemas becomes unpalatable for a large audience (probably Aristotle,
and part of the Peripatetic tradition, and mainstream Stoicism), and for serious logical
reasons. If we add to this the indisputable datum that such a reconstruction fts Sextus
actual wording rather poorly anyway, I hope I have constructed a compelling case for
sticking to a purely dialectical interpretation of T20s which does not require
that Protagoras thesis is proved to be logically false because it entails its own contra-
dictory. But, then, Burnyeats presupposition that dialectically self-refuting propositions
are (also) somehow falsifed becomes dubious, and in the light of what I have explained
about cm and ta his suggestion that, generally, in a argument starting out
with p we deduce not-p and so conclude that the original premise was false,1 which
sounds like a description of cm, turns out to be suspect too.2
The absence of any distinction between the dialectical nature of and a log-
ical law like cm is a frst aspect of Burnyeats analysis which might require some revi-
sion. Another point of disagreement, which will emerge more clearly in the next sec-
tion, concerns the existence, allowed by Burnyeat, of some arguments in
which the dialectical aspect plays a limited role, or no role at all: as we have seen, di-
alectical is, according to Burnyeat, only one species of ancient self-refutation,3
along with others, like absolute and pragmatic self-refutation.4 I believe, instead, that
all has a character not reducible to purely formal argument, without there-
by denying that distinct species of ancient self-refutation can be usefully classifed.5
Before passing to the inspection of some arguments allegedly amounting
to absolute self-refutation,6 let us consider briefy three other Sextan passages which
replicate and confrm some fundamental features which I have detected in T20:
T26 They will say () either that all things which appear are true, or that some are. If all,
their incurs reversal (); for it appears to some people that nothing is true.
(ph 2.88)
T27 Either all things which appear and all those unclear are true, or some things which appear
and some unclear. Now if all are, their will again be reversed (), it be-
ing granted as true ( ) also that nothing is true. (ph 2.91)
(c) qr Either <you answer that> cause (sign, proof ) exists or <you answer
that> cause (sign, proof ) does not exist
(d) p In any case <you must admit that> cause (sign, proof ) exists
1 Burnyeat 1976a, p. 48 (italics mine).
2 Contra also Barnes 1997, p. 31: [] turns on the exotic truth that anything which implies its own
negation is itself false.
One could object that I have shown, at most, that logical , intended as something akin to cm, was
only available to some philosophical schools which did not endorse ta; however, Burnyeat himself maintains, I
think correctly, that evidently peritrope is a commonplace of later Greek controversy, available to disputants of any
persuasion to confute the other side of the debate (1976a, p. 57; italics mine).
3 In Burnyeats own terms, it is an extension of the notion of self-refutation (1976a, p. 59), and not a correc-
tion or replacement of it.
4 In his analysis Burnyeat makes it clear that pragmatic often occurs in dialectical contexts, but the
fact that dialectical self-refutation is chosen as the name of a species of risks obscuring this very im-
portant point.
5 Burnyeat juxtaposes, without stressing their diferences, the general, non-dialectical account of
which I have criticised above and the following dialectical account: For precisely what self-refutation consists in
is a reversal whereby advancing a proposal commits one to its contradictory opposite (1976a, p. 49).
6 For reasons which should now be clear, I do not agree with McPherrans analysis of T20s argument as an
instance of absolute self-refutation (1987, p. 293n8).
58 luca castagnoli
T28 And if the something is true, everything will be true; from which it is in turn concluded
that nothing is true, since this itself, I mean that nothing is true, being a thing, is true. (ph
In T26 a dialectical exchange is unquestionably presupposed: is the result of
one saying that all are true, and thus what gets reversed, presumably into its
own contradictory, is what one has said ( ).
In T27 the reason why the that all appearances are true incurs reversal seems
to be, simply, that one who proposes it is thereby conceding as true also something which
clearly amounts to its contradictory (that nothing is true). Nothing is suggested about
the possible implications for the truth-value of that .
In T28 Nothing is true cannot be meant to be the true conclusion of a complete
proof by self-refutation of the assumption that Everything is true, but must be an un-
welcome consequence of that undischarged assumption: if everything were true, then
nothing would be true. Once again, no further step involving cm (e.g. therefore, not
everything is true) is presented or suggested. This time no dialectical framework is
present either, but if one explores the broader context of T28 that framework can be
easily supplied:
T29 Moreover, the something, which is, they [sc. the Stoics] say, the highest genus of all, is ei-
ther true or false or neither false nor true or both false and true. If then they will say that
it is false, they will be admitting that everything is false. [] T28 And if the something is
true, everything will be true
Although at the beginning the four logical possibilities are listed in an abstract way, sub-
sequently it is not the consequence of the frst possibility that is analysed, but the con-
sequence of the Stoics accepting it and saying that their summum genus, the something,
is false. Symmetry requires us to understand the same also in the case of the second
possibility in T28: And if <the Stoics will say that> the something is true, <they will
be thereby admitting that> everything is true; from which it is in turn concluded that
<they must concede that> nothing is true. We had already accepted this sort of ellip-
ticity as a plausible explanation of certain otherwise problematic features of the Dissoi
logoi argument (section 2) and of Socrates frst dilemma against md in the Theaetetus
(section 4. 2. 2); more crucially, we shall appreciate soon that such conciseness is a fea-
ture of Sextus own usus scribendi.
6. 2. Nothing is true: two diferent approaches
I have already noted that according to Burnyeat some instances of (single-
premise reversals) ft the description of absolute self-refutation: they exemplify the
strictly self-refuting case of a thesis falsifed by its own content (1976a, p. 59) through
the pattern of a Consequentia Mirabilis of the form (pnot-p)not-p (p. 49). I shall
examine now two of these arguments in Sextus, to verify whether the modern con-
ception of absolute self-refutation, both in its bare bones, as just described, and in the
fuller details of Mackies infuential formalisation, really captures their logic.1 Let us
start from the frst:
1 McPherran (1987, p. 292) agrees with Burnyeat that they are cases of absolute self-refutation.
self-refutation arguments from democritus to augustine 59
T30 Now, we have shown above that those who say that all things are false (f b
) incur reversal (). For if all things are false, All things are
false, being one of all things, will be false. And if All things are false is false, its contra-
dictory, Not all things are false, will be true. Therefore, if all things are false, not all things
are false. (M8.55)
I believe that actually this passage contains strong evidence against interpreting
as a form of absolute self-refutation. To begin with, it is not the proposition
All things are false that is charged with reversal, but it is those who say that all things
are false that are turned about, reversed. Given what follows in T30 and what we
have learnt about in general, this should mean that they are reversed into say-
ing that not all things are false.1 Sextus phrasing would be devious if all he meant were
that those who maintain that all things are false are mistaken because their thesis is
demonstrably false, by . However, one could still defend the plausible idea
that such a dialectical manoeuvre is made possible by (and mirrors) a particular logical
property of the thesis asserted: those who say that all things are false can be forced in-
to admitting the contradictory of their own thesis because the falsehood of that thesis
can be established through a proof along the lines of Consequentia Mirabilis.2 Perhaps
ancient always comes with an essential dialectical dress, but it would still be,
intrinsically, a form of absolute self-refutation.
This conciliatory approach has its drawbacks. If we follow carefully the steps of T30s

(1) If all things are false, All things are false is false; by universal instantiation
(2) If All things are false is false, Not all things are false is true; FpTp
(3) Therefore, if all things are false, not all things are false. from (1) and (2), by transi-
tivity and Tpp
we notice that the argument halts one inference short of Consequentia Mirabilis: no con-
clusion is inferred from the conditional (3) of the form pp. Should we presume that
this happens because the intended conclusion, p (Not all things are false), is only too
clear and thus implicitly meant? On the basis of what I have argued about the shadowy
status of cm in antiquity, to take such an answer for granted will not do, and we should
also be cautious before interpreting (3) as a true conditional proposition which might
function as the premiss of a Consequentia Mirabilis, given what we have learnt about ta.
But, then, the proposal to catalogue this argument under the heading absolute self-
refutation becomes questionable, and even more so when we notice that the starting
move of the of T30 is the recognition of the self-reference of All things are
false, and not Mackies law of T-prefxability.3
Given the kind of ellipticity we seem to fnd in Sextus, projecting the dialectical
framework suggested by the frst sentence of T30 into the subsequent part of the
passage is not doing violence to the text. On my reading, T30 might be paraphrased as
follows: Now, as to those who assert that all things are false, we proved above that they
1 Cf. e.g. M8.295.
2 Analogously, one might say that those who present such and such a thesis are reduced to the impossible, i.e.
are bound to admit an impossibility, because their thesis does imply an impossibility.
3 For a similar argument cf. also Epict. disc. 2.20.1-3, pace Barnes prejudice that Epictetus argument is based
on cm, but is not formulated as cleanly as modern logicians would like (1997, pp. 30-31).
60 luca castagnoli
necessarily end up by admitting the contradictory thesis. For if they assert that all things
are false, they must admit that All things are false, being one of all things, is itself false.
And if they admit this, they cannot help conceding that its contradictory, Not all things
are false, is true. Therefore, if they assert that all things are false, they are inexorably
reversed into admitting that not all things are false. Once again, Everything is false
and its proponents would be unmasked as hopeless dialectical losers. But if
is conceived in such a dialectical framework and its outcome is a denial of the original
thesis by its own supporters, bringing cm into the picture, even supposing it possible,
would be redundant.
Is such a dialectical tactic unbeatable? As I have noticed earlier, step (2) might be ques-
tionable in this context: although it rests on a fundamental and apparently uncontrover-
sial principle governing our notions of truth and falsehood (FpTp), it is likely to be
implicitly rejected by someone who dares to suggest that Everything is false. There-
fore, he could protest that that principle should not be exploited against him, question-
beggingly, in a dialectical self-refutation argument, which should aim at defeating him
by relying only on the consequences of his own position. How successful such a defen-
sive manoeuvre would be is, however, far from obvious.1
At the beginning of T30, Sextus writes as if the argument he is about to
present had already been illustrated before, probably referring to this passage of the pre-
vious book:2
T31 For if all the appearances are false and nothing is true, Nothing is true is true. If, there-
fore, nothing is true, a truth exists; and in this way Xeniades was driven round to the op-
posite of his original position ( ), when he said that
all appearances are false and that absolutely nothing true exists in reality. For, in general, it
is impossible to say of any particular thing that is false without also afrming that some-
thing is true. For example, when we say that A is false, we are predicating the existence of
that very falsity of A, and we are afrming that A is false, so that what we virtually assert
is something like this: It is true that A is false. Simultaneously, then, with saying a thing
to be false we are necessarily afrming the existence of a truth. (M 7.399)
In lieu of the usual vocabulary, we fnd here the phrase
. This is further valuable verifcation of the meaning of
: to be reversed must be equivalent to to be driven round to the opposite
of ones original position, where ones original position () is what one has
said. Once again, it is noteworthy that this turnabout is not explicitly meant to falsify
Xeniades thesis, but to show why one must say () [] that not all <appear-
ances> are false (7.398).
But let us inspect T31s argument more closely. Actually its pattern difers signifcantly
from that of T30s :3 it does not rely on self-reference,4 and has a form we have
never encountered in ancient texts so far. Undeniably, Sextus If nothing is true, Noth-
ing is true is true; if, therefore, nothing is true, a truth exists bears a striking resem-
blance to Mackies proof of the absolute self-refutation of Nothing is true:
1 One might elicit the admission that not everything is false directly from the admission that Everything is
false is false (by straightforward semantic descent Fpp). I have explained on p. 45 that having to reject a prin-
ciple like FpTp would be a deeply embarrassing escape route; by rejecting also the principle Fpp the op-
ponent would make his position even more unintelligible. 2 So, e.g., Bett 2005, p. 99n18.
3 As Bett (2005, p. 99n18) notices. Either Sextus backwards reference was careless, or he could not appreciate
the diference between the two argument patterns. 4 Pace McPherran 1987, p. 292.
self-refutation arguments from democritus to augustine 61
Mackie Sextus Empiricus
(1) (p)(pTp) T-prefxability
(2) (p)TpT((p)Tp) From (1), by substitution If nothing is true, Noth-
ing is true is true
(3) T((p)Tp)(p)Tp Existential generalisation If Nothing is true is true,
a truth exists
(4) (p)Tp((p)Tp) From (2),(3), by transitivity If nothing is true, a truth
and double negation exists
(5) ((p)Tp) From (4), by cm
My contention is that this surface resemblance does not survive careful scrutiny. The
problem is not only that, once again, the fnal inferential step by cm delivering the con-
clusion (5) of Mackies absolute self-refutation does not feature in the lines of Sextus
Greek (the idea that this is fortuitous becomes less and less convincing as long as we
continue stumbling across more and more cases in which the same supposed chance
consistently recurs). What is even more important is that in Sextus the justifcation un-
derlying (2) is not Mackies T-prefxability, on the use of which in this context I have ex-
pressed my perplexity in section 2, but a diferent general () principle, accord-
ing to which it is impossible to say () of any particular thing that is false without
also afrming that something is true. Sextus explanation makes it abundantly clear
that this is not an abstract law concerning propositions, but governs our ordinary prac-
tice of making assertions and the nature of this speech act: asserting that p is tantamount
to asserting p is true, i.e. making an assertion is not only predicating something of
some subject but also, at the same time, positing this whole propositional content as
true.1 The structure of T31s reversal now becomes clear: when someone, like Xeniades,
asserts that Nothing is true, he is, by this very speech act, unwittingly committing
himself to the truth of what he is asserting; but, then, he must grant that at least one
truth does exist, thus admitting the contradictory of his original position and conced-
ing defeat. This is why one had better not say that Nothing is true; nothing is added
about the truth or falsehood of the propositional content of that assertion.
Sextus explanation of the logic of this self-refutation argument also casts new light
on other passages we have analysed above. For in the case of T31 it is impossible to de-
ny that, whereas Sextus wording of the argument could make one easily think of some
kind of non-dialectical proof by cm of the logical truth of Something is true, that for-
mulation is actually meant to be an extremely elliptical reminder of the various stages
of a which clearly presupposes a dialectical exchange in the background. But
the compressed formulation of T31s strongly resembles, in tone and context,
those we have found in T20, T28 and T30; therefore, T31 provides valuable corrobora-
tion to my proposed reading of those texts too. We started with the impression of hav-
ing fnally discovered an ancient instance of absolute self-refutation unequivocally
matching Burnyeats and Mackies description; we end up with a confrmation that an-
1 Burnyeats claim that the principle There are no truths entails its contradictory opposite (1976a, p. 50;
italics mine) should be emended, therefore, to the assertion of the principle There are no truths entails its
contradictory opposite, where the notion of entailment is not that of a logical consequence, but resembles that
pragmatic implication discussed, e.g., in the literature listed in the second part of n. 2 on p. 13.
62 luca castagnoli
cient , in its nature and purposes, should not be identifed with, or reduced
to, that modern form.
It might be added that in this case the ancient approach is not only diferent from, but
also more promising than, the modern one. For, as I have already noticed, that nothing
is true is, at least on some possible understandings of the fuzzy nothing, something
which, while not possibly-true, is still possible.1 In section 2 I criticised Mackies approach
for barring this logical possibility; T31s argument seems more successful in the more
modest task of showing that the commonsense view that some truth exists can resist
any attack in ordinary dialectical contexts,2 because any denial of it will amount, at the
same time, to an admission of its truth.3
7. Augustinian turn? The Soliloquia argument
for the imperishability of truth
We are approaching the end of our journey through the ancient philosophical tradition
in the hunt for early instances of self-refutation arguments against theses such as
Everything is true and Nothing is true. One fnal witness deserves our full attention,
since he appears to testify to the existence of an argumentative pattern which difers
from all those encountered so far and interestingly resembles that of Mackies absolute
We are now at the beginning of the second book of Augustines Soliloquia. In spite
of its title, this short writing has the form of a dense dialogue between Augustine and
a character, Ratio, who presumably represents Augustines own reason, some kind of
inner intellectual voice. The question is raised whether our souls are immortal and thus
will retain (and indeed increase) their knowledge forever. As the frst step towards an
answer, Ratio argues for the imperishability of truth:
T31 R.: When it [sc. the world] will have perished, if its going to perish, will it not then be true
that the world has perished? For as long as its not true that the world has perished, it
hasnt perished. There is a confict, therefore, between the world having perished and
it not being true that the world has perished.
A.: I grant this too.
R.: And what about this: does it seem to you that its possible that something is true, while
truth does not exist?
A.: In no way.
R.: Therefore there will be truth, even if the world perishes.
A.: I cannot deny that.
R.: And what about this: if truth itself perishes, will it not be true that truth has perished?
(si ipsa veritas occidat, nonne verum erit veritatem occidisse?)
A.: Who would deny that either?
R.: But there cannot be something true if truth doesnt exist.
1 In such a case, the assertion that nothing is true would be Liar-paradoxical. Cf. pp. 12-13.
2 A supporter of the thesis that nothing is true could object, however, that his speech acts must not be intended
as carrying the ordinary import, and that the normal rules of dialectic do not hold good of them: his utterances lack
assertoric force, but have some diferent kind of rhetorical or performative function (cf. p. 27 and p. 41n1).
3 We might say that, rather than absolutely self-refuting, Nothing is true is, in Mackies own jargon, opera-
tionally self-refuting, i.e. cannot be coherently asserted because what is implied by making an assertion contradicts
the asserted content (for a similar analysis cf. Passmore 1961, p. 68). From this perspective, some analogy exists
between the assertion of Nothing is true and the so-called Moorean assertions, such as This is a table, but I do
not believe that or This is a table, but that is not true (cf. the literature listed in the second part of n. 2 on p. 13).
self-refutation arguments from democritus to augustine 63
A.: I have just granted that.
R.: Therefore in no way will truth perish (nullo modo igitur occidet veritas). (sol. 2.2.2)
The elegant structure of Ratios argument is transparent:
(1) If truth (veritas) perishes, after it perishes it will be true (verum) that truth has perished.
(2) If <at any time> something is true, truth will exist <at that time>.
<3> <Even if truth perishes, truth will exist after it perishes.> From (1) and (2)
(4) Therefore, in no way will truth perish. From <3>
Premiss (1) looks like an instantiation of a temporalised version of Mackies law of T-
prefxability (P
x)): in the frst half of T32 Ratio had applied the same law
to the world perishes in order to illustrate its validity. Premiss (2) is secured by the def-
nition of veritas: truth is that by which whatever is true (verum) is true (2.15.29), and
whereas veritas persists when something verum passes away (1.15.28), nothing could
ever be verum if veritas did not exist. It is not difcult to reconstruct how the imper-
ishability of truth can be derived from these premisses. The unstated intermediate step
<3> follows from (1) and (2), and the conclusion that in no way will truth perish is a
straightforward consequence of it: if not even on the assumption that it perishes (i.e.
ceases to exist) can veritas cease to exist, then there is no conceivable way veritas could
perish. Mackies proof of the necessary falsehood, by absolute self-refutation, of Noth-
ing is true was based on T-prefxability and cm; Augustines proof of the imperisha-
bility of truth, while not making explicit reference to Consequentia Mirabilis, seems to
rely on an analogous inferential step (from <3> to (4)), and clearly starts from the ap-
plication of a version of T-prefxability.1 It seems that we have come, fnally, as close as
we could have hoped to a genuine antecedent for what in Mackies analysis is an ab-
solute self-refutation argument. Although it is presented by Ratio within her dialogue
with Augustine, no dialectical context seems to be required for the argument itself to
function; the vocabulary of assertion, admission, concession, so conspicuous in most
of the earlier instances of , is absent from the various steps of Ratios argu-
ment.2 Not only would it be very hard, given the textual details, to interpret Ratio as
making the point that if one asserts that veritas does not exist (because it has perished)
then one is involuntarily and self-defeatingly conceding the existence of at least some
verum, i.e. the content of ones own assertion, and thus the existence of veritas itself;3
by such a dialectical strategy she could never hope to prove what Augustine wants to
be established, the impossibility that truth will ever perish. Ratio is trying to demon-
strate a fundamental and atemporal truth about reality, and not to clarify what we can
or cannot coherently think or successfully assert now. Is Augustines innovation a real
step forward? Appraisal of this issue is far from easy.4 To begin with, it is not obvious
1 And not on self-reference, pace Charron and Doyle, 1993, p. 247.
2 That language does occur in T32, where Augustine gives his assent to the various steps of the argument pro-
posed by Ratio, but it is not part of the argument itself, as the summary at 2.15.28 confrms.
3 Contra DAgostini 2002, pp. 45-46. Augustine employs a similar kind of dialectical strategy, instead, in his De
vera religione (39.73): if one doubts the existence of any verum, and thus of veritas, at least one cannot doubt that
one is doubting (the point that understanding, and thus presumably not doubting, that one is doubting is one of
the necessary conditions for genuine doubt is made at trin. 10.10.14). Therefore, everyone who doubts the exis-
tence of veritas is certain of some verum, that he is doubting the existence of veritas. But being certain of some
verum is inconsistent with doubt about veritas. For a critical analysis of this argument cf. Kirwan 1983, pp. 219-220.
4 Although I shall criticise Augustines argument, I disagree with Abercrombies dismissive remark that we
are conscious that this is a verbal engine, of no ontological efcacy (1938, p. 63n1).
64 luca castagnoli
whether we are really entitled to speak of innovation here: Augustine himself neither
shows any awareness of his being original and reshaping an old, diferent dialectical ma-
noeuvre, nor reveals any consciousness of applying a generally valid argumentative pat-
tern (notice that he attaches no self-refutation label to the argument).
Second, when scrutinised carefully Augustines argument betrays dangerous ambi-
guities, similar to those which I detected in Mackies formulation of the absolute self-
refutation of Nothing is true in section 2. In T32 Augustine consistently uses verum
with infnitive clauses: this might suggest, but by no means proves, that verum is adopt-
ed as a kind of proposition-forming operator, and not as a predicate of sentences, or
whatever the truth-bearers are.1 But how does this operator work? What is its mean-
ing? The attempt to settle these issues on the basis of Augustines usage of verumin the
rest of the Soliloquia delivers disappointing results. At 1.15.28, shortly before T32, verum
is treated as a predicate expressing a property of objects: a true tree (vera arbor) is an
existing, real tree, but this existential use of verum seems to be of little help to decide
the sense of verum in T32. This comes as no surprise when one examines the second
book of the Soliloquia in its entirety: a very large portion of it is devoted to a dazzling
quest for the meaning of verum and falsum, which ends with very few certainties.
Throughout the discussion down to 2.9.16 the subjects of the predicates verum and fal-
sumremain objects. At 2.11.19-21 disciplines, in particular grammar and dialectic, are said
to be true; since disciplines include defnitions and divisions and processes of reason-
ing, we could welcome at this stage a frst implicit recognition of the application of
verumto propositional items. Such recognition becomes explicit only some sections lat-
er: Medeas fight on snakes, being something entirely non-existent, cannot even be
called false, but the sententia describing it can be said to be false, and there is a great
diference between the things which are said and those about which we say something
(2.15.29). Only towards the end of the Soliloquia and well after T32, then, are proposi-
tional items like sententiae (declarative sentence-tokens) explicitly accepted among the
bearers of falsehood and truth: however, no account of verum fully relevant to its ap-
parent use in T32 has emerged.
On the basis of what we have just seen, let us examine two possible analyses of pre-
miss (1), the kernel of Ratios argument:
(1a) If truth perishes, <the sentence> Truth has perished will then be true;
(1b) If truth perishes, it will then be the case that truth has perished.
(1a) relies on the possibility of taking verum as a predicate expressing a property of sen-
tentiae, although we have seen that this is not the most immediate reading of Augus-
tines own wording. (1b) is based, instead, on a possible extension of the ontological
notion of truth as real existence, applied by Augustine only to objects, to entities like
facts or obtaining states of afairs: in our case, the state of afairs described by the past
infnitive veritatem occidisse (that truth has perished).
I suggest that on both interpretations Ratios starting move is open to serious criti-
cism. On reading (1a), it is not difcult to imagine a possible scenario which would
falsify this conditional: on the assumption that verum is a predicate of sentence-tokens,
if all veritas perished (let us say, because the whole world and mankind are annihilated),
1 This is not to deny that it is often difcult to determine the supposition of an expression, that is whether it
is intended to be concerned with a real state of afairs, the proposition or concept formed in the mind in thinking
of the state, or autonymously, with itself as a linguistic item (Charron, Doyle 1993, p. 245).
self-refutation arguments from democritus to augustine 65
there would not be any existing sentence like Truth has perished thereafter, and so, a
fortiori, no such true sentence.1
Reading (1b) looks more promising. If vera are existing objects and obtaining states
of afairs, the complete obliteration of reality is one would think what would be re-
quired for the antecedent of (1b) (Truth perishes) to become true; on such a hypoth-
esis, however, the consequent would still be true, since it would then be the case that
reality and truth have perished. Things being so, veritas ipsa, which, on this reading,
would be the principle by which whatever is the case is the case, could never perish. De-
spite its apparent smoothness, this train of reasoning risks derailing, however. To begin
with, the notion of tensed states of afairs (that truth perished) to which it appeals
sounds inherently odd (one could even protest that the universal annihilation presup-
posed by (1b)s antecedent would not spare time, any reference to it after the hypothet-
ical destruction of veritas thus becoming illicit).2 A second, related concern regards a
tacit presupposition on which Augustines argument would rely: that the past is im-
mutable and cannot be erased. Without such a presupposition, which, although emi-
nently reasonable and widespread in ancient thought,3 is not a logical truth, the idea
that veritas could never become non-existent unless it frst perished is no longer unas-
sailable. For one could postulate a sudden annihilation of all the vera, present and past:
thereafter (so to speak), it would not be the case that truth has perished, because its past
existence would have been cancelled too, and what never existed cannot have perished.
One might try to overcome these problems and strengthen Ratios argument by sug-
gesting that Veritas has perished is not to be taken literally, as a description of a past
event, but is equivalent to Veritas does not exist (or, what amounts to the same, Noth-
ing is verum); this would make Ratios reasoning even more similar to Mackies, with
the elimination of the signifcant diference in the use of tenses. I believe this would not
be a commendable strategy, since Mackies own version of the argument is actually
weaker than Augustines. If verum is what exists or is the case, as opposed to what does
not exist or is not the case, on the hypothesis that veritas does not exist and nothing is
verum reality is, so to speak, a complete blank. To postulate the existence of some kind
of second-order verum, that nothing (or, avoiding self-reference, nothing else) is verum,
means to empty verum and veritas of their supposed role: although of course nothing
precludes this kind of prefxability from a merely formal point of view, we must consider
what it would amount to. Consider the following example. Socrates is not alive, does
no longer exist (Augustine would perhaps be ready to say that Socrates is not verus); is-
nt there a state of afairs which now obtains, i.e. Socrates not being alive? One might
reasonably reply that there is no state of afairs which obtains today which makes the
sentence Socrates is alive false: Socrates is alive is false today because the state of af-
fairs that Socrates is alive does not obtain, and not because some mysterious negative
state of afairs (that Socrates is not alive) obtains. To attach the adjective verum to nega-
tive states of afairs, albeit formally unimpeachable, is to empty it of the meaning and
function we are attributing to it here: for, as a consequence of this move, veritas would
become a principle of both existence (it is because of veritas that it is the case that I am
1 Watson (1990, p. 184) takes reading (1a) for granted and proposes an analogous criticism. This objection is
similar to that which I have formulated in sect. 2 against Mackies argument on reading (b).
2 Augustine could have been sympathetic towards such a line of reasoning, if we judge by his argument for
the non-eternity of time in Confessions 11 and De civitate 11, according to which it makes no sense to ask what God
did before creating the world and time (cf. p. 67n4). 3 Cf. e.g. Arist. ne vi 2, 1139b8-11.
66 luca castagnoli
alive) and of non-existence (it is because of veritas that it is the case that Socrates is not
alive). Moreover, to crowd ones ontology with a swarm of negative states of afairs ob-
taining (being vera) whenever the corresponding positive ones do not obtain seems to
be an unavoidable toll for Augustines argument on this reading of its main premiss:1
but in the presence of such an unparsimonious ontology we need not be fanatic Ock-
hamists to shiver (not to speak of the oddness of describing non-linguistic items like
state of afairs as negative).2 Ratios move, on interpretation (1b) of its crucial premiss,
would be questionable also for a contextual reason. In the Soliloquia Augustines ulti-
mate goal when arguing for the imperishability of veritas is establishing the immortali-
ty of the soul: if veritas is imperishable, and veritas must dwell, inseparably, in our souls,
then our souls must themselves be immortal (2.19.33). But the existence of that veritas
established on interpretation (1b), i.e. the existence of at least some verumin the formal
sense just discussed, cannot guarantee the existence of souls, since that only verum
which is guaranteed to exist forever is an entity which need not (and perhaps cannot)
dwell in our souls (the supposedly true fact that everything, including us, has been an-
nihilated). Augustine seems to ofer some revision, when he claims, late in the second
book, that bona fde vera are only the objects of the disciplinae, e.g. geometrical entities
like squares and circles. But this idea is impossible to reconcile with the use of verum
which Ratio has made in her self-refutation argument in T32. If that argument works,
it guarantees the eternal existence of one and only one verum which is irrelevant to the
proof of the immortality of the soul: for its validity would be perfectly consistent with
a scenario in which the whole of reality has been obliterated, and with it all the vera, ex-
cept for one single verum, the fact that every other verum has perished. To make a com-
parison, the inscription on a sheet of paper All writing has been cancelled from this
sheet cannot be true, being itself a surviving instance of writing, but this by no means
proves that there must be some other inscription on that sheet of paper (or that it is im-
possible to erase all the other inscriptions originally written on it).3
Augustine himself might have been aware of at least some of the difculties of Ra-
tios strategy I have denounced. After a summary of T32 at 2.15.28, his plea for more
time to assess the merits of the argument and his promise that, although he himself
could not fnd any objection, he will make sure that learned and prudent men read
these things and correct any rashness of ours there may be could be interpreted as
hints at some genuine perplexity or dissatisfaction.4
1 Notice that Mackie faces analogous problems on some interpretations of his T-prefxability law (cf. sect. 2).
2 For example, on the basis of Metaph. 29, 1024b17-21 Crivelli convincingly argues that the only states of
afairs recognised by Aristotle are afrmative states of afairs (2004b, pp. 49-50). This is not to say that an on-
tology which allows the existence of negative facts (obtaining negative states of afairs) corresponding to false
propositions and making them false is sheer nonsense. Such an ontology was defended, for example, by Russell
in his 1918 lectures on logical atomism (cf. Russell 1956, p. 211).
3 The parallel is not perfect, because in my example also the sentence All writing has been cancelled from this
sheet could be cancelled, whereas the pivot of our self-refutation argument is precisely that the fact that veritas
has perished could never be cancelled even if veritas perished.
4 Ratios argument exhibits some structural resemblance with this argument for the eternity of time: if time
was created, there was a time before then in which no time existed; if time will perish, there will be a time after
then in which time will not exist; therefore time neither was created nor will perish (cf. e.g. s.e. m 10.189). Since
Augustine rejected this line of reasoning (cf. e.g. civ. dei 11.4-6; 12.16; conf. 11.13.15; 11.30.40), he could have been
doubtful about endorsing something analogous about truth.
Augustines appeal to docti atque prudentes viri could not lapse unheard: in the Middle Ages a handful of illus-
trious readers of his Soliloquia eagerly returned to Ratios self-refutation argument, either to endorse it (e.g.
self-refutation arguments from democritus to augustine 67
8. Conclusion
We have fnally reached the end of our exploration of the ancient history and logic of
the self-refutation charge against Everything is true, Everything is false, and cog-
nate theses.1 We started from Mackies infuential account of absolute self-refutation,
on the form and content of which I have expressed some perplexities (sect. 2). We have
then discovered that, contrary to what is often maintained in the literature, Mackies
approach based on T-prefxability and Consequentia Mirabilis fnds no clear parallel in the
ancient texts, with the unique, remarkable and late exception of Augustines Soliloquia
(sect. 7).2 I have argued that this conspicuous absence does not betray a defect of logi-
cal rigour which prevented the ancients from achieving full consciousness of the logi-
cal form of their self-refutation arguments, and thus from moulding them into more
precise, fully formalised Mackie-style shape. Through careful re-examination of the
textual evidence in sects. 3-6 I have tried to prove something which has been obscured
by the uncritical adoption of the modern paradigm as the only guidance to our inter-
pretation of the ancient testimonies: the ancient approach to self-refutation, which on-
ly from the Hellenistic age came to be widely identifed by the label but has
been revealed to be quite unitary in some of its basic features, is not a muddled ances-
tor of the modern one, but difers from it in philosophically interesting ways. Not on-
ly was Mackies strategy in the pattern of Consequentia Mirabilis never adopted before
Augustine as a matter of fact, but it could not have been adopted by anyone accepting
certain fundamental non-classical features entrenched in the two most prominent an-
cient logical systems, the Aristotelian and the Chrysippean (sect. 6). Ancient self-refu-
tation arguments do not aim at establishing, in vacuo, the truth-value (necessary false-
hood) of the propositional content of theses like Everything is true or Nothing is
true; they criticise such theses as dialectical losers, because whoever dares to propose
and defend them can be forced into admitting their contradictories and rejecting them
as a direct consequence of proposing or trying to defend them. In other terms, the an-
cient self-refutation charge, at least in so far as it was applied to the kind of extremist
theses which have been the subject-matter of our inquiry here,3 does not aim at estab-
lishing by logic some absolute truth about the world, but at clarifying what can and can-
Anselm, Thomas) or to criticise it (e.g. Scotus). I cannot analyse this fascinating history here (for which cf. Char-
ron, Doyle 1993 and Castagnoli 2005). I only signal that
(1) one of the keenest supporters of the argument, Bonaventure of Bagnoregio, seems to misinterpret it in such
a way as to make it extraordinarily similar to Sextus diferent dialectical argument in T31 (cf. e.g. Myst. Trinit. q. i,
art. 1, ad n. 5 e contrario; contra Charron, Doyle 1993, pp. 250-252);
(2) at Ordin. i, d. 2, p. 1, q. 2, n. 45 John Duns Scotus rejected a proof for the necessary existence of veritas sim-
ilar to Augustines with considerations analogous to those I have illustrated above. The idea that if nothing exist-
ed and nothing were true then there would be no intellect which could truly judge this (and thus no bearer of
what Scotus calls formal truth) matches my criticism of the soundness of Ratios argument on the assumption
that verum is a predicate of sentence-tokens. Scotus remark that if nothing existed then veritas in re (identifed with
existence and founding reality) would not exist either seems to be based on a tacit refusal to attribute any posi-
tive ontological status, deserving the label verum, to wholly negative facts like the fact that nothing exists.
1 For an overview of the same subject in medieval and modern times cf. DAgostini 2002.
2 This exception is even more signifcant in the light of the fact that Augustine fathered another non-dialecti-
cal self-refutation argument, the famous Si fallor, sum, based on an implicit application of cm (cf. lib. arb. 2.3.7;
civ. dei 11.26; trin. 15.12.21), on which cf. Matthews 1992, pp. 32-33, Harrison 1999 and Castagnoli 2005.
3 I only mention, dogmatically, that my broader research in Castagnoli 2005 showed that the same conclu-
sion holds good for ancient self-refutation arguments quite generally.
68 luca castagnoli
not be successfully entertained in dialectical exchange, the original locus of philosoph-
ical inquiry. The dialectical context, thus, is not simply the broad and natural background
on which self-refutation arguments happened to be formulated, but it is typically a nec-
essary condition for them to work properly. Although ancient self-refutation arguments
cannot falsify our most radical adversaries views (and defuse our own most hyper-
bolical doubts) by proving that what they envisage is logically impossible, they can si-
lence them, by delimiting the area of constructive philosophical inquiry and debate.
And even if silencing your adversary does not necessarily amount to proving the truth
of your own position, it can be something extremely valuable if your position is already
the default one, and therefore you do not need to win new ground, but only to with-
stand the siege. However, the self-refutation arguments themselves were not perceived
as philosophical wonders by the ancients; it is the self-refuting positions that were seen
as amazing in their hopelessness.
A thorny question is what level of consciousness we are entitled to attribute to the an-
cients themselves of anything like my distinction between an absolute proof of the false-
hood of a certain proposition and a dialectical silencer of its supporters. Although no
such distinction is explicitly articulated in any of our sources, I have suggested that it
might be refected in the caution with which the results of self-refutation are often cast
in terms of the admissions or concessions to which the proponents of certain theses are
fnally bound. However, we have also noticed an opposite tendency to elliptical formu-
lations which are likely to blur that distinction: it remains hard to decide whether such
a tendency refects some confusion between the two distinct levels or testifes exactly
to the opposite, i.e. a dialectical setting for self-refutation was so obvious that making
it fully explicit was sometimes felt as unnecessary.
I have praised the ancient approach to self-refutation for its modesty in setting it-
self a dialectical goal which can be achieved with available dialectical resources. This
does not mean that all the self-refutation manoeuvres we have encountered in sects.
3-6 easily fulfl their proposed task. A supposed advantage of self-refutation arguments
is that they exploit only what ones opponent has already admitted, or is already com-
mitted to conceding, in virtue of his presenting his own position; to agree on the pre-
cise extent of ones dialectical commitments, however, can be no less difcult than
agreeing on the truth-value of the premisses of an ordinary direct refutation. I have
often signalled the possibility for the target of the self-refutation charge to protest that
the argument deployed against him is actually question-beggingly foisting upon him
tacit assumptions to which not only has he never committed himself, either explicitly
or implicitly, but which he has actually rejected, at least by implication, in the very act
of presenting his revisionist views. The idea that all self-refutation arguments could,
or should, avail themselves exclusively of the self-refuting thesis is illusory; their suc-
cess will depend on the degree to which the further presuppositions involved are per-
ceived by all the parties to the debate (including the audience) as the immovable back-
ground conditions for any discussion on a certain subject to take place and remain an
intelligible and genuine discussion on that subject, and, correspondingly, on the degree
to which any attack on these presuppositions is bound to appear as a desperate and
merely ad hoc tactic to avoid admitting an only too clear defeat.
Department of Classics and Ancient History Durham University
self-refutation arguments from democritus to augustine 69
Abercrombie N., Saint Augustine and French Classical Thought, Oxford, Clarendon Press, 1938.
Aubenque P., Sur linauthenticit du livre K de la Mtaphysique, in P. Moraux, J. Wiesner (eds.),
Zweifelhaftes im Corpus Aristotelicum: Studien zu einigen Dubia. Akten des 9. Symposium Aris-
totelicum (Berlin, 7-16 September 1981), Berlin-New York, de Gruyter, 1983, pp. 318-344.
Baltzly D. C., Aristotle and Platonic Dialectic in Metaphysics 4, Apeiron, 32 (1999), pp. 171-202.
Barnes J., The catalogue of Chrysippus logical works, in K. A. Algra, P. W. van der Horst, D. T. Ru-
nia (eds.), Polyhistor: Studies in the History and Historiography of Ancient Philosophy, Leiden-New
York-Kln, Brill, 1996, pp. 169-184.
Barnes J., Logic and the Imperial Stoa, Leiden-New York-Kln, Brill, 1997.
Barnes J., Argument in Ancient Philosophy, in D. Sedley (ed.), The Cambridge Companion to Greek
and Roman Philosophy, Cambridge, Cambridge University Press, 2003, pp. 20-41.
Barnes J., What is a disjunction?, in D. Frede, B. Inwood (eds.), Language and Learning: Philosophy
of Language in the Hellenistic Age, Cambridge, Cambridge University Press, 2005, pp. 274-298.
Bellissima F., Pagli P., Consequentia Mirabilis. Una regola logica tra matematica e flosofa, Firen-
ze, Olschki, 1996.
Bemelmans R., Why Does Protagoras Rush Of? Self-Refutation and Haste in Plato, Theaetetus 169a-
171d, Ancient Philosophy, 22 (2002), pp. 75-86.
Bett R., Sextus Empiricus. Against the Logicians, Cambridge, Cambridge University Press, 2005.
Bobzien S., Stoic Syllogistic, Oxford Studies in Ancient Philosophy, 14 (1996), pp. 133-192.
Bolton R., Metaphysics as a Science, in T. Scaltsas, D. Charles, M. L. Gill (eds.), Unity, Identity, and
Explanation in Aristotles Metaphysics, Oxford, Clarendon Press, 1994, pp. 321-354.
Bonney W. L., Operational Self-Refutation, Philosophical Quarterly, 16 (1966), pp. 348-351.
Bostock D., Platos Theaetetus, Oxford, Clarendon Press, 1988.
Boyle J. M. Jr., Self-Referential Inconsistency, Inevitable Falsity and Metaphysical Argumentation,
Metaphilosophy, 3 (1972), pp. 25-42.
Burnet J., Platonis Opera: Tomus i, tetralogias i-ii continens, Oxford, Oxford University Press, 1900.
Burnyeat M. F., Protagoras and Self-Refutation in Later Greek Philosophy, Philosophical Review,
85 (1976), pp. 44-69. [1976a]
Burnyeat M. F., Protagoras and Self-Refutation in Platos Theaetetus, Philosophical Review, 85
(1976), pp. 172-195. [1976b]
Burnyeat M., The Theaetetus of Plato, translation by M. J. Levett, Indianapolis-Cambridge,
Hackett, 1990.
Burnyeat M. F., Antipater and Self-Refutation: Elusive Arguments in Ciceros Academica, in B. In-
wood, J. Mansfeld (eds.), Assent and Argument: Studies in Ciceros Academic Books. Proceedings
of the 7
Symposium Hellenisticum (Utrecht, August 21-25, 1995), Leiden, Brill, 1997, pp. 277-310.
Burnyeat M. F., Dissoi logoi, in E. Craig (ed.), Routledge Encyclopaedia of Philosophy in cd-rom,
London, Routledge, 1998.
Burnyeat M. F., Plato on how not to speak of what is not: Euthydemus 283a-288a, in P. Pellegrin, M.
Canto-Sperber (eds.), Le Style de la Pense, Paris, Belles Lettres, 2002, pp. 40-66.
Campbell L., The Theaetetus of Plato, Oxford, Clarendon Press, 18832.
Cassin B., Il senso di Gamma e la strategia di Aristotele contro i Presocratici in Metafsica iv, Rivista
di Filosofa Neo-scolastica, 85 (1993), pp. 533-565.
Cassin B., Narcy M., La dcision du sense: Le livre Gamma de la Mtaphysique dAristote, Paris,
Vrin, 1989.
Castagnoli L., Self-bracketing Pyrrhonism, Oxford Studies in Ancient Philosophy, 18 (2000), pp.
Castagnoli L., Protagoras Refuted: How Clever is Socrates Most Clever Argument at Theaetetus
171a-c?, Topoi, 23 (2004), pp. 3-32. [2004a]
Castagnoli L., Review of Nasti 2002, Elenchos, 25 (2004), pp. 182-195. [2004b]
70 luca castagnoli
Castagnoli L., Il condizionale crisippeo e le sue interpretazioni moderne, Elenchos, 25 (2004), pp.
353-395. [2004c]
Castagnoli L., The Logic of Ancient Self-Refutation: From Democritus to Augustine, Ph.D. thesis,
Cambridge, 2005 (University of Cambridge Library and Faculty of Classics Library).
Castagnoli L., crisippea e tesi di Aristotele, in La logica nel pensiero antico (Atti del i
Colloquio, Roma, 28-29 Novembre 2000), Napoli, Bibliopolis, 2008 (forthcoming).
Cavini W., Chrysippus on Speaking Truly and the Liar, in K. Dring, T. Ebert (eds.), Dialektiker und
Stoiker: Zur Logik der Stoa and ihrer Vorlufer, Stuttgart, Steiner, 1993, pp. 85-109.
Cavini W., Principia Contradictionis: Sui Principi Aristotelici della Contraddizione ( 1-3), Anti -
quorum Philosophia, 1 (2007), pp. 123-169.
Cavini W., Principia Contradictionis: Sui Principi Aristotelici della Contraddizione ( 4), Anti -
quorum Philosophia, 2 (2008) (forthcoming).
Champlin T. S., Refexive Paradoxes, London-New York, Routledge, 1988.
Chance T. H., Platos Euthydemus: Analysis of What Is and Is not Philosophy, Berkeley-Los Ange-
les-Oxford, University of California Press, 1992.
Chappell T. D. J., Reading Platos Theaetetus, Indianapolis-Cambridge, Hackett, 2005.
Chappell T. D. J., Reading the : Theaetetus 170c-171c, Phronesis, 51 (2006), pp. 109-139.
Charron W., Doyle J. P., On the Self-Refuting Statement There is No Truth: A Medieval Treatment,
Vivarium, 31 (1993), pp. 241-266.
Cherniss H. F., Aristotles Criticism of Presocratic Philosophy, Baltimore, Johns Hopkins, 1935.
Conley T. M., Dating the so-called Dissoi Logoi: A Cautionary Note, Ancient Philosophy, 5 (1985),
pp. 59-65.
Cooper G. L. iii, Attic Greek Prose Syntax, after K. W. Krger, 2 vols., Ann Arbor, University of
Michigan Press, 1988.
Cornford F. M., Platos Theory of Knowledge: The Theaetetus and the Sophist of Plato translated
with a running commentary, London, Routledge & Kegan Paul, 1935.
Crivelli P., Aristotle on the Liar, Topoi, 23 (2004), pp. 61-70. [2004a]
Crivelli P., Aristotle on Truth, Cambridge, Cambridge University Press, 2004. [2004b]
DAgostini F., Disavventure della verit, Torino, Einaudi, 2002.
Dancy R. M., Sense and Contradiction: A Study in Aristotle, Dordrecht, Reidel, 1975.
De Praetere T., The Demonstration by Refutation of the Principle of Non-Contradiction in Aristotles
Metaphysics, Book iv, Logique et Analyse, 36 (1993), pp. 343-358.
Denyer N., Language, Thought and Falsehood in Ancient Greek Philosophy, London-New York,
Routledge, 1991.
Dis A., Platon: Oeuvres complte, Tome viii - 2e Partie: Thtte, Paris, Les Belles Lettres, 1924.
Duke E. A., Hicken W. F., Nicoll W. S. M., Robinson D. B., Strachan J. C. G., Platonis Ope-
ra: Tomus i, tetralogias i-ii continens, Oxford, Oxford University Press, 1995.
Emilsson E. K., Platos Self-Refutation Argument in Theatetus 171A-C Revisited, Phronesis, 39
(1994), pp. 136-149.
Fine G., Platos Refutation of Protagoras in the Theaetetus, Apeiron, 31 (1998), pp. 201-234.
Finnis J. M., Scepticism, Self-Refutation, and the Good of Truth, in P. M. S. Hacker, J. Raz (eds.), Law,
Morality, and Society, Oxford, Clarendon Press, 1977, pp. 247-267.
Ford A., Protagoras Head: Interpreting Philosophic Fragments in the Theaetetus, American Journal
of Philology, 115 (1994), pp. 199-218.
Gottlieb P., The Principle of Non-Contradiction and Protagoras: The Strategy of Aristotles Meta-
physics iv 4, Proceedings of the Boston Area Colloquium in Ancient Philosophy, 8 (1992),
pp. 183-198.
Graham A. C., Disputers of the Tao: Philosophical Arguments in Ancient China, LaSalle, Open
Court, 1989.
Grant C. K., Pragmatic Implication, Philosophy, 33 (1958), pp. 303-324.
Green M. S., Williams J. N. (eds.), Moores Paradox: New Essays on Belief, Rationality, and the First
Person, Oxford, Oxford University Press, 2007.
self-refutation arguments from democritus to augustine 71
Harbsmeier C., Science and Civilisation in China, Vol. 7, Part 1: Language and Logic, Cambridge,
Cambridge University Press, 1998.
Harrison S., Do we have a Will? Augustines Way in to the Will, in G. B. Matthews (ed.), The Augus-
tinian Tradition, Berkeley-Los Angeles-London, University of California Press, 1999, pp. 195-205.
Hawtrey R. S. W., Commentary on Platos Euthydemus, Ann Arbor, Edwards Brothers, 1981.
Herrnstein Smith B., Unloading the Self-Refutation Charge, in R. T. Ames, W. Dissanayake (eds.),
Self and Deception: A Cross-Cultural Philosophical Enquiry, Albany, suny Press, 1996, pp. 143-160.
Hintikka J., Knowledge and Belief, Ithaca, Cornell University Press, 1962.
Hughes G. E., John Buridan on Self-reference: Chapter Eight of Buridans Sophismata, with a Trans-
lation, an Introduction, and a philosophical Commentary, Cambridge, Cambridge University
Press, 1982.
Irwin T. H., Aristotles First Principles, Oxford, Clarendon Press, 1988.
Jaeger W., Aristotle: Fundamentals of the History of his Development, Oxford-London, Oxford
University Press, 1934.
Johansson I., Performatives and Antiperformatives, Linguistics and Philosophy, 26 (2003), pp.
Johnstone H. W. jr., Self-Refutation and Validity, Monist, 48 (1964), pp. 467-485.
Johnstone H. W. jr., Self-Application in Philosophical Argumentation, Metaphilosophy, 20 (1989),
pp. 247-261.
Jrgensen J., Some Refections on Refexivity, Mind, 63 (1953), pp. 289-300.
Kahn C. H., Some Puzzles in the Euthydemus, in T. M Robinson, L. Brisson (eds.), Plato: Euthy-
demus, Lysis, Charmides. Proceedings of the v Symposium Platonicum, Sankt Augustin, Ac-
ademia Verlag, 2000, pp. 88-97.
Ketchum R. J., Platos Refutation of Protagorean Relativism: Theaetetus 170-171, Oxford Studies
in Ancient Philosophy, 10 (1992), pp. 73-105.
Kirwan C., Augustine against the Skeptics, in M. Burnyeat (ed.), The Skeptical Tradition, Berkeley-
Los Angeles, University of California Press, 1983, pp. 205-223.
Kneale W., Aristotle and the Consequentia Mirabilis, Journal of Hellenic Studies, 77 (1957), pp.
Koyr A., The Liar, Philosophy and Phenomenological Research, 6 (1946), pp. 344-362.
Lear J., Aristotle and Logical Theory, Cambridge, Cambridge University Press, 1980.
Lee E. N., Hoist With his Own Petard: Ironic and Comic Elements in Platos Critique of Protagoras
(Tht. 161-171), in E. N. Lee, A. P. D. Mourelatos, R. M. Rorty (eds.), Exegesis and Argument: Stud-
ies in Greek Philosophy presented to Gregory Vlastos, Assen, Van Gorcum, 1973, pp. 225-261.
Lee M.-K., Epistemology After Protagoras: Responses to Relativism in Plato, Aristotle, and Democritus,
Oxford, Clarendon Press, 2005.
Levi A. J., On Twofold Statements, American Journal of Philology, 61 (1940), pp. 292-306.
Long A., Refutation and Relativism in Theaetetus 161-171, Phronesis, 49 (2004), pp. 24-40.
ukasiewicz J., Aristotles Syllogistic: From the Standpoint of Modern Formal Logic, Oxford, Claren-
don Press, 19572.
ukasiewicz J., Remarks on Many-Valued Systems of Propositional Logic, in O. Wojtasiewicz (ed.),
Jan ukasiewicz. Selected Works, Amsterdam-London, North-Holland, 1970, 153-178 [English
translation of Philosophische Bemerkungen zu den mehrwertigen Systemen des Aussangenkalkulus,
Monatshefte fr Mathematik und Physik, 37 (1930), pp. 51-77].
Mackie J. L., Self-Refutation A Formal Analysis, Philosophical Quarterly, 14 (1964), pp. 193-203
[reprinted in J. L. Mackie, P. Mackie (eds.), Logic and Knowledge, Oxford, Clarendon Press, 1985,
pp. 54-67].
Matthews G. B., Thoughts Ego in Augustine and Descartes, Ithaca-London, Cornell University
Press, 1992.
McCabe, M. M, Platos Individuals, Princeton, Princeton University Press, 1994.
McCabe M. M., Silencing the Sophists: The Drama of the Euthydemus, Proceedings of the Boston
Area Colloquium in Ancient Philosophy, 14 (1998), pp. 139-168.
72 luca castagnoli
McCabe M. M., Plato and His Predecessors: The Dramatisation of Reason, Cambridge, Cambridge
University Press, 2000.
McCall S., Connexive Implication, Journal of Symbolic Logic, 31 (1966), pp. 415-433.
McDowell J., Plato. Theaetetus, Oxford, Clarendon Press, 1973.
McPherran M. L., Skeptical Homeopathy and Self-refutation, Phronesis, 32 (1987), pp. 290-328.
Mignucci M., Aristotele. Analitici Primi, Napoli, Lofredo, 1969.
Mignucci M., The Liar Paradox and the Stoics, in K. Ierodiakonou (ed.), Topics in Stoic Philosophy,
Oxford, Clarendon Press, 1999, pp. 54-70.
Moore G. E., Reply to my Critics, in P. A. Schilpp (ed.), The Philosophy of G. E. Moore, Evanston
(Ill.), Northwestern University, 1942.
Morel P.-M., Dmocrite et la recherche des causes, Paris, Klincksieck, 1996.
Narcy M., A qui la parole? Platon et Aristote face Protagoras, in B. Cassin (ed.), Positions de la
sophistique, Paris, Vrin, 1986, pp. 75-90.
Narcy M., Platon revu et corrig, in B. Cassin, M. Narcy, La dcision du sens: Le livre Gamma de
la Mtaphysique dAristote, Paris, Vrin, 1989, pp. 61-110.
Narcy M., Platon: Thtte, Paris, Flammarion, 19952.
Nasti De Vincentis M., Logica Scettica e Implicazione Stoica, in G. Giannantoni (ed.), Lo scetti-
cismo antico. Atti del convegno organizzato dal Centro di Studio del Pensiero Antico del
c.n.r. (Roma, 5-8 novembre 1980), 2 vols., Napoli, Bibliopolis, 1981, vol. ii, pp. 501-532.
Nasti De Vincentis M., Logiche della connessivit, Bern, Haupt, 2002.
Newman J., The Recoil Argument, Apeiron, 16 (1982), pp. 47-52.
Nuchelmans G., Theories of the proposition: Ancient and medieval conceptions of the bearers of truth
and falsity, Amsterdam-London, North Holland, 1973.
Nuchelmans G., Dilemmatic arguments: Towards a history of their logic and rhetoric, Amsterdam-
Oxford, North Holland, 1991.
Page C., On Being False by Self-Refutation, Metaphilosophy, 23 (1992), pp. 410-426.
Passmore J. A., Philosophical Reasoning, New York, Scribners, 1961.
Patzig G., Aristotle and Syllogisms from False Premisses, Mind, 68 (1959), pp. 186-192.
Polansky R. M., Philosophy and Knowledge: A Commentary on Platos Theaetetus, Lewisburg-Lon-
don, Bucknell University Press-Associated University Press, 1992.
Politis V., Aristotle and the Metaphysics, London-New York, Routledge, 2004.
Prior A. N., Epimenides the Cretan, Journal of Symbolic Logic, 23 (1958), pp. 261-266.
Prior A. N., The Possibly-True and the Possible, Mind, 78 (1969), pp. 481-492.
Rankin H. D, O , in G. B. Kerferd (ed.), The Sophists and Their Legacy, Wies-
baden, Steiner, 1981, pp. 25-37.
Reale G., Aristotele. La Metafsica, 3 vols., Milano, Vita e Pensiero, 1993.
Robinson T. M., Contrasting Arguments: An Edition of the Dissoi Logoi, New York, Arno Press,
Ross W. D., Aristotles Metaphysics, 2 vols., Oxford, Clarendon Press, 1924.
Russell B., The Philosophy of Logical Atomism, in R. C. Marsh (ed.), Logic and Knowledge: Essays
1902-1950, London, Allen & Unwin, 1956, pp. 177-281.
Rstow A., Der Lgner: Theorie, Geschichte, und Aufsung, Leipzig, Teubner, 1910.
Schreiber S. G., Aristotle on False Reasoning: Language and World in the Sophistical Refutations,
Albany, suny Press, 2003.
Sedley D. N., The Midwife of Platonism: Text and Subtext in Platos Theaetetus, Oxford, Claren-
don Press, 2004.
Smith R., Aristotles Prior Analytics, Indianapolis, Hackett, 1989.
Sorensen R. A., Blindspots, Oxford, Clarendon Press, 1988.
Sprague R. K., Platos use of fallacy: A study of the Euthydemus and some other dialogues, London,
Routledge & Kegan Paul, 1962.
Stack M., Self-Refuting Arguments, Metaphilosophy, 14 (1983), pp. 327-335.
Stopper M. R., Schizzi Pirroniani, Phronesis, 28 (1983), pp. 265-297.
self-refutation arguments from democritus to augustine 73
Stroud B., Transcendental Arguments, Journal of the History of Philosophy, 65 (1968), pp. 241-256
Vailati G., A proposito di un passo del Teeteto e di una dimostrazione di Euclide, Rivista di flosofa
e scienze afni, 6 (1904), pp. 5-15.
Vanderveken D., Illocutionary Logic and Self-Defeating Speech Acts, in J. R. Searle, F. Kiefer, M.
Bierwisch (eds.), Speech Act Theory and Pragmatics, Dordrecht, Reidel, 1980, pp. 247-272.
Waterfield R. A. H., Plato. Theaetetus, Harmondsworth, Penguin, 1973.
Waterlow S., Protagoras and Inconsistency: Theaetetus 171a6-c7, Archiv fr Geschichte der
Philosophie, 59 (1977), pp. 19-36.
Watson G., Saint Augustine. Soliloquies and Immortality of the Soul, Warminster, Aris & Phillips,
Wedin M. V., Some Logical Problems in Metaphysics , Oxford Studies in Ancient Philosophy,
19 (2000), pp. 113-162.
Wedin M. V., A Curious Turn in Metaphysics Gamma: Protagoras and Strong Denial of the Principle
of Non-Contradiction, Archiv fr Geschichte der Philosophie, 85 (2003), pp. 107-130.
Wedin M. V., Animadversions on Burnyeats Theaetetus: On the Logic of the Exquisite Argument, Ox-
ford Studies in Ancient Philosophy, 29 (2005), pp. 171-191.
White F. C., Self-refuting Propositions and Relativism, Metaphilosophy, 20 (1989), pp. 84-92.
Wohlrab M., Platonis Theaetetus, Leipzig, Teubner, 1891.
74 luca castagnoli
Louis-Andr Dorion
es commentateurs ont depuis longtemps relev que le Gorgias, notamment le pas-
sage 521c-522e, contient des allusions assez claires laccusation de corruption de
la jeunesse, au procs de Socrate et, partant, lApologie.1 Mais on ne semble pas stre
aperu que les rapports entre le Gorgias et lApologie sont beaucoup plus troits et quils
ne se limitent pas ces quelques allusions au procs de Socrate. En mme temps quil
expose les raisons de lhostilit de Platon lendroit de la rhtorique, le Gorgias expose
les principales raisons pour lesquelles Socrate refuse lavance davoir recours la rh-
torique pour assurer sa dfense. Mais quen est-il au juste de la dfense de Socrate dans
lApologie? Refuse-t-il vraiment davoir recours au type dargumentation rhtorique
quil condamne dans le Gorgias? Je me propose de montrer que la dfense de Socrate,
dans lApologie, gagne tre lue la lumire des distinctions que Socrate tablit, dans le
Gorgias, entre elenchos rhtorique et elenchos dialectique.
1. Inutilit de la rhtorique pour l

homme juste injustement accus

Lors de son entretien avec Polos, Socrate soutient (480a-d) que la rhtorique nest dau-
cune utilit pour lhomme coupable dun crime, car ce dernier ne doit pas chercher
chapper la peine qui sanctionne son injustice. Le coupable doit au contraire recon-
natre sa culpabilit et aller au-devant du chtiment. Pour un tel homme, la rhtorique
ne prsente aucune utilit, moins quil ny fasse appel pour saccuser lui-mme, ou en-
core pour accuser ses proches qui se seraient rendus coupables dactes injustes. Lutilit
que Socrate est prt concder la rhtorique est pour le moins inattendue.
Si lon conoit aisment que Socrate ne reconnaisse pas lutilit de la rhtorique pour
lhomme coupable qui cherche chapper aux rigueurs de la justice, lon stonne quil
ne la considre pas dans le cas dun homme juste qui serait injustement accus. Ce cas
de fgure, qui nest pas voqu lors de lentretien avec Polos, est fnalement soulev par
Callicls (486a-b) et il est difcile de ne pas songer au cas de Socrate, mme si trs peu
de commentateurs ont tabli un rapprochement entre ce cas de fgure et la situation de
Socrate dans lApologie.2 Or, l encore, et de faon plus inattendue, Socrate refuse de
reconnatre lutilit de la rhtorique. Ce refus tient essentiellement trois raisons.
Premirement, la vie bonne ne consiste pas vivre le plus longtemps possible (511b-c,
512d-e) et, de toute faon, le plus grand mal nest pas de subir une injustice (469b-c).
* Une version courte de cette tude est parue dans M. Erler & L. Brisson (eds.), Gorgias-Menon, Selected Papers
from the Seventh Symposium Platonicum, Sankt Augustin, Academia Verlag, 2007, pp. 284-289.
1 Cf. Dodds 1959, p. 372 ad 522d7; Irwin 1979, p. 240 ad 521c-d; Canto 1987, p. 302 n. 245; Dalfen 2004, pp.
2 Cf. Montuori 1981, p. 215: Gorgias thus joins the Apology in taking up the same basic theme of the moral
reasons which led Socrates to refuse to defend himself in a way that would please his judges.
Cette premire raison est galement prsente dans lApologie: Socrate y soutient quil
faut dabord et avant tout veiller ne pas commettre linjustice, plutt que chercher
se soustraire la mort par tous les moyens (39a), et que, de toute faon, la mort nest
probablement pas un mal. lexemple de ce quil soutient dans le Gorgias (522e), il
considre que la mchancet et linjustice sont beaucoup plus craindre que la mort
(28b-d, 32b-d, 39a-b).
Cette premire raison qui motive le refus de la rhtorique est galement formule
par Socrate dans lHippias majeur. Tout la fn du dialogue, Hippias soutient que ce qui
est beau et digne dune grande estime (e d ), le voici: cest dtre
capable de produire de bonne et belle faon un discours qui emporte la persuasion
() devant un tribunal ( ), devant le Conseil ou devant nimporte
quelle autre autorit laquelle on est susceptible dadresser un discours , et de se reti-
rer en ayant gagn non le moindre, mais le plus grand des prix (a a
): le salut de notre propre personne ( ), de nos biens et de nos
amis.1 La rponse de Socrate Hippias est tout fait conforme la position quil
nonce dans le Gorgias et lApologie:
Mais quand [] ayant t persuad par vous, je dis comme vous que ce qui lemporte sur tout,
cest dtre capable de produire de bonne et belle faon un discours pour lemporter devant un
tribunal ( ) ou devant nimporte quelle assemble, je nentends alors que des in-
sultes, venant aussi bien de la part des autres qui sont ici que de cet homme qui ne cesse de me
rfuter (d e ). Car cet homme, qui se trouve
tre un parent trs proche, vit dans la mme maison que moi (
J d ), et ds que je rentre chez moi et quil mentend parler de la sorte, il
me demande si je nai pas honte davoir laudace de discuter de ces belles conduites alors que jai
t si manifestement rfut propos du beau et que je ne sais mme pas ce quest la chose mme
dont je parle.2
Si cet homme3 soppose ce que Socrate suive les conseils des orateurs, cest parce
que Socrate prsumerait ainsi quil sait reconnatre un beau discours, alors quil ne
sait pas en quoi consiste le beau. Or vivre dans lignorance du beau, cest un tat pire
que la mort (304d-e). La mort nest donc pas un mal auquel il faut chapper tout
Deuximement, dans la mesure o elle est essentiellement une forme de fatterie (463a)
qui vise persuader un auditoire par lexpos de ce qui est agrable ce dernier, la rh-
torique fait ncessairement f de la vrit. Or Socrate ne cherche jamais plaire par son
langage, car ce nest pas le plaisir ou lagrable quil a en vue, mais le bien, ce qui ne
manquera pas, le cas chant, dirriter ses juges et de provoquer sa propre condamna-
tion (521d). Cette deuxime raison justifant le refus de la rhtorique est galement pr-
sente dans lApologie: Socrate y soutient (38d-e) que ce nest pas en raison de son inca-
pacit prononcer des discours rhtoriques quil a t condamn mort, mais parce
quil na pas eu lefronterie ni limpudence ncessaire pour sabaisser prononcer le
genre de discours, et user du genre de moyens (prires, larmes, supplications) qui plai-
sent aux juges au plus haut point ( x i b q , 38d) et
1 304a-b (trad. Pradeau). 2 304c-d (trad. Pradeau).
3 Lhomme qui adresse ainsi des reproches Socrate, et que ce dernier prsente comme un parent trs
proche, est en ralit une sorte de double fctif que Socrate se forge (Pradeau 2005, p. 25) et qui joue le rle
dun interlocuteur anonyme depuis 286c.
76 louis-andr dorion
qui les apitoient.1 Lopposition est donc trs nette, aussi bien dans le Gorgias que dans
lApologie, entre dune part une rhtorique fatteuse qui cherche uniquement plaire au
mpris de la vrit et, dautre part, une forme de discours qui satisfait aux exigences de
la vrit et qui ne cherche pas plaire. De fait, Socrate insiste de nombreuses reprises,
dans lApologie, sur la vrit de son discours2 et sur le fait que loin de chercher plaire
aux juges, il les irrite trs souvent (17d, 20e, 21a, 27b, 30c, 34c-d), ce qui est un indice,
parmi dautres, de son refus de la rhtorique.
Troisimement, lhomme juste injustement accus qui ferait appel la rhtorique pour
se dfendre utiliserait forcment des procds et des moyens qui sont non seulement in-
dignes de lui et des juges, mais aussi contraires la justice dans la mesure o ils risquent
dinciter les juges violer leurs serments de juger en conformit avec la loi (Apol. 35c).
Lhomme juste serait ainsi en contradiction avec lui-mme puisquil consentirait user
de moyens injustes pour se dfendre. Cette troisime raison justifant le refus de la rh-
torique est absente du Gorgias, mais elle nest pas incompatible avec le Gorgias. En efet,
comme le bon orateur, suivant Socrate, doit tre juste et savant en matire de justice
(508c), et que lhomme juste ne saurait commettre dinjustice (460b-c), il sensuit que le
bon orateur, ou tout homme juste, ne pourrait se rsoudre, dt-il le payer de sa vie,
employer des moyens injustes, ou de nature causer une injustice, pour se dfendre.
Si Socrate refuse dadmettre que la rhtorique puisse tre utile lhomme injuste (ac-
cus justement), de mme qu lhomme juste (injustement accus), est-ce dire quil
devrait lui-mme renoncer se dfendre sil tait injustement accus? La rponse cette
question est dj formule dans le Gorgias. Callicls qui soutient que la rhtorique est
la seule faon de se dfendre contre une accusation injuste, Socrate rtorque que si la
rhtorique est le seul moyen de se dfendre contre une telle accusation il naura rien
dire devant le tribunal (521d-e).3 Sa dfense, qui est aux yeux de Socrate la meilleure,
consistera rappeler quil na jamais commis dinjustice:
[Callicls] Et tu trouves, Socrate, que cest l un beau rle pour un homme de ne pouvoir se d-
fendre lui-mme ( J ) dans sa patrie! [Socrate] Oui, Callicls, la
condition quil possde cet autre moyen de dfense que tu as toi-mme reconnu plusieurs re-
prises, qui est de navoir aucune faute se reprocher, en paroles ou en actes, ni envers les dieux
ni envers les hommes; car cette manire de se dfendre soi-mme ( a
), ainsi que nous en sommes plusieurs fois tombs daccord, est la meilleure de toutes
(). (522c-d; trad. Croiset)
1 Cf. Gorgias 522d-e: mais si ma mort avait pour seule cause mon ignorance de la fatterie oratoire (
), je suis certain que tu me verrais accepter mon sort avec tranquillit. (trad. Croiset). Cette position
est conforme celle de Socrate dans lApologie, cette difrence prs: dans lApologie (38d), Socrate attribue sa
condamnation mort non pas lignorance de la rhtorique, mais son refus dy avoir recours.
2 Cf. 17b, 20d, 22a, 22b, 24a, 28a, 28d, 31c, 31e, 32a, 33c, 34b. On pourrait mobjecter que linsistance souli-
gner la vrit dun discours est elle-mme un procd rhtorique, de sorte que ces passages ne traduisent pas une
opposition la rhtorique, puisque Socrate se comporte lui-mme en rhteur en insistant sur la vrit de son dis-
cours. Mais dans la mesure o il considre que la rhtorique ne cherche pas connatre, mais uniquement plaire
et convaincre par tous les moyens, Socrate est dune certaine faon justif dinsister sur la vrit de son discours,
puisquil ne cherche pas plaire et quil nusera pas des moyens habituels par lesquels la rhtorique engendre la
persuasion. Lofce de lorateur est de dire la vrit (Apol. 18a), quand bien mme cette vrit se retournerait
contre lui.
3 je naurai rien dire au tribunal ( , 521e). Voir aussi 527a, o Socrate
fait nouveau allusion son incapacit de se dfendre et au mutisme dont il serait frapp au tribunal: tu [scil.
Callicls] resteras bouche be () et la tte perdue, pareil l-bas [scil. le tribunal des fls de Zeus] ce que
je serais moi-mme ici. (trad. Croiset)
elenchos dialectique et elenchos rhtorique 77
Il est vrai que Socrate rappelle, dans lApologie (32a-d), quil na jamais commis dinjus-
tice, mais sa dfense ne se rduit pas ce rappel, tant sen faut. Est-ce dire que le Gor-
gias, qui laisse entendre que la dfense de Socrate serait minimale et ngative, en ce sens
quelle se bornerait souligner quil na pas commis dinjustice, est daucune utilit pour
comprendre la faon dont Socrate se dfend au cours de son procs? Je crois au
contraire, et cest prcisment ce que je meforcerai de dmontrer dans les deux pro-
chaines sections, que le Gorgias permet de comprendre une dimension essentielle de la
dfense de Socrate dans lApologie.
2. La dialectique et le refus des tmoignages
Alors quil discute avec Polos, Socrate oppose lun lautre deux types delenchos (471e-
472c), soit lelenchos rhtorique qui est habituellement employ devant les tribunaux, et
lelenchos dialectique dont Socrate semble considrer quil est le seul le pratiquer. Sauf
erreur de ma part, aucun commentateur na jusqu maintenant relev limmense int-
rt de cette distinction pour la comprhension du type de dfense prsent par Socrate
loccasion de son procs.1 Voici comment Socrate caractrise lelenchos rhtorique:
Mon trs cher, tu essaies de me rfuter par des procds de rhtorique (
), comme le font ceux qui croient rfuter devant les tribunaux (
). L, un orateur croit rfuter ( ) son ad-
versaire quand il peut produire en faveur de sa thse des tmoins nombreux et considrables
( fd ) alors que lautre nen a quun seul ou point du tout. Mais
ce genre de rfutation (y b ) est sans valeur pour dcouvrir la vrit (e
e c ); car il peut arriver quun innocent succombe sous de faux
tmoignages nombreux et qui semblent autoriss (e d ). (471e-472a;
trad. Croiset)
Lelenchos rhtorique a donc pour caractristique essentielle dtre un discours dont
lautorit se fonde sur le nombre et le prestige2 des tmoins, et qui, pour cette raison
mme, est impuissant dcouvrir la vrit, puisque la vrit dun discours nest pas
fonction du nombre ni du prestige des hommes qui sont prts tmoigner en sa faveur.
Qui plus est, lelenchos rhtorique peut contribuer linjustice dans la mesure o il peut
arriver quun innocent soit condamn en raison dune multitude de faux tmoignages
ports contre lui. Cela confrme que lune des raisons pour lesquelles lhomme juste
injustement accus doit sabstenir de la rhtorique est que celle-ci risque de favoriser
linjustice. Dans la suite du Gorgias, Socrate fait de nombreuses allusions linutilit
dun grand nombre de tmoins pour fonder un elenchos (cf. 473d, 474a, 475e-476a, 523c,
523e). Le mpris de Socrate pour une dfense fonde sur des tmoignages est galement
manifeste dans le mythe eschatologique sur lequel se clt le dialogue. Avant la rforme
1 lexception de Dalfen (2004, p. 278), les commentateurs du Gorgias (Dodds 1959, Irwin 1979) et de lApo-
logie (Brickhouse & Smith 1989, Reeve 1989, de Strycker & Slings 1994) que jai consults ntablissent aucun
rapprochement entre ce passage du Gorgias et laspect formel de la dfense de Socrate dans lApologie. Dalfen fait
un bref rapprochement entre ce passage du Gorgias et lelenchos auquel Socrate soumet Mltos dans lApologie,
mais il ne soulve pas la question de savoir si la dfense de Socrate, dans lensemble de lApologie, est de bout en
bout fdle la distinction tranche quil tablit dans le Gorgias entre lelenchos rhtorique et lelenchos dialectique.
2 Cf. aussi Rp. ii 366b. Le fondement de lelenchos rhtorique le nombre et la rputation des tmoins rap-
pelle les deux critres qui servent, daprs Aristote, identifer une opinion autorise (), savoir le nom-
bre des hommes qui soutiennent une opinion et/ou leur rputation de sagesse (cf. Top. i 1, 100b21-23).
78 louis-andr dorion
du tribunal de lau-del, de nombreux hommes, ayant des mes mauvaises (a
a , 523c), sont accompagns de nombreux tmoins attestant quils ont
vcu selon la justice (d , , 523c).
Ces nombreux tmoins produisent donc autant de faux tmoignages qui contribuent
linjustice des jugements qui sont rendus. Aprs la rforme du tribunal ordonne par
Zeus, lme du dfunt comparat seule devant le juge (523e), sans tre assiste par des
tmoins. On ne saurait donc trop insister sur le rejet sans appel, de la part de Socrate,
des dfenses fondes sur des tmoignages.1
Ce mpris des tmoignages, dans un contexte dialectique, nest pas exclusif au Gor-
gias. Dans lHippias majeur, par exemple, Hippias est persuad que la position nonce
par Socrate est irrfutable dans la mesure o il peut compter sur le tmoignage de tous
les auditeurs: Comment pourrais-tu tre rfut, Socrate ( a , t ,
), si ce que tu dis est ce que tout le monde pense et si tous tes auditeurs en
attestent lexactitude (d
;)?2 Socrate ne prend pas la peine de contester Hippias sur ce point, mais il ne fait
aucun doute quil ne partage pas son avis. Lorsque Socrate invoque un tmoignage, que
ce soit le sien ou celui dautrui, pour tayer une position, ou encore pour en reconna-
tre la vrit, on peut demble le souponner de faire preuve dironie. Cest le cas, en
particulier, des deux passages suivants:
Je peux apporter mon tmoignage la vracit de ce que tu afrmes (
): votre technique a rellement fait progresser laptitude matriser de concert
les afaires publiques comme celles qui sont dordre priv. (Hippias maj. 282b; trad. Pradeau)
[Socrate] Tu vois, Hippias, que je dis la vrit quand jafrme que je suis insistant dans les
questions que je pose aux savants; et il est possible que je naie que ceci de bon, et que je sois
pour le reste bien faible, car je me trompe quant la condition des choses et je ne sais pas com-
ment elles sont. Une preuve sufsante en est que, lorsque je me trouve prs de lun dentre vous,
qui tes si clbres pour votre savoir ( d 0), comme tous les Grecs
peuvent en tmoigner ( E ), il devient manifeste que je ne sais
rien; car sur rien, pour ainsi dire, je nai les mmes opinions que vous: quelle preuve plus grande
dignorance que dtre toujours en dsaccord avec les savants? (Hippias min. 372b-c; trad. Fron-
Il est pour le moins douteux, dans le premier cas, que Socrate pense rellement que la
technique des sophistes a contribu faire progresser la matrise des afaires prives ou
publiques, et il nest pas moins clair, dans le second cas, que Socrate ne souscrit pas au
tmoignage unanime des Grecs et que ce tmoignage nest en aucune faon une preuve
ou une dmonstration de la comptence des sophistes.
Un court change entre Polmarque et Clitophon, au livre i de la Rpublique, confr -
me que le tmoignage des auditeurs est superfu lorsque linterlocuteur lui-mme
convient () de ce que lui soumet Socrate:
Par Zeus, Socrate, scria Polmarque, cest clair comme le jour. Oui, si tu lui apportes ton t-
moignage (), intervint Clitophon. En quoi Socrate a-t-il besoin de tmoignage
( )? continua Polmarque. Thrasymaque lui-mme convient () que les
1 Sur le mpris des tmoignages dans le cadre dune argumentation, cf. Lois i 638d. Platon est ambivalent len-
droit des tmoignages, puisquil lui arrive de les invoquer pour confrmer la valeur dun homme (cf. Time 20a-b).
2 288a (trad. Chambry).
elenchos dialectique et elenchos rhtorique 79
gouvernants prescrivent parfois des choses qui leur sont prjudiciables et quil est juste que les
sujets les excutent. (340a; trad. Chambry)
Revenons au Gorgias. lelenchos rhtorique pratiqu par Polos et qui est en vogue de-
vant les tribunaux, Socrate oppose le type delenchos quil prtend tre le seul prati-
quer et quil caractrise ainsi:
tu produis seulement contre moi une foule de faux tmoins ( ) pour
tcher de marracher mon bien et la vrit ( ). Moi, au contraire, si je nobtiens pas
ton propre tmoignage (), et lui seul, en faveur de mon afrmation, jestime navoir rien
fait pour la solution de notre dbat, non plus que toi du reste, si tu nobtiens pas lappui de mon
tmoignage, seul entre tous (a c x J ), et si tu ne renvoies pas tous
tes autres tmoins. Voil donc deux sortes de rfutation ( ), la premire laquelle
tu crois comme bien dautres, la seconde qui est la mienne. (472b-c; trad. Croiset modife)
Alors que lelenchos rhtorique ne permet aucune interaction entre les deux thses en
prsence, puisquil se contente de faire appel des tmoignages nombreux en faveur
dune thse et quil vise convaincre les juges, et non pas celui qui soutient la thse op-
pose, lelenchos dialectique se droule tout entier dans cette interaction, pour autant
quil ne peut pas progresser sans que chacun des deux interlocuteurs tmoigne, cest-
-dire donne son accord (, 472b7), en faveur de la proposition que lautre
lui soumet. Seul ce type delenchos semble propice la dcouverte de la vrit (486e). Un
peu plus loin, dans un passage dont nous verrons limportance pour le procs de So-
crate, Socrate oppose nouveau lelenchos de Polos et le sien:
Ne me demande donc pas maintenant de faire voter les auditeurs ( f ).
Si tu nes pas capable de mieux me rfuter ( c ), alors, comme
je te lai propos, laisse-moi le faire ta place, comme cela, tu auras lexprience de ce que doit
tre, daprs moi, une rfutation (d x g r r). En efet,
je ne sais produire quun seul tmoignage () en faveur de ce que je dis, cest celui de mon
interlocuteur, et jenvoie promener tous les autres; en outre, un seul homme, je sais le faire vo-
ter (d ), mais quand il y a plus de gens, je ne discute pas avec eux
( b b ).1
Ce que Socrate oppose, lorsquil dnonce lelenchos rhtorique, ce ne sont pas deux fa-
ons de produire un elenchos devant un tribunal, mais le type delenchos en usage devant
les tribunaux dune part, et lelenchos dialectique quil est en train de pratiquer avec Po-
los dans un contexte extra-judiciaire dautre part. Socrate nenvisage donc pas la ques-
tion du type delenchos quil choisirait sil avait se dfendre devant un tribunal, mais
cest une question que nous sommes en droit de nous poser et que nous pouvons tu-
dier dans le contexte de lApologie. tant donn que Socrate rejette sans appel lelenchos
rhtorique en usage devant les tribunaux, il ne devrait donc pas, du moins en principe,
1 474a (trad. Croiset). Epictte cite ce passage du Gorgias pour dcrire la mthode propre Socrate: Comment
donc agissait () Socrate? Il forait son interlocuteur tmoigner pour lui ( ) et navait nul be-
soin dun autre tmoin ( \ e ). Aussi pouvait-il dire: Bien du plaisir tous les au-
tres! Mon contradicteur me suft toujours moi comme tmoin (); quant aux autres, je ne rclame pas
leur sufrage ( ), je ne rclame que celui de mon interlocuteur. (ii 12, 5 = ssr i c 521; trad. Souilh).
Voir aussi ii 26, 6: Voil pourquoi Socrate, plein de confance en cette facult, disait: Je nai pas lhabitude
dinvoquer dautre tmoin de mes paroles ( z ), et je me contente toujours de celui qui discute
avec moi, je rclame son sufrage ( ), jen appelle son tmoignage (d ), et lui
seul il remplace pour moi tous les autres.
80 louis-andr dorion
y avoir recours pour dmontrer la fausset des accusations dont il est lobjet. De fait,
cest prcisment laide de lelenchos dialectique que Socrate entend dmontrer, dans
lApologie (24d-28a), que les accusations retenues contre lui sont sans fondement. Lors
de linterrogatoire de Mltos, Socrate ne fait appel aucun tmoin externe et lelenchos
dialectique auquel il soumet Mltos pour rfuter les accusations portes contre lui ne
considre quun seul tmoin, soit Mltos lui-mme.
Au terme de linterrogatoire de Mltos, Socrate estime tre parvenu dmontrer
que son accusateur se contredit sur chacun des trois chefs daccusation, de sorte quil se
croit justif de dclarer: Cela tabli, Athniens, il nest pas besoin dune dfense plus
longue ( r ) pour prouver que je ne suis pas coupa-
ble de ce dont maccuse Mltos dans sa plainte; ce que je viens de dire suft (a
a d ).1 premire vue, je ne vois aucune bonne raison de ne pas prendre
cette dclaration de Socrate au pied de la lettre. Au vu de la distinction et de lopposi-
tion quil tablit dans le Gorgias entre deux types delenchos, il suft en efet que Socrate
soit parvenu faire tmoigner son accusateur contre lui-mme, cest--dire lui faire
reconnatre quil se contredit, pour dmontrer du mme coup linanit des accusations
portes contre lui. Certes, nous pouvons lever des doutes sur la lgitimit ou la vali-
dit de telle ou telle tape dans la rfutation de Mltos, mais cela ne change rien au fait
que cette rfutation a valeur, aux yeux de Socrate, de dmonstration de la fausset des
accusations portes contre lui. De ce point de vue, je vois une remarquable continuit
entre, dune part, le Gorgias, o Socrate expose les raisons pour lesquelles la rhtorique
nest daucun secours lhomme qui doit se dfendre contre des accusations justes ou
injustes, et dautre part lApologie, o il expose ouvertement les motifs de son refus de
la rhtorique (34c-35d) et o il emploie, pour se dfendre, le seul type delenchos dont il
reconnat la lgitimit dans le Gorgias.
Mais sil est vrai, ainsi que Socrate le souligne dans le Gorgias, que lelenchos rhto-
rique est en vogue devant les tribunaux et quil est le seul pratiquer lelenchos dialec-
tique, il sensuit que la rfutation de Mltos, dans lApologie, est une procdure inhabi-
tuelle. Il est en efet possible de montrer que lApologie nest pas un texte reprsentatif
de ce qui avait habituellement cours devant les tribunaux, prcisment parce que la r-
futation de Mltos ne repose pas sur des tmoins, mais plutt sur la procdure de
l, laquelle navait pas la faveur des logographes en raison de son caractre im-
prvisible.2 Comme il tait impossible de prvoir lavantage de quel parti tournerait
l, lon estimait plus sage de fonder lelenchos sur des tmoins ainsi que lat-
teste abondamment la littrature judiciaire3 plutt que sur linterrogatoire de la par-
tie adverse. Or Socrate fait exactement le contraire et cest en ce sens quil subvertit
le mode dobtention habituel dun elenchos devant un tribunal.
1 28a (trad. Brisson). Socrate emploie la mme expression, la fn de la premire section (24b), pour signifer
quil en a termin avec les premiers accusateurs: Sur les accusations portes contre moi par mes premiers accu-
sateurs, ce que jai dit pour ma dfense doit vous sufre ( c e ).
2 Je renvoie ici le lecteur ce que je me suis eforc de dmontrer dans une tude antrieure (cf. Dorion 1990,
pp. 328-332, 336-341).
3 Cf. Antiphon vi 25; Ise vi 65; Lysias vii 11, vii 20; Dmosthne xxxv 27 et les nombreux passages que jai
signals dans une tude antrieure (cf. Dorion 1990, pp. 323-327).
elenchos dialectique et elenchos rhtorique 81
3. Lappel aux tmoins dans la dfense rhtorique de Socrate
Si la dfense de Socrate se limitait la rfutation de Mltos, laccord entre lApologie et
le Gorgias serait parfait. Mais comme la rfutation de Mltos nest que la deuxime
tape dune dfense en trois volets (18b-24a, 24b-28a, 28a-34b), il reste examiner si les
deux autres moments de la dfense de Socrate respectent la position dveloppe dans
le Gorgias.
La premire tape de la dfense de Socrate est la rfutation des premiers accusa-
teurs, cest--dire ceux qui ont t les premiers calomnier Socrate et qui ont puis-
samment contribu ternir sa rputation dans lopinion publique. Cest dailleurs pour-
quoi Socrate les considre plus redoutables encore que les accusateurs de 399 (18b).
Avant dentreprendre la rfutation de ces premiers accusateurs, Socrate se dsole de
leur absence car cela aura pour consquence quil ne pourra pas vraiment les rfuter:
Impossible, en efet, de faire monter cette tribune aucun dentre eux ni de le rfuter
( ). Mais, pour me dfendre (), je me trouve tout bonnement
contraint de me battre contre des ombres et de me lancer dans une rfutation sans per-
sonne pour me rpondre ( e ).1 Ce passage gagne tre
lu la lumire de la distinction que le Gorgias tablit entre deux elenchoi. Ce dont So-
crate se dsole, au seuil de sa dfense, cest de ne pas pouvoir soumettre ses premiers
accusateurs un elenchos dialectique. Comme lelenchos rhtorique se fonde essentielle-
ment sur des tmoignages, et quil peut trs bien faire lconomie dun change ou
dune interaction avec lautre parti, labsence des premiers accusateurs nest pas un obs-
tacle la conduite dun elenchos rhtorique. En revanche, comme lelenchos dialectique
requiert le tmoignage, cest--dire laccord de celui que lon cherche rfuter, lab-
sence des premiers accusateurs contraint Socrate rfuter sans personne pour lui r-
pondre, ce qui revient dire quil ne pourra pas sagir, proprement parler, dune r-
futation dialectique. Je conclus de ce passage que si les premiers accusateurs avaient t
prsents, et que la procdure le lui avait permis, Socrate naurait pas hsit les rfuter
dialectiquement, comme il le fera avec Mltos. Mais comme ils sont absents, et que So-
crate doit nanmoins se dfendre, il doit se rsigner rfuter sans personne pour lui r-
pondre, cest--dire sans faire tmoigner et voter ses adversaires.
La section 19a-24b se prsente bel et bien comme une rfutation des premiers
accusateurs. Il est dailleurs remarquable quau dbut et la fn de cette section, Socrate
emploie la mme expression dtruire la calomnie ( c c, 19a1;
c c , 24a3) pour exprimer lobjectif quil poursuit.
Comme Socrate est contraint de sengager dans un elenchos non dialectique, la question
se pose videmment de savoir sil consent user de lelenchos rhtorique quil dnonce
dans le Gorgias. Or force est de constater que Socrate nhsite pas invoquer des t-
moignages en sa faveur. Il invoque en efet des tmoins trois reprises (19d, 20e, 21a),
mais dans deux cas les tmoins invoqus ne me paraissent pas trs concluants dun point
de vue rhtorique. Lorsque Socrate invoque le tmoignage dApollon pour attester la
sagesse qui est la sienne, cela ressemble fort de la provocation ce dont Socrate semble
dailleurs conscient (cf. 20e4-5: c , a ).
Dun point de vue rhtorique, je doute quApollon soit un trs bon tmoin car il lui est
1 18d (trad. Brisson).
82 louis-andr dorion
videmment impossible de tmoigner en personne. Le deuxime tmoin que Socrate
invoque nest pas non plus trs persuasif: comme Chrphon est mort et quil ne peut
pas se porter lui-mme tmoin de la vracit du rcit de Socrate sur loracle, Socrate in-
voque le tmoignage du frre de Chrphon (21a), dont on ne sait pas sil tait lui-mme
prsent Delphes. Ce tmoignage semble donc assez fragile.1 Enfn, ces deux tmoi-
gnages ne satisfont pas lun des deux critres noncs dans le Gorgias, soit le critre du
nombre des tmoins.
Le troisime tmoignage me parat plus efcace que les deux prcdents. Alors quil
rcuse les allgations suivant lesquelles il se serait livr des recherches impies sur les
phnomnes clestes, Socrate fait appel des tmoins:
Mais en vrit, Athniens, ce sont des sujets dont je nai rien faire, et cest au tmoignage per-
sonnel de la plupart dentre vous que jen appelle ( b s f f
). Oui, je vous demande de tirer entre vous cette afaire au clair, vous tous qui une fois
ou lautre mavez entendu discourir; et parmi vous ils sont nombreux () ceux-l. Deman-
dez-vous les uns aux autres si jamais peu ou prou lun dentre vous ma entendu discourir sur de
tels sujets; cela vous permettra de vous rendre compte que tout ce que peuvent raconter la
plupart des gens sur moi est du mme acabit. (19c-d; trad. Brisson)
Ce tmoignage ne peut pas tre plac sur le mme pied que les deux prcdents, non
seulement parce quil nest ni ironique ni invrifable, mais aussi parce quil satisfait au
critre du nombre et celui de la rputation des tmoins. Qui plus est, Socrate ne fait
pas appel des tmoins extrieurs qui ne sont pas directement impliqus dans le pro-
cs, mais ceux-l mmes qui devront voter pour ou contre sa condamnation.
En raison de limpossibilit o il se trouve de soumettre ses premiers accusateurs
un elenchos dialectique, Socrate est donc contraint dlaborer un elenchos rhtorique.
Mais les concessions quil fait lelenchos rhtorique me paraissent trs limites pour
deux raisons: a) bien quil y fasse appel, les tmoignages ne sont pas le principal fonde-
ment de largumentation quil dveloppe pour rfuter ses premiers accusateurs; b) alors
que lelenchos rhtorique sinscrit dans une stratgie qui a pour but de fatter les juges et
de leur complaire, Socrate ne se soucie pas de plaire ses juges, ainsi quen tmoignent
les passages de la premire partie de sa dfense (cf. 17d, 20e, 21a) do il ressort claire-
ment quil les irrite et les met en colre.
Il nous reste examiner la troisime tape de la dfense de Socrate, qui stend de
28a 34b. Cette section est parfois prsente comme une digression,2 mais ceux-l
mmes qui la qualifent ainsi sentendent reconnatre quelle fait partie intgrante
de la dfense de Socrate contre les accusations de 3993 et quelle constitue le volet
1 La position de Vlastos, sur ce point, me parat contradictoire. Aprs avoir afrm que dans lApologie a well
qualifed witness is in court to attest the story (1991, p. 289), il soutient ensuite, dans lhypothse o Chrphon
aurait consult loracle selon la procdure de la clromancie, que Chaerephon could have kept it [scil. la r-
ponse de loracle] to himself and divulged it only to Socrates. (1991, p. 289; je souligne). Si la rponse de loracle
na t divulgue qu Socrate, comment le frre de Chrphon pourrait-il tre un bon tmoin? En outre,
comme on a limpression que Socrate raconte lhistoire de loracle pour la premire fois le jour de son procs,
ainsi que Vlastos (p. 289) le souligne avec justesse, cela renforce lhypothse que cet oracle ntait connu que de
Socrate et Chrphon.
2 Cf. Burnet 1924, p. 197; De Strycker & Slings 1994, pp. 23-24; Brisson 1997, pp. 76-77.
3 Cf. Burnet 1924, p. 197: Having disposed of Meletus, Socrates makes his serious defence. In form, it is a
digression; in fact, it is the most important part of the speech. [] In any case, we have here what Plato himself
regarded as the true answer to the accusation. De Strycker & Slings 1994, p. 128: The general content of the
Digression is quite new when compared with that of the Refutation, although close examination reveals that the
elenchos dialectique et elenchos rhtorique 83
positif 1 dune dfense dont la rfutation de Mltos reprsente le volet ngatif. Nous
avons dautres exemples, dans la littrature judiciaire du ive sicle, dapologies qui se
composent de deux parties analogues celles que nous retrouvons galement dans
les Mmorables,2 soit une premire partie qui seforce de surmonter et de rfuter di-
rectement les accusations lances par des adversaires, et une deuxime partie, plus po-
sitive, qui sapplique mettre en lumire et illustrer les qualits de laccus qui suf-
sent montrer que les accusations lances contre lui sont dpourvues de fondement
et quelles sont sans objet.3 Socrate conclut cette prtendue digression par une ob-
servation qui confrme hors de tout doute, me semble-t-il, que cette section est un
prolongement de sa dfense: Eh bien, citoyens, ce que je pourrais allguer pour ma
dfense (L b g i ) se rduit ces observations et dau-
tres du mme genre.4 Contrairement ce que Socrate laisse entendre en 28a, sa d-
fense contre les accusations de 399 ne semble donc pas prendre fn avec la rfutation
de Mltos. Cette troisime section sapparente la premire un double point de
vue: ce nest pas un elenchos dialectique et Socrate y fait nouveau appel des t-
moignages, que je me propose dtudier attentivement pour dterminer si cette troi-
sime section peut tre considre comme un elenchos rhtorique. Il y a cependant
une difrence fondamentale entre la premire section de la dfense et la troisime.
Dans la premire section, Socrate est contraint davoir recours lelenchos rhtorique
en raison de labsence des premiers accusateurs. Cette raison nest cependant pas va-
lable pour la troisime section, puisquelle sattaque aux mmes accusations que la
deuxime section, de sorte que Socrate ne peut pas sautoriser dune contrainte ex-
trieure, dans cette troisime section, pour avoir recours un elenchos rhtorique.
Comme dans la premire section de sa dfense, Socrate invoque trois tmoignages
en sa faveur, dont un seul, cette fois-ci, me parat douteux sur le plan rhtorique. Pour
dmontrer quil sest consacr entirement sa mission, au sacrifce mme de ses int-
rts personnels, Socrate fait valoir ce qui suit:
mes accusateurs, qui ont lefronterie damasser contre moi dautres griefs, se sont trouvs im-
puissants pousser leur efronterie au point de produire un tmoin () qui atteste quil
mest arriv dexiger ou de solliciter une rmunration. En efet, il me suft (e ), jima-
gine, de produire le tmoin qui atteste que je dis vrai (g e
): cest ma pauvret. (31b-c; trad. Brisson)
Si lon peut concder Socrate quil est sans doute signifcatif que ses accusateurs
naient pas song produire des tmoins pour lui reprocher davoir pratiqu une forme
denseignement rmunr, sa pauvret, en revanche, nest certainement pas un t-
digressions frst section proves implicitly that Socrates is a most pious man and its second section that his infu-
ence upon the young has been eminently benefcial.
1 Cf. Reeve 1989, p. 108: Having exposed the falsehood and explained the genesis of the ancient caricature,
having refuted Meletus formal charges, Socrates turns to a broader characterization and defense of the philoso-
phic life he has led. Previously, he has been attacking the false Socrates of rumour and writ. Now he turns to a po-
sitive account of what he is, to the true Socrates.
2 On sentend reconnatre que les Mmorables comportent deux parties, soit i 1-2 et i 3 iv 8 (cf. Dorion 2000,
pp. clxxxiv-clxxxv).
3 Erbse (p. 1961) a le mrite davoir mis en lumire cette structure bifde qui joue le rle dun vritable schma
rhtorique. Il analyse lapplication de ce schma dans certains discours de Lysias et dIsocrate (pp. 267-269), et
conclut de faon convaincante que la structure bifde des Mmorables ressortit au mme schma (cf. p. 274 n. 1,
284-285). 4 34b (trad. Brisson).
84 louis-andr dorion
moin au sens o lentend lelenchos rhtorique. De plus, il est pour le moins tonnant
que Socrate invoque un tmoin pour dmontrer la vrit de son discours, puisquil a
lui-mme expliqu Polos que les tmoignages, fussent-ils nombreux et rputs, ne ser-
vent rien pour tablir la vrit (cf. supra, 471e-472a).
Il y a cependant deux passages o Socrate invoque srieusement, dans la plus pure
tradition rhtorique, des tmoins nombreux lappui de ses dires. Dans le premier de
ces passages, alors quil vient de rappeler quil a refus de se faire le complice de lar-
restation de Lon de Salamine, Socrate conclut en ces termes: Cela maurait sans doute
valu la mort, si le rgime navait t trs rapidement renvers. Voil des faits qui vous
seront attests par de nombreux tmoins ( d ).1
Le second passage est le plus important de tous ceux que nous avons considrs
jusqu maintenant. Pour se dfendre contre laccusation de corruption de la jeunesse
ce qui confrme que la troisime section nest pas quune simple digression , So-
crate fait valoir non seulement quaucun des jeunes quil aurait corrompus ne la accus
dune telle chose, mais quil ny a pas non plus un seul de leurs parents qui ait song
le faire. Mltos na dailleurs cit aucun deux comme tmoin et sil sagit l dun ou-
bli de sa part, Socrate lautorise faire comparatre comme tmoin lun ou lautre des
nombreux parents qui assistent au procs et qui se sont dplacs pour lui prter main
forte (34a7: d ; 34b4: ). Il est frappant de
constater que Socrate nprouve aucun malaise ni aucune honte de ce quune assistance
nombreuse se soit dplace pour lui prter main forte (), alors quil blme jus-
tement Callicls pour son incapacit se secourir lui-mme (d
x , 526e) lorsquil comparatra, aprs sa mort, devant le tri-
bunal o la rhtorique et labondance des tmoins ne sont daucun secours. Rappelons
que le thme du secours () que lon se porte soi-mme, au tribunal, est lob-
jet dun dbat entre Callicls et Socrate. Callicls blme Socrate dtre incapable, cause
de son refus de la rhtorique, de se secourir lui-mme ( J ,
522c5-6). La rplique de Socrate consiste rappeler que la meilleure faon de se porter
secours soi-mme ( a , 522d2-3), devant un
tribunal, est de navoir aucune injustice se reprocher. Or force est de constater que
Socrate, dans lApologie, ne parvient pas, mme sil na pas commis dinjustice, se se-
courir lui-mme par ses propres moyens et quil fnit par accepter le secours dautrui.
Dans une longue tirade qui rappelle irrsistiblement le passage du Gorgias (472a-b) o,
se moquant de Polos, il numre les nombreux Athniens qui pourraient tmoigner en
faveur de sa position,2 Socrate dresse la liste des nombreux parents prsents au procs
et qui se sont dplacs pour lui prter assistance.
En tout cas, beaucoup () de ces gens-l sont venus au tribunal, et je peux les voir. Et
dabord, Criton qui a mon ge et qui est du mme dme, le pre de Critobule que voici. Et en-
1 32d-e (trad. Brisson). La dimension rhtorique de ce passage est pleinement reconnue par Socrate lui-mme,
puisquil introduit son rcit en prcisant quil va parler sans discrtion la mode des plaideurs ( b
a b d ), mais en disant la vrit.
2 En fait, sur lexemple allgu par toi [scil. Polos], tu auras, ou peu sen faut, lappui de tous les Athniens et
de tous les trangers, si tu les appelles tmoigner () contre la vrit de ce que jafrme: tu auras pour
tmoins ( ), si tu le dsires, Nicias fls de Nicratos et avec lui tous ses frres, dont on voit les tr-
pieds rangs en bel ordre dans le sanctuaire de Dionysos; tu auras, si tu le veux, Aristocrats fls de Skellios, le do-
nateur de cette belle ofrande quon admire Delphes; et si tu le veux encore, la famille entire de Pricls, ou,
dans Athnes, toute autre grande famille quil te plaira de choisir. (trad. Croiset)
elenchos dialectique et elenchos rhtorique 85
suite, Lysanias de Sphettos, le pre dEschine que voici. Et encore notre Antiphon de Cphise, le
pre dEpigne. Dautres encore que voici dont les frres mont frquent, Nicostrate, fls de
Thozotids et frre de Thodote comme Thodote est mort, il ne pourrait donc pas emp-
cher Nicostrate de parler contre moi , puis Paralios, le fls de Dmodocos et qui avait pour frre
Thags. Voici encore le fls dAriston, Adimante de qui Platon, ici prsent, est le frre, et Aan-
todore, dont japerois le frre Apollodore. Et je puis vous en nommer encore beaucoup dau-
tres ( ), dont tel ou tel aurait d tre cit avant tout autre comme tmoin
() par Mltos dans son rquisitoire. Mais sil a oubli de le faire, quil les cite mainte-
nant, je lui cde la place; oui, sil arrive trouver un tmoin de ce genre, quil le cite. Mais, tout
au contraire, citoyens, vous trouverez que tous ces gens sont prts mapporter leur appui
( d ) moi qui les corromps, moi qui fais du mal leurs proches,
suivant ce que prtendent Mltos et Anytos. (33d-34b; trad. Brisson)
Certes, Socrate ne fait tmoigner aucun deux directement, mais il nest pas vraiment
ncessaire quil le fasse pour quil obtienne le rsultat escompt: convaincre les juges
que sa position est endosse par de nombreux tmoins. Il mapparat donc difcile
dchapper la conclusion que Socrate pratique lui-mme, dans lApologie, le type
delenchos rhtorique quil dnonce dans le Gorgias. On pourrait mobjecter que cette
conclusion contredit lafrmation de Socrate suivant laquelle il na pas fait appel aux
procds de la rhtorique pour se dfendre (Apol. 34c-35d). Or lorsque Socrate se d-
fend davoir eu recours la rhtorique, il fait mention des pleurs, des prires et des la-
mentations (34c), mais pas des tmoins. Cette absence des tmoins est sans doute si-
gnifcative. En efet, lorsque Socrate, immdiatement aprs quil eut pris connaissance
du vote qui le condamne mort (38c), fait nouveau mention des procds rhtoriques
(pleurs, gmissements, paroles et actes indignes dun homme libre) quil condamne et
quil a refus, au prix de sa vie, demployer pour se dfendre, il fait nouveau limpasse
sur le recours aux tmoins et, partant, il ne les compte pas au nombre de ces procds
infamants (38c-d). Certes, Socrate fait ailleurs rfrence aux petits enfants, aux proches
et aux amis (34c) que certains nhsitent pas faire monter devant le tribunal, alors que
lui a refus de le faire, mais il ne refuse pas pour autant les tmoins, puisque ce nest pas
en guise de tmoins que lon fait ainsi appel aux enfants et aux proches, mais unique-
ment pour attendrir et apitoyer les juges. Or dans la mesure o lappel aux tmoins est
lun des principaux moyens dont lorateur dispose pour tablir une preuve,1 et que So-
crate nhsite pas recourir ce procd rhtorique, force est de conclure que Socrate
est loin dtre aussi dtermin quil le prtend ne pas faire appel la rhtorique pour
se dfendre.
Sil est exact, ainsi que le soutiennent plusieurs commentateurs,2 que la vraie dfense
de Socrate ne se situe pas en 24d-28a, alors quil soumet Mltos une rfutation dia-
lectique qui est dclare sufsante, mais plutt dans la digression de 28a-34b, cest
le statut de cette digression, sur le plan de largumentation, qui dterminera si Platon
confe la dialectique, ainsi que lon pourrait sy attendre, ou au contraire la rhto-
rique la portion la plus signifcative de la dfense de Socrate. Or mon analyse de cette
1 Dans la Rhtorique, Aristote compte les tmoignages au nombre des preuves () extra-techniques (cf. i
2, 1355b35-37; i 15, 1375a22-24) et il consacre un long dveloppement aux difrents types de tmoins (i 15, 1375b26
2 Cf. supra, p. 83 n. 3 et De Strycker & Slings, p. 124: the is not the section of the Apology where
we should look for Socrates real defence. ( je souligne); p. 128: The does not contain the true defence of
Socrates nor does it constitute the core of the work. ( je souligne)
86 louis-andr dorion
digression, o lappel rpt aux tmoins joue un rle dterminant, mincite conclure
que la troisime partie de la dfense de Socrate sapparente davantage au type delenchos
rhtorique dnonc dans le Gorgias qu la dialectique rfutative. Ma position, concer-
nant la troisime section, est donc aux antipodes de celle dfendue par de Strycker &
Structurally the Digression consists of two distinct , imaginary objections, each of
which gets an extensive answer. [] Through this literary device [scil. la rponse des objections
fctives], Plato avoids putting long rhetorical and self-asserting speeches into Socrates mouth.
Instead, he makes Socrates introduce an imaginary objector and have an everyday conversation
with him. So the true defence of Socrates takes the form of dialectic and not of oratory.1
Vu limportance de lenjeu il sagit de rien de moins que de dterminer si la vraie d-
fense de Socrate est de nature rhtorique ou dialectique , je discuterai en dtail la po-
sition de Slings & de Strycker. Il est exact que Socrate introduit un contradicteur fctif
en 28b et quil lui attribue cette objection: Peut-tre bien me dira-t-on (I \ i s
): Nas-tu pas honte, Socrate, davoir adopt une conduite qui aujourdhui
texpose la mort?.2 La rponse de Socrate nest pas une rfutation, puisquil ny a pas
et ne peut y avoir dchange entre Socrate et lobjecteur anonyme, mais une simple r-
plique lobjection: cela je serais en droit de faire cette rponse: Mon bon (t
), ce nest pas parler comme il faut que dimaginer, comme tu le fais, quun
homme qui vaut quelque chose, si peu que ce soit, doive, lorsquil pose une action, met-
tre dans la balance ses chances de vie et de mort, au lieu de se demander seulement si
laction quil pose est juste ou injuste, sil se conduit en homme de bien ou comme un
mchant.3 La rplique de Socrate linterlocuteur anonyme prend fn en 28d et il ny
a plus aucune trace de cet interlocuteur fctif jusqu la fn de la digression, soit 34b.
Cest dire que lartifce littraire (literary device) qui permettrait Platon de transfgurer
en argumentation dialectique un long monologue qui a, en ralit, toute lapparence
dun discours rhtorique, noccupe quune trs courte section (28b-d) de la digression
(28b-34b). Dans la plus longue section de la digression, Socrate ne sadresse donc pas au
contradicteur fctif qui intervient ponctuellement, en 28b, pour formuler une seule ob-
jection, mais bien aux juges qui devront passer au vote ds que Socrate aura termin
son discours. Les rfrences aux membres du tribunal sont nombreuses,4 mais, lex-
ception dun passage (29c-d) o il imagine un bref change entre ses juges et lui, So-
crate ne sadresse pas aux juges comme des interlocuteurs qui prennent part, avec lui,
un change dialectique, et encore moins comme des rpondants qui soutiennent une
position que Socrate chercherait rfuter. Le fait que Socrate sadresse aux juges
confrme en ralit le caractre rhtorique de son discours, puisque lorateur, dans un
contexte rhtorique, cherche prcisment convaincre celui qui doit trancher, par son
vote, en faveur de lune ou lautre des deux thses qui safrontent.
Il me parat donc abusif de parler de dialectique en un sens aussi vague5 et je consi-
dre quune argumentation ne peut tre qualife de dialectique que sil y a change de
1 1994, p. 128 ( je souligne). 2 28b (trad. Brisson). 3 28b (trad. Brisson).
4 Cf., entre autres, les passages o Socrate sadresse aux juges en les appelant Athniens (t \A,
28d6, e1, 29b3, d2-3, 30b7, c2, d5, 31a2, d6-7, 32a9, e5, 33c1, c7, 34b6).
5 La conception que Slings se fait de la dialectique est en vrit aussi foue et imprcise que la conception de
lelenchos quil expose dans son commentaire du Clitophon (1999) et dont jai dj eu loccasion de critiquer les in-
cohrences et les insufsances (cf. Dorion 2000, pp. cxxviii-cxxxiii).
elenchos dialectique et elenchos rhtorique 87
questions et de rponses entre deux interlocuteurs prsents lentretien et qui partici-
pent efectivement cet change. Si la troisime section est dialectique, comment
faut-il alors qualifer la rfutation de Mltos? ce sujet, je remarque au passage que de
Strycker & Slings parlent de la rfutation de Mltos comme du seul morceau labor
de dialectique dans lApologie (the only developped piece of dialectic in the Apology,
p. 126), auquel cas ils semblent reconnatre que la digression, qui est pourtant plus
longue que la rfutation de Mltos (24b-28a), nexpose pas une argumentation dialec-
tique. Que dirait le Socrate du Gorgias dune argumentation dialectique qui ne fait pas
voter son interlocuteur et qui se fonde trois reprises sur des tmoignages? Enfn, le
fait de rpondre des objections nest pas un trait distinctif de la dialectique, puisque
lorateur doit galement tre en mesure de rpondre des objections.1 Sil est vrai que
la digression constitue la vraie dfense de Socrate, il faut en conclure, selon mon ana-
lyse, que Platon confe lelenchos rhtorique, plutt que dialectique, le soin dassumer
la partie la plus dterminante de la dfense de Socrate contre les accusations de 399.
Alors que Socrate dans le Gorgias prne lusage exclusif de lelenchos dialectique et ne
semble reconnatre aucune lgitimit lelenchos rhtorique, non seulement il nhsite
pas les utiliser lun et lautre loccasion de son procs, mais il rserve mme lelen-
chos rhtorique un rle qui supplante celui qui est assum par lelenchos dialectique.
4. Conclusion: vote et dialectique
Dans la deuxime section de cette tude, jai soulign lintrt dun passage du Gorgias
(474a) o Socrate prie Polos de ne pas lui demander de faire voter lassistance car la seule
personne quil est dispos faire voter en tant que tmoin, cest son interlocuteur. Ce
nest pas le seul passage o Socrate soppose ce que lon ait recours au vote pour tran-
cher un dbat. Dans le Lachs (184d-e), alors que Nicias sest prononc en faveur de lho-
plomachie et Lachs contre, Socrate refuse de voter pour lune ou lautre des thses en
prsence et il choisit plutt de reprendre lexamen de cette question depuis le com-
mencement, en faisant appel la dialectique. De mme, dans un beau passage du livre
I de la Rpublique, Socrate explique Glaucon que lavantage de la dialectique sur la rh-
torique est quelle na pas besoin de juges ou darbitres pour trancher le dbat, puisque
les participants lentretien, par leur accord commun, font eux-mmes ofce de juges:
Si donc, dis-je, ramassant nos forces et opposant discours discours ( a ), nous
numrons tous les avantages qu son tour comporte la justice, et quil rplique et que nous r-
pondions, il faudra compter les avantages et mesurer ce que nous aurons dit lun et lautre dans
nos deux discours respectifs, et il nous faudra ds lors des arbitres pour trancher le dbat
( ). Si au contraire nous examinons les choses
comme tout lheure, en nous mettant daccord (), nous serons nous-mmes
la fois juges et orateurs (d d ).2
Or la procdure que Socrate refuse dans tous ces textes, celle dun vote ou dun arbi-
trage des personnes qui assistent lentretien, est celle-l mme qui lui est impose dans
lApologie. Dans un contexte dialectique, il suft Socrate que son interlocuteur t-
1 Sur la nature et le rle de lobjection () dans largumentation rhtorique, cf. Aristote, Rht. ii 22,
1397a5; 25, 1402a31-b35; 26, 1403a28-31.
2 i 348a-b (trad. Chambry lgrement modife). Cf. Apol. 17b, o Socrate explique quelles conditions il se-
rait dispos accepter quon puisse le considrer comme un orateur ().
88 louis-andr dorion
moigne et vote contre lui-mme, cest--dire reconnaisse quil se contredit, pour que sa
position soit par le fait mme dfaite. Il ne procde pas autrement lors de son interro-
gation de Mltos: il lui suft davoir mis en lumire les contradictions inhrentes au
discours de Mltos pour considrer que les accusations portes contre lui sont nulles
et non avenues. Si linterrogation de Mltos stait droule dans un cadre dialectique,
Socrate aurait pu considrer que cette rfutation tait en efet sufsante. Or cette rfu-
tation sinscrit dans un cadre indit pour Socrate, celui dun procs dont lissue, quil le
veuille ou non, est scelle par le vote de ceux qui assistent, en loccurrence les juges. So-
crate est ainsi dpossd du verdict prononc sur lissue dune rfutation dialectique et
cest bien l, me semble-t-il, que rsident les limites de sa dfense. Il ne fait aucun doute
quil sest dfendu, et que sa dfense est limage mme de sa faon de pratiquer la phi-
losophie, mais cest la premire fois que le verdict prononc lissue dune rfutation
dialectique lui chappe. De ce point de vue, lelenchos rhtorique est sans doute plus ap-
propri que lelenchos dialectique au contexte judiciaire; comme les tmoins sur lesquels
se fonde lelenchos rhtorique sont dune certaine faon autant de votes en faveur dune
thse, le recours aux tmoins est en un sens une prfguration et une anticipation du
vote des juges. Autrement dit, comme lelenchos rhtorique repose sur le nombre et que
son fondement est identique celui du vote, il jouit certainement dun avantage consi-
drable sur lelenchos dialectique dans le cadre dun procs o les juges sont nombreux.
Socrate a prvenu Polos que lorsque lassistance est nombreuse, il ne discute mme pas
avec elle ( b b ), cest--dire que son interlocuteur lui suft
et quil ne peut pas faire voter ni tmoigner, dans le cadre dun entretien dialectique, de
nombreuses personnes. De fait, Socrate ne peut pas engager un entretien avec ses juges,
mais ce sont nanmoins eux, nen dplaise Socrate, qui dcident de lissue dun elen-
chos dialectique. Le fait mme que Socrate nait pas hsit pratiquer une forme delen-
chos rhtorique dans la troisime partie de sa dfense invite penser quil tait conscient
des limites de lelenchos dialectique et que sa rfutation de Mltos ntait probablement
pas sufsante.
Universit de Montral
Brickhouse T. C. & Smith N. D., Socrates on trial, Oxford, o.u.p., 1989.
Brisson L., Platon: Apologie et Criton, Paris, gf Flammarion, 1997.
Burnet J., Plato. Euthyphro, Apology of Socrates and Crito, Oxford, Clarendon Press, 1924.
Canto M., Platon: Gorgias, Paris, gf Flammarion, 1987.
Dalfen J., Platon: Gorgias (Platon Werke, vi 3), Gttingen, Vandenhoeck & Ruprecht, 2004.
De Strycker E. & Slings S. R., Platos Apology of Socrates, Leiden, Brill, 1994.
Dodds E. R., Plato: Gorgias, Oxford, Clarendon Press, 1959.
Dorion L.-A., La subversion de lelenchos juridique dans lApologie de Socrate, Revue philoso-
phique de Louvain, 88 (1990), pp. 311-344.
Dorion L.-A. & Bandini M., Xnophon: Mmorables, tome i: Introduction gnrale, Livre i [intro-
duction, traduction et notes par L.-A. Dorion; texte tabli par M. Bandini], Paris, Les Belles
Lettres, 2000.
Erbse H., Die Architektonik im Auf bau von Xenophons Memorabilien, Hermes 89 (1961), pp. 257-
Irwin T., Plato: Gorgias, Oxford, Clarendon Press, 1979.
Montuori M., Socrates: physiology of a myth, Amsterdam, J. C. Gieben, 1981.
elenchos dialectique et elenchos rhtorique 89
Pradeau J.-F. & Fronterotta F., Platon: Hippias majeur, Hippias mineur, Paris, gf Flammarion,
Reeve C. D. C., Socrates in the Apology. An essay on Platos Apology of Socrates, Indianapolis,
Hackett, 1989.
Slings S. R., Plato: Clitophon, Cambridge, c.u.p., 1999 (Cambridge Classical Texts and Com-
mentaries, 37).
Vlastos G., Socrates: ironist and moral philosopher, Ithaca (n.y.), Cornell u.p., 1991.
90 louis-andr dorion
201C ET 201E
Luc Brisson
e voudrais dans cet article analyser de faon dtaille un passage du Banquet (199c-201c)
o Socrate soumet Agathon une rfutation laquelle il avait lui-mme auparavant
t soumis par Diotime (201e-203a). Cette rfutation1 remet en cause la reprsentation
traditionnelle dros, pour en proposer une nouvelle qui considre ros non plus
comme un dieu, mais comme un daimon, intermdiaire privilgi entre les dieux et les
hommes, entre le savoir et lignorance, et lassociant ainsi la philosophie dfnie
comme lan vers le savoir. Le but de cet article est de montrer que, pour Socrate, lim-
portant dans la rfutation, cest plus son efet thique,2 qui peut se propager dun in-
terlocuteur un autre, que la validit de son argumentation logique,3 qui ne peut tre
remise en cause ici en dpit de son manque de rigueur technique.
Agathon,4 qui joue ici le rle de rpondant, dfend une thse qui devient la cible de
la rfutation. Cette thse, qui reste implicite tout au long de la rfutation, a cependant
t clairement formule au dbut de son discours, en 195a5-7:
Je dclare donc que, parmi les dieux, qui tous sont heureux, ros, sil est permis de le dire sans
inciter au ressentiment, est le plus heureux, car il est le plus beau et le meilleur.5
Socrate, lui, va amener Agathon admettre des propositions qui sont acceptes sans
discussion, mais qui vont contredire la thse quil soutenait.
Dans le cas prsent, la situation se trouve complique, par le fait que Socrate joue sur
les mots et enchane les propositions les unes aux autres.
Socrate commence par poser Agathon cette question particulirement ambigu:
x r , j ; (199d1-2). Dans cette ques-
tion, le terme peut tre interprt a) comme un nom propre, E ou b) comme
1 On trouvera une analyse de ce passage dans le contexte dune discussion entre difrents auteurs de langue
anglaise (cits dans son article) par A. Payne, The refutation of Agathon: Symposium 199c-201c, Ancient Philosophy
19 (1999), pp. 235-253. Lauteur formule ainsi son objectif: I hold that Socrates uses an invalid argument to convince
Agathon that Eros is not beautiful, but along the way he gives a valid and sound argument for a diferent and per-
haps more interesting position, that Eros lacks and is in need of the beautiful objects of his desire. The refutation
of Agathon deserves close attention both to show that Diotimas speech still deals with eros and to clarify the
structure of Socrates argument. (p. 236). Je partage pour lessentiel ce point de vue, mais je pense que Platon qui
ne donne pas un cours de logique lve, mesure quavance le dialogue, les ambiguts qui sont nombreuses dans
le propos de Socrate.
2 Voir aussi L.-A. Dorion, Socrate, Que sais-je? 899, Paris, Puf, 2004, pp. 55-66; et surtout Le destin ambivalent du
Sophiste: lexemple dAristote et de Clment dAlexandrie, dans B. Melkevik et J.-M. Narbonne (eds), Une philosophie
dans lhistoire. Hommages Raymond Klibansky, Qubec, Presses de lUniversit Laval, 2000 (Collection Ztsis,
3), pp. 47-63. Dans cet article, L.-A. Dorion, montre que de la fgure du rfutateur, Aristote ne retient que laspect
logique, tandis que Clment dAlexandrie ne conserve que laspect thique.
3 G. Vlastos, The socratic elenchus, Oxford Studies in Ancient Philosophy, 1, 1983, pp. 27-58.
4 En 194d1-e3, Phdre avait empch Socrate de poser des questions Agathon. Socrate revient la charge,
mais en se justifant cette fois. Comme on le verra tout lheure, ce questionnement prend la forme dun vrita-
ble elenchos. Les traductions du Banquet et de lApologie sont de moi.
5 Il est un kalos kagathos, lidal du citoyen athnien, au plus haut degr.
un nom commun, . Socrate sempresse dliminer la premire interprtation
(199d2-e5).1 Reste la seconde possibilit, sur laquelle portera la rfutation. Cest en ce
sens, me semble-t-il, que Socrate reformule sa question: , j
; (199e6-7). Alors apparat un nouveau problme: celui de savoir si les gnitifs
et sont des masculins ou des neutres. Dans un cas, il sagit dune personne,
dans lautre dune chose ou dune proprit. Prendre position dans un sens ou dans un
autre na aucune incidence sur la structure logique de la rfutation en tant que telle;
mais cela entrane un certain nombre de consquences du point de vue de la doctrine
qui la sous-tend, comme je vais le montrer. Quil sufse pour linstant de rappeler que
Socrate fait, dabord, reconnatre Agathon qu, quil sagisse de lamour comme
notion ou que cette notion soit personnife, est un terme relatif une personne, une
chose ou une proprit.2
Socrate passe alors la seconde proposition, o il associe lamour () un dsir
S. Tout ce que je veux savoir, cest si lamour prouve, oui ou non, le dsir de ce dont il est
amour ( y , j ;). A. Assurment, il en
prouve le dsir. (200a2-4)
Ds lors, lamour est un dsir qui peut avoir pour objet une personne, une chose ou une
Puis vient une troisime proposition: on ne possde pas ce que lon dsire ou ce que
lon aime (200a5-6). Lamour et le dsir dnoncent donc un manque (200a5-7): on
naime, on ne dsire que ce dont on manque: e y , j
c , a c b ; (200a8-b1). Plusieurs exemples sont voqus: lhomme
grand ou fort qui ne dsire pas tre grand ou fort (200b3-8); lhomme rapide qui ne d-
sire pas tre rapide (200b9-10); lhomme en bonne sant qui ne dsire pas tre en bonne
sant (200b10-c1). Et Socrate apporte plusieurs exemples, pour bien faire comprendre
Agathon ce que signife la proposition quil vient daccepter (200c1-e6). On remarquera
que ces exemples ne font jamais rfrence une personne ou une chose possdant
telle ou telle proprit,4 mais une proprit, quil sagisse de la grandeur, de la rapi-
dit ou de la sant. Ces exemples lvent donc limprcision qui frappait la troisime pro-
position; lamour, quil soit personnif ou non, porte sur une proprit, la beaut.5
Mme si lon admet cette interprtation, on peut penser quil subsiste encore une
faute logique dans cette proposition: ce nest pas parce quros dsire le beau comme
proprit, quil est ncessairement dpourvu de beaut.6 ce stade de la dmonstra-
1 Mais pas totalement, comme on le verra; dune part il fait remarquer Agathon qutre pre, mre ou frre
relve de la relation comme lindique le gnitif; et la fn, il reviendra sur les origines dros.
2 Toutes ces difcults avaient t releves par R. E. Allen, A note on the elenchus of Agathon: Symposium199c-
201c, Monist, 50 (1966), pp. 460-463.
3 Ce passage laisse entendre que les distinctions tablies par D. Hyland (

E, et , Phro-
nesis, 13 [1968], pp. 32-46) sont moins tranches quil ne le pense.
4 Il sagit de la principale difcult sur laquelle achoppent les commentateurs anglais et amricains, que cite
A. Payne la p. 236 de son article.
5 Cest la position de C. D. C. Reeve, Telling the truth about love: Platos Symposium, Proceedings of the Bos-
ton Area Colloquium in Ancient Philosophy, 8 (1992), p. 99.
6 Even if Eros desires the property beauty, even if Eros desires the property more than any other beautiful
thing, it need not follow that Eros lacks the property in the sense of not being beautiful. What follows is that Eros
does not have the beautiful object of his desire in the sense of not having ready access to what he desires. (A.
Payne, Ancient Philosophy, 19 [1999], p. 247).
92 luc brisson
tion le doute se maintient, mais plus bas il est lev par Diotime:1 ne pas tre beau si-
gnife non pas tre laid, mais manquer de beaut. Entre possder une proprit et en
tre totalement dpourvu, il y a des intermdiaires. Et cest sur ce point quinsiste Dio-
time, suivie par Socrate.
Cest alors que Socrate peut conclure en rappelant les deux propositions incrimines:
Poursuivons donc, dit Socrate, et rcapitulons les points sur lesquels nous sommes tombs
daccord dans la discussion. Nest-il pas vrai premirement quros recherche quelque chose et
deuximement quil recherche quelque chose dont il2 est dpourvu pour linstant. (200e7-9)
Cette conclusion admise par Agathon entre en contradiction fagrante avec la thse
quil dfendait jusque-l. Puisquros est amour du beau et du bien, il doit manquer de
beaut et de bont.
Agathon est dabord forc dadmettre quros manque de beaut (201a2-b10), prenant
ainsi le contre-pied de la thse quil a dveloppe dans son discours sur ros.3 Do cette
remarque piteuse dAgathon: Je risque fort, Socrate, davoir parl sans savoir ce que je
disais. (201b11-12). Puis il est amen par un Socrate, qui ironiquement loue la beaut de
son discours, reconnatre quros manque de ce qui est bon (201c1-5). Cest alors
quAgathon admet son impuissance sopposer cette conclusion, capitulant ainsi sur
la question de la beaut et sur celle de la bont.
On sest interrog sur la question de savoir si, en rvlant que sur un mme sujet un
rpondant a donn des rponses contradictoires, on dmontrait que les opinions du r-
pondant taient fausses. Cela pour Socrate ne semble gure faire de doute;4 il nen reste
pas moins que ses rponses contradictoires dnoncent lignorance du rpondant; un ex-
pert, quelquun qui prtend un savoir, ne peut tenir sur un mme sujet des discours
Cest par ce biais que la dimension logique de la rfutation est subordonne sa fna-
lit morale. Socrate ne pratique pas la rfutation pour le plaisir de rfuter, et donc pour
faire honte au rpondant, mais pour rendre, par ce sentiment de honte, son interlocu-
teur meilleur en lui faisant prendre conscience de son ignorance, et donc en le purifant
dun faux savoir. Cest dailleurs ce que nont pas compris les citoyens dAthnes, comme
le fait remarquer Socrate lui-mme, aprs avoir voqu les rfutations quil avait fait
subir aux hommes politiques, aux potes5 et ceux qui travaillent de leurs mains:
Cest prcisment cette enqute, Athniens, qui ma valu des inimitis si nombreuses qui pr-
sentaient une virulence et une gravit dune telle importance quelles ont suscit maintes ca-
lomnies et mont valu de me voir attribuer ce nom, celui de savant. (Apologie, 22e6-23a3)
1 Dans le passage suivant qui sera cit plus bas: Pas de blasphme, reprit-elle. Timagines-tu que ce qui nest
pas beau doive ncessairement tre laid? S. Certainement. D. Timagines-tu de mme que celui qui nest pas un ex-
pert est stupide? Nas-tu pas compris que, entre science et ignorance, il y a un intermdiaire. (Banquet 201e10-202a3)
2 Traduction qui prend provisoirement position sur lambigut relative , on ne peut savoir si ren-
voie celui qui aime, lamoureux ( ) ou ros ( E). Si on pense quil sagit de celui qui aime,
il faudrait traduire ainsi: Nest-il pas vrai premirement que lamoureux recherche quelque chose et deuxime-
ment, quil recherche quelque chose dont il est dpourvu pour linstant.
3 En 197b, Agathon a dit quros est amour de la beaut; et en 201a, Socrate linvitera sen rappeler. Cest pr-
cisment ce point que Socrate va ici remettre en cause.
4 A. En ce qui me concerne, Socrate, je ne suis pas de taille engager avec toi la controverse; quil en soit
comme tu le dis. S. Non, trs cher Agathon, cest avec la vrit que tu ne peux engager la controverse; avec So-
crate ce nest vraiment pas difcile. (201c6-9).
5 On peut considrer que fait partie de ce groupe la rfutation laquelle Socrate soumet Agathon.
banquet (199c-201c et 201e-203a) 93
Socrate, lui, voyait les choses dun autre il:
Mais que je sois, moi, le genre dhomme que la divinit ofre la cit en cadeau, les considra-
tions suivantes vous permettront de vous en convaincre. Aucun motif humain ne semble devoir
expliquer que je nglige toutes mes afaires personnelles, et que jen supporte les consquences
dans ladministration de ma maison depuis tant dannes dj, et cela pour moccuper en per-
manence de vous, en jouant auprs de chacun de vous en particulier le rle dun pre ou dun
frre plus g, dans le but de le convaincre davoir souci de la vertu. (Apologie, 31a2-b5)
La chose est dimportance puisquaucun homme ne pourra devenir vertueux que si son
me a t au pralable purife des faux savoirs quelle recle son insu.1
Ce processus de purifcation est dclench par un sentiment de honte. Linterlocuteur
rfut doit avoir honte de ses contradictions et donc de son ignorance. Dans le cas pr-
sent, Agathon est bien oblig, on la vu, dadmettre ceci: Je risque fort, Socrate, davoir
parl sans savoir ce que je disais. (201b11-12). Et cet aveu est encore plus douloureux
quand il nest pas arrach la suite dun dbat oratoire, comme ce serait le cas pour un
sophiste qui recherche la victoire au moyen du discours par exemple, mais quand il se
prsente comme une obligation lgard de la vrit elle-mme, comme le fait remar-
quer Socrate Agathon: en ce cas, cest la reprsentation que se fait de soi-mme le r-
pondant et de son rle dans la cit qui sont en cause. On se trouve donc l, avec le dia-
logue entre Socrate et Agathon, devant une rfutation qui prsente la forme logique
classique, et qui atteint sa fnalit morale en dbouchant sur un aveu dimpuissance hon-
teuse de la part du rpondant.
Mais comme Socrate est lhte dAgathon, qui se conduit de faon exquise envers
tous,2 il va tenter de redresser la situation, en se prsentant de faon trs habile, comme
la premire victime de la rfutation quil vient de lui faire subir:
coutez plutt le discours sur ros que jai entendu un jour de la bouche dune femme de Man-
tine, Diotime, qui tait experte en ce domaine comme en beaucoup dautres (201d1-3)
En mettant son discours sous le patronage de Diotime, Socrate peut justifer les propos
contradictoires quil a tenus. Dabord il sest adress ryximaque, qui propose de pren-
dre ros pour thme des discours dans les termes suivants: [ta proposition] ne ren-
contrera dopposition ni chez moi, qui dclare ne rien savoir sauf sur les sujets qui re-
lvent dros (Banquet 177d6-7) Puis il est revenu sur sa dclaration:
Cest alors (aprs avoir cout le discours dAgathon), oui, que jai compris que jtais ridicule
lorsque je vous promettais de faire mon tour, de conserve avec vous, un loge dros, et quand
je dclarais que jtais redoutable sur les sujets qui relvent dros, alors je ne savais rien sur la
manire dont il convient de faire un loge. (Banquet 198c5-d3)
1 La sixime dfnition dans le Sophiste (226a-231c) aborde ce thme.
2 Comme le montrent ces troix exemples. 1) Agathon admet que Socrate arrive la fn du repas (174c-175d). 2)
Il se montre aimable avec ses esclaves, quand il fait cette remarque malicieuse: Eh bien, soit, laissons-le (Socrate),
si tel est ton avis, reprit Agathon. Mais nous autres, il faut, garons, nous apporter manger. Vous servez tou-
jours ce quil vous plat, sil arrive quil ny ait personne pour vous surveiller, ce que personnellement je nai ja-
mais fait. Aujourdhui donc, faites comme si ctait vous qui nous aviez invits souper, moi-mme et les autres
convives, et traitez-nous de faon mriter nos loges. (175b1-c1). 3) LorsquAlcibiade avin fait irruption dans la
salle (212c-214a), il laccueille tout de mme avec une grande courtoisie.
94 luc brisson
Certes cette position de retrait ne semble toucher que la forme du discours; mais la pre-
mire dclaration de Socrate reste nigmatique, car cest la seule exception quil sem-
ble faire sa revendication dignorance.
Mais lnigme est bientt leve, car ce savoir sur ros ne lui appartient pas vraiment,
puisquil vient de Diotime:
S. Je soutenais quros tait un grand dieu, et quil faisait partie de ce qui est beau.1 Et elle me
rfutait en faisant valoir les mmes arguments prcisment que ceux que je viens dutiliser avec
Agathon, savoir quros nest ni beau ni bon, comme je viens de le dire. Je lui rpliquai: Que
dis-tu l, Diotime? Si tel est le cas, ros est laid et mauvais.2 D. Pas de blasphme, reprit-elle.
Timagines-tu que ce qui nest pas beau doive ncessairement tre laid? S. Certainement. D.
Timagines-tu de mme que celui qui nest pas un expert est stupide? Nas-tu pas compris que,
entre science et ignorance,3 il y a un intermdiaire?4 (201e5-202a3)
Bref, pour attnuer la honte quentrane cette rfutation, Socrate explique que, ayant
lhabitude de soutenir sur ros une position similaire celle dAgathon, il avait t r-
fut par Diotime qui lui avait fait sentir son ignorance. Et pour attnuer le caractre ap-
paremment blasphmatoire de la rfutation, il explique que le manque de beau et de
bien qui caractrise ros est conforme sa nature dintermdiaire.
Socrate, en efet, peut continuer prtendre quil ne sait rien, et donc quil nenseigne
rien qui vienne de lui, tout en critiquant une reprsentation traditionnelle dros pour
en proposer une nouvelle;5 ros est un daimon, et non un dieu. Puisquil a pour pre
Poros (Expdient) et pour mre Pnia (Pnurie), ros est un tre intermdiaire, qui
recherche le beau et le bon. Or, le beau et le bon, comme va le montrer Diotime, ne se
situent pas seulement dans le sensible: dun beau corps, il faut remonter aux beaux
corps, aux belles occupations, aux belles mes et fnalement la Beaut elle-mme, qui
se trouve dans lintelligible. Do ces deux consquences, lune touchant lamour
comme dsir, et lautre, ros comme patron de la philosophie.
Au niveau des corps, le dsir est caus par un manque, ce qui permet de lassocier au
plaisir et la douleur. En efet si un manque se produit rapidement et contre nature, il
est ressenti comme une douleur, alors que sil est restaur rapidement et suivant la
nature, il est ressenti comme un plaisir, suivant lexplication quon trouve dans le Time
1 Je modife ici la traduction qui est la mienne (Banquet de Platon, Paris, Flammarion, 1999, Collection gf,
987). La question est de savoir comment interprter b . Faut il sous-entendre un ou un
avant le . Dans un cas, ros est amour des belles choses, tandis que dans lautre, il fait partie des belles
choses. Je penche maintenant plutt vers la seconde solution.
2 On se trouve ici en prsence dun vritable problme de logique; sont contraires deux classes exclusives, cest-
-dire deux classes dont les lments ne peuvent se trouver dans lune et dans lautre la fois. Suivant cette df-
nition, il y a trs peu de vritables contraires; par exemple, un / plusieurs, repos / mouvement. En revanche, beau
et laid ne sont pas des contraires, puisque lun et lautre peuvent comporter des degrs. Do la solution de Dio-
time: ros est un intermdiaire.
3 En 203d-204b, les termes , , sont considrs comme des synonymes, qui ont pour
antonyme .
4 Sur la notion de , cf. le livre ancien, mais essentiel, de J. Souilh, La notion platonicienne dintermdiaire
dans la philosophie de Platon [1919], New York, Garland, 1987 (coll. Greek and Roman Philosophy, 37). Cette no-
tion dintermdiaire joue un rle de premier plan dans la suite du discours de Diotime.
5 Il ne me semble pas que lon puisse parler de conficting religious discourses dans le Banquet, comme le fait
S. Robinson, The contest of wisdom between Socrates and Agathon in Platos Symposium, Ancient Philosophy, 24
(2004), pp. 81-100. La mythologie grecque laissait une trs large marge de manuvre qui voulait raconter un
mythe traditionnel en laccommodant sa convenance.
banquet (199c-201c et 201e-203a) 95
(64a-65b). Par suite, alors que laversion se trouve dfnie comme une fuite devant la
douleur, le dsir lest comme la recherche du plaisir. Cela dit, le dsir peut porter non
seulement sur les corps, mais aussi sur lincorporel,1 car il existe un dsir de la science
(Politique 272d4), de la vertu (Lois vii 827d2), et mme de lintelligible comme on peut
le constater dans la Banquet, lorsquil sagit de la Beaut. On est alors en mesure de
dresser un parallle entre le manque dont soufre le corps, et celui dont ptit lme. Mais
il y a une difrence entre les plaisirs sensibles et les plaisirs purs, qui sont ceux de lme.
Alors que le plaisir sensible est toujours ml de douleur, comme le rappelle Socrate
dans le Phdon (60b-c), le plaisir pur, en plus de durer, ne comporte aucune douleur,2
comme cela est expliqu dans le Philbe (50e-53c). En tant que daimon, ros se trouve
mi-chemin entre les dieux et les hommes, et donc entre le savoir qui caractrise les dieux
et lignorance dont sont victimes les hommes, qui cependant peuvent chercher
acqurir ce savoir: Aucun dieu ne tend vers le savoir ni ne dsire devenir savant (
d \ e ), car il lest ( ); or, si lon est
savant, on na pas besoin de tendre vers le savoir ( , )
(204a1-2). Cest donc parce quil est amour du savoir, dsir du savoir, quros peut
prsider la philosophie. On comprend mieux ainsi pourquoi dans le Phdre (248d3), le
est associ au .
Ces conclusions mettent en vidence les trois traits suivants qui caractrisent Socrate.
Elles font apparatre limportance de la rfutation socratique comme pratique philoso-
phique. En rapportant Diotime3 la rfutation laquelle il soumet Agathon, elles
montrent comment, une fois de plus, Socrate enracine sa pratique de la philosophie
dans la tradition religieuse sans bien videmment ly subordonner.4 Enfn, comme
lamour dont parle Socrate mne en dernire analyse lamlioration de lme, on com-
1 Chez Platon, le domaine de lincorporel, cest essentiellement celui de lintelligible et de lme qui se situe
entre lintelligible et le sensible, qui est le domaine du corps.
2 Lme prouve les douleurs de lenfantement (Banquet 206e1), lorsquelle est la recherche de la vrit, de la
ralit vritable. Voir aussi Phdre 251e-252a.
3 Voici en quels termes Socrate prsente son discours dans le Banquet. coutez plutt le discours sur ros que
jai entendu un jour de la bouche dune femme de Mantine, Diotime, qui tait experte en ce domaine comme en
beaucoup dautres, et qui, un moment donn, dix ans avant la peste, avait amen les Athniens ofrir des sa-
crifces qui ont permis de reculer de dix ans la date du fau. Oui, cest elle qui ma instruit des choses concernant
lamour. (Banquet 201d1-5). Si la peste quvoque Socrate est bien celle de 430, celle qui clata au dbut de la guerre
du Ploponnse et dont Pricls fut lune des victimes, cela implique que Socrate a entendu Diotime vers 440, soit
24 ans plus tt, alors quil navait quune trentaine dannes. Diotime prsente plusieurs traits commun avec pi-
mnide, dont il est dit dans les Lois (i 642d-e) que, dix ans avant les faits, il avait prdit les guerres mdiques et la
dfaite des Perses. Elle sinscrit donc dans une tradition religieuse dont il est impossible de dterminer lorienta-
tion. On notera cependant que, en Grce ancienne, les femmes pouvaient jouer un rle en ce domaine, comme
en tmoigne Platon dans le Mnon (81a).
4 En voici trois exemples. 1) Dans lApologie, Socrate justife son activit de rfutateur en lui donnant pour ori-
gine la rponse de lOracle de Delphes, qui avait dclar que personne ntait plus savant que Socrate (Apologie
20c-21a); et cest pour comprendre cette rponse que Socrate va chercher (Apologie 21b-22e), mais en vain, un v-
ritable savant (sur tout cela, voir, mon Introduction Platon, Apologie de Socrate et Criton, Paris, Flammarion, 1997
[gf, 848], pp. 65-74). 2) Une voix divine infchit ngativement le comportement de Socrate en intervenant dans
les dtails les plus insignifants de sa vie; comme je lexplique dans Socrates and the divine signal according to
Platos testimony: philosophical practice as rooted in religious tradition, Apeiron, 28, 2 [special issue: Socrates
and divine sign, ed. by P. Destre and N. D. Smith] [2005], pp. 1-12 et 3) ici dans le Banquet, le seul savoir quil admet
possder (Banquet 177d6-7), le savoir relatif aux choses de lamour, Socrate dit le tenir de Diotime, une femme ex-
perte en matire religieuse.
96 luc brisson
prend pourquoi Socrate est un sducteur si droutant,1 comme lexplique Alcibiade
dans son magnifque loge (Banquet 214a-222a).
Paris cnrs
P. S. Je remercie Monsieur Frdric Plin qui ma aid corriger les preuves de ce texte.
1 Lorsque Socrate, au terme dune rfutation, amne un jeune homme reconnatre son ignorance, il suscite
en lui le dsir du savoir et donc du bien; or, tout naturellement ce dsir, le jeune homme, inversant par ailleurs le
sens de la relation amoureuse habituelle, o cest lamant plus g qui poursuit de ses avances laim plus jeune
(sur le sujet, voir L. Brisson, Agathon, Pausanias, and Diotima in Platos Symposium: Paiderastia and Philosophia,
in J. Lesher, D. Nails and F. Shefeld (edd.), Platos Symposium. Issues in Interpretation and Reception, Washington
d.c., Center for Hellenic Studies, 2006, pp. 229-251), le reporte sur un Socrate (voil ce quexplique trs bien L.-A.
Dorion, dans son Socrate, pp. 89-94.) qui physiquement ne devait pas rpondre aux canons de la beaut de lpoque.
banquet (199c-201c et 201e-203a) 97
Giuseppe Cambiano
olti dialoghi platonici mettono in scena la rappresentazione scritta di confuta-
zioni di opinioni espresse nello scambio tra interlocutori in una discussione ora-
le. Per lo pi Socrate a confutare i suoi interlocutori, ma talvolta Socrate stesso ad
essere confutato, per esempio da Parmenide nel dialogo omonimo. stato tuttavia pi
volte notato che in alcuni degli ultimi dialoghi se si accetta la cronologia tradizionale
di essi Platone presenta anche la confutazione orale di dottrine formulate in testi scrit-
ti da parte di autori assenti alla discussione. Il problema che intendo afrontare se e co-
me possa essere confutato, nella prospettiva platonica, un testo scritto. Accenner quin-
di preliminarmente a due punti: quali siano secondo Platone le condizioni di fatto, non
formali e logiche, afnch una confutazione sia possibile; quali siano le considerazioni
svolte nellultima parte del Fedro sulla possibilit e sulle modalit della confutazione di
un testo scritto e della sua difesa dalle confutazioni. Partendo di qui prender in esame
il dialogo nel quale, a mio avviso, rappresentata pi articolatamente la confutazione
di un testo scritto, il libro di Protagora, ossia il Teeteto.
La confutazione socratica, qual presentata nei primi dialoghi, indipendentemente dal
fatto che essa possa far leva su contraddittoriet interne o sul contrasto con fatti o con
opinioni generalmente condivise, richiede che il confutato sia presente in carne ed os-
sa a rispondere alle domande di Socrate. Solo a questa condizione possibile raggiun-
gere passo passo e nella conclusione laccordo, homologia, tra Socrate e il suo interlocu-
tore.1 Se tale homologia non raggiunta ad ogni passo della discussione, non si pu dire
che si sia pervenuti ad una confutazione. Ci si pu allora chiedere se sia possibile con-
futare un testo scritto, nel caso che lautore sia assente e quindi non possa rispondere
alle domande, formulando altre credenze ed esprimendo di volta in volta il proprio con-
senso o dissenso. In prima istanza sar necessario prendere in considerazione le rifes-
sioni esplicite di Platone su questo punto: il testo essenziale naturalmente il Fedro, in
particolare lultima parte nella quale sono svolti rilievi critici nei confronti della scrittu-
ra. Qui infatti Platone sottolinea che lo scritto presenta un punto di somiglianza con la
pittura: esso consiste nellindurre a credere che i suoi prodotti si ergano davanti al let-
tore come esseri viventi, ma se si pone loro qualche domanda, tacciono solennemente.
1 Per una descrizione degli aspetti della confutazione socratica e del suo carattere personale sempre essen-
ziale Robinson 19532, in particolare pp. 7-19; cfr. anche pp. 75-84 sulla procedura per domande e risposte. Non ri-
levante per la mia questione afrontare il problema se la confutazione socratica mirasse soltanto a dimostrare lin-
coerenza e incompatibilit fra le credenze sostenute dallinterlocutore oppure fosse anche uno strumento per la
ricerca della verit in sede morale. Su questo punto e sulla necessit che linterlocutore sia sincero nelle sue ri-
sposte fondamentale Vlastos 1983; cfr. anche Vlastos 1991, pp.107-131. Di qui si sviluppato anche negli anni
successivi un ampio dibattito sugli Oxford Studies in Ancient Philosophy, ad opera di R. Kraut, T. Brickhouse e
N. D. Smith, R. Polansky e altri. Per una critica al punto di vista di Vlastos cfr. Benson 1995.
Si potrebbe credere che essi parlino come se pensassero (phronein) qualcosa, ma se vo-
lendo apprendere qualcosa di quanto detto nello scritto, si pone qualche domanda, lo
scritto signifca sempre soltanto la stessa cosa (hen ti semainei monon tautn ae) (275 d
4-9). Un rilievo analogo svolto da Platone anche nel Protagora, dove oratori che pro-
nunciano lunghi discorsi, come Pericle, sono paragonati a libri, che se interrogati su
qualche punto, non hanno nulla da rispondere o da dire essi stessi, anzi se qualcuno li
interrogasse anche su qualche piccolo punto delle cose dette, come i bronzi percossi ri-
suonano a lungo e protraggono il loro suono, se non li si ferma. Qui lanalogia col li-
bro appare in un contesto nel quale sono contrapposti i lunghi discorsi continui degli
oratori e il kat brachy dialegesthai, attraverso brevi domande e risposte, privilegiato da
Socrate. Lopposizione dunque fra due tipi di oralit, ma il libro fatto corrisponde-
re ad uno di essi, ossia al lungo discorso continuo, caratterizzato appunto dalla fssit
semantica (iterazione dello stesso suono) e dallincapacit di fornire risposte ad even-
tuali domande, mediante esplicazioni o integrazioni.1 Entrambi i passi del Protagora e
del Fedro esprimono, anche se in modi diversi, quella che potremmo chiamare la mo-
nosemanticit stabile del libro, la quale contrasta con la nozione moderna dellinesau-
ribilit infnita delle interpretazioni possibili di uno stesso testo. Credo sia necessario
non presupporre come ovvio anche in Platone questo principio tipico di certe correnti
dellermeneutica moderna.2 Che lidea di monosemanticit dello scritto non fosse
uninvenzione platonica potrebbe essere confermato da uniscrizione alla base di una
statuetta votiva del vi a.C., trovata sullAcropoli, nella quale si legge: A qualsiasi uomo
me lo chieda, io rispondo sempre la stessa cosa, ossia che Androne, fglio di Antifane,
mi ha dedicata.3 Ma se lo scritto itera sempre la stessa cosa e non risponde alle do-
mande, com possibile confutarlo? Scrittura e confutazione parrebbero collocarsi su
due piani totalmente incomunicabili.
Nello stesso Fedro Platone sottolinea a pi riprese che lo scritto circola nelle mani di
chiunque, ma, se lautore assente, esso non capace di difendersi e di portare aiuto a
se stesso.4 Ma da che cosa deve difendersi lo scritto? La possibilit di una sua circola-
zione indiscriminata anche presso destinatari non scelti dallautore stesso pu dar luo-
go a un fraintendimento e a un vilipendio dello scritto. Per descrivere questa situazio-
ne Platone usa i verbi plemmelein e loidorein (275 e 3-5): il primo indica propriamente lo
stonare, landare contro il melos (il suono, il metro o il ritmo), il secondo loltraggiare,
vilipendere, ingiuriare. Loggetto di ci sar il contenuto del libro, pi che il libro come
oggetto materiale, anche se non chiaro in che consista il vilipendio: in una critica fuo-
1 Protagora 328 e 5-329 a 7. Sullimpossibilit di interrogare Omero e i poeti in generale cfr. anche Ippia minore
365 d 1, Protagora 347 e 3-7, dove si accenna anche alla discordanza tra gli interpreti nel comprendere che cosa pen-
sasse o intendesse dire (noein) il poeta.
2 Ferrari 1987, p. 221 aferma che la fssit del testo scritto un miraggio e che, dopo tutto, ci saranno tante
interpretazioni diferenti del suo signifcato quanti sono gli interpreti; per Platone la vera stabilit invece di ci
che scritto nelle anime, cio convinzione personale. Questultima osservazione esatta, ma lidea della molte-
plicit delle interpretazioni una proiezione moderna, non contenuta nel testo platonico (dove si parla piuttosto
di plemmelein e loidorein, ossia di fraintendimenti di un testo scritto), e nulla dice che per Platone fosse unovviet,
come sembra essere per noi. 3 Lazzarini 1976, n. 658.
4 Fedro 275 e 3-5, 276 c 7-9. Per converso il discorso scritto nellanima di chi apprende ed accompagnato da
episteme capace di difendere se stesso (276 a 5-7). Sul tema del boethein, del prestare soccorso a un discorso orale
o scritto ha insistito Szlezak 1985, ma nella prospettiva del problema dellesistenza di dottrine orali attribuibili a
Platone e allinterno dei dialoghi platonici stessi. A me interessa invece il problema del boethein a proposito di scritti
non di Platone. In generale per la critica platonica alla scrittura cfr. Ferrari 1987, p. 218 sgg., anche per osserva-
zioni critiche sullinterpretazione di Derrida.
100 giuseppe cambiano
ri luogo o in un semplice fraintendimento da parte di lettori incompetenti o ostili? In
ogni caso si tratta di qualcosa che non rispetta il testo per quello che n gli rende giu-
stizia. Ma difcile scorgere in ci qualcosa che abbia a che fare con la confutazione,
almeno nel senso in cui lintende Socrate e con lui, pare, Platone. Il contesto sembra ri-
levare che gli autori di queste operazioni scorrette nei confronti dello scritto siano de-
stinatari inadeguati di esso e, quindi, non si pu pensare a flosof, per i quali invece la
confutazione una faccenda seria. Non a caso il tema della confutazione invece in-
trodotto in seguito nel Fedro, quando Socrate indica quali siano le condizioni perch
autori di generi letterari tradizionali esemplifcati con Lisia e i compositori di discor-
si, Omero e i poeti, Solone e gli scrittori di leggi1 possano essere considerati anche f-
losof. Tra queste c anche la condizione che lautore sia in grado di portare aiuto al
suo scritto afrontando un elenchos riguardante le cose che ha scritto e si dimostri capa-
ce di parlare (legon) lui stesso e di dimostrare che le cose che ha scritto sono dappoco
(phaula) (278 b 6-e 2). Il rispetto delle condizioni flosofche, ossia conoscere la verit
(condizione antecedente allo scrivere) e capacit di difendere lo scritto dopo che sta-
to composto, sembrano azzerare le diferenze tra questi generi letterari, riconducen-
doli a un comune denominatore flosofco. Ma la cosa da sottolineare che in questo
passo si parla di confutazione dello scritto, ma in un contesto orale, nel quale lautore
capace di prender la parola in difesa del suo scritto, non della difesa mediante un altro
scritto successivo.
Ma che cosa succede se lautore assente? Si potrebbe pensare che la confutazione
consista nel mettere in luce contraddizioni fra parti di una stesso scritto, ma Platone
non dice nulla in proposito. Nel Fedro Socrate rileva che lorazione di Lisia, che stata
letta da Fedro, allinizio non ha defnito correttamente eros e, proprio partendo da que-
sta assunzione arbitraria iniziale, Lisia ha strutturato (syntaxamenos) tutto il discorso si-
no alla fne (263 d 1-e 2). Per controllare questo punto letto una seconda volta linizio
del discorso di Lisia e ci lascia intendere che per controllare se lo scritto stato com-
posto in modo tecnicamente corretto, il banco decisivo di prova dato dallo scritto stes-
so, di cui presupposta la stabilit. La rilettura dellinizio del discorso manifesta, agli
occhi di Socrate, uninversione di parti rispetto alla sequenza logica necessaria (ek tinos
anankes ephexes) in cui dovrebbe esser composto il discorso, ossia uno scambio dellini-
zio con la fne, con il resto gettato alla rinfusa, come ci che veniva di volta in volta in
mente allautore nello scrivere (264 a 4-b 8). Ma si pu dubitare che il mettere in rilievo
questo tipo di disordine dello scritto costituisca una vera e propria confutazione. Ci su
cui Fedro pu convenire che ogni discorso deve essere una totalit di parti organica-
mente connesse tra loro: Socrate paragona il discorso ad un essere vivente, uno zoon,
dotato di un corpo proprio, con testa, piedi e parti intermedie, ossia con parti scritte
congruenti tra loro e con il tutto secondo un certo ordine naturale (264 c 2-5).2 chia-
ro che ci costituisce lesatto opposto dellinterscambiabilit indiferenziata delle parti
entro lordine complessivo del discorso, come avviene nellepigramma posto sulla tom-
ba di Mida, i cui versi possono essere letti in qualsiasi ordine (264 c 7-e 2). Al discorso di
Lisia Socrate contrappone il caso del suo secondo discorso, che era stato strutturato
consequenzialmente con una defnizione iniziale (eros come mania), alla quale aveva fat-
1 Cfr Kuehn 2000, pp. 13-26, 71-73.
2 Per passi paralleli cfr. De Vries 1969, pp. 211-2, nonch Gorgia 503 e-504 per una descrizione del lavoro del-
lartigiano in termini analoghi.
dal fedro al teeteto di platone 101
to seguito una similitudine, con la quale forse si era giunti a toccare qualcosa di vero
(265 b 1-c 3). Inoltre esso aveva mostrato di far uso delle procedure in cui si articola
lindagine dialettica. Ma questa contrapposizione metodica, operata da Socrate, del
proprio discorso al discorso scritto di Lisia, privo di organicit, non pu certo essere
scambiata per una confutazione in senso proprio. Unaltra via invece oferta dalla si-
tuazione in cui lautore sia assente, ma siano presenti altri (discepoli o ammiratori) che
ne seguano le orme e quindi possano rispondere in difesa dello scritto dellautore.
Quando parla dei discorsi accompagnati da scienza e seminati nelle anime di destinata-
ri adatti e capaci di germogliare in esse dando luogo ad altri discorsi seminati in altre
anime e cos via, Socrate indica, tra le caratteristiche di questi discorsi, anche quella di
essere capaci di portare aiuto a se stessi e al phyteusanti, ossia al seminatore di questo di-
scorso orale (276 e 7 - 277 a 1). Qui per non si fa cenno a scritti, ma soltanto a discorsi
orali. possibile estendere questa attivit di difesa anche nei confronti di eventuali
scritti di questo seminatore di discorsi orali nelle loro anime? Poco prima Socrate ave-
va parlato di discorsi scritti per gioco a scopo mnemonico per s, ossia per lautore, ma
anche per chiunque segua la medesima orma (276 d 4-5). Non si pu dunque esclude-
re che destinatari che si avvertano e siano in grado di proseguire per le tracce indicate
dallautore dello scritto possano anche essere in grado di difendere non solo i discorsi
orali seminati nelle loro anime, ma anche gli scritti composti da tale seminatore a sco-
po mnemonico anche per essi.
Si tratta ora di vedere se sia possibile rintracciare nei dialoghi platonici indicazioni che
si muovano lungo le linee direttive tracciate nel Fedro. Nei Memorabili (I, 6, 14) di Seno-
fonte messa in bocca a Socrate, che sta discutendo sui valori dellamicizia, la seguen-
te afermazione: i tesori degli antichi sapienti, che essi lasciarono dopo averli scritti in
libri, io, srotolandoli anelitton, con evidente riferimento al papiro in comune con gli
amici, li percorro e se vediamo qualcosa di buono lo scegliamo. Il verbo vedere sem-
bra riferirisi al percorrere con lo sguardo il testo scritto (non solo allascoltarne la let-
tura); comunque non una lettura solitaria, ma in comune, che approda ad una sele-
zione di contenuti giudicati positivi. Essa appare dunque come componente dello stare
insieme di Socrate con amici a flosofare: ci fa gi capire che anche in un contesto ora-
le non era assente la pratica di far riferimento a testi scritti. Nei dialoghi platonici nu-
merosi sono i casi nei quali si fa riferimento a versi di poeti citati a memoria, a quanto
pare, da Omero a Simonide,1 ma anche a testi di flosof e naturalmente non sar qui
possibile condurre unindagine esaustiva di essi. Inoltre non tutte le rappresentazioni
platoniche di discussioni di libri possono essere ricondotte nel solco della confutazione
vera e propria. Mi limito a due esempi di libri con autori, rispettivamente presenti e as-
senti, nei quali emerge come il riferimento a libri possa dar luogo a rilievi critici nei lo-
ro confronti, che per non sono riconducibili alla confutazione. Nella parte iniziale del
Parmenide si fa riferimento al libro di Zenone di Elea, portato per la prima volta ad Ate-
ne e letto da Zenone in persona davanti ad ascoltatori desiderosi di ascoltarlo. Lunico
che aveva gi ascoltato in precedenza tale lettura efettuata da Zenone era Pitodoro, he-
tairos di Zenone (127 c 1-d 5). Si tratta di una lettura pubblica del proprio scritto da par-
1 Per la discussione dei versi di Simonide nel Protagora cfr. Giuliano 1991, con bibliografa.
102 giuseppe cambiano
te dellautore, la quale non coincide tout court con la pubblicazione di esso, dato che
se non altro gi Pitodoro laveva ascoltato. Si pu parlare di pubblicazione nel senso
che attraverso questa lettura esso era reso noto per la prima volta ad un pubblico ate-
niese. Dopo aver udito leggere lo scritto, Socrate chiede che sia riletta la prima ipotesi
del primo argomento, ossia contro la molteplicit e, una volta riletta, chiede: Che in-
tendi dire con questo? Che, se gli enti sono molti, allora devono essere simili e dissimi-
li, ma questo appunto impossibile, dato che n i dissimili possono essere simili n i si-
mili dissimili? Non dici cos? (127 d 6 - e 4). La domanda di Socrate ha una portata
ambivalente: da una parte si tratta di accertarsi di aver capito che cosa ha detto Zeno-
ne; dallaltra di capire se Zenone intendeva dire ci che egli, Socrate, ha riformulato o
parafrasato con la sua domanda, dal momento che la domanda non la semplice ripe-
tizione o citazione delle parole di Zenone che sono state appena lette. Rispondendo
cos (127 e 5) Zenone accetta come corretta la formulazione di Socrate, che allora pu
procedere a riformulare anche il secondo momento dellargomentazione di Zenone,
chiedendo: se dunque impossibile che i dissimili siano simili e i simili dissimili,
anche impossibile che gli enti siano molti? Infatti se fossero molti, subirebbero cose im-
possibili (127 e 6-8). A questo punto Socrate chiede se questa sua riformulazione ri-
specchi lintentio auctoris, espressa dal verbo boulontai, e se quindi egli stia assegnando il
signifcato corretto alle parole di Zenone: forse questo che vogliono le tue argo-
mentazioni, nientaltro che sostenere contro tutto ci che si dice comunemente, che gli
enti non sono molti? (127 e 8-10). In tal modo Socrate chiede se questa sia la legittima
conclusione del ragionamento, ossia la contraddittoria rispetto allassunto ipotetico ini-
ziale se i molti sono. Che le domande di Socrate abbiano lintento di individuare lin-
tentio auctoris confermato da ci che segue immediatamente: ed proprio di questo
che tu credi (oiei) che ciascuna delle tue argomentazioni sia una prova (tekmerion), tan-
to da ritenere anche che quante siano le argomentazioni che hai scritto, altrettante sia-
no le prove da te fornite che gli enti non sono molti? Dici cos o sono io che non capi-
sco correttamente?, dove la comprensione riguarda evidentemente ci che Socrate ha
udito leggere (127 e 8 - 128 a 1). In questo contesto viene dunque riconosciuta la possibi-
lit di fraintendere uno scritto che appena stato letto, nel senso di non cogliere con
esattezza lintenzione dellautore, che cosa intendeva dire. Ci che Socrate vuol capire
se Zenone consideri prove della tesi che gli enti non sono molti i discorsi che ha scrit-
to, ossia se Zenone conferisca ad essi lo statuto di dimostrazioni. Il vantaggio qui da-
to dal fatto che presente lautore dello scritto, il quale pu dunque confermare se la
riformulazione socratica o interpretazione (diremmo noi) di proposizioni del suo scrit-
to corrispondono al suo intendimento, ossia dimostrare che gli enti non sono molti.
Proprio questo ci che Zenone conferma, ossia che Socrate ha compreso bene lin-
tendimento dellintero scritto (holon to gramma ci che bouletai) (128 a 2-3). Siamo quin-
di in presenza di un caso di non fraintendimento di uno scritto, confermato dallautore
dello scritto stesso, per cui Socrate si ricordino le osservazioni del Fedro apparirebbe
qui come un destinatario adatto dello scritto, la cui lettura-pubblicazione non avrebbe
dato luogo a una distorsione o vilipendio di esso.
Ma la discussione dello scritto di Zenone non si arresta qui, perch linterrogazione
mira ora a individuare se, al di l o al di sotto dellintenzione dellautore quale emer-
sa nellaccordo raggiunto fra Zenone e Socrate non sia presente una intentio pi pro-
fonda, deliberatamente nascosta e non immediatamente visibile nello scritto. Socrate si
rivolge ora a Parmenide, non a Zenone, manifestando di aver capito (manthano) unal-
dal fedro al teeteto di platone 103
tra cosa, una seconda intenzione dello scritto, ossia che Zenone vuole (di nuovo boule-
tai) ribadire i suoi legami di afnit con Parmenide non solo in generale, ma anche con
il suo scritto. E spiega questa sua afermazione in questo modo: infatti in certo modo
egli ha scritto la stessa cosa (tautn) che hai scritto tu, ma con una modifcazione (meta-
ballon) cerca di ingannarci (exapatan, quindi di farci credere) che egli stia dicendo qual-
cosa di diverso (128 a 4-8). La cosa interessante che per avere conferma di questa se-
conda intenzione dello scritto di Zenone, Socrate non interroga Zenone, ma
Parmenide. questultimo che pu confermare se il contenuto dello scritto di Zenone
identico al suo e se le modifcazioni introdotte da Zenone siano soltanto un espediente
per mascherare questa identit e quindi ingannare i lettori del proprio scritto, che sa-
rebbero indotti a considerarlo qualcosa di diverso e di nuovo rispetto a quello di Par-
menide. Socrate non accenna al fatto che queste modifcazioni siano puramente for-
mali, per esempio costituite dal passaggio da una formulazione in poesia (il poema di
Parmenide) a una in prosa (lo scritto di Zenone). Si pu notare tra laltro che Socrate
qui chiama sygramma lo scritto di Parmenide, che pure in versi, mentre poche righe
dopo parla di poiemata di Parmenide, il che pu confermare che il punto rilevante non
la distinzione tra poesia e prosa. Infatti Socrate fornisce una diversa spiegazione della
identit di contenuti tra i due scritti, al di l dei mutamenti introdotti da Zenone: Tu,
Parmenide, nei poiemata dici che il tutto uno e ne fornisci prove di nuovo tekmeria
ottime ed eccellenti, mentre Zenone dal canto suo dice che gli enti non sono molti e
fornisce anchegli numerosissimi e amplissimi tekmeria di ci. Dunque che uno afermi
lunit e laltro invece neghi la molteplicit e che ciascuno di voi due parli in modo da
produrre lapparenza di non aver detto nulla di identico, pur dicendo praticamente le
stesse cose, d limpressione che le cose dette da voi siano dette per passare sopra noi
altri (128 a 8 - b 6). Assumendo che uno e molti siano contraddittori e quindi che uno
sia equivalente a non molti e viceversa, si pu concludere che le due asserzioni di Par-
menide e di Zenone abbiano lo stesso contenuto, anche se in due descrizioni diverse.
Di qui Socrate pu inferire la sostanziale identit delle tesi espresse nei due scritti, al di
l di diferenze puramente apparenti, anzi pu addirittura insinuare il sospetto che tali
diferenze siano volute e deliberatamente ingannevoli. Socrate chiede conferma di que-
sta sua conclusione allautore dello scritto cronologicamente pi antico, Parmenide, pi
vecchio di Zenone, aspettandosi che proprio da lui la sua tesi possa venir confermata e
che quindi Parmenide possa rivendicare o sottolineare la sua priorit e originalit. in-
teressante per che il padre del discorso pi antico tace e risponde invece lautore dello
scritto pi recente, Zenone, il quale di primo acchito sembra accettare lanalisi di So-
crate, ma precisa immediatamente che Socrate non ha percepito la verit del suo scrit-
to in ogni suo aspetto, bench, come le cagne della Laconia, egli sia bravo a inseguire
le tracce delle cose dette (128 b 6 - c 2). Mentre prima aveva ammesso la correttezza del-
lanalisi interna del suo scritto, condotta da Socrate, e della sua fnalit consistente nel
dimostrare che gli enti non sono molti, ora Zenone sostiene che Socrate non ha capito
la vera motivazione del suo scritto in rapporto a quello di Parmenide. Egli parla di ale-
theia del proprio scritto, quasi che la verit autentica di esso consistesse, pi che nella
dimostrazione a s stante che gli enti non sono molti, nel contribuire a raforzare lale-
theia dello scritto di Parmenide. Zenone imputa a Socrate di non aver colto il vero in-
tento, la dianoia del suo scritto, ossia di averlo frainteso, e quindi in quanto autore pre-
sente pu andare in soccorso del suo scritto. Egli nega che il suo scritto intendesse
presentarsi come dotato di unautorit venerabile (semnynetai) la vera autorevolezza
104 giuseppe cambiano
pertiene piuttosto a Parmenide e di aver voluto nascondere il suo reale intento, cer-
cando di renderlo impercepibile allesterno: il suo scritto non aveva alcun intento di pre-
sentarsi come una grande impresa rispetto allo scritto di Parmenide (128 c 2-5). Il vero
intento del suo scritto era piuttosto di essere un soccorso (boetheia) a quello di Parme-
nide contro quelli che tentavano di porlo in commedia, sostenendo che se il tutto uno,
il discorso di Parmenide veniva a subire molte conseguenze ridicole e contraddittorie
(128 c 5 - d 2).
Descrivere se stesso come autore di uno scritto in soccorso di quello di Parmenide si-
gnifca autoqualifcarsi stando alle indicazioni del Fedro come uno che condivide le
tesi dello scritto di Parmenide, si pone sulle sue orme e costruisce il proprio scritto co-
me mezzo di difesa per quello di Parmenide, che riveste dunque il ruolo primario e fon-
dante. Ma, a diferenza del caso prospettato nel Fedro, abbiamo qui uno scritto che va in
difesa di un altro scritto, non un soccorso orale in difesa di un testo scritto. Zenone
presenta come intendimento del suo scritto non quello di dimostrare qualcosa di nuo-
vo e di diverso da quanto sostenuto da Parmenide, quanto di dimostrare che sono le as-
sunzioni degli avversari di Parmenide, ossia della molteplicit, che portano a conclu-
sioni ridicole e contraddittorie, per cui dallaccertamento della falsit della loro tesi
usciva ulteriormente confermata la verit della tesi di Parmenide che il tutto uno (128
d 2-6). Lambito in cui Zenone colloca il suo scritto quello agonistico dellantilegein,
latteggiamento psicologico che aveva sorretto questo suo scritto giovanile era stato la
philonikia, ma tale scritto gli era stato rubato e ci gli aveva impedito di decidere se ren-
derlo pubblico. Questo fatto esclude ai suoi occhi che nel comporlo egli fosse guidato
invece da philotimia, il desiderio, proprio di uno pi anziano, di acquistare reputazione,
come invece aveva insinuato Socrate (128 d 6 - e 4). Socrate accetta queste considerazio-
ni di Zenone (128 e 5-6). Si pu dunque dire, rimanendo al nostro problema, che il di-
scorso di Zenone era apparso come un soccorso orale, una boetheia riuscita per giusti-
fcare il vero intendimento del suo scritto, contro il fraintendimento di Socrate, proprio
argomentando che il suo scritto non era stato altro che un soccorso a quello di Parme-
nide. Ma chiaro, in ogni caso, che le considerazioni di Socrate sullo scritto di Zenone
non possono esser considerate una confutazione di esso e pertanto questa autodifesa di
Zenone non pu, a sua volta, essere considerata una difesa da una confutazione ri-
guardante i contenuti del suo scritto, ma solo una difesa delle sue intenzioni nello scri-
vere tale libro. piuttosto subito dopo che Socrate cerca di argomentare contro un pun-
to dellargomentazione di Zenone, ossia sullimpossibilit che i simili siano dissimili e i
dissimili simili, introducendo il riferimento alle idee e alla nozione di partecipazione,
ma a questo punto, come si sa, sar Socrate stesso a subire la confutazione da parte di
Un caso di presenza di un libro in prosa il cui autore assente rappresentato dal li-
bro di Anassagora nel Fedone, ma anchesso non oggetto di confutazione. DallApolo-
gia di Socrate (26 d 6 - e 2) sappiamo che si trattava di un libro abbastanza difuso, acqui-
stabile ad Atene ad un prezzo non altissimo, una dracma. Il Fedone introduce il libro di
Anassagora come una tappa importante nello sviluppo intellettuale di Socrate, narrato
da Socrate stesso. Insoddisfatto dalle risposte discordanti su questioni di flosofa della
natura, Socrate sente leggere da un libro di Anassagora che lintelletto ordinatore e
causa di tutte le cose (97 b 8 - c 2). Si tratta dellascolto soltanto di questa parte del libro
di Anassagora, non di eventuali altre parti contenenti argomentazioni a sostegno di que-
sta tesi o applicazioni di essa nellindagine su questioni particolari. Tuttavia sulla base
dal fedro al teeteto di platone 105
di questo semplice enunciato generale Socrate inferisce quelle che a lui appaiono con-
seguenze di esso, ossia conferisce a tale tesi lo statuto di premessa o antecedente di un
conseguente: se lintelletto causa, allora lintelletto ordina e dispone ciascuna cosa nel
modo migliore. La premessa implicita che una causa intelligente non pu che pro-
cedere intelligentemente e quindi disporre le cose nel modo migliore. Il secondo mo-
mento dellinferenza di Socrate caratterizzato dal ritenere che la premessa generale
enunciata da Anassagora e intesa in questo senso fnalistico debba poi trovare applica-
zione anche nella spiegazione delle singole questioni; altrimenti non ci sarebbe coeren-
za tra lenunciato generale e i punti particolari sottoposti a indagine da Anassagora (97
c 2 - d 1). In base a questa inferenza che si tratti di uninferenza sembra confermato da
logizomenos in 97 d 6 Socrate si sente legittimato a considerare Anassagora il maestro
sulla causa degli enti e a nutrire laspettativa di trovare nel libro di Anassagora spiega-
zioni causali di singoli fenomeni coerenti con lenunciato generale della sua teoria (97
d 5 - 98 a 1). chiaro che questa aspettativa pu emergere solo sulla base dellinterpre-
tazione che Socrate ha dato di questo enunciato e delle conseguenze che ne ha tratto.
A mostrare che si trattava di uninterpretazione e di uninferenza erronea sar proprio
la lettura diretta dellintero libro di Anassagora, che presenter enunciati, ossia spiega-
zioni causali non conformi al carattere fnalistico che Socrate aveva ritenuto di poter in-
ferire dallenunciato generale del libro, che aveva sentito leggere (98 a 1 - c 2). Lo scritto
di Anassagora, non contenendo le conseguenze che Socrate aveva ritenuto di poter in-
ferire dalla premessa generale, si mostrava quindi a lui come logicamente incoerente
nel suo insieme: appariva vera la tesi generale, che per rimaneva a s stante, senza tro-
vare applicazione nelle spiegazioni particolari, che non si conformavano a questa pre-
messa. Anassagora per non era presente per poter eventualmente difendere la coe-
renza del suo scritto, contro il fraintendimento socratico derivato da una sua erronea
inferenza. Per Socrate tuttavia lo scritto di Anassagora, rivelandosi incoerente, cessava
di essere quel potenziale contenitore di verit che il Socrate, ascoltatore di una lettura
solo parziale dello scritto, aveva sperato di trovare nellintero scritto. Anassagora non
poteva propriamente essere il suo maestro, perch di fatto continuava a rimanere im-
pigliato nel terreno sul quale si era mossa la peri physeos historia, che sfociava necessa-
riamente in forme di diaphonia. Il libro di Anassagora non aveva svolto una funzione di-
rimente rispetto alle spiegazioni discordanti dei vari fenomeni naturali. Il suo libro per
aveva avuto il merito di risvegliare in Socrate una questione, anche se tale questione non
aveva trovato la sua soluzione autentica nel libro stesso. Formalmente potrebbe sem-
brare che il rifuto del libro di Anassagora fosse il risultato di una confutazione, se per
confutazione sintende la dimostrazione dellincoerenza fra le credenze espresse da un
individuo. Efettivamente il libro di Anassagora mostrava uno scollamento e una di-
scordanza fra tesi generale e spiegazioni particolari, lungo una linea che pu ricordare
le considerazioni svolte nel Fedro sullo scritto come totalit organica paragonabile a un
essere vivente. Ma non si pu dire che laccertamento di questa incoerenza fosse il ri-
sultato di uninterrogazione ripetuta del testo scritto: in realt era bastata la lettura del-
lo scritto a mostrare a Socrate ma non ad Anassagora, lautore assente questa in-
coerenza, ma ci era dipeso appunto dalla interpretazione che Socrate stesso aveva dato
della tesi generale di Anassagora e dalle conseguenze che egli aveva ritenuto che si do-
vessero logicamente ricavare da essa. Proprio questo punto Anassagora, a causa della
sua assenza, non poteva contestare e, quindi, con le sue risposte sconfermare o confer-
mare limputazione che Socrate muoveva al suo scritto.
106 giuseppe cambiano
In situazione pi favorevole per afrontare la questione della confutabilit di uno scrit-
to ci possiamo trovare a proposito di altri dialoghi, per esempio nel Sofsta, dove a con-
futazione sono sottoposti non solo Parmenide, ma anche molti esponenti della tradi-
zione flosofca. Il dialogo in cui per queste tematiche sono sviluppate pi a lungo e
pi articolatamente il Teeteto in relazione al libro di Protagora. Anche in questo caso
lautore assente, ma vengono escogitate una serie di procedure che tentano di ren-
derlo interlocutore in una discussione dialettica per domande e risposte. Si potrebbe
pensare che una conferma del collegamento del Teeteto con le rifessioni del Fedro, che
abbiamo descritto, possano trovare conferma nelle analisi stilometriche, che da varie
prospettive sembrerebbero confermare una vicinanza cronologica tra i due dialoghi.1
Ma preferibile mantenere un certo scetticismo verso le considerazioni stilometriche,
tra laltro basate sullassunzione esplicita o implicita che Platone non abbia mai riscrit-
to i suoi dialoghi anche a distanza di tempo e che comunque questa eventuale riscrit-
tura non abbia portato a radicali modifcazioni nellimpiego delle particelle, la cui fre-
quenza duso funziona da indicatore cronologico. Un punto di connessione tra i due
dialoghi potrebbe piuttosto essere dato dal fatto che il Teeteto esplicitamente presen-
tato come un dialogo caratterizzato da una funzione ipomnematica. Tra le condizioni
positive per scrivere in modo corretto Platone, come si visto, enuncia nel Fedro, ac-
canto alla conoscenza della verit e al rivolgersi a destinatari adatti, anche la funzione
di fungere da supporto mnemonico per la dimentica vecchiaia dellautore stesso e per
chiunque si ponga sulla sua stessa orma (276 d 1-5).2 Ora, il Teeteto appunto presen-
tato nel primo prologo come la scrittura della memorizzazione di una conversazione
dialogica. Si di fronte a una serie di incastri: Platone mette per iscritto una conversa-
zione nella quale Euclide di Megara racconta a Terpsione come ha messo per iscritto il
suo ricordo di una conversazione intercorsa fra Socrate, Teeteto e Teodoro (142 c 3-8).
Questa scrittura si era scandita in una serie di fasi: il punto di partenza era il suo ricor-
do (anemnesthen) di questo evento, al quale faceva seguito lascolto ad Atene dellespo-
sizione orale da parte di Socrate dei discorsi svolti con Teeteto (142 c 8 - d 3). Euclide non
in grado di riferire a memoria (ap stomatos) a Terpsione questi discorsi, ma lo infor-
ma che appena tornato a Megara scrisse hypomnemata di quanto gli aveva riferito So-
crate. Ci sottolinea che la fonte autentica dello scritto di Euclide ci che egli aveva
memorizzato non tanto della conversazione originaria, quanto del resoconto socratico
di questa conversazione originaria. Ma questo hypomnema immediato era poi sottopo-
sto da Euclide ad una riscrittura a proprio agio sulla base del suo richiamare alla me-
moria (kat scholn anamimneskomenos) il racconto socratico (142 d 6-143 a 2). Non del
tutto chiaro se questo lavoro di rielaborazione dellhypomnema originario consistesse in
integrazioni o anche in sostituzioni. Ci che certo che Euclide, ogni volta che si re-
ca ad Atene, pone ripetute domande a Socrate stesso su ci che non ricordava e, torna-
to a Megara, corregge il suo scritto sulla base delle risposte di Socrate. Il suo scritto era
1 Cfr. Brandwood 1990 e 1992, con una propensione a considerare il Teeteto anteriore al Fedro, e Kahn 2002,
di cui per non condivido lo scetticismo sul fatto che uneventuale riscrittura potesse modifcare il tasso di ricor-
renza delle particelle. Devo a Jaap Mansfeld la segnalazione dei saggi di P. Keyser in Bryn Mawr Classical Review,
2, 1991, pp. 402-27 e 3, 1992, pp. 58-74, nei quali sono svolte importanti considerazioni sui limiti delle analisi fonda-
te su presupposti stilometrici. 2 Sulla funzione ipomnematica cfr. Cerri 1991, pp. 87-92.
dal fedro al teeteto di platone 107
il risultato fnale di questo processo scandito nel tempo: esso si presentava come reso-
conto fedele, che aveva come fonte e termine di controllo Socrate stesso (143 a 3-5).1 Eu-
clide dichiara infne di aver trasformato la forma indiretta, con Socrate narrante, in un
dialogo diretto, senza gli incisi narrativi (143 b 5-c 5). Anche altri dialoghi come il Sim-
posio o il Parmenide sono presentati nel prologo come degli hypomnemata, ma diretta-
mente di conversazioni dialogiche o comunque di una catena di trasmissioni orali, sen-
za alcun riferimento a scritti che riportino tali conversazioni. Pi complessa invece la
situazione del Teeteto, dove Platone lautore nascosto di uno scritto che riporta lo scrit-
to di Euclide, frutto di una memorizzazione di questultimo, ma sottoposta alle inte-
grazioni e al controllo di Socrate stesso. Lo scritto di Euclide viene dunque presentato
come il resoconto memorizzato non di una narrazione orale di una conversazione, ma
di questa conversazione orale stessa, che costituisce nella quasi totalit il dialogo che ab-
biamo. Ma a monte dello scritto di Euclide c lo scritto di Platone che ingloba quello
di Euclide come suo contenuto: pur sempre Platone che conduce le fla, anche se at-
traverso la fnzione di cedere il suo ruolo di autore a Euclide. come se Platone voles-
se distogliere lattenzione da s e richiamare lattenzione sul fatto che questo dialogo
era dovuto, in ultima analisi, alla scrittura che ne aveva fatto Euclide. lunico caso, nel-
lintero corpus platonico, in cui com stato notato un dialogo viene presentato co-
me la lettura ad alta voce da parte di uno schiavo di un testo scritto (143 b 3).2
In questa cornice di dialogo diretto sinserisce quindi la confutazione non solo degli
interlocutori di Socrate, ma anche di uno scritto, quello di Protagora.3 Non analizzer
la struttura logica della confutazione di Protagora, su cui esiste una letteratura ormai
sempre pi sterminata;4 mi sofermer invece sui modi in cui rappresentato e de-
scritto il processo di questa confutazione. Il riferimento esplicito al libro di Protagora
fatto da Socrate subito dopo che Teeteto ha tracciato lidentit fra conoscenza e perce-
zione (dora innanzi cp). Socrate rileva che c il rischio (kindyneueis) che Teeteto abbia
pronunciato un logos sullepisteme non da poco, anzi che questo logos sia quello che fu
detto anche da Protagora, che in un certo altro modo (tropos) disse queste stesse cose
(t aut tauta). A comprovare questo fatto Socrate cita lasserzione protagorea sulluo-
mo misura di tutte le cose (dora innanzi hm), che in seguito risulter essere linizio
dello scritto di Protagora (152 a 1-4).5 Che si tratti di citazione da uno scritto sembra con-
fermato dalla domanda di Socrate a Teeteto se labbia letto (anegnokas), cui Teeteto ri-
1 Velardi 2000, p. 134, ha giustamente rilevato che le fasi di questo lavoro di rielaborazione scritta del rac-
conto sono scandite con precisione dalle tre forme verbali: laoristo egrapsamen segnala limmediatezza della pri-
ma stesura; la processualit della seconda fase espressa dal durativo egraphon; il prodotto nella sua versione de-
fnitiva presentato dal risultativo gegraptai: il passaggio dallhypomnema (prima fase) al biblion. Qui hypomnema
signifca specifcamente stesura provvisoria (cfr. in generale Dorandi 2007, pp. 65-81. Sul prologo del Teeteto cfr.
anche Narcy 1994, pp. 22-30.
2 Cfr. Friedlaender 2004, p. 869, per il quale il prologo serve a sottolineare laccuratezza storica del racconto.
3 tesi generale di Sedley 2004 che il Socrate del Teeteto rappresenti non lattuale flosofa di Platone, ma un
ritorno al Socrate dei primi dialoghi, che funziona da levatrice del neonato platonismo. Ci rende anche conto
della centralit della confutazione nel ritratto del Socrate del Teeteto. Il merito di aver richiamato lattenzione sul-
le connessioni del Teeteto con la discussione fnale del Fedro sulla scrittura va riconosciuto a Ford 1994.
4 Cfr. da ultimo in questo numero di Antiqvorvm Philosophia larticolo di Castagnoli con i suoi ampi riferi-
menti bibliografci.
5 Probabilmente pou alle righe a 2 e a 4 signifca in qualche punto del suo scritto (cfr. Denniston 1934, pp.
490-495). Secondo lantico commentatore anonimo al Teeteto (col. lxii, 8-17 Bastianini-Sedley) Socrate sapeva che
la doxa di Protagora diferiva da quella di Teeteto, ossia hm da cp e per questo diceva kindyneueis t aut. Ma cre-
do che il kynduneueis serva piuttosto a sottolineare che il discorso di Teeteto (cp) non da poco: a comprovare ci
Socrate adduce (ironicamente?) il fatto della sua coincidenza con hm.
108 giuseppe cambiano
sponde di averlo letto sovente (152 a 4-5). Il verbo anagignoskein comporta il riconoscere
lettere e parole scritte, anche se la lettura avviene ad alta voce e non in silenzio; in que-
sto caso sembra trattarsi della lettura diretta di un testo, magari ad alta voce, pi che il
sentir leggere un testo da altri.1 Ed verosimile che lammissione di Teeteto, con linsi-
stenza sul pollakis, non riguardi solo la lettura della frase iniziale dello scritto di Prota-
gora, ma lintero scritto, che dunque ben noto a Teeteto, il quale tra laltro presen-
tato come dotato delle qualit di essere capace di apprendere facilmente, eumath (144
a 3) e di buona memoria, mnemon (144 a 7). Il commentario anonimo al Teeteto (col. lx,
26-40 Bastianini-Sedley), collega il fatto che Teeteto si era impegnato seriamente sugli
scritti di Protagora al suo essersi associato a Teodoro. Ed efettivamente Teodoro di Ci-
rene presentato ripetutamente come hetairos di Protagora, legato da stretti rapporti
con Protagora e si pu quindi inferire che anchegli fosse al corrente dei contenuti del-
lo scritto di Protagora.2 Questo punto importante perch serve a sottolineare che la
discussione e confutazione del libro di Protagora avviene con interlocutori che lo co-
noscono bene e sono quindi in grado di controllare la correttezza della descrizione so-
cratica dei suoi contenuti e la congruit dei suoi tentativi di confutazione.
Il primo problema di chiarire in che senso lhm esprima lo stesso contenuto di cp,
anche se in forma o modo diverso. A tale scopo Socrate chiede a Teeteto se Protagora
in un certo modo (qui pos equivale al tina allon tropon di poco prima) non dica che qua-
le ciascuna cosa appare (phainetai) a me, tale per me e quale a te, tale per te, ma tu
e io siamo entrambi uomo e Teeteto conferma ci, ossia la correttezza dellattribuzio-
ne di questa afermazione a Protagora (152 a 6-9). Socrate invita a tener dietro a Prota-
gora, che essendo un sapiente verosimilmente non fa chiacchiere, e introduce lesem-
pio del vento freddo per alcuni e non freddo per altri (152 b 1-3), dove non chiaro se
epakolouthein signifchi condividere la tesi di Protagora o seguirlo anche in ci che sca-
turisce dalla sua tesi. Sembrerebbe pi probabile questultimo senso, in quanto Socrate
opera una ridescrizione del metron nel senso di apparire a, la quale essenziale per
portare alla successiva identifcazione di phainesthai con aisthanesthai. Ammettere che
il vento per uno freddo e per un altro no signifca prestar credito (peisometha) a Prota-
gora (152 b 5-7). Ma a ci Teeteto risponde con un semplice sembra (eoiken) (152 b 8),
il che potrebbe far pensare che egli non trova riscontro diretto di ci nello scritto di Pro-
tagora, anche se ne ammette la compatibilit. In ogni caso, grazie allesempio del ven-
to, Socrate pu appunto compiere lultimo passo per arrivare a riconoscere lidentit di
contenuto tra hm e cp. Esso consiste nel chiedere se phainetai equivalga a aisthane-
sthai e anche in questo caso Teeteto d la sua conferma (152 b 9-12). In altri termini,
Teeteto che ha letto il libro di Protagora, non riscontra nellassimilazione operata da
Socrate alcuna incompatibilit col libro di Protagora. E alla conseguenza, tratta da So-
crate, che la percezione ha sempre per oggetto ci che e non mai falsa, in quanto
conoscenza, Teeteto risponde ancora con un phainetai (152 c 5-7). Socrate ha in tal mo-
do riformulato lenunciato scritto dellhm in termini di percezione e di relativit della
percezione dei singoli individui, anche se in seguito si aprir lo spazio per uninterpre-
tazione allargata dellhm, non ristretta al solo ambito percettivo. Che linterpretazione
nellhm di uomo in senso distributivo (ciascun singolo uomo) e di metron in relazione
ad apparire a, possa essere ricondotta al Protagora storico e al suo libro potrebbe es-
1 Su anagignoskein come lettura in pubblico cfr. Mansfeld 1994, pp. 193-194.
2 Sui personaggi di Teeteto e Teodoro cfr. Blondell 2002, pp. 251-313.
dal fedro al teeteto di platone 109
sere confermato dal Cratilo (385 e 4-386 a 4), a meno che anche in questo caso non si trat-
ti di una ridescrizione platonica, con altri termini linguistici, dellhm protagoreo.1 Di fat-
to comunque il Teeteto tende a presentare la riformulazione socratica come una descri-
zione corretta del contenuto dellhm e da questa riformulazione scaturisce lidentit di
contenuto di hm e cp. Lassunto di Socrate che il contenuto dellhm protagoreo sia lo
stesso di cp, anche se espresso in una forma linguistica diversa, che per non modifca
tale contenuto, anzi ne mostra lidentit. Laccettazione di questa identit, non conte-
stata da Teeteto che ha letto pi volte lo scritto protagoreo, essenziale per lo svolgi-
mento e laccettazione della successiva confutazione, che riguarder sia la formulazio-
ne letterale dellhm, sia la riformulazione di esso presentata da Socrate, a prescindere
dalla considerazione del contesto e dellorizzonte pragmatico entro il quale si colloca-
va originariamente lhm e che a noi rimane sostanzialmente ignoto. Naturalmente si
posto agli interpreti il problema se efettivamente tra cp e hm ci sia identit di conte-
nuti o se uno implichi laltro o entrambi si implichino reciprocamente.2
Ma prima di procedere alla discussione dellhm nella formulazione relativistica ri-
stretta alla percezione, qual qui presentata, Socrate compie unulteriore passaggio, an-
cora una volta costruito intorno alla sottolineatura della sapienza (passophos) di Prota-
gora (152 c 8-9). Questo passaggio consiste nellattribuire a Protagora una dottrina
esposta in segreto ai suoi discepoli, mentre a noi, accozzaglia molteplice, egli lespres-
se per enigmi (enixato) (152 c 9-10). Si tratta, come noto, della tesi del mobilismo uni-
versale (dora innanzi mu). Socrate descrive il modo in cui essa fu comunicata da Pro-
tagora, distinguendo due tipi di destinatari, i discepoli, ai quali questa dottrina sarebbe
stata trasmessa a chiare lettere, ma in maniera riservata (oralmente?) solo ad essi, e tut-
ti gli altri esclusi da questa trasmissione diretta, ma per i quali sarebbe stata messa in
opera una forma di trasmissione indiretta per enigmi. Il verbo ainittesthai da ainos, pa-
rola importante, carica di signifcato ha il senso di dire cose signifcative e perci dif-
cili da comprendere e si specializza nel signifcato di parlare per enigmi. Si tratta cio di
un parlare in cui il contenuto rilevante non emerge chiaramente, ma va decifrato.3 Se
si ammette che si sia trattato di una trasmissione, ma solo per via di enigmi, si apre
unulteriore alternativa: oralmente o per iscritto? Cornford notava che la frase con cui
1 Sul problema della storicit della presentazione platonica della tesi protagorea in senso ampio o ristretto al-
le percezioni cfr. da ultimo Sedley 2004, pp. 49-53. Barnes 1979, vol. ii, pp. 239-342 accredita a Platone la lettura re-
strittiva, in chiave percettiva, dellhm. Per una rassegna delle interpretazioni dellhm cfr. Guthrie 1969, pp. 183-191.
2 Secondo il commentatore anonimo (col. lxvi, 11-43 Bastianini-Sedley) largomentazione sarebbe conforme
al terzo schema, dove la prima premessa (come le cose appaiono a ciascuno, tali sono anche per lui) sarebbe posta
da Protagora, mentre la seconda (cio phainetai = aisthanetai) sarebbe posta da Platone. La nota di Bastianini-Se-
dley, pp. 550-551, chiarisce come non si tratti della terza fgura di sillogismo aristotelico (come pensava Moraux),
ma che lautore aveva in mente una serie di relazioni di identit. Cornford 1935, pp. 30-31, interpreta il rappor-
to fra cp e hm come combinazione dialettica operata da Platone allo scopo a portare alla luce il pieno signifca-
to del semplice enunciato cp. Secondo Gulley 1962, p. 77 (cfr. anche p. 80) cp implicherebbe laccettazione di hm;
infatti da hm seguirebbe che la percezione infallibile, il che sarebbe assunto nella tesi che la conoscenza sia per-
cezione. Per McDowell 1973, p. 121, lassezione dellidentit di hm e cp richiederebbe almeno che ciascuna delle
due dovrebbe implicare laltra. Ma mentre in una direzione ci sembra ovvio, ossia se pc, allora hm vera, pi
dubbia limplicazione inversa, ossia che, se vero hm, allora la percezione sempre veridica. Secondo Burnye-
at 1990, p. 11-12, il passo 152 bc presenterebbe unargomentazione volta a mostrare che da hm possiamo derivare
la proposizione che ogni percezione conoscenza, mentre non sarebbe per ora chiaro linverso, ossia se ogni co-
noscenza sia percezione. Per una discussione delle relazioni fra cp e hm cfr. anche Hardy 2001, p. 51 e Chappell
2004, pp. 58-59, secondo cui cp implica hm.
3 Ford 1994, pp. 207-208, rinvia allinterpretazione allegorica, ma non mi pare che nel testo platonico vi siano
accenni in questa direzione.
110 giuseppe cambiano
introdotta questa dottrina segreta di Protagora implica che essa non era reperibile ta-
le quale nel libro di Protagora e non esiste quindi alcun fondamento per inferire che
Protagora fosse un eracliteo. Di fatto, secondo Cornford, nessuno pu ritenere seria-
mente che tale dottrina fosse realmente insegnata da Protagora; invece Socrate che la
costruisce mediante una combinazione dialettica di elementi tratti, con importanti mo-
difcazioni e restrizioni, da Protagora e Eraclito.1 Nel dialogo Socrate presentato co-
me non facente parte della schiera dei discepoli di Protagora e, usando lespressione
noi, include se stesso e anche Teeteto nellaccozzaglia molteplice, sicch chiaro che
egli non ha conoscenza diretta della dottrina protagorea del mu. Eppure egli espone
questa dottrina segreta, evidentemente perch riuscito a decifrarla in base a quanto
Protagora aveva esposto o nel suo insegnamento orale pubblico (non quello riservato
ai discepoli) o in quanto aveva scritto. Non abbiamo per notizia di uninsegnamento
orale enigmatico di Protagora, che sarebbe estraneo del resto alla pratica sofstica del-
lepideixis. Resta dunque la possibilit che lo scritto di Protagora potesse contenere cen-
ni enigmatici che ad un buon decifratore sarebbero potuti apparire come rinvii alla dot-
trina del mu. Stando a come Platone lo presenta, il mu sarebbe dunque qualcosa di
implicito e presupposto nel libro che enuncia lhm. Il lettore avvertito che lesposi-
zione del mu che seguir frutto dunque delloperazione di deciframento dello scritto
protagoreo compiuta da Socrate e che questa esposizione socratica non altro che
lesplicitazione di contenuti impliciti nel libro protagoreo. Si tratterebbe dunque di una
dottrina strettamente correlata allhm, verosimilmente di una necessaria premessa im-
plicita di esso, il nucleo nascosto di un insegnamento, che nello scritto trovava espres-
sione solo indiretta e obliqua nella dottrina dellhm. Si tratterebbe cio dellontologia
nella quale lhm troverebbe piena giustifcazione. Se assumiamo che mu sia avanzato co-
me aiuto a sostegno di hm, avremmo allora che il soccorso prestato allo scritto di Pro-
tagora comincerebbe ben prima della cosiddetta apologia di Protagora e questo soc-
corso sarebbe avanzato da Socrate stesso.
Occorre tuttavia osservare che mu non presentato da Socrate come prerogativa
esclusiva di Protagora, bens come una tesi sulla quale convergono molti sapienti, an-
che antecedenti cronologicamente a Protagora stesso, da Omero a Eraclito, con luni-
ca eccezione di Parmenide (152 d 2 - e 9).2 Questa inclusione di Protagora in una schie-
ra pi ampia di pensatori una mossa essenziale per precisare che ci che Socrate
1 Cornford 1935, pp. 34-36, 39, 49: lefetto di questo innesto di eraclitismo sarebbe di modifcare hm, nel sen-
so di sostituire diviene ad nellespressione ci che appare a me, per me, per cui hm varrebbe nellambito
del sensibile, tesi con cui Platone andrebbe oltre Protagora. Per E. Heitsch 1988, pp. 26-27, una certa arbitrariet
dellidentifcazione di cp con hm e il mu sarebbe confermata da 183 b 7 - c 3. Anche per McDowell 1973, pp. 121-
122, la dottrina segreta non va presa sul serio, ma Platone lassumerebbe come unimplicazione della dottrina espli-
cita di Protagora. Cfr. anche Burnyeat 1990, p. 12, per cui il mu fornirebbe invece condizioni necessarie e suf-
cienti per cp e hm; cfr. anche Sedley 2004, pp. 39-44. Mi-Kyoung Lee 2000, p. 52, sostiene che la relazione tra hm
e la dottrina segreta (che non semplicemmente una dottrina del fusso universale, ma un insieme di pi tesi con-
nesse) non sarebbe di implicazione mutua. La dottrina segreta avrebbe invece lo status di una ipotesi indipenden-
te usata da Platone per difendere Protagora e spiegare che cosa signifca per le cose avere propriet solo relative
ai percipienti. Plato quindi non argomenterebbe che cp implica hm e che questo implica mu, bens che mu impli-
ca hm e questo cp. Ma anchegli riconosce che Platone la chiama segreta presumabilmente per riconoscere che
Protagora non espose mai una tale dottrina, che sarebbe invece escogitata da Platone per difendere lhm, senza
che ci comporti che si tratti di implicazioni dedotte (pp. 47-48, 53-54).
2 Ci non signifca che agli altri sostenitori del mu si debba attribuire una forma di relativismo protagoreo, co-
me sembrerebbe afermare Mi-Kioung Lee 2000, p. 84. Sul precedente di Ippia nel costruire questi accostamenti
di poeti e flosof mediante passi tematici paralleli cfr. soprattutto Mansfeld 1986, pp. 28-31.
dal fedro al teeteto di platone 111
esporr sono i tratti generali di una dottrina condivisa da molti pensatori, incluso Pro-
tagora. Resta dunque aperta la possibilit che lesposizione socratica possa far leva non
soltanto sui pochi cenni enigmatici eventualmente presenti nel libro di Protagora, ma
anche su un patrimonio dottrinale pi ampio. In ogni caso quello che ora Socrate espo-
ne non un phaulos logos (152 d 2), proprio come non lo era la defnizione di Teeteto di
cp (151 e 8). Tutti gli esempi che in seguito Socrate fornisce a supporto della dottrina se-
greta del mu (153 a 5 sgg.) non sono mai presentati come risalenti specifcamente a Pro-
tagora. E la stessa cosa vale per le articolazioni riguardanti il mu, per esempio sulla ge-
nesi di percezioni a partire dallincontro di movimenti e cos via. Quando, nel contesto
di queste esposizioni, Protagora viene invocato soltanto per ricordare che cosa po-
trebbe dire lui e chiunque cerchi di dire le sue stesse cose nellipotesi appena nega-
ta che le cose fossero dotate di qualit stabili (154 b 6-8) oppure quali domande egli po-
trebbe porre (154 c 7). Ossia Protagora viene introdotto come interlocutore potenziale
fttizio nel quadro di una discussione dialettica tra Socrate e Teeteto, ma chiaro che
qui il libro di Protagora e ci che esso dice esplicitamente o anche per enigmi abban-
donato. In ballo la discussione della tesi generale del mu, come conferma la risposta
di Teeteto ad una domanda che Protagora potrebbe porre: interessante che Teeteto
formuli la sua opinione, ma al tempo stesso dica anche quale dovrebbe essere la rispo-
sta coerente con lammissione del mu, per non contraddirsi (154 c 10 - d 2). E che in gio-
co non sia qui Protagora direttamente sembra ulteriormente confermato dal fatto che
Socrate esclude che lui e Teeteto possano esser chiamati sophoi (154 d 8-9), come invece
ripetutamente era prima qualifcato Protagora, e dal fatto che ci che sar sottoposto
ad esame una serie di phasmata presenti in noi (155 a 2) che generano contraddizio-
ni nelle nostre anime. A 155 d 5-7 Socrate ribadisce che le cose che sono state dette deri-
vano dalle (ex) cose che noi diciamo che Protagora dice, dove chiaro che si tratta di
inferenze tratte da Socrate a partire da quello che Socrate stesso aferma essere il det-
tato originario protagoreo. Non stupisce quindi che Teeteto confessi di non rendersi an-
cora conto di ci (155 d 8). Si tratter allora da parte di Socrate di investigare con Tee-
teto la verit nascosta ulteriore allusione al titolo dellopera protagorea1 di questa
dottrina che non si trova esplicitamente nello scritto, tanto vero che si dice che essa
la verit della dianoia di quelluomo o piuttosto di uomini rinomati (155 d 9 - e 1), cio
tutti i mobilisti, segno che ancora una volta oggetto di discussione la tesi generale del
mu, pi che una dottrina esclusiva del solo Protagora.
Sin qui abbiamo assistito allesposizione di tre dottrine, presentate come connesse tra
loro, anche se in modi diversi: cp, hm e mu. Poi si comincia a passare alla confutazione,
dapprima dellhm, ossia del contenuto esplicito del libro di Protagora, anche se nella ri-
formulazione o ridescrizione socratica (157 e 1 sgg.). Per tale confutazione essenziale
che ci sia qualcuno che risponda alle domande poste da Socrate: questo ruolo svolto
da Teeteto, che come sappiamo ha letto spesso il libro di Protagora e quindi, si pu pre-
sumere, possa rappresentare lautore, ma che di fatto espone ci che egli pensa. Si ve-
da ad esempio 158 a 8-b 4, dove Teeteto, accettando i controesempi avanzati da Socrate
sulle percezioni in stati di sogno, malattie e mania, sconfessa la propria tesi cp, ma da-
ta lequivalenza della sua tesi cp con hm, di fatto accetta che i controesempi socratici
1 Gi il commento anonimo al Teeteto scorgeva, sulla base del dialogo, un riferimento allo scritto di Protago-
ra dal titolo per aletheias (col. ii, 1-11 Bastianini-Sedley, col loro commento a p. 481). Secondo Untersteiner 19962,
p. 24, probabile che tale titolo fosse stato assegnato per la prima volta da Platone stesso. Cfr. anche Heitsch 1969.
112 giuseppe cambiano
valgano da confutazione anche dellhm e quindi di una tesi centrale del libro di Prota-
gora. Socrate assume allora su di s lonere di presentare le risposte a queste obiezioni,
risposte che potrebbero dare la costruzione con an e lottativo i sostenitori della
verit delle cose che appaiono per colui cui appaiono di volta in volta, ossia di hm (158
e 5-6). Qui ovviamente il riferimento anche al libro di Protagora, ma essendo Prota-
gora assente, Socrate che comincia ad assumersi il ruolo di difesa di esso dai tentativi
di confutarlo. Tale difesa procede per domande, volte allo scopo di articolare meglio il
discorso sulla relativit delle percezioni. Il logos esposto da Socrate conferma la persi-
stenza del suo signifcato relativistico (cfr. 160 b 8 - c 2), ossia che la mia percezione ve-
ra per me ed io sono krits kat tn Protagoran delle cose che sono per me, che sono e di
quelle che non sono per me, che non sono (160 c 7-9), dove chiara la sostituzione del-
luomo metron con io sono giudice. La risposta di Teeteto eoiken mostra ancora una
volta che egli non contesta la congruit dellinterpretazione dellhm in termini di rela-
tivit delle percezioni. Di qui diventa poi possibile il passaggio allultima conclusione ri-
guardante lidentit di hm con cp, ossia con la tesi sotenuta inizialmente da Teeteto (160
d 1-3). A questo punto Socrate ribadisce che Teteto ha detto benissimo che episteme non
altro che percezione e che sono venuti a coincidere (eis tautn sympeptoken) il mu (per
cui si fanno i nomi di Omero, Eraclito e tutto la schiera citata in precedenza, entro cui,
come sappiamo era inclusa anche la dottrina segreta di Protagora), lhm del libro di Pro-
tagora e la tesi di Teeteto cp (160 d 5-e 2). Questo risultato presentato da Socrate co-
me se fosse parto di Teeteto e dovuto alla sua opera di ostetrico (160 e 2-3) e, in ultima
analisi, come generazione dovuta ad entrambi (egennesamen, 160 e 5). Ma che signifca
che coincidono eis tautn? Si tratta di identit di contenuti, anche se in formulazioni lin-
guistiche diverse o di connessione e, in tal caso, di quale tipo di connessione? Su questo
punto esiste un ampio dibattito,1 ma qui interessa in primo luogo il fatto che il dialogo
avanza la tesi di una connessione stretta fra lhm contenuto nel libro di Protagora e cp,
da una parte, e mu, dallaltra, ma secondo modalit diverse. Mentre per hm e cp si par-
la di identit di contenuti, mu sembra presentato come un presupposto in gran parte
implicito nel libro di Protagora e non solo in Protagora, ma tale da implicare hm. In al-
tri termini la confutazione della tesi di Teeteto comporta la mobilitazione dellHM del
libro protagoreo come tesi sostanzialmente equivalente a cp, ma anche della tesi del mu
come implicante hm. Queste operazioni richiedono ridescrizioni dei contenuti del libro
di Protagora e insieme mobilitazioni di tesi e argomenti non contenuti esplicitamente
a chiare lettere in esso. E la confutazione che seguir viene presentata come riguardante
non specifcamente il solo libro di Protagora, ma il parto opera congiunta di Teeteto
e Socrate, ossia le tre tesi coincidenti cp, hm e mu.
Ma prima di arrivare a tale confutazione entra in scena Teodoro, cui Socrate dopo
aver ribadito di esser capace non di produrre propri logoi, ma di assumere e accogliere
solo il logos di un altro sophs (161 a 7-b 5) comunica di essersi meravigliato dellinizio
dellAletheia di Protagora, dove questi non disse che di tutte le cose misura il porco o
il cinocefalo e qualche altro essere capace di percezione. Traendo da ci linferenza che
Protagora, sul piano della phronesis non meglio di un girino di rana (161 c 2-d 2), di fat-
to Socrate perviene a un vero e proprio vilipendio di Protagora. Si noti che il vilipendio
1 Per una serie di interpretazioni diverse del collegamento fra le tre tesi, in termini di combinazione dialettica,
di identit o di implicazione cfr. Cornford 1935, p. 31; Gulley 1962, pp. 78-80; McDowell 1973, pp. 121-122; Bur-
nyeat 1990, pp. 9; Fine 2003, pp. 132-159 (cfr. anche pp. 160-183); Mi-Kioung Lee 2000, in particolare pp. 53-54.
dal fedro al teeteto di platone 113
ha di mira proprio lincipit dello scritto di Protagora, non una sua dottrina orale, quasi
a confermare lattendibilit della descrizione del Fedro sui pericoli della circolazione in-
discriminata del libro. Ed signifcativo che tale vilipendio intervenga appena entra in
ballo Teodoro, di cui si sottolinea che Protagora era hetairos (161 b 8-9), implicitamente
presupponendo, sempre in conformit alle considerazioni del Fedro, che in quanto he-
tairos Teodoro dovrebbe essere in grado di portare aiuto allo scritto di Protagora e sal-
varlo da tale vilipendio. Come si sa, obiezione cruciale che, se vero hm, Protagora
non pu presentarsi come maestro di altri a chiedere compensi (161 d 3-e 3). Ma signi-
fcativo che di fronte a ci Teodoro riconosce che Protagora suo philos e che quindi
egli non accetterebbe di vederlo confutato attraverso le sue ammissioni, ossia se fosse
lui a dover rispondere alle domande di Socrate in nome di Protagora, una possibilit che
era prevista, come si sa, nel Fedro. Daltra parte per egli fa proprio il requisito di sin-
cerit che secondo Vlastos richiesto agli interlocutori di Socrate nel formulare le
sue risposte in modo che non vadano contro la propria doxa. Ma ci lascia aperto lo spa-
zio alla possibilit che la sua doxa non coincida con ci che ha detto o potrebbe dire Pro-
tagora e quindi il suo ruolo di difensore lo costringerebbe, per opporsi a Socrate e di-
fendere Protagora, anche ad enunciare opinioni diverse dalle proprie. Proprio per
evitare questa situazione egli invita Socrate a riprendersi Teeteto come interlocutore
(162 a 4-8), rifutando in tal modo di assumere il ruolo di difensore del libro di Protago-
ra contro i tentativi di confutarlo. Per altro verso, altrettanto rilevante per il nostro di-
scorso unaltra osservazione di Socrate, ossia che se vera lAletheia di Protagora, cio
lhm pronunciato ek ths adytou, dal santuario inacessibile del libro, con le sue implica-
zioni relativistiche, diventa impossibile lintera attivit di Socrate stesso e, precisamen-
te, il dialegesthai, inclusa la confutazione delle doxai, dal momento che hm comporta
che esse siano vere per chi le ha (161 e 4-162 a 3).1
Teeteto sembra colpito dalle obiezioni di Socrate, in particolare dallosservazione che
stando allhm, che riguarda uomini o di, lui stesso Teeteto verrebbe ad essere per nul-
la inferiore sul piano del sapere a qualsiasi altro uomo o a un dio, e quindi sta quasi per
abbandonare hm con le sue implicazioni relativistiche (162 c 7 - d 2). allora nuovamente
Socrate ad assumere il ruolo di difensore, ma ipotizzando che cosa dir lautore assen-
te, Protagora, o qualcun altro, in sua vece e in suo favore (hyper autou). Egli imputer ai
suoi obiettori di non far uso di dimostrazioni necessarie (apodeixis e ananke), ma sol-
tanto di ci che verosimile (eikoti), ossia di ci che se Teodoro o qualche altro geo-
metra usasse non varrebbe neppure un soldo; si tratta invece di questioni importanti
per le quali non ci si pu accontentare di discorsi puramente persuasivi e verosimili (pi-
thanologia e eikosi) (162 e 4-163 a 1). Questa controobiezione attribuita da Socrate ipote-
ticamente a Protagora o a un suo sostituto assai signifcativa, dal momento che gli in-
terlocutori nella discussione sono Teodoro e Teeteto, entrambi geometri. Ci che la
controbiezione mette in rilievo luso confutatorio di procedure metodicamente ina-
deguate, che non dovrebbero soddisfare i geometri con cui Socrate sta tentando di con-
1 Sullinconfutabilit di Protagora, dato il carattere privato, disorganico e relativistico del suo mondo e delle
sue credenze, cfr. soprattutto Mc Cabe 2000, pp. 23-138, che tuttavia non discute la questione con esplicito riferi-
mento al problema della confutabilit del libro scritto, che in qualche modo assicura identit e persistenza allau-
tore di esso.
114 giuseppe cambiano
futare il libro di Protagora. Infatti Teeteto riconosce subito che non giusto accogliere
obiezioni costruite su queste basi inattendibili (163 a 2-3). Allora Socrate tenta unaltra
strada, riproponendo la domanda se la conoscenza sia percezione o se siano cose diverse
(163 a 7-10), ossia ritornando dallhm al quesito iniziale riguardante cp, avanzata da Tee-
teto, che aveva messo in movimento tutte le successive connessioni. Si tratta di una se-
rie di domande volte a mostrare linsostenibilit di cp, con un uso del noi che coin-
volge interrogante e rispondente (cfr. 163 b 1, b 3, b 7, b 9), con la mobilitazione di un
interrogatore immaginario (163 d 1-4) e con Teeteto che risponde in prima persona.
interessante per che Socrate non presenta queste obiezioni come decisive e defnitive,
ossia come una confutazione pienamente riuscita: non cantiamo vittoria prima del tem-
po, egli dice (164 c 4-5). Grazie a questa serie di domande venuto a morte il mito di
Protagora (hm) e insieme quello di Teeteto (cp) (164 d 4-10),1 ma non sarebbe andata co-
s se fosse ancora vivo il padre del primo mito, che avrebbe respinto molti attacchi; ora
invece che esso orfano, noi dice Socrate lo copriamo di fango (164 e 2-4) e neppu-
re i tutori lasciati da Protagora uno dei quali Teodoro vogliono venire in soccorso
di esso, anzi c il rischio che siamo noi stessi a venire in soccorso (boethein) di esso per
ragioni di giustizia (164 e 4-6). Come si vede, ci troviamo esattamente nella situazione
descritta dal Fedro: lo scritto, nel caso di autore assente, cio orfano di padre, incapa-
ce di difendere se stesso da vilipendi (qui si parla di coprire di fango) e da tentativi di
confutazione. Questo aiuto potrebbe essere fornito da chi si pone sulle stesse tracce del-
lautore, ma Teodoro precisa appunto che da tempo ha abbandonato queste tracce, nel
senso che ha lasciato i logoi psiloi per convergere apeneusamen termine geometrico
verso la geometria (164 e 7 - 165 a 3). La cosa dovrebbe fnire qui: n lautore (in quanto
assente) n altri legati da amicizia con lautore intervengono nella difesa dello scritto,
lasciato quindi in balia degli attacchi altrui, in questo caso di Socrate.
La mossa di Platone di non chiudere qui la confutazione del libro di Protagora, ma
di conferire a Socrate stesso il ruolo di difensore, capace di formulare controobiezioni
alle obiezioni del confutatore (cio di Socrate stesso). In tal modo Socrate assume il du-
plice ruolo di confutatore e di difensore da queste sue stesse confutazioni. Si tratta di
uscire da una zona confnante col vilipendio per individuare una strategia di difesa, ma
chiaro che, se anche questa sar smantellata, lobiettivo della confutazione uscir raf-
forzato. inutile sottolineare il raforzamento di persuasivit della confutazione rag-
giungibile mediante questa procedura di assunzione del ruolo di difensore che cerca di
rendere piena giustizia al discorso contenuto nel libro di Protagora. Ma la difesa dello
scritto di Protagora non pu procedere introducendo argomenti diversi e nuovi non ri-
conducibili in qualche modo a Protagora o non connettibili alle tesi da lui espresse nel-
lo scritto, perch in tal caso non avremmo una difesa del suo scritto, che non si di-
mentichi secondo Platone signifca sempre soltanto ununica identica cosa. Si tratter
allora di fornire una ridescrizione dellhm capace di resistere alle obiezioni mosse in pre-
cedenza, ma insieme ancora riconoscibile come protagorea. La strategia consiste nel
provare a dire che cosa Protagora dir a sostegno delle proprie tesi: la ben nota apo-
1 Ford 1994, pp. 209-210 n. 18 mette in parallelo questo passo con la conclusione della Repubblica, dove si dice
(621 d 8-9) che in tal modo il mito stato salvato e non perito, ipotizzando una contrapposizione fra gli hazards
della trasmissione scritta (i testi di Protagora sarebbero stati bruciati, secondo la leggenda) e il modo magico in
cui il racconto vero di Er sfuggito alla morte e alloblio ed stato poi incorporato nella Repubblica. A ci si po-
trebbe aggiungere che Er torna tutto intero in vita a riferire il racconto, mentre di Protagora si ipotizzer in se-
guito (171 c 10-d 5) che emerger solo sino al collo.
dal fedro al teeteto di platone 115
logia di Protagora, che per Socrate dice che tocca a lui enunciare (164 e 4-5), dal mo-
mento che Protagora assente. Esplicitamente Socrate la presenta come le cose che sa-
r lui a pronunciare in difesa di Protagora, ma come se fossero dette da Protagora stes-
so in prima persona (165 e 7 - 166 a 2). Ci che messo immediatamente in dubbio
proprio il fatto che le obiezioni mosse da Socrate a Teeteto nel ruolo di rispondente pos-
sano essere considerate attendibili. Viene cio contestato che le obiezioni presentate sia-
no una vera e propria confutazione proprio perch, a prescindere dai contenuti di vol-
ta in volta presi in considerazione, non avvenuta attraverso linterrogazione diretta di
lui, Protagora. Secondo Protagora, infatti, Socrate ha spaventato Teeteto e in tal modo
ha condizionato la sua risposta, perch Teeteto non era in grado di prevedere le conse-
guenze della sua risposta e in tal modo aveva fatto apparire ridicolo Protagora stesso
(166 a 2-6). A questo punto messa in bocca a Protagora lindicazione delle condizioni
da rispettare afnch si possa considerare riuscita la confutazione di un autore assente:
quando esamini mediante domande qualcuna delle mie tesi, se linterrogato, dando in
risposta le cose che appunto io risponderei, sbaglia, allora sono io a essere confutato,
ma se risponde cose diverse, lo stesso interrogato ad esserlo (166 a 6-b 1). Da ci pos-
siamo inferire che la confutazione di un testo scritto e, pi in generale, di qualsiasi tesi
enunciata da qualcuno non pi presente, pu esser considerata riuscita soltanto a con-
dizione che le risposte di chi fa le veci dellautore siano le stesse che darebbe lautore se
fosse presente. Se questa condizione non rispettata, rimane sempre aperto il dubbio
che la confutazione non sia reale, perch colpirebbe un bersaglio diverso dallautore.
Lespediente socratico dellapologia messa in bocca a Protagora avrebbe quindi lobiet-
tivo di individuare le cose che Protagora stesso, non un suo sostituto, direbbe, per cui
se le asserzioni contenute nellautodifesa di Protagora riuscissero, a loro volta, anches-
se confutate, si potrebbe conferire maggiore attendibilit a questultima confutazione
e considerarla per certi versi decisiva. Infatti dopo aver elencato una serie di cose che sa-
ranno o no concesse rispetto alla discussione precedente, Protagora respinge il vilipen-
dio dellhm attraverso il riferimento a porci e cinocefali e quindi del suo scritto (166 c 6
- d 1), ma torna a invitare Socrate a confutare la tesi della relativit delle percezioni (166
c 2-6), di fatto facendo propria linterpretazione relativistica dellhm sul piano della per-
cezione. Non solo ma riaferma la validit dellhm, la quale fondata sul riconosci-
mento della diversit tra i singoli individui, dovuta al fatto che ad uno appaiono e sono
per lui certe cose e a un altro altre. Questa la verit comegli lha scritta (166 d 1-4). In
tal modo Protagora ribadisce che il suo scritto signifca sempre soltanto ununica cosa,
come precisato nel Fedro. Tuttavia questa verit contenuta nel suo scritto non in-
compatibile con la possibilit di distinguere fra chi sapiente e chi non lo , ma ci di-
pende dal fatto che a sapiente occorre dare un signifcato diverso da quello corrente,
ossia interpretarlo come colui che opera trasformazioni non dal falso al vero, ma dal
dannoso allutile e dal peggio al meglio. Ci che Protagora imputa a Socrate di aver
perseguito il suo discorso attaccandosi alle parole e il discorso che ora egli pronuncer
avr lobiettivo di far comprendere pi chiaramente (saphesteron) a Socrate ci che egli
dice (166 d 8 - e 1; cfr. anche 168 b 7 - c 1).
Il nuovo discorso di difesa non apporta dunque un mutamento dei contenuti delle te-
si contenute nel suo scritto, ma costituisce soltanto un nuovo modo di esporli, un mo-
do che dovrebbe risultare pi chiaro a Socrate, facendogli comprendere linfondatezza
delle sue obiezioni. Quel che non detto se questo chiarimento comporti anche del-
le aggiunte compatibili col contenuto dello scritto, o consista soltanto in una ridescri-
116 giuseppe cambiano
zione di questo in termini diversi. In assenza di documentazione inutile speculare se
lapologia contenga elementi rintracciabili nel Protagora storico. Restano tuttavia due
fatti salienti: il primo che essa rappresenta una riformulazione dellhm e quindi del
contenuto dellAletheia protagorea e il secondo che lapologia contiene almeno un ri-
ferimento ad uno scritto di Protagora. interessante infatti che questo discorso imma-
ginario di difesa, oltre ad afermare una distinzione sul piano della sapienza tra uomo e
animali, mobilita anchesso un riferimento a uno scritto di Protagora e, precisamente,
allincipit del suo libro sugli dei (b4), dove diceva di non poter dire degli dei n che sia-
no n che non siano (162 d 4-e 4).1 Non si pu dunque escludere che lapologia fosse co-
struita da Platone anche con lutilizzazione di elementi desumibili da scritti protagorei,
il che naturalmente avrebbe accresciuto autorevolezza a tale apologia per quanti fosse-
ro a conoscenza degli scritti protagorei.2 In ogni caso lapologia porta a precisare il senso
da attribuire alla parola sapiente e, in tal modo, giustifca la possibilit di distinguere
tra sapiente e non sapiente, contrariamente al precedente tentativo di confutazione, che
mirava appunto a mostrare come questa possibilit venisse ad essere negata dalle im-
plicazioni relativistiche dellhm. Grazie a questo nuovo signifcato di sapiente, anche
Socrate deve rassegnarsi, lo voglia o no, ad essere metron e quindi il logos di Protagora
ossia lhm, contenuto nel suo scritto salvato (167 d 1-4). Di qui parte il nuovo invito
da parte di Protagora a confutarlo, ma secondo unetica non agonistica della discussio-
ne e dellinterrogare che non comporti ingiustizia nei confronti dellinterrogato (167 d
4 - 168 b 2). Tali norme riguardano un contesto orale di discussione fra un interrogante
e un rispondente, ma chiaro che sono estendibili anche al caso in cui chi risponde ri-
sponde per un altro o per il testo scritto di un altro. Queste norme illustrate da Prota-
gora si possono agevolmente trovare indicate in vari dialoghi platonici, soprattutto del
primo periodo, da parte di Socrate stesso. Perch Platone le mette in bocca a questo
Protagora redivivo che viene in soccorso al suo scritto? Ovviamente non possibile da-
re una risposta sicura a ci, ma dal punto di vista di una strategia di persuasione si trat-
ta di una mossa importante, perch prepara ad accogliere la successiva confutazione,
che Socrate condurr, come una confutazione che rispetta precisamente i presupposti
etici della discussione e dellinterrogazione dialettica attribuiti a Socrate da Platone nel-
larco della sua opera. Si tratter quindi di una confutazione in piena regola anche sul
piano etico e pertanto accreditabile ad essere accolta come tale. Credo tuttavia che Pla-
tone sia consapevole e intenda rendere il lettore consapevole del fatto che lapologia
messa in bocca a Protagora non pu essere scambiata tout court con le parole di Pro-
tagora stesso.3 Platone infatti mette in bocca a Socrate stesso losservazione che queste
sono le cose che lui, Socrate, ha fornito in difesa dellhetairos di Teodoro: si badi, non di-
ce che Protagora hetairos di lui, Socrate, che in tal caso sarebbe un difensore per cos
dire autorizzato. E queste cose sono state avanzate da Socrate secondo le sue capacit,
smikr ap smikrn; ma se fosse stato vivo Protagora in persona ben maggiore sarebbe
stato il soccorso che egli avrebbe fornito alle proprie tesi (168 c 2-5). Non da esclude-
re una portata ironica in questa asserzione, quasi a voler dire che Protagora stesso non
1 Sulle implicazioni di questo incipit per linterpretazione dellhm cfr. Mansfeld 1981.
2 Per una rassegna di opinioni sulla presenza o meno di autentici elementi protagorei nellapologia cfr.
Untersteiner 19962, pp. 79-85 e 102 n. 1. Congetturale il tentativo di Cole 1966 di rintracciare nellapologia due
nuclei teorici opposti, di cui uno riconducibile al Protagora storico, cio allhm, e laltro, consistente in tesi utili-
taristiche, attribuitogli da Platone. Cfr. anche Sedley 2004, pp. 55-56.
3 Blondell 2002, p. 254, parla di ventriloquing.
dal fedro al teeteto di platone 117
avrebbe potuto fornire una difesa migliore di quella apprestata da Socrate; resta co-
munque il fatto che lapologia in ultima analisi riconducibile a Socrate e non a Prota-
gora in carne ed ossa e lascia quindi ancora aperto il margine alla possibilit di una mi-
glior difesa da parte di Protagora stesso.
La conclusione di questa apologia messa in bocca a Protagora invita Socrate a con-
futare anche il mu, non solo lhm (nella nuova formulazione in cui presentato); non
solo, ma si tratter anche di esaminare, partendo da queste cose, se vera cp oppure
no (168 b 1 - c 2). Socrate accoglie il rimprovero di aver spaventato un ragazzino e linvi-
to a prendere invece sul serio il logos di Protagora (168 c 8 - d 6).1 Naturalmente un mo-
do di procedere seriamente dovrebbe consistere nel condurre lesame di tale logos me-
diante domande e risposte tra adulti (168 d 8 - e 3), Ancora una volta Teodoro cerca di
dirottare su Teeteto il ruolo di rispondente, ma Socrate questa volta non recede e coin-
volge Teodoro nellesame sullhm (168 e 6 - 169 c 6), nella nuova formulazione presen-
tata nellapologia (166 d 3-8). Ma non irrilevante che Socrate torni a insistere sulla dif-
ferenza di situazione in una discussione con lautore presente o lautore assente. Egli
ribadisce infatti che se fosse presente Protagora stesso a dare il suo assenso (homologein)
e non fossimo invece noi, venendo in suo aiuto, a concederlo a nome suo, non ci sa-
rebbe alcun bisogno di riprendere nuovamente la tesi di Protagora per renderla stabile;
ora per forse qualcuno potrebbe stabilire che noi non siamo autorizzati a concedere la
homologia a nome suo. Perci meglio concordare a fondo (diomologesasthai) pi chia-
ramente su questo punto, dato che la diferenza non piccola se la cosa sta cos o di-
versamente (169 d 10 - e 5).2 Questa osservazione batte su un punto decisivo per il pro-
blema della confutazione di un libro: poich lautore assente, la questione cruciale
data dalla homologia fornita dallinterlocutore a chi interroga e dalla congruit di questa
con il libro dellautore assente. Si tratta cio di valutare se questa homologia corrispon-
da a quella che darebbe Protagora se fosse presente. Non un caso allora che a questo
punto linterlocutore sia Teodoro, lamico di Protagora, e non pi Teeteto, e che So-
crate dia una precisa indicazione metodica sulla via da percorrere: Assumiamo dunque
la homologia non mediante altri, ma a partire (ek) dal logos di Protagora hos di brachyta-
ton (169 e 7 - 170 a 1). La discussione deve dunque assumere come punto di partenza il
logos protagoreo e di qui trarre le conseguenze che esso comporta; non escludei che lek
riguardi appunto una procedura di inferenza di conseguenze dalle premesse consegna-
te nello scritto di Protagora. Che il riferimento riguardi tale scritto mi pare conferma-
to dal fatto che subito dopo presentata una brevissima rifomulazione dellhm: il di
brachytaton la frase inziale dello scritto in cui Protagora enuncia la sua tesi. Su questa
base da assumere la homologia, ossia desumendo ci che compatibile (non contrasta)
con essa. Ma in questa riformulazione si viene ad aprire uno spazio anche per la doxa,
oltre che alla percezione. Essa infatti cos presentata: egli dice pou (in qualche luogo
o modo) che ci che pare (dokoun) a ciascuno questo anche per colui cui pare (170 a
3-4), dove dokein introduce appunto anche lambito della doxa in generale, non della so-
la percezione. Teodoro conferma che si tratta di unasserzione di Protagora stesso (170
a 5). Ma aprendo la strada alla doxa, si viene a ricuperare anche la distinzione fra vero e
1 In 168 d 3 Protagora descritto come aposemnymon lhm, espressione che richiama il tacere assai semnos del
libro scritto nel Fedro. Si noti come ripetutamente Protagora iteri la citazione dellincipit del suo libro nel Teeteto.
2 Secondo Cornford 1935, p. 77, con questa afermazione Platone intendeva forse evitare limputazione di
attribuire a Protagora unasserzione che non compariva in questa stessa forma nei suoi scritti.
118 giuseppe cambiano
falso (170 b 8-9). Ci che Socrate obietta che dallhm deducibile la conseguenza ne-
cessaria (cfr. ananke in 170 d 1) che impossibile opinare il falso e che nessuno pi sa-
piente di un altro, ma ci contrasta con quanto anche secondo Teodoro nessuno del-
la cerchia (amphi) di Protagora consentirebbe a sostenere, ossia che nessuno ritenga che
altri siano ignoranti e opinino il falso (170 c 5 - d 2). Che il tentativo di confutazione ope-
rato da Socrate poggi su proposizioni deducibili necessariamente dal logos di Protagora
confermato da Teodoro (170 e 6). Si apre allora lalternativa o che Protagora stesso e
tutti gli altri non credevano nellhm, e allora la verit scritta da Protagora non tale per
nessuno; oppure che Protagora ci crede, ma i molti no e allora ben pi numerosi sono
coloro cui tale verit non dokei e quindi tanto pi essa verit che non (170 e 7 - 171 a
3). Anche in questo caso Teodoro ammette la necessit di tale conclusione (171 a 4-5).
Lintera argomentazione costruita su una serie di conseguenze deducibili dalla pre-
messa dellhm contenuta nello scritto di Protagora e concesse da Teodoro, difensore di
Protagora pi autorizzato di Teeteto: ad esempio Protagora deve ammettere che, se
vero lhm, allora vera anche lopinione di quanti hanno unopinione contraria e quin-
di ritengono che Protagora sia nel falso (171 a 6-9) e che gli altri non concedono di esse-
re nel falso, mentre egli ammette che anche questa loro doxa sia vera a partire dalle co-
se che scrisse (ex on gegraphen) (171 b 4-7). La conclusione fnale che, poich
contestata da tutti, lAletheia di Protagora non sar vera per nessuno, n per qualcun al-
tro n per lui stesso (171 c 5-7).
Non qui il luogo per discutere se la confutazione socratica sia riuscita; quel che im-
porta che essa presentata come il risultato di una serie di inferenze dallhm, conte-
nuto nello scritto protagoreo, le quali sarebbero in contraddizione con lhm stesso.
Di fatto nella confutazione condotta da Socrate, per quanto ne sappiamo, non sem-
brano mai addotte citazioni esplicite di altre parti dello scritto di Protagora per mostrare
che esse contraddicono lhm, n si fa riferimento al contesto e alla situazione pragma-
tica in cui Protagora formulava la sua tesi e ai destinatari ai quali si rivolgeva. Nel mo-
do in cui Platone la rappresentava, la confutazione di uno scritto sembrava invece pog-
giare la sua validit sulla correttezza di inferenze tratte a partire dallo scritto stesso,
anche se non contenute in esso (per lo meno, non sappiamo se fossero contenute o se,
per lo meno, esso contenesse elementi mediante i quali costruire tali inferenze). Il Tee-
teto mostrerebbe quindi come sia possibile confutare uno scritto, in assenza di un auto-
re e ponendo domande a un interlocutore non ostile verso lautore dello scritto: non
un caso che questa sezione del dialogo abbia appunto per interlocutore Teodoro, hetai-
ros di Protagora, e non il giovane Teeteto, e che non a caso Teodoro si mostri lieto di
svolgere la funzione di rispondente dopo che si conclusa la confutazione dellhm (183
c 5-7). Rimane per sempre la cautela di base. Infatti Tedororo rileva immediatamente
che essi stanno incalzando troppo il suo hetairos e anche Socrate riconosce che oscu-
ro se essi non stiano andando al di l di ci che corretto ed anzi verosimile (eiks) che
Protagora, essendo pi vecchio, sia pi sapiente di Socrate e Teodoro stesso. Ancora
una volta Socrate ipotizza che se Protagora fosse presente, ossia emergesse improvvi-
samente sino al collo, su molte cose confuterebbe me che chiacchiero, com verosi-
mile, e te che fai ammissioni e poi, immergendosi, se ne andrebbe via (171 c 9 - d 3).1
1 Su questa rappresentazione cfr. in particolare Ford 1994, che tra laltro suggerisce che il cenno allemergere
della sola testa possa rinviare allincipit di uno scritto, dov gi contenuto il nucleo essenziale (kephalaion) delle
tesi di esso.
dal fedro al teeteto di platone 119
Socrate per riconosce che si tratta di una situazione impossibile e che occorre invece
contare sulle proprie forze ed esprimere sempre la propria opinione (171 d 3-5). Di fatto
lHM, che la base su cui si regge (histatai) il logos di Protagora, anche nella riformula-
zione che ne stata tracciata (hypegrapsamen)1 venendo in aiuto a Protagora, ossia nel-
lapologia, viene a trovarsi in contraddizione con lammissione che uno sia pi sapien-
te di un altro ossia che esistano competenti nei vari campi (171 d 9 - e 8). Linterrogazione
di Protagora non si conclude a questo punto, ma viene ancora posta in seguito, quan-
do a proposito dellutilit delle leggi che riguardano anche il futuro, si pone a Prota-
gora o a chiunque altro dica le sue stesse cose, la domanda se lhm, che comporta lesi-
stenza in ciascun soggetto percipiente del kriterion della verit delle sue percezioni di
qualit sensibili, comporti lesistenza di questo stesso criterio anche per la verit delle
sue credenze riguardanti il futuro (178 b 2 c 2), dove evidente che in questi casi, per
esempio in casi di malattie, il criterio sar presente nellesperto, ossia nel medico. Ma
ancora una volta il riconoscimento che uno pi sapiente di un altro e che misura ap-
punto chi pi sapiente e non uno privo di scienza come Socrate, mentre il discorso
prounciato da Socrate in difesa di Protagora e in vece sua costringeva anche Socrate a
essere misura, lo volesse o no, presentato come unammissione che Protagora deve
necessariamente concedere (179 a 10 - b 5). E Teodoro a questo punto esprime lopinio-
ne (dokei) che soprattutto in questo modo il logos di Protagora sia stato preso in trap-
pola, cio sia stato confutato (179 b 6-9). Seguendo linvito formulato nella conclusione
dellapologia di Protagora, il dialogo continua poi con la confutazione del mu, che, co-
me si sa, porta alla conseguenza dellimpossibilit di parlare di qualsiasi cosa e infne
passa a confutare la tesi di Teeteto (cp) introducendo la discussione sulle propriet co-
muni (koin). Siamo qui in un orizzonte che si colloca ormai fuori del discorso conte-
nuto nello scritto di Protagora e di una discussione diretta di esso, anche se Socrate ave-
va sottolineato la connessione fra le tre tesi. Resta da osservare che, se anche per il
Teeteto vale la tesi del Fedro sulla fssit e univocit semantica dello scritto, lo scritto di
Protagora rappresentava di fatto, come fssazione su un supporto stabile, una contrad-
dizione pragmatica rispetto a un eraclitismo, che la dottrina segreta tendeva ad attri-
buire anche a Protagora. La confutazione della tesi di Teeteto (cp) doveva dunque pro-
cedere per una strada diversa da quella intrapresa per confutare il libro di Protagora. In
ultima analisi per la confutazione del libro di Protagora, riuscita o no, era esibita in
uno scritto e, quindi di fatto, si presentabva come una confutazione scritta. Essa inclu-
deva come passaggio fondamentale la tecnica di cercare di argomentare in difesa dello
scritto come avrebbe potuto argomentare lautore stesso. Non escluderei che questa
prassi simponesse nelle scuole flosofche, almeno in alcune di esse. Gi Aristotele, nei
Topici (viii 5.159 b 27-35), prevedeva una situazione nella quale ci si esercita a difendere
una tesi flosofca, senza necessariamente condividere la tesi in questione, ma per cer-
care di argomentare come avrebbe argomentato il sostenitore della tesi stessa. Da allo-
ra questa diventata una procedura standard nella discussione, confutazione e difesa di
scritti flosofci.2
Scuola Normale Superiore Pisa
1 Sul signifcato di hypographomen, come sketched as in dictation from Protagoras cfr. Ford 1994, p. 212 n, 23.
2 Sono assai grato a Geofrey Lloyd e Jaap Mansfeld per le loro preziose osservazioni, che mi hanno salvato da
alcuni errori e mi hanno consentito di precisare pi punti. Naturalmente la responsabilit rimane in ogni caso mia.
120 giuseppe cambiano
Riferimenti bibliografici
Barnes J., The Presocratic Philosophers, London-New York, Routledge & Kegan, 1979.
Benson H. G., The Dissolution of the Problem of the Elenchus, osap, 13, 1995, pp. 45-112.
Blondell R., The Play of Character in Platos Dialogues, Cambridge, Cambridge University Press,
BrandwoodL., The Chronology of Platos Dialogues, Cambridge, Cambridge University Press, 1990.
Brandwood L., Stylometry and Chronology, in R. Kraut, ed., The Cambridge Companion to Plato,
Cambridge, Cambridge University Press, 1992, pp. 90-120.
Burnyeat M. L., The Theaetetus of Plato, Indianapolis-Cambridge, Hackett, 1990.
Cerri G., Platone sociologo della comunicazione, Milano, Il Saggiatore, 1991.
Chappell T., Reading Platos Theaetetus, Sankt Augustin, Academia Verlag, 2004.
Cornford F. M., Platos Theory of Knowledge, London, Routledge & Kegan, 1935.
Denniston J. D., The Greek Particles, Oxford, Clarendon, 1934.
De Vries G. J., A Commentary on the Phaedrus of Plato, Amsterdam, Hakkert, 1969.
Dorandi T., Nellofcina dei classici. Come lavoravano gli autori antichi, Roma, Carocci, 2007.
Ferrari G. A. F., Listening to the Cicadas. A Study of Platos Phaedrus, Cambridge, Cambridge Uni-
versity Press, 1987.
Fine G., Plato on Knowledge and Forms. Selected Essays, Oxford, Clarendon, 2003.
Ford A., Protagoras Head: Interpreting Philosophic Fragments in Theaetetus, American Journal of
Philology, 115, 1994, pp. 199-218.
Friedlaender P., Platone, trad. it., Milano, Bompiani, 2004.
Giuliano F. M., Esegesi letteraria in Platone: la discussione sul carme simonideo nel Protagora, Studi
classici e orientali, 41, 1991, pp. 105-190 = Studi di letteratura greca, Pisa, Giardini editori e
stampatori in Pisa, 2004, pp. 1-86.
Gulley N., Platos Theory of Knowledge, London, Methuen, 1962.
Guthrie W. K. C., A History of Greek Philosophy, vol. iii, Cambridge, Cambridge University Press,
Hardy J., Platons Theorie des Wissens im Theaitet, Gttingen, Vandenhoeck & Ruprecht, 2001.
Heitsch E., Ein Buchtitel des Protagoras, Hermes 97, 1969, pp. 292-296 = C. J. Classen, ed.,
Sophistik, Darmstadt, Wissenschaftliche Buchgesellschaft, 1976, pp. 298-305.
Heitsch E., Ueberlegungen Platons im Theaetet, Wiesbaden, Steiner, 1988.
Kahn Ch., The Platonic Chronology, in J. Annas, C. Rowe, edd., New Perspectives on Plato, Modern
and Ancient, Washington, Center for Hellenic Studies, 2002, pp. 93-127.
Kuehn W., La fn du Phdre de Platon. Critique de la rhtorique et de lcriture, Firenze, Olschki, 2000.
Lazzarini M. Le formule delle dediche votive nella Grecia arcaica, Memorie dellAccademia Nazio-
nale dei Lincei, 19, 1976.
Mansfeld J., Protagoras on Epistemological Obstacles and Persons, in G. B. Kerferd, ed., The Sophists
and their Legacy, Wiesbaden, Steiner, 1981, pp. 38-53.
Mansfeld J., Aristotle, Plato, and the Preplatonic Doxography and Chronography, in G. Cambiano,
ed., Storiografa e dossografa nella flosofa antica, Torino, Tirrenia Stampatori, 1986, pp. 1-59 =
Studies in the Historiography of Greek Philosophy, Assen-Maastricht, Van Gorcum, 1990.
Mansfeld J., Prolegomena. Questions to be settled before the study of an Author or a Text, Leiden, Brill,
McCabe M. M., Plato and His Predecessors. The Dramatisation of Reason, Cambridge, up, 2000.
McDowell J., Platos Theaetetus, Oxford, Clarendon, 1973.
Mi-Kyoung Lee, The Secret Doctrine: Platos Defence of Protagoras in Theaetetus, osap, 19, 2000,
pp. 47-86.
Narcy M., Intr. a Platon, Thtte, Paris, Garnier-Flammarion, 1994.
Robinson R., Platos Earlier Dialectic, Oxford, Clarendon, 19532.
dal fedro al teeteto di platone 121
Sedley D., The Midwife of Platonism. Text and Subtext in Platos Theaetetus, Oxford, Clarendon,
Szlezak T. A., Platon und die Schriftlichkeit der Philosophie, Berlin, De Gruyter, 1985.
Untersteiner M., I sofsti, Milano, Bruno Mondadori, 19962.
Velardi R., Scrittura e tradizione dei dialoghi di Platone, in G. Casertano, ed., La struttura del dialogo
platonico, Napoli, Lofredo, 2000, pp. 108-139.
Vlastos G., The Socratic Elenchus, osap 1, 1983, pp. 27-58 = Socratic Studies, Cambridge, Cam-
bridge University Press, 1994, pp. 1-37.
Vlastos G., Socrates. Ironist and Moral Philosopher, Cambridge, Cambridge University Press, 1991.
122 giuseppe cambiano
Walter Cavini
d ,
Arist. Int. 6, 17a33-341
a contraddizione ha un ruolo rilevante nella teoria aristotelica dellargomentazio-
ne sia dialettica sia dimostrativa, cio scientifca. Per Aristotele la forma privilegia-
ta di argomentazione dialettica la confutazione ();2 e la confutazione aristo-
telica il sillogismo della contraddizione (o della contraddittoria), cio la deduzione
formalmente valida (sillogismo) di una conclusione che contraddice la tesi iniziale del-
linterlocutore.3 Ma se la contraddizione gioca un ruolo rilevante nellargomentazione
dialettica come conclusione dell, gioca un ruolo altrettanto rilevante nella scel-
ta delle premesse sia dialettiche sia dimostrative. Aristotele defnisce infatti la premes-
sa () dialettica o dimostativa come una delle due parti della contraddizione
( e )4 e distingue la premessa dialettica da quella dimostrati-
va in base al criterio di assunzione della premessa: il dialettico assume indiferente-
mente () una qualsiasi delle due parti della contraddizione, mentre chi dimostra
* Si ritenuto opportuno pubblicare qui solo le prime tre parti dellarticolo ( 1-3). La quarta e ultima parte
( 4) uscir nel secondo numero di Antiqvorvm Philosophia. Dedico questo lavoro alla memoria di Vincenza
Celluprica e Mario Mignucci.
1 Per i nomi degli autori greci e i titoli delle loro opere seguo le abbreviazioni del lsj da integrarsi con quelle
di G. W. H. Lampe, A Patristic Greek Lexicon, per gli autori cristiani e con quelle del siglario di K. Ziegler per i Moralia
di Plutarco. Per i nomi degli autori latini e i titoli delle loro opere seguo le abbreviazioni dellOxford Latin Dictio-
nary a cura di P. G. W. Glare, da integrarsi, quando occorra, con quelle del Thesaurus Linguae Latinae.
2 Cfr. Arist. se 3, 165b12-16: c d -
. b e , d d d e d
e e ; questo elenco che, a mio avviso, si applica per Aristotele al-
largomentazione dialettica in generale e non solo a quella forma degenere di dialettica che largomentazione
sofstica chiaramente una climax discendente: b a ,
b , , b [] b e
e (b18-22).
3 Cfr. Arist. se 2, 165b3-4: d [scl. ] \ d . Nel-
l aristotelico la tesi iniziale dellinterlocutore non fa parte delle premesse della confutazione. L
aristotelico una forma di confutazione diretta: cfr. Gobbo 1997, pp. 351-356, che corregge Cavini 1993a, pp.
4 Arist. APo. i 2, 72a8-9: leggo invece di , che la lezione dei codici, accogliendo
lemendazione proposta da Giorgio Colli (1955, pp. 893-896) e da Jonathan Barnes (1975, p. 102, e 1993, p. 98) (pace
Mignucci 1975, pp. 32-33). Barnes (1993, p. 98) sospetta anche che la successiva defnizione di (72a11-12)
sia una glossa aggiunta dopo la corruzione di in . La difcolt era gi stata avvertita da
Colli (1955, pp. 895-896), che, invece di leggere b , accoglie la varian-
te del codice d al posto della lezione degli altri codici. Segue Barnes 1993 (ma senza men-
zionare Colli 1955) anche Lemaire 2005, pp. 166-170.
ne assume determinatamente () solo una parte perch vera.1 La con-
traddizione infne gioca un ruolo essenziale in quella forma di argomentazione indi-
retta che la dimostrazione per assurdo: se da una premessa deriva necessariamente
una conclusione manifestamente falsa, allora la premessa falsa e, per il principio del
terzo escluso e il Modus Tollendo Ponens (mtp),2 la sua contraddittoria vera.
Aristotele non stato solo il primo per noi a introdurre il termine tecnico contrad-
dizione () nel lessico flosofco greco, e quindi nel nostro lessico flosofco, ma
della contraddizione ha saputo anche declinare lucidamente le propriet logiche essen-
ziali. Questo saggio ha per tema appunto la teoria aristotelica della contraddizione e in
particolare quelle che per Aristotele sono le sue propriet logiche essenziali, i principi
aristotelici della contraddizione.
Il saggio si divide in quattro parti. Aristotele defnisce la contraddizione come una
coppia di enunciati dichiarativi opposti, cio lafermazione e la sua negazione logica:
la prima parte del saggio ( 1) dedicata dunque a quello che lingrediente fonda-
mentale della contraddizione aristotelica, cio lenunciato dichiarativo o (al-
tro termine tecnico di conio aristotelico), e alla sua propriet logica fondamentale, quel-
la di essere sempre o vero o falso, che Jan ukasiewicz (1921) ha chiamato Principio di
Bivalenza (pb).
Nella seconda parte ( 2) esamino invece le due defnizioni esplicite che Aristotele d
di contraddizione o e in particolare la propriet logica (l) che per Ari-
stotele distingue lopposizione contraddittoria dagli altri modi dellopposizione, per
esempio dallopposizione contraria, cio il principio per cui, di una coppia di contrad-
dittori, necessariamente uno vero e laltro falso, principio che C. W. A. Whitaker
(1996) ha chiamato Regola delle Coppie Contraddittorie (rcc).
Tale regola che presuppone il principio di bivalenza, ma non va confusa con esso,
come spesso accaduto e accade un corollario, cio una conseguenza immediata,
della congiunzione di due principi aristotelici della contraddizione, (a) quello per cui i
contraddittori non possono essere veri insieme (cio allo stesso tempo, ) e (b) quel-
lo per cui i contraddittori non possono essere falsi insieme. Nella terza parte del saggio
( 3) esamino tali principi semantici (in cui cio ricorrono i predicati semantici vero e
falso), stabilendo la loro equivalenza, per Aristotele, rispettivamente con quelli che
chiamiamo Principio di (Non) Contraddizione (pnc) e Principio del Terzo Escluso
(pte): per il principio di non contraddizione i contraddittori non possono essere veri in-
sieme, per quello del terzo escluso non possono essere falsi insieme; dalla loro con-
giunzione deriva immediatamente la regola delle coppie contraddittorie, ovviamente
assumendo per i contraddittori la validit del principio di bivalenza, in quanto la con-
traddizione aristotelica vale solo per lafermazione e la negazione, cio per enunciati
dichiarativi veri o falsi. I commentatori greci di Aristotele hanno chiamato tali principi
semantici, sia collettivamente (la loro congiunzione), sia distributivamente (ciascuno di
1 Arist. APo. i 2, 72a9-11 (cfr. anche APr. i 1, 24a22-25): questo testo avr, come vedremo alla fne del saggio (
4), unimportanza decisiva per comprendere la natura del principio aristotelico di non contraddizione come prin-
cipio di tutti i principi dimostrativi e credenza ultima da cui dipende ogni dimostrazione e quindi, per Aristote-
le, la nostra conoscenza scientifca del mondo.
2 Cfr. Arist. APo. i 11, 77a22-23: e \ j e : il princi-
pio per cui ogni cosa < necessario> afermarla o negarla ( j ) appunto, come vedremo,
una delle due formulazioni aristoteliche di quello che chiamiamo principio del terzo escluso, logicamente equi-
valente a uno dei due principi aristotelici della contraddizione, quello per cui i contraddittori non possono essere
falsi insieme o allo stesso tempo. Per il mtp cfr. 4.
124 walter cavini
essi), assioma della contraddizione ( ). Nella terza parte del
saggio mostro anche come da tale formula greca e dalluso che ne fanno i commenta-
tori greci di Aristotele derivi quella latina di principium contradictionis, che alle origini
della logica moderna ( Jungius 1638) designa un solo principio di cui il principio di non
contraddizione la versione negativa e quello del terzo escluso la versione afermativa.
Sar solo a partire da Christian Wolf (1729) e Alexander Baumgarten (1739), con lin-
troduzione di un nome ad hoc per il principio del terzo escluso (principium exclusi tertii
seu medii), che la formula principium contradictionis designer soltanto il principio di non
contraddizione. Di qui il nostro principio di contraddizione e insieme, come vedre-
mo, curiosamente, la sua variante sinonimica principio di non contraddizione (con o
senza trattino).
Infne, nella quarta e ultima parte del saggio ( 4), prendo in esame due propriet che
Aristotele attribuisce specifcamente al principio di non contraddizione in quanto prin-
cipio dimostrativo e che ne fanno per natura il principio anche di tutti gli altri assiomi
o principi dimostrativi, cio il suo essere, da un lato, c (Metaph. 3,
1005b14) e, dallaltro, (Metaph. 3, 1005b33). Per c suggeri-
sco uninterpretazione per cui un mutuato verosimilmente dal vi
libro della Repubblica di Platone (510b6-7, 511b5) non designa la natura non-ipotetica
o incondizionata del principio aristotelico, ma rinvia implicitamente allc
di Platone ed esplicitamente alla distinzione aristotelica fra e
propria degli Analitici Secondi (I 2, 72a14-24). Per , la credenza ul-
tima che sta alla base di ogni dimostrazione, e quindi della nostra conoscenza scienti-
fca del mondo, suggerisco invece un rinvio implicito alla propriet logica della con-
traddizione corrispondente al principio aristotelico di non contraddizione, cio alla
negazione della co-verit dei contraddittori.
Unultima considerazione pi generale prima di iniziare: per lesperto di logica anti-
ca e moderna i principi aristotelici della contraddizione risulteranno senzaltro ovvi e
scontati, ma questo non dovrebbe in alcun modo infciare limportanza del risultato rag-
giunto da Aristotele. Prima di Aristotele, per quanto antichi e moderni abbiano ricer-
cato precursori dei principi aristotelici di non contraddizione e del terzo escluso, non
esisteva in realt nessuna teoria al riguardo che connettesse tali principi alle propriet
logiche fondamentali della contraddizione. Questo senzaltro merito di Aristotele. E
dopo Aristotele, se guardiamo alle molte traduzioni, commenti e interpretazioni di te-
sti aristotelici e non aristotelici (come, per esempio, il De Fato ciceroniano), in cui spes-
so anche oggi i principi di bivalenza, di non contraddizione e del terzo escluso sono si-
stematicamente confusi e fraintesi, allora sembra legittimo pensare che forse non sia
del tutto inutile continuare a occuparsi di queste importanti banalit.
1. \A
Fra i tanti meriti di Jan ukasiewicz come interprete della logica antica, vi senzaltro
quello di aver sottolineato limportanza fondamentale, per la logica antica e moderna,
del principio che ogni enunciato dichiarativo vero o falso, cio del principio che a par-
tire da ukasiewicz chiamiamo Principio di Bivalenza. Scrive ukasiewicz al riguardo:
All systems of logic known so far, both Aristotelian logic and Stoic logic, both traditional formal
logic and modern symbolic logic, have been based on the principle that every proposition is
either true or false. That principle, which has served so far as the foundation of all logic, will be
sui principi aristotelici della contraddizione 125
called the principle of bivalence, and the logic which assumes that there are two and only two
logical values will be called two-valued.1
Secondo ukasiewicz,2 Aristotele enuncia il principio di bivalenza in un passo del De
T1 e b , \ > e j
ma non ogni <enunciato> dichiarativo, se non quello cui appartiene lessere vero o falso,3
Arist. Int. 4, 17a2-3
cio considera T1 equivalente a:
(i) Ogni enunciato dichiarativo un enunciato vero o falso,
secondo quella che linterpretazione prevalente del passo.4 Ma contro tale interpreta-
zione si possono sollevare almeno due obiezioni: (a) che il passo in esame di per s
compatibile con uninterpretazione pi debole, cio con la conversa di (i):
(ii) Ogni enunciato vero o falso un enunciato dichiarativo,
dal momento che T1 dice solo che dichiarativo lenunciato cui appartiene lessere ve-
ro o falso; e (b) che Aristotele sembra ammettere almeno due casi di enunciati dichia-
rativi n veri n falsi, cio il Mentitore e gli enunciati singolari contingenti al futuro; di
conseguenza, si dovrebbe preferire linterpretazione pi debole, che insieme compa-
tibile col testo e con lammissione da parte di Aristotele di enunciati dichiarativi n ve-
ri n falsi.5
Queste due obiezioni sono, a mio avviso, infondate e quindi la conseguenza che ne de-
riva, cio linterpretazione debole di T1, senzaltro da rifutare in favore dellinterpre-
tazione forte della tradizione. Secondo la prima obiezione, T1 sarebbe compatibile con
linterpretazione secondo cui essere (un enunciato) vero o falso condizione sufcien-
te ma non necessaria per essere (un enunciato) dichiarativo. Ma ci in evidente con-
trasto con quello che sembra essere il signifcato pi plausibile da attribuire a dichiara-
tivo () nel passo in esame. Il termine non attestato prima di Aristotele e
anche in Aristotele compare solo nel De Interpretatione in sei occorrenze concentrate fra
Int. 4, 17a2, e Int. 5, 17a15. Quella del nostro passo dunque la sua prima occorrenza nel
corpus aristotelicum, e dal momento che Aristotele non ritiene necessario spiegarne il si-
gnifcato, questo doveva risultare senzaltro evidente. Ora laggettivo e il
sostantivo corrispondente , anchesso attestato solo a partire da Aristotele, so-
no entrambi deverbativi di , verbo di cui largamente in uso nel greco comu-
ne come in quello di Aristotele6 la locuzione c seu
, manifestare o dichiarare la propria opinione su qualcosa. Lenunciato di-
chiarativo pertanto quello che manifesta o dichiara lopinione del parlante, ci che il
parlante crede, cio ritiene vero, a proposito di qualcosa.7 Ovviamente un enunciato di-
1 ukasiewicz 1921/1970, p. 91.
2 ukasiewicz 1930/1970, p. 176: The law of bivalence, i.e. the law according to which every proposition is
either true or false, was familiar to Aristotle, who explicitly characterized a proposition, , as discourse
which is either true or false. Per le diferenze fra il moderno principio di bivalenza e la tesi crisippea che ogni
vero o falso (o disgiunzione esclusiva), cfr. Barnes 2007, pp. 1-6.
3 Le traduzioni dal greco, salvo indicazione contraria, sono mie. 4 Cfr. Crivelli 2004a, p. 87 n. 31.
5 Cfr. Crivelli 2004a, pp. 86-87; vedi anche Gourinat 2006, p. 51, e infra, p. 149 n. 3.
6 Cfr. per es. Po. 6, 1450a7, b12.
7 Cfr. Nuchelmans 1973, pp. 26-27, 107, 120, 131, 133 (che traduce con statement e con
statement-making); Cavini 1991, p. 37 n. 7; Crivelli 2004a, p. 87 (che tuttavia traduce con assertion
126 walter cavini
chiarativo pu essere falso se il parlante ha una credenza falsa, cio in errore, o insin-
cero se il parlante dichiara come vero ci che ritiene falso; ma in ogni caso un enuncia-
to dichiarativo quello che manifesta una credenza, ci che il parlante ritiene o fnge di
ritenere vero, e una credenza, ci che uno ritiene vero, appunto vera o falsa; dunque
lenunciato dichiarativo corrispondente sar anchesso vero o falso.
Quanto ai due casi di enunciati dichiarativi n veri n falsi, cio il Mentitore e gli
enunciati singolari contingenti al futuro, che Aristotele ammetterebbe come eccezioni
al pb, si tratta, com noto, di una materia assai controversa, che non possibile
afrontare qui nei dettagli. Tuttavia non afatto certo che Aristotele ammettesse tali
eccezioni e, alla luce del signifcato aristotelico di enunciato dichiarativo, senzaltro
improbabile: un enunciato dichiarativo n vero n falso equivarrebbe alla manifesta-
zione di una credenza n vera n falsa, cio a ritenere vero qualcosa che non n vero
n falso; ma dubbio che in questo caso si possa parlare ancora di credenza: sarebbe
come ritenere vero un comando o una domanda.
Delle due possibili eccezioni, quella del Mentitore aristotelico (se 25, 180b2-7) mi sem-
bra senzaltro la pi dubbia. Si tratti o meno del celebre paradosso, il sofsma come lo
presenta Aristotele quello dello stesso uomo che insieme dice il falso e dice il vero, e
ci che rende intrattabile (, b5) questo sofsma che non facile stabilire qua-
le dei due contrari (dire il vero o dire il falso) sia da prendere in senso assoluto ()
e quale in senso relativo, cos da risolvere il sofsma e ristabilire il principio violato di
non contrariet (due predicati contrari non possono applicarsi simultaneamente allo
stesso soggetto). Secondo linterpretazione tradizionale, per quanto sia difcile stabili-
re quale dei due contrari sia da prendere in senso assoluto e quale in senso relativo, que-
sta comunque la soluzione che ha in mente Aristotele. Ma linterpretazione tradizio-
nale notoriamente non in grado di dire quale dei due contrari sia da prendere in senso
assoluto e quale in senso relativo. Cos rimane aperta la possibilit che vadano presi o
entrambi in senso assoluto, violando in tal modo il principio di non contrariet (e quin-
di anche il principio di non contraddizione, da cui quello discende secondo Aristotele),
o entrambi in senso relativo, cio nessuno dei due in senso assoluto; ma in questo caso,
come scrive Paolo Crivelli,1 failing to hold without qualifcation is failing to hold in the
proper sense: di conseguenza quello del Mentitore risulterebbe essere un enunciato di-
chiarativo n vero n falso, e quindi violerebbe il pb.
Questa interpretazione senzaltro plausibile se i predicati dire il vero e dire il fal-
so sono presi in senso logico, cio come sinonimi rispettivamente di essere nel vero e
di essere in errore. In tal caso non si dovrebbe parlare propriamente di Mentitore, bens
di Falsidico. Ma i verbi e signifcano anche dire il vero e dire il falso
in senso morale, cio essere veritiero (o sincero) e mentire;2 e chiaramente si pu esse-
re sinceri ma dire il falso, cio essere in errore, e mentire ma dire il vero. Se il Mentito-
re aristotelico preso in senso morale e non in senso logico, allora il fatto che lo stesso
e con assertoric, termini che non hanno nulla a che vedere col senso dichiarativo della parola gre-
ca; per asserzione come traduzione di , cfr. infra, p. 142). Per la corrispondenza fra e cfr. in
particolare Int. 14, 24b1-3: d , d b d
, e ukasiewicz 1910b, p. 18/2000, p. 392. Heidegger, invece, com noto, ha interpretato
l del discorso dichiarativo non come un manifestare la propria opinione su qualcosa, ma come un
mostrare la cosa stessa su cui verte il discorso, far vedere un ente, : a partire dallente stesso (Heidegger
1925/26, p. 133; cfr. Cavini 1991, p. 37).
1 Crivelli 2004a, p. 33. 2 Cfr. Crivelli 2004a, p. 31 n. 11.
sui principi aristotelici della contraddizione 127
uomo insieme menta e sia veritiero non in senso assoluto ma in senso relativo non vio-
la il pb pi di quanto non lo violi il fatto che una sfera per met bianca e per met nera
sia insieme in parte bianca e in parte nera. E un indizio che il Mentitore aristotelico va-
da preso in senso morale e non in senso logico dato dalla seconda parte del passo di
se 25 relativo al sofsma, cio 180b5-7, dove, come nota acutamente Crivelli,1 Aristotele
introduce al posto dei verbi e gli aggettivi corrispondenti e
, che in quanto riferiti a persona non possono che avere il signifcato morale di
veritiero e mentitore. Crivelli ritiene che in questo modo Aristotele voglia distinguere
the real Liar, quello della prima parte del passo formulato coi verbi e
presi in senso logico, dal pi innocuo paradosso formulato con gli aggettivi
e presi in senso morale. Ma questo sembra improbabile per almeno tre
ragioni: (a) i verbi e hanno comunemente il signifcato di dire il ve-
ro e dire il falso in senso morale: vi sarebbe quindi continuit fra la prima e la seconda
parte del passo; (b) sia coi verbi sia con gli aggettivi la soluzione prospettata quella di
un senso assoluto (falso o mentitore in senso assoluto) e di un senso relativo (veritiero
o sincero in senso relativo); infne (c) non si vede perch, come sostiene senzaltro Cri-
velli, la particella di 180b5, che introduce la seconda parte del passo, debba avere va-
lore avversativo: non correlata a un antecedente e quindi pu avere semplice-
mente valore continuativo. In conclusione, non ci sono ragioni forti per considerare il
Mentitore aristotelico come un sofsma riguardante dire il vero e dire il falso in senso
logico, e quindi per considerare lenunciato corrispondente (qualunque esso sia) come
uneccezione al pb.2
Quanto al caso degli enunciati singolari contingenti al futuro di De Interpretatione 9,
che tali enunciati rappresentino per Aristotele una eccezione al pb tesi comune alle
varie versioni della cosiddetta interpretazione antirealista dellargomento aristoteli-
co, per la quale cio il futuro secondo Aristotele non ancora reale e quindi gli enun-
ciati contingenti al futuro non sono ancora n veri n falsi. Questa tesi ha un pedigree
storico lunghissimo che risale almeno a Epicuro, come testimonia Cicerone nel De Fa-
to (21 e 37), e pi in generale al dibattito delle scuole ellenistiche sul problema del de-
terminismo logico e causale. Ma che sia anche una tesi di Aristotele un punto sen-
zaltro controverso. Come ha dimostrato, a mio avviso, inequivocabilmente Whitaker
nel suo fondamentale commento al De Interpretatione, gli enunciati singolari contingenti
al futuro non sono una eccezione al pb, che riguarda l, cio lenunciato di-
chiarativo, ma alla Regola delle Coppie Contraddittorie (rcc), che riguarda non
l ma l, cio lopposizione contraddittoria di afermazione e nega-
zione, di cui l solo una parte. Tale regola stabilisce che di ogni coppia di
enunciati contraddittori un enunciato vero e laltro falso; e largomento aristotelico di
Int. 9 conclude appunto che tale regola non vale per le coppie contraddittorie di enun-
ciati singolari contingenti al futuro, cos come in precedenza non valeva per le coppie
contraddittorie di enunciati non universali su universali, che possono essere entrambi
veri (Int. 7), e per quelle degli enunciati multipli, che possono essere entrambi falsi (Int.
8). La conclusione di Int. 9 assolutamente chiara al riguardo:
1 Cfr. Crivelli 2004a, p. 150; Crivelli 2004b, pp. 66b-67a.
2 Per ragioni, in parte, diverse anche Paolo Fait (2007, pp. 209-211) nega che il paralogismo aristotelico (che Fait
chiama il Falso) sia una versione del Mentitore e implichi una limitazione del pb.
128 walter cavini
T2 d c b
c b r
Di conseguenza chiaro che non necessariamente di ogni afermazione e negazione opposte
[scl. contraddittorie] luna vera e laltra falsa
Arist. Int. 9, 19a39-b2
e francamente sorprende che gli interpreti vi abbiano visto una eccezione al pb, dal
momento che Aristotele esplicitamente si riferisce non a ogni enunciato afermativo o
negativo ma a ogni coppia contraddittoria di enunciati afermativi e negativi. E anche
quando si interpreta correttamente la conclusione come una eccezione alla rcc, si fa
discendere tale eccezione dalla negazione del pb:
with some contradictory pairs of future-tense assertions, sometimes it is not the case that one
member is true and the other false. This is because with some contradictory pairs of future-tense
assertions, sometimes both members are neither true nor false.1
Ma che la rcc non valga nel caso degli enunciati singolari contingenti al futuro perch
entrambi i membri delle coppie contraddittorie non sono n veri n falsi, chiaramen-
te smentito da Aristotele nel passo che precede la conclusione e da cui la conclusione
immediatamente discende:
T3 [] a b b r j ,
j \ \ , d b c ,2 j
[] in questi casi infatti [scl. riguardo alle cose che non sempre sono o non sempre non sono]
necessariamente una delle due parti della contraddizione vera o falsa, ma non questa o
quella in particolare, bens come capita in un modo o nellaltro, e una delle due s pi vera,
ma non gi vera o falsa.
Arist. Int. 9, 19a36-39
Comunque si interpreti questo passo molto controverso, in esso non si sostiene la tesi
che le parti o i membri di una coppia contraddittoria di enunciati singolari contingenti
al futuro non sono n veri n falsi, si dice anzi che sono necessariamente luno vero e
laltro falso (necessariamente una delle due parti della contraddizione vera o falsa),
e in questo senso rispettano la rcc, ma la violano in quanto non determinato quale
dei due membri vero o falso, dal momento che tali enunciati predittivi riguardano un
futuro contingente. ukasiewicz, citando questo passo come eccezione al pb, riconosce
tuttavia che Aristotele non si esprime chiaramente al riguardo.3 In realt Aristotele in
Int. 9 non menziona mai il pb, che vale per il singolo enunciato dichiarativo, e non fa
derivare il determinismo logico da tale principio, come accadr invece nel dibattito del-
le scuole ellenistiche; e quindi non confuta il determinismo logico limitando lapplica-
zione del principio di bivalenza. Aristotele ha sempre presente l, non
1 Crivelli 2004a, p. 218 (corsivo mio).
2 Qual il referente del termine femminile c (cfr. Gourinat 2001, p. 73)? Hermann Weidemann (2002,
p. 16) interpreta: die eine (der beiden kontradiktorisch entgegengesetzten Aussagen), ma Aussage traduce
, che maschile. Suggerisco invece come referente implicito nel senso, ben attestato in Aristote-
le, di parte della contraddizione o contraddittorio (cfr. per es. Int. 7, 18a11; 8, 18a26-27; Bonitz 1870, 67b13-20) o di
una o laltra parte della contraddizione (cfr. per es. Int. 7, 18a10-11; 9, 18a28-29, 18a34, 18a37-38). Per il comparativo
b vedi Int. 9, 19a19-22.
3 Cfr. ukasiewicz 1930/1970, p. 176: his [scl. di Aristotele] way of putting the matter is not quite clear.
sui principi aristotelici della contraddizione 129
l, e i due principi logici che menziona riguardano appunto lopposizione
contraddittoria: da un lato la rcc, che il tema dei capp. 7-9 del De Interpretatione, dal-
laltro il principio del terzo escluso (pte) (19a27-32).1 Ed appunto dallapplicazione in-
condizionata della rcc agli enunciati singolari contingenti al futuro che Aristotele fa de-
rivare il determinismo logico: se di ogni coppia contraddittoria di enunciati singolari
contingenti al futuro lafermazione o la negazione vera determinatamente, allora il fu-
turo sempre necessario.2 La confutazione del determinismo logico richieder per-
tanto una restrizione della rcc: se ammettiamo che vi sia anche un futuro indetermi-
nato, quello della contingenza e del caso, allora le coppie contraddittorie di enunciati
predittivi che lo ritraggono saranno tali che una parte della contraddizione sar vera e
una parte falsa ma in modo indeterminato, come il futuro che in esse viene predetto.3
In conclusione, non vi motivo di ammettere eccezioni al pb in Aristotele: tale prin-
cipio condizione insieme necessaria e sufciente dell, in quanto appunto
enunciato dichiarativo, tale cio da manifestare o dichiarare una credenza vera o falsa
del parlante.4 Ma tale principio, come vedremo, rimane in Aristotele limitato alla def-
nizione di enunciato dichiarativo e non entrer a far parte di quelli che per Aristotele
sono propriamente assiomi o principi. Questi ultimi riguarderanno non l ma
l, di cui l solo una parte; non lenunciato dichiarativo vero o fal-
so, ma la coppia di enunciati dichiarativi contraddittori. I principi logici aristotelici so-
no principi della contraddizione.5
2. \A
Come il termine e laggettivo corrispondente , cos il termine
e lavverbio corrispondente sono attestati solo a partire da Ari-
stotele e costituiscono una parte integrante del suo lessico flosofco.6
1 Il sogno di Aristotele (Aristotles fantasy, cfr. Quine 1976, p. 19) secondo cui in Int. 9 egli accetterebbe il
pte ma non il pb, cio riterrebbe necessariamente vera la disgiunzione Domani ci sar una battaglia navale o
domani non ci sar una battaglia navale, ma n veri n falsi i disgiunti che la compongono, appunto un sogno,
ma dei suoi interpreti. Cfr. Gourinat 2001, p. 75 n. 18; e per una variante epicurea del sogno di Aristotele vedi
infra, p. 136 (T11).
2 a j c j , d j j c (Int. 9,
3 a (Int. 9, 19a33). Errata la traduzione di Colli (1955, p. 69): i discorsi
sono veri analogamente a come lo sono gli oggetti (corsivo mio) e di Marcello Zanatta (1992, p. 99): i discorsi
sono veri in modo simile a come lo sono i fatti (corsivo mio): i fatti evidentemente non sono veri, ma rendono
veri i discorsi; i fatti sono invece necessari o contingenti, e di conseguenza i discorsi corrispondenti saranno
necessariamente o contingentemente veri.
Per determinatamente come antonimo di come capita in un modo o nellaltro cfr. anche Cat. 10, 12b38-13a3,
4 Cfr. anche Cat. 4, 2a7-8 (ogni afermazione ritenuta essere o vera o falsa); Cat. 5, 4b8-10 ( perch la cosa
o non che anche lenunciato detto essere vero o falso); Int. 1, 16a9-11; 10, 20a34 (la negazione sempre neces-
sariamente vera o falsa); de An. iii 6, 430b4-5 (falso o vero non solo che Cleone bianco, ma anche che era o sar
[!] <bianco>); de An. iii 6, 430b26-27 (ogni afermazione [] e negazione [ Torstrik] vera o falsa).
5 ukasiewicz considera il pb come [l]a legge pi fondamentale della logica (1930/1970, p. 165), tesi che, com
noto, era sostenuta con forza nellantichit in particolare da Crisippo: fundamentum dialecticae est, quidquid
enuntietur (id autem appellant , quod est quasi efatum) aut verum esse aut falsum (Cic. Luc. 95 [svf ii
196 = fds 880 = fr. 183 Dufour]); contendit omnis nervos Chrysippus, ut persuadeat omne aut verum esse
aut falsum (Cic. Fat. 21 [svf ii 952 = fds 884 = fr. 958 Dufour]).
6 Non cos il verbo : Aristotele ne fa un uso flosofcamente signifcativo solo a proposito della tesi
paradossale di Antistene, (non possibile contraddire) (Top. i 11, 104b20-21; Metaph. 29,
1024b32-34), e nel De Caelo (ii 13, 294b9-10) nel senso di obiettare a se stesso (d a e
i y e ).
130 walter cavini
In flosofa oggi per contraddizione si intende sia un enunciato contraddittorio sia
una coppia di enunciati contraddittori. Un enunciato contraddittorio pu essere o sem-
plice, cio una negazione della forma:
(1) S non S,1
o non-semplice, cio una congiunzione della forma:
(2) p
Una coppia di enunciati contraddittori costituita invece da due enunciati di cui luno
la negazione logica dellaltro:
(3) {p, p}.
Il termine contraddizione pertanto ambiguo e il suo signifcato varia, in particolare,
fra una formulazione collettiva (o congiuntiva), la (2), e una formulazione distri -
butiva, la (3). La diferenza fondamentale che nel caso della formulazione collettiva i
due enunciati contraddittori sono asseriti congiuntamente, cio insieme o allo stesso
tempo, mentre nel caso della formulazione distributiva possono essere distribuiti nel
tempo, come per esempio avviene nell aristotelico, nel sillogismo della con-
traddizione (o della contraddittoria), dove si ha una tesi iniziale p, quella del rispon-
dente, e unargomentazione deduttiva valida (un sillogismo), fra le cui premesse non
compare la tesi iniziale del rispondente e la cui conseguenza necessaria appunto la
negazione della tesi iniziale, p.2
Aristotele d due defnizioni esplicite di contraddizione:
T4 d , d []
E sia questo la contraddizione: lafermazione e la negazione opposte []
Arist. Int. 6, 17a33-34
T5 \ b ( ,
w , b ) []
Fra gli opposti della contraddizione non c intermedio (questo infatti la contraddizione:
lopposizione di cui una delle due parti presente in una cosa qualsiasi, senza avere alcun
intermedio) []
Arist. Metaph. I 7, 1057a33-36
Per T4 la contraddizione una coppia di opposti, per T5 una opposizione. La coppia
di opposti di T4 quella dellafermazione e della negazione opposte: lopposto di unaf-
fermazione la sua negazione, lopposto di una negazione lafermazione corrispon-
dente. Afermazione e negazione sono qui enunciati dichiarativi veri o falsi, in partico-
lare enunciati dichiarativi predicativi, afermativo (d a ) e negativo (d e
1 Questa formulazione della contraddizione, che i medievali chiamavano contradictio in terminis, spiega come
mai il principio di non contraddizione corrispondente, S non non S, sia stato nella tradizione, e in particolare
a partire da Kant, associato al principio di identit nella forma S S. Cfr. infra, pp. 163-164. (T43).
2 Cfr. Varzi 2004, pp. 93-94; Berto 2006, pp. 22-23, che oltre alle formulazioni sintattiche, collettiva e distribu-
tiva, della contraddizione elenca anche le due formulazioni semantiche corrispondenti alla formulazione sintatti-
ca collettiva (dove V sta per vero che e F per falso che):
(4) Vp
(5) Vp
sui principi aristotelici della contraddizione 131
), cio enunciati dichiarativi semplici aventi la forma Soggetto (S)-Predicato (P).
Lopposizione di T5 invece fra due predicati contraddittori, ed caratterizzata come
quella che esclude un intermedio fra gli opposti, cio non possibile che entrambi i pre-
dicati contraddittori non appartengano a un soggetto qualsiasi, come accade per altri
tipi di predicati opposti, per esempio i contrari. T5 caratterizza dunque lopposizione
contraddittoria come quella che esclude la co-falsit degli opposti, ovvero, come ve-
dremo, come quella per la quale vale il principio che sar detto del terzo escluso.1
Le due defnizioni aristoteliche di contraddizione convergono nel considerare la con-
traddizione non come un enunciato contraddittorio, semplice o non semplice, ma co-
me una coppia contraddittoria, le cui parti sono o enunciati afermativi e negativi ve-
ri o falsi, o predicati afermativi e negativi veri o falsi di un soggetto. La formulazione
aristotelica della contraddizione non collettiva, ma distributiva.2
Quali sono secondo Aristotele le propriet logiche di una coppia contraddittoria?
Quella stabilita in T5, che esclude la co-falsit degli opposti contraddittori (sia enun-
ciati predicativi veri o falsi, sia predicati veri o falsi di un soggetto: unequivalenza
costante in Aristotele), una condizione necessaria ma non sufciente: lopposizione
contraddittoria non solo esclude la co-falsit degli opposti (a diferenza dellopposi-
zione contraria), ma anche la loro co-verit (come lopposizione contraria), cio di
ogni coppia contraddittoria necessariamente una parte vera e una parte falsa.
Questa propriet dellopposizione contraddittoria chiaramente enunciata da Aristo-
tele nel cap. 10 delle Categorie dedicato ai quattro modi dellopposizione: relativi, con-
trari, privazione e possesso, afermazione e negazione.3 I contrari non si oppongono
come i relativi; privazione e possesso non si oppongono n come i relativi n come i
contrari; infne afermazione e negazione non si oppongono secondo nessuno di
questi modi:
T6 d a d e b b e b r.
Infatti solo nel caso di questi <opposti> sempre necessario che uno di essi sia vero e laltro
Arist. Cat. 10, 13b2-3
Questa propriet appunto il tratto distintivo dellopposizione contraddittoria, il suo
,4 come ribadisce Aristotele alla fne del capitolo con una variante ellittica della for-
mula impiegata in T6:
T7 d i e d b j r,
d .
Di conseguenza solo nel caso di questi <opposti>, quanti cio si oppongono come laferma-
zione e la negazione, varrebbe la propriet di essere sempre luno o laltro vero o falso.
Arist. Cat. 10, 13b33-35
Infatti gli opposti secondo gli altri modi dellopposizione o non sono n veri n falsi,
quando sono detti senza connessione, o potranno essere entrambi falsi, nel caso degli
opposti contrari e privativi, quando sono detti secondo connessione: se il soggetto non
1 Su Metaph. I 7 cfr. ora Chiaradonna 2005. 2 Cfr. Code 1987, pp. 131-132; Wedin 2004, p. 244.
3 Il termine non mai usato nelle Categorie, a diferenza degli altri trattati dellOrganon, dove invece
largamente attestato: lopposizione contraddittoria nelle Categorie sempre indicata come lopposizione di afer-
mazione e negazione, e i contraddittori sono designati con la perifrasi d .
4 Per come tratto distintivo cfr. Metaph. 4, 1008a26-27.
132 walter cavini
esiste saranno entrambi falsi. Gli opposti contraddittori invece sono sempre veri o falsi
e non possono mai essere entrambi falsi. Sono sempre veri o falsi in quanto sono detti
secondo connessione (Cat. 4, 2a4-10), cio non sono termini ma o enunciati (Cat.
10, 12b6-8),1 in particolare, come preciser il De Interpretatione, enunciati apofantici o di-
chiarativi, che manifestano una credenza vera o falsa del parlante. Non possono mai es-
sere entrambi falsi perch anche quando il soggetto non esiste la negazione sar vera e
lafermazione falsa. La conseguenza che ne trae Aristotele nelle Categorie che solo gli
opposti contraddittori godono della propriet di essere sempre uno vero e laltro falso.
E che questo per Aristotele sia l della contraddizione confermato proprio dai ca-
pitoli centrali del De Interpretatione, dove, come ha dimostrato Whitaker, dopo aver de-
fnito la contraddizione si studiano tre possibili eccezioni alla rcc, cio alla propriet lo-
gica della contraddizione stabilita nelle Categorie.
L che distingue lopposizione di afermazione e negazione dagli altri modi del-
lopposizione in Cat. 10 coincide alla lettera con quella che Whitaker ha chiamato Re-
gola delle Coppie Contraddittorie e che ha dimostrato essere al centro dei capitoli 7-9
del De Interpretatione.2 Ma nel De Interpretatione, a diferenza delle Categorie, non si trat-
ta di distinguere lopposizione contraddittoria dagli altri tipi di opposizione, bens di
esaminare lopposizione contraddittoria come tema centrale di una teoria del discorso
apofantico. L delle Categorie tacitamente presupposto e la sua prima occorrenza
nel De Interpretatione (7, 17b26-33) compendia i due modi di formulazione della regola gi
incontrati nelle Categorie:
T8 (a) b s d , c r j
, d d \ , x
(b) \ d c , d b c b a
e d , d
e d []
(a) Quanti dunque dei contraddittori riguardano gli universali in modo universale, di ne-
cessit luno o laltro vero o falso, e <cos anche> quanti riguardano le cose singole, per
esempio Socrate bianco / Socrate non bianco; (b) quanti invece riguardano gli uni-
versali in modo non universale, non sempre uno vero e laltro falso: vero infatti dire in-
sieme che luomo bianco e che luomo non bianco, e che luomo bello e che luomo
non bello []
Arist. Int. 7, 17b26-33
La clausola (a) introduce la regola nella sua formulazione ellittica (dei contraddittori
[] di necessit luno o laltro vero o falso) come in T7, con la sola diferenza che qui
compare il nuovo termine tecnico assente nelle Categorie; la clausola (b) for-
mula leccezione alla regola nella sua variante estesa (non sempre uno vero e laltro
falso) come in T6: lunica diferenza, a parte loccorrenza o meno del termine tecnico
1 Richard Bods (2002, p. 145) , a mio avviso, in errore quando aferma, a proposito di Int. 4, 16b26, che
Aristotele chiama anche lexpression dun terme simple, ni vraie ni fausse, qui entre dans une formule
dclarative de nature complexe: invece una c composta di parti, di cui almeno una signi-
fcativa per s, cio o un sintagma attributivo come e (Int. 2, 16a22), n vero n falso, o un enunciato
dichiarativo o non dichiarativo, vero o falso o n vero n falso.
2 Juliette Lemaire (2005, pp. 90-91 e 93 n. 70) considera invece (e non la sola) il principio secondo cui di due
contraddittori necessariamente uno vero e laltro falso come un des noncs du principe du tiers-exclu. Ma il
principio del terzo escluso equivale alla negazione della co-falsit dei contraddittori, non allafermazione della
verit delluno e della falsit dellaltro.
sui principi aristotelici della contraddizione 133
, che, mentre nelle Categorie prima compare la formulazione estesa e poi
quella ellittica, nel De Interpretatione introdotta per prima la formulazione ellittica e
subito dopo quella estesa.
Le due formule sono dunque chiaramente sinonime, e in particolare quella ellittica
non va scambiata, come spesso avvenuto e tuttora avviene, per una formulazione del
pb. Le due formule si alternano nel De Interpretatione in un numero quasi uguale di oc-
correnze. Quella ellittica ricompare una volta alla fne di Int. 7 ( c j c
, 18a10-11), dove riprende la clausola (b) di T8, a riprova della sinonimia delle
due formulazioni, e quattro volte in Int. 9: tre allinizio del capitolo ( c
j c j r, 18a28-29; j
c j , 18a34; <j > c j , 18a37-38),1 e
una alla fne ( b b r j , 19a36-
37).2 Quella estesa ricompare una volta alla fne di Int. 8 ( c b c b
r , 18a26-27) e tre volte in Int. 9: due allinizio del capitolo (d c
b c b [scl. c j c ], 18a29-30;
d [] r c b c b
, 18b27-29), e una alla fne ( d -
c b c b r, 19b1-2).3
Da un punto di vista logico (della logica classica o bivalente), pb e rcc sono principi
equivalenti e interderivabili, cio equipollenti come ogni tautologia; ma, da un punto
di vista semantico, non sono sinonimi. Il pb riguarda l, lenunciato dichiara-
tivo afermativo o negativo, mentre la rcc riguarda l, la coppia di enunciati
dichiarativi opposti, cio lafermazione e la sua negazione logica. In Int. 9 il paradosso
del determinismo logico non imputabile al pb ma alla rcc: che ogni enunciato di-
chiarativo contingente al futuro sia vero o falso non implica che sia vero (potrebbe es-
sere falso) e che il futuro sia predeterminato in un senso piuttosto che in un altro; men-
tre se di due enunciati contraddittori contingenti al futuro, necessariamente uno vero
e laltro falso, allora uno dei due sar vero e il futuro risulter predeterminato in un sen-
so piuttosto che in un altro. La soluzione aristotelica del paradosso del determinismo
logico appunto quella di sospendere la rcc per tali enunciati conservando il pb: si pu
discutere del signifcato di tale soluzione, che presenta aspetti senzaltro oscuri,4 ma che
1 La lezione j attestata dalla traduzione siriaca () e da Ammonio (), ed accolta da
Pacius, Buhle e Weise (cfr. Crivelli 2004a, p. 202 n. 12). Whitaker (1996, p. 115) segue tale lezione e argomenta con-
vincentemente a favore della tesi che si tratti, anche in questo caso, di una delle abbreviazioni standard della rcc
(cfr. in particolare 18a34): Aristotle is arguing that if one person afrms what the other denies, then one or the
other must be telling the truth, if of every afrmation and negation one is true and the other false. The claim that
every assertion is either true or false would not imply that either assertion needed to be true. pb cannot be the
rule that is invoked. Crivelli (2004a, pp. 201-202), seguendo la lezione d attestata dalla
traduzione armena () e accolta da Bekker, Dbner e Cooke, ritiene invece che si tratti del pb: the frst [deter -
ministic argument] (1834-18b4) concludes that Bivalences holding for future-tense singular assertions entails that
at every time one member of a contradictory pair of future-tense singular assertions is true.
2 La formula ellittica attestata, a mio avviso, anche in Metaph. 7, 1011b28: d r j c
j , almeno se vale la ricostruzione dellargomento aristotelico di Metaph. 7, 1011b23-29, da me tentata
in Cavini 1998.
3 Una versione abbreviata della formula estesa anche in Metaph. 8, 1012b10-11: a
r .
4 Il problema principale forse quello di trovare una risposta soddisfacente alla domanda che Richard Gaskin
(1995, p. 167 n. 80) ha rivolto a Whitaker: But if it is not fxed which member of the pair is true, how can it ne-
vertheless be the case that one member is true?. La lettura non vero-funzionale del pb (come anche della rcc di
Whitaker) suggerita da Gaskin mi sembra tuttavia insoddisfacente. Quanto allinterpretazione che Gerhard Seel
134 walter cavini
in gioco per Aristotele vi sia la rcc e non il pb, e che tali principi non siano sinonimi,
questo dovrebbe essere, a mio avviso, del tutto evidente.
La derivazione del determinismo (logico e causale) dal pb invece tesi stoica attesta-
ta a partire da Cicerone, che nel De Fato la attribuisce espressamente a Crisippo:
T9 Concludit enim Chrysippus hoc modo: Si est motus sine causa, non omnis enuntiatio, quod
dialectici appellant, aut vera aut falsa erit; causas enim efcientis quod non habebit,
id nec verum nec falsum erit; omnis autem enuntiatio aut vera aut falsa est; motus ergo si-
ne causa nullus est. Quod si ita est, omnia, quae funt, causis funt antegressis; id si ita est,
fato omnia funt; efcitur igitur fato feri, quaecumque fant.
Dunque Crisippo conclude in questo modo: Se esiste un moto incausato, allora non ogni
enunciato, che i dialettici chiamano , sar o vero o falso, poich ci che non avr una
causa efciente non sar n vero n falso; ma tutti gli enunciati sono o veri o falsi: quindi non
esiste alcun moto incausato. E se le cose stanno cos, tutto ci che accade, accade per opera
di cause antecedenti, e quindi accade per opera del fato. Ne consegue dunque che tutte le
cose che accadono, accadono per opera del fato.
Cic. Fat. 20-211
Ma la tesi senzaltro precede Crisippo, perch ripetutamente Cicerone cita la reazione
di Epicuro a tale tesi, cio la sua negazione sia del principio che ogni movimento abbia
una causa sia del pb per gli enunciati contingenti al futuro (cfr. Fat. 18-19 e 21).2 La testi-
monianza di Cicerone al riguardo particolarmente interessante, perch Cicerone, do-
po aver attribuito a Epicuro la negazione del pb, gli attribuisce anche quella della rcc,
senza in apparenza distinguere fra i due principi:3
(2001) ha dato di vero in modo indeterminato in Ammonio e Boezio come vero ma non necessariamente ve-
ro, a mio avviso non coglie un aspetto cruciale del commento di Ammonio e Boezio, come anche del testo ari-
stotelico, cio, come scrive R. W. Sharples (2003, p. 9), that Ammonius and Boethius do not simply take one di-
sjunct and assert that it is true, though indefnitely; rather, they regularly refer [come fa anche Aristotele] to the
two disjuncts dividing the true and the false, but doing so indefnitely. Ma dubito che il suggerimento di Sharples
di interpretare vero ora come an admission that the truth[-value] may change, in other words that it is indef-
nite (p. 7), cio di interpretare in modo indeterminato (indefnite) come changeable, possa essere una risposta
soddisfacente alla domanda di Gaskin.
1 La traduzione di riferimento del De Fato quella di Francesca Antonini (1994): nel passo citato la traduzione
(p. 65) contiene un refuso: invece di .
2 In una nota alla sua traduzione di Fat. 19, Antonini (1994, p. 62 n. 53) scrive: Epicuro [] rifut il principio
di non-contraddizione (che ogni enunciato debba essere o vero o falso) per le proposizioni riguardanti il futuro.
La confusione dei due principi, di non contraddizione e di bivalenza, sistematica in questa traduzione (cfr. p. 15
n. 20, pp. 33-34), ma ha un pedigree illustre che risale almeno a Carl Prantl (1855-1870, i, p. 450 e pp. 450-451 n. 136).
Stessa confusione anche in Paolillo 1966, pp. 16-17, 42 n. 12, 53 n. 19, 75 n. 37. Domenico Pesce (1970), da parte sua,
attribuisce a Epicuro prima la negazione del principio di non contraddizione (p. 61, cfr. p. 64 n. 70), subito dopo
quella del terzo escluso (p. 62 n. 60), confondendo entrambi i principi fra loro e col principio di bivalenza; infne,
a commento di Fat. 37, confonde il pnc con la rcc: Altro infatti il principio logico di non contraddizione, altro
il principio fsico di causalit. Per il primo, date due proposizioni contraddittorie, luna deve essere vera e laltra
falsa n fanno eccezione le proposizioni che si riferiscono ad un evento futuro (p. 81). Correttamente invece Al-
do Magris (1994, p. 91 n. 73), seguendo Talanga 1986, p. 109, parla di principio di bivalenza, anche se attribuisce ad
Aristotele la tesi epicurea secondo cui nessuna delle due proposizioni opposte riguardanti il futuro sarebbe a prio-
ri vera o falsa, ma solo lalternativa sarebbe vera.
3 Susanne Bobzien (1998, pp. 76-77) considera il principio logico di Fat. 37 a curious intermediate between Prin-
ciple of Bivalence and semantic Excluded Middle (p. 77). Unanaloga sinonimia fra pb e rcc si riscontrer, come
vedremo, in Leibniz, che chiama principio di contraddizione sia il pb: Le principe de contradiction est en g-
nral: une proposition est ou vraie ou fausse (Nouveaux Essais sur lEntendement Humain [1705/1765], iv ii 1), sia la rcc:
il y a deux grands principes de nos raisonnements: lun est le principe de la contradiction, qui porte que de deux
propositions contradictoires, lune est vraie, lautre fausse (Essais de Thodice [1710], 44).
sui principi aristotelici della contraddizione 135
T10 Necesse est enim in rebus contrariis duabus (contraria autem hoc loco ea dico, quorum al-
terum ait quid, alterum negat), ex iis igitur necesse est invito Epicuro alterum verum esse,
alterum falsum [].
Infatti necessario che fra due contrari (qui chiamo contrari due enunciati dei quali uno af-
ferma una cosa e laltro la nega) fra questi dicevo necessario, anche se Epicuro lo nega,
che uno sia vero e laltro falso.
Cic. Fat. 37
In T10 chiaramente non abbiamo a che fare con un enunciato vero o falso, ma con una
coppia di enunciati contraddittori, di cui necessariamente uno vero e laltro falso, se-
condo la formulazione estesa della rcc che abbiamo incontrato nelle Categorie e nel De
Interpretatione. Ma la testimonianza di Cicerone al riguardo si complica ulteriormente
perch da un lato, subito dopo, egli attribuisce non pi a Epicuro ma agli Epicurei la tesi
(una variante del cosiddetto sogno di Aristotele) secondo cui per gli enunciati
contingenti al futuro sono vere le disgiuntive dei contraddittori (cio vale il pte), ma
nessuno dei due contraddittori vero:
T11 Nisi forte volumus Epicureorum opinionem sequi, qui tales enuntiationes nec veras nec
falsas esse dicunt aut, cum id pudet, illud tamen dicunt, quod est inpudentius, veras esse ex
contrariis diiunctiones, sed, quae in his enuntiata essent, eorum neutrum esse verum.
A meno che, per caso, non vogliamo seguire lopinione degli Epicurei, che dicono che tali
enunciati non sono n veri n falsi, oppure, vergognandosi di dire questo, dicono questal-
tra cosa, ancor pi vergognosa, che sono vere le disgiuntive di contrari, ma che nessuno dei
due enunciati vero.
Cic. Fat. 37
mentre, dallaltro, in due opere di poco precedenti il De Fato, cio il Lucullus e il De Na-
tura Deorum, egli attribuisce a Epicuro non il sogno di Aristotele ma la negazione pu-
ra e semplice della verit del pte per gli enunciati contingenti al futuro:
T12 Etenim cum ab Epicuro, qui totam dialecticam et contemnit et inridet, non inpetrent ut ve-
rum esse concedat quod ita efabimur aut vivet cras Hermarchus aut non vivet, cum dia-
lectici sic statuant, omne quod ita disiunctum sit quasi aut etiam aut non <non> modo
verum esse sed etiam necessarium [] cum hoc igitur dialectici pugnent, id est Antiochus
et Stoici; totam enim evertit dialecticam: nam si e contrariis disiunctio (contraria autem ea
dico, cum alterum aiat alterum neget) si talis disiunctio falsa potest esse, nulla vera est.
Infatti, essi non ottengono da Epicuro, il quale deride e spregia lintera dialettica, che egli
ammetta la verit di una proposizione che formuleremo cos: Ermarco domani o sar vivo
o non sar vivo, mentre i dialettici afermano schiettamente che ogni enunciazione fatta
nella forma disgiuntiva O s o no non soltanto vera ma anche necessaria []. Ebbene
dunque, i dialettici, cio Antioco e gli Stoici, combattano pure con costui, giacch egli
abbatte tutta quanta la dialettica: infatti, se lopposizione disgiuntiva procedente da enun-
ciati contrari (e chiamo contrari quelli luno dei quali aferma, laltro nega) se, dico, tale
disgiunzione pu essere falsa, non ve n nessuna che sia vera.
Cic. Luc. 971
T13 Idem facit contra dialecticos; a quibus cum traditum sit in omnibus diiunctionibus, in quibus
aut etiam aut non poneretur, alterum utrum esse verum, pertimuit ne, si concessum esset
1 Traduzione di Rafaello Del Re (1976).
136 walter cavini
huius modi aliquid aut vivet cras aut non vivet Epicurus, alterutrum feret necessarium:
totum hoc aut etiam aut non negavit esse necessarium; quo quid dici potuit obtusius?
In modo analogo [Epicuro] si comporta quando discute coi dialettici. Questi insegnarono
che in tutte le proposizioni disgiuntive in cui si d unalternativa del tipo o o non una
delle due vera; egli ha temuto che, se venisse accettata una proposizione del tipo: domani
Epicuro vivr o non vivr, una delle due alternative sarebbe stata necessaria: neg in bloc-
co la necessit delle proposizioni del tipo o o non ; quale afermazione pu essere pi
Cic. n.d. i 701
La posizione di Epicuro e degli Epicurei secondo la testimonianza di Cicerone appare
dunque complessa e controversa, e non pu essere approfondita qui. Ma quanto ad Ari-
stotele, i tre principi in gioco: bivalenza, terzo escluso e coppie contraddittorie risulta-
no essere con tutta chiarezza, se ci si attiene strettamente ai testi, principi distinti e con
signifcati diversi. E come tali vanno studiati, senza applicare ad Aristotele, come pure
si tende a fare qualche volta negli studi di logica antica, leggi logiche che Aristotele o
non conosce o comunque non applica esplicitamente. bene che la maieutica dei testi
non si risolva in una mantica. Aristotele non solo conia il termine tecnico ,
ma inaugura uno studio sistematico della contraddizione che non ha precedenti nella
storia della logica antica, e ci che vi di interessante in questo non sono i precorri-
menti della logica classica ma il metodo seguito da Aristotele nellesplorare le proprie-
t logiche della contraddizione e i risultati che ottiene al riguardo.
La rcc dunque l che distingue lopposizione di afermazione e negazione da-
gli altri modi dellopposizione. Ma quali sono le propriet logiche dellopposizione
contraddittoria compendiate in tale regola? Nel passo delle Categorie (10, 13a37-b35) si
sottolinea in particolare il fatto che negli altri tipi di opposizione entrambi gli opposti
possono risultare falsi: per esempio, se Socrate esiste, allora gli enunciati contrari
Socrate sano e Socrate malato saranno uno vero e laltro falso, dal momento che
sano e malato sono contrari senza intermedio ( escluso che, se Socrate esiste, possa
essere n sano n malato);2 ma se Socrate non esiste, saranno entrambi falsi. Questo
non pu mai avvenire nel caso degli enunciati contraddittori: se Socrate non esiste,
allora non godr di nessuna propriet che si applica agli esseri viventi e quindi sar
vera (a fortiori) la negazione Socrate non malato e sar falsa lafermazione Socra-
te malato. In generale, dunque, gli opposti contraddittori non possono essere en-
trambi falsi.3
Questa tuttavia solo met della storia: perch valga la rcc non basta escludere la co-
falsit dei contraddittori, si deve anche escludere la loro co-verit. Gli opposti contrad-
dittori, per essere sempre e necessariamente uno vero e laltro falso, non possono esse-
re n insieme falsi n insieme veri. Il passo delle Categorie deriva esplicitamente la rcc
dalla negazione della co-falsit degli opposti contraddittori, perch questo che distin-
gue lopposizione contraddittoria dagli altri modi dellopposizione, mentre la negazio-
ne della co-verit vale anche, per esempio, per i contrari, bench questo non sia detto
1 Traduzione di Cesare Marco Calcante (1992), modifcata. 2 Cfr. Cat. 10, 11b38-12a9, 12b27-32.
3 In generale, perch, come ha mostrato Whitaker (1996, pp. 95-108), in Int. 8 le asserzioni multiple come
Luomo e il cavallo sono bianchi rappresentano una eccezione alla rcc in quanto i contraddittori possono essere
entrambi falsi. Per la negazione della co-falsit cfr. anche Int. 9, 18b17, e Metaph. 8, 1012b10-11.
sui principi aristotelici della contraddizione 137
nel De Interpretatione e, come vedremo in seguito dettagliatamente, in Metafsica
che Aristotele menziona esplicitamente laltra met della storia, cio la negazione della
co-verit degli opposti contraddittori. Il primo testo al riguardo il passo gi citato di
Int. 7 dove insieme (a) introdotta per la prima volta la rcc e (b) si formula la prima
eccezione a tale regola, cio la co-verit di due enunciati contraddittori non universali
su universali:
T8 (a) b s d , c r j
, d d \ , x
(b) \ d c , d b c b a
e d , d
e d []
(a) Quanti dunque dei contraddittori riguardano gli universali in modo universale, di ne-
cessit luno o laltro vero o falso, e <cos anche> quanti riguardano le cose singole, per
esempio Socrate bianco / Socrate non bianco; (b) quanti invece riguardano gli uni-
versali in modo non universale, non sempre uno vero e laltro falso: vero infatti dire in-
sieme che luomo bianco e che luomo non bianco, e che luomo bello e che luomo
non bello []
Arist. Int. 7, 17b26-33
Il punto ribadito in Int. 9:1
T14 a b b c e e , e
, <j > c j a
d .
Se infatti uno afermer che qualcosa sar e laltro negher questa stessa cosa, chiara-
mente necessario che uno dei due dica il vero, se ogni afermazione <o negazione> vera
o falsa, perch entrambe le cose non sussisteranno insieme in tali casi.
Arist. Int. 9, 18a35-39
Per Aristotele dunque gli opposti contraddittori non sono n insieme veri (analoga-
mente agli opposti contrari) n insieme falsi (a diferenza degli opposti contrari),2 ma
necessariamente luno vero e laltro falso (fatte salve le tre eccezioni esaminate in Int. 7-
9: entrambi veri [Int. 7]; entrambi falsi [Int. 8]; luno vero e laltro falso, ma in modo
indeterminato [Int. 9]). La rcc dunque una conseguenza immediata (un corollario)
della negazione congiunta della co-verit e della co-falsit degli opposti contraddittori.
Ammonio, commentando T14, chiama assioma della contraddizione la negazione
congiunta della co-verit e della co-falsit degli opposti contraddittori:
T15 b e \
d d c ,
c . b
e b a a
d [] e b b a ,
Quindi muove dallassioma della contraddizione dicendo che necessariamente delle pro-
posizioni singolari e contingenti prese al tempo futuro una delle due vera, poich non
1 Cfr. Ammon. in Int. 140.4-6 b., Boeth. in Int. i 109.6-7 m.
2 Cfr. anche Arist. Metaph. I 4, 1055b1-3.
138 walter cavini
possibile n che siano insieme entrambe false n che siano insieme entrambe vere. E che
non siano entrambe vere detto chiaramente qui con le parole entrambe le cose non sus-
sisteranno insieme in tali casi [] mentre che non sia neanche possibile che esse siano in-
sieme false, lo aggiunger in seguito.1
Ammon. in Int. 139.32-140.11 b.2
Il termine assioma della contraddizione ha una lunga storia, a parte ante e a parte post
rispetto ad Ammonio, che merita di essere ricostruita nelle sue alterne vicende, antiche
e moderne, e che sar loggetto della terza parte di questo lavoro. Essa riguarda in par-
ticolare linterpretazione di Metafsica e degli assiomi introdotti e discussi in questo
libro da Aristotele, cio il principio di non contraddizione (pnc) e il principio del terzo
escluso (pte). Aristotele enuncia i due assiomi, ma non li nomina. I commentatori gre-
ci di Aristotele, a partire almeno da Alessandro di Afrodisia, chiameranno assioma del-
la contraddizione sia la congiunzione dei due assiomi aristotelici sia ciascuno di essi
singulatim. Di qui, come vedremo, deriveranno fra laltro le peripezie dei moderni prin-
cipi di non contraddizione e del terzo escluso.
3. \A \A
3.1. Un tratto peculiare della metafsica di Aristotele in quanto scienza dei principi di
tutte le cose che i principi di cui si occupa non si limitano a quelli dell (di ci che
vi realmente nel mondo, cio della sostanza), ma si estendono anche a quelli delle di-
mostrazioni (cio ai principi della nostra conoscenza scientifca del mondo). Nel libro
Bdella Metafsica la seconda delle quindici aporie sollevate a proposito della scienza ri-
cercata si chiede appunto
T16 a a a d j d d
z , x e d d
j , d d []
se sia proprio della scienza in questione considerare solo i principi della sostanza o anche
rifettere sui principi a partire dai quali tutti dimostrano, come per esempio se sia possibile
insieme afermare e negare la stessa e identica cosa oppure no, e sugli altri principi del
genere []
Arist. Metaph. B 1, 995b6-10
Lesempio di principio dimostrativo implicitamente oferto in questo testo appunto
una delle formulazioni aristoteliche del pnc:
(1) Non possibile insieme afermare e negare la stessa cosa,
formulazione che ritroviamo anche altrove (con varianti che, a seconda dei casi,
ampliano o riducono la formula) sia nella Metafsica3 sia nel De Interpretatione4 e negli
Analitici Secondi,5 e di cui sono possibili due interpretazioni, a seconda che con afer-
mare e negare la stessa cosa si intenda afermare e negare lo stesso enunciato:
1 Cfr. Arist. Int. 9, 18b17-18.
2 Cfr. anche Ammon. in Int. 223.2-4 b.: a e
3 4, 1008a36-b1 ( i e e d ); 6, 1011b20-21 (
d ). 4 12, 21b20 (e e d a ).
5 i 11, 77a10 (e b c d ).
sui principi aristotelici della contraddizione 139
(1a) Non possibile insieme afermare e negare lo stesso enunciato,
o afermare e negare lo stesso predicato (sottintendendo, in questo caso, dello stesso
(1b) Non possibile insieme afermare e negare lo stesso predicato <dello stesso soggetto>.
La prima interpretazione si estende a ogni enunciato dichiarativo, predicativo o non
predicativo, ed quella pi confacente alla universalit del pnc, che appunto vale per
ogni enunciato dichiarativo. Ma Aristotele sembra avere in mente piuttosto la seconda
interpretazione, quella ristretta agli enunciati predicativi. E questo per almeno due ra-
gioni: in primo luogo, nel passo del De Interpretatione, dove presente unanaloga
formulazione (12, 21b20), compare chiaramente laggiunta a ; in secondo
luogo, come vedremo, anche le altre tre formulazioni aristoteliche del principio sono
tutte predicative.1
La seconda formulazione del pnc si trova, insieme a quella del pte, nella ripresa del-
la seconda aporia in Metaph. B 2:
T17 a c d d , d j ,
( b a a a z )
x j j , d r d c r, d
, d j , i c
, c c .
Ma anche a proposito dei principi dimostrativi controverso se spettino a una sola scienza
o a pi (chiamo principi dimostrativi le credenze comuni a partire dalle quali tutti quanti
dimostrano), come per esempio che ogni cosa necessario o afermarla o negarla e che
impossibile insieme essere e non essere, e tutte quante le altre proposizioni del genere:
<laporia > se unica la scienza di essi e della sostanza o diversa, e qualora non sia unica,
quale delle due debba essere chiamata col nome di quella ora ricercata.
Arist. Metaph. B 2, 996b26-33
Questa formulazione ellittica ricorre solo qui e due volte in Metaph. (4, 1006a3-4; 5,
1010a35-36), ma con varie integrazioni o omissioni quella pi frequente nel libro del-
la Metafsica,2 dove in due casi compare anche in forma predicativa esplicita:
T18 e \ , e e r d c r e
, a e .
1 Sulla equivalenza in Aristotele delle due interpretazioni cfr. Mignucci 2003a, p. 78: Tuttavia, nelle intenzio-
ni di Aristotele, la limitazione del principio [a proposizioni che abbiano forma predicativa] non voluta, dato che
egli sembra supporre che tutte le proposizioni abbiano forma predicativa. Quindi nella sua prospettiva le due
ipotesi si equivalgono.
2 Cfr. 3, 1005b23-25 ( a e r d c r, b
^H); 1005b29-30 (e e e r d c r e ):
Giovanni Reale (2004, p. 145) traduce solo in questo caso r con esistere, ma la traduzione immotivata, men-
tre Carlo Augusto Viano (1974, p. 273) scambia e e per e e ( evidente che impossibile credere nello
stesso tempo che la medesima cosa sia e non sia la medesima cosa); 4, 1005b35-1006a2 (Ed ,
, e e r d c r, d ); 1006b18 ( r
d c r e e); 5, 1012a25 ( \ b ^H , r d c r,
). Vedi anche K 5, 1061b36-1062a1 ( e e \ d e e r d c
r), a7-8, 16-17.
140 walter cavini
Il problema non se possibile che la stessa cosa insieme sia e non sia un uomo quanto al
nome, ma quanto alla cosa.
Arist. Metaph. 4, 1006b20-22
T19 b r e e r d c r .
Non dunque possibile che sia vero insieme dire che la stessa cosa un uomo e non un
Arist. Metaph. 4, 1006b33-34
cio nella forma predicativa diretta Soggetto (S)-Predicato (P):
(2) Non possibile che S sia P e insieme S non sia P,
rispetto alla forma predicativa inversa Predicato (P)-Soggetto (S) della prima formula-
(1b) Non possibile afermare P di S e insieme negare P di S.
Le altre due formulazioni predicative sono entrambe predicative inverse. La prima la
celebre enunciazione del principio di non contraddizione in Metaph. 3:
T20 e a e d c d a e (d
\ , e a a ) []
impossibile infatti che la stessa cosa insieme appartenga e non appartenga alla stessa co-
sa e sotto lo stesso rispetto (e quante altre precisazioni potremmo aggiungere, si conside-
rino aggiunte in risposta alle capziosit dialettiche) []
Arist. Metaph. 3, 1005b19-22
dove la relazione predicativa inversa espressa tramite la relazione di appartenenza
propria degli Analitici Primi e della teoria del sillogismo categorico, cui fa eco pi avanti
una formulazione analoga, in cui la relazione di appartenenza sostituita da quella di
T21 a .
impossibile che si predichino insieme i <predicati> contraddittori <dello stesso soggetto>.
Arist. Metaph. 4, 1007b181
La seconda invece esprime la relazione di predicazione tramite quella di essere vero di:
T22 c a []
impossibile che i <predicati> contraddittori siano veri insieme dello stesso <soggetto> []
Arist. Metaph. 6, 1011b16-172
1 Errata la traduzione di Antonio Russo (1973, p. 101), che interpreta a come asserzioni con-
traddittorie e non come predicati contraddittori: impossibile che asserzioni contraddittorie vengano usate si-
multaneamente come predicato (anche se, ovviamente, impossibile che unasserzione venga usata come pre-
dicato!). Ugualmente errata la traduzione di Viano (1974, p. 279): impossibile che le proposizioni contraddittorie
vengano afermate contemporaneamente.
2 Anche in questo caso le traduzioni di Russo e di Viano fraintendono il signifcato di : impossibi-
le che le asserzioni contrarie [!] su una medesima cosa siano nello stesso tempo vere entrambe (Russo 1973, p.
114); impossibile che proposizioni contraddittorie siano contemporaneamente vere della stessa cosa (Viano
1974, p. 292). Cfr. anche Metaph. 4, 1007b18-19: a .
sui principi aristotelici della contraddizione 141
Si danno pertanto nel testo aristotelico le seguenti formulazioni predicative del pnc:
(1b) Non possibile afermare P di S e insieme negare P di S;
(2) Non possibile che S sia P e insieme S non sia P;
(3) Non possibile che Pappartenga a Se insieme e sotto lo stesso rispetto Pnon appartenga a S;
(4) Non possibile che P sia vero di S e insieme non-P sia vero di S.
Di queste formulazioni predicative del pnc, tre inverse ([1b], [3] e [4]) e una diretta ([2]),
quella ufciale, cio la (3), tale in quanto tiene conto anche della restrizione a e
(sotto lo stesso rispetto) e implicitamente degli altri possibili parametri restritti-
vi, che Aristotele declina esplicitamente nelle Confutazioni Sofstiche (5, 167a26-27),1 se-
guendo con tutta probabilit lesempio platonico.2
Resta da esaminare unultima formulazione, che ritroviamo alla fne di Metaph. 6,
nel rsum conclusivo che Aristotele d della trattazione del pnc:
T23 O b s e c r a , d
, d a , []
Che dunque la credenza pi certa di tutte sia quella che le asserzioni opposte [scl. contrad-
dittorie] non sono vere insieme, e che cosa ne consegue per coloro che dicono cos,3 e per
quale ragione dicono cos, riguardo a ci basti quanto stato detto fn qui.
Arist. Metaph. 6, 1011b13-154
Questa lunica formulazione aristotelica non predicativa ma enunciativa del pnc, in
quanto in questo contesto non pu che signifcare asserzione come genere le cui
specie sono lafermazione () e la negazione () predicative, cio si-
nonimo di (termine che non compare mai nella Metafsica). Le asserzioni op-
poste sono appunto lafermazione e la sua negazione logica, e da questo punto di vista
la locuzione a semplicemente ricapitola le formulazioni predi-
cative precedenti, dal momento che afermazione e negazione per Aristotele sono
enunciati predicativi; subito dopo infatti, in T22, Aristotele ritorna alla formulazione
predicativa inversa. Abbiamo pertanto una quinta formulazione del principio:
(5) Le asserzioni contraddittorie non sono vere insieme.
a (in questo caso anche Reale [2004, p. 155], dopo aver tradotto correttamente
a con i contraddittori, traduce erroneamente con le afermazioni contraddittorie,
il che assurdo: non si danno per defnizione afermazioni contraddittorie; la contraddizione appunto lopposi-
zione tra lafermazione e la negazione corrispondente; Reale cade nello stesso errore traducendo a
di Metaph. 6, 1011b14, con le afermazioni contraddittorie).
1 Cfr. anche Metaph. 6, 1011a17-24, a34-b1.
2 Pl. r. iv 436b-437a; Sph. 230b7-8. Cfr. Dorion 1995, pp. 239-240, e Fait 2007, p. xlix e n. 70.
3 Intendo, con David Ross (1908), che cosa ne consegue per coloro che dicono che le asserzioni contradditto-
rie sono vere insieme; la traduzione di Barbara Cassin e Michel Narcy (1989, p. 153): ce qui arrive ceux qui di-
sent ces noncs [scl. les noncs opposs] , a mio avviso, non grammaticale (in francese come in italiano: che
signifca dire un enunciato?!) e si presta a fraintendimenti: non si tratta di dire gli enunciati opposti ma di as-
serire che gli enunciati opposti sono veri insieme.
4 Come gi segnalato supra, pp. 141-142 n. 2, la traduzione di Reale di a con le afer-
mazioni contraddittorie senzaltro errata: signifca afermazione ogni volta che opposta esplicitamente
ad (e questo a partire almeno da Platone, Sph. 263e12, e Alcidamante ap. D.L. ix 54; cfr. Nuchelmans
1973, p. 30, e per luso aristotelico, Bonitz 1870, 813a17-23), come invece qui non avviene; e comunque non si dan-
no afermazioni opposte, cio contraddittorie. Viano invece non trova di meglio che tradurre con pro-
142 walter cavini
ukasiewicz, nel suo celebre saggio Del principio di contraddizione in Aristotele (1910),
chiama ontologiche le formulazioni (2) e (3), perch in esse si parlerebbe di oggetti
e attributi e non di giudizi e verit, e propone quindi di formulare in questi ter-
mini il principio ontologico di non contraddizione:
Nessun oggetto pu possedere e non possedere uno stesso attributo nello stesso tempo.1
Chiama invece logica la formulazione (5), cio T23, perch riguarda il valore di verit
dei giudizi,2 cio parla di giudizi e non di oggetti e attributi, mentre non pren-
de mai in considerazione le altre formulazioni logiche (?), cio (1b) e (4), in cui non si
parla n di giudizi n di oggetti e attributi, ma di ci che si predica o vero di qual-
cosa, cio di soggetti e predicati comunque intesi. ukasiewicz pertanto propone la se-
guente formulazione del principio logico di contraddizione:
Non possono essere veri nello stesso tempo due giudizi, dei quali uno assegna alloggetto pro-
prio quellattributo che dallaltro gli viene negato.3
La tesi di ukasiewicz, sostenuta nel secondo capitolo del suo libro, che le due for-
mulazioni non sono sinonime ma sono logicamente equivalenti.4 Non sono sinonime
perch [n]el principio ontologico di contraddizione si parla degli oggetti, in quello lo-
gico dei giudizi.5 Sono invece logicamente equivalenti, anche se Aristotele non ha mai
enunciato espressamente questa tesi, perch interderivabili. ukasiewicz fonda la in-
terderivabilit delle due formulazioni e quindi la loro equivalenza logica su un passo di
Int. 9, che qui riporto nella sua interezza:
1 ukasiewicz 1910a/2003, p. 19: nella formulazione di ukasiewicz, che compare allinizio del i capitolo del
suo libro, omessa la restrizione a e ; inoltre, nella traduzione italiana si legge principio ontologico di
non contraddizione (corsivo mio), mentre in quella francese troviamo principe ontologique de contradiction
(ukasiewicz 1910a/2000, p. 47). Come vedremo fra breve, questa curiosa duplicit del nome del principio di
(non) contraddizione non sar senza signifcato per il nostro lavoro. Segnalo infne, sempre allinizio del i capito-
lo, un evidente errore di greco da parte di ukasiewicz (come desumo dal fatto che presente in entrambe le tra-
duzioni consultate): ukasiewicz parla a un certo punto della proposizione logica, [sic], ma chiara-
mente non si tratta della negazione, bens dellenunciato dichiarativo () o asserzione ().
2 Suppongo, per il principio di carit ermeneutica, che il termine originale sia giudizi, come nella traduzio-
ne francese (ukasiewicz 1910a/2000, p. 48), e non afermazioni, come nella traduzione italiana (ukasiewicz
1910a/2003, p. 20), dove Metaph. 6, 1011b13-14, reso (seguendo evidentemente Reale e non ukasiewicz): La
nozione pi salda di tutte sia questa: che le afermazioni [sic] contraddittorie non possono essere vere insieme, e
dove, subito dopo, si pu leggere: Aristotele intende per afermazioni contraddittorie, 0 [sic], -
[sic] , lafermazione, , e la negazione, [sic], sullo stesso oggetto e sotto lo stes-
so rispetto: a parte i ripetuti errori di greco, abbiamo afermazioni contraddittorie, che logicamente impos-
sibile, e afermazioni che sono negazioni, che una contraddizione in termini, un ferro di legno.
3 ukasiewicz 1910a/2003, p. 20: la formulazione di ukasiewicz tiene giustamente conto del fatto che i
giudizi aristotelici hanno struttura predicativa. Ovviamente non condivisibile la sua traduzione di con
4 Com noto, ukasiewicz distingue, seguendo (senza citarlo) Heinrich Maier (1896-1900, i pp. 43-47), una ter-
za formulazione del principio, che chiama psicologica e che non sinonima di quelle ontologica e logica, per-
ch in essa si parla di convinzioni o atti di credenza: Due convinzioni, a cui corrispondono giudizi contraddittori,
non possono sussistere nello stesso tempo nella stessa mente (ukasiewicz 1910a/2003, p. 21). Ma il passo di Aristotele
a cui ukasiewicz si riferisce, cio Metaph. 3, 1005b23-32, non contiene una formulazione del pnc, bens un ar-
gomento a sostegno della tesi, enunciata in precedenza, che il pnc, nella sua versione ontologica, il principio pi
certo di tutti i principi ( d , 1005b22-23). Per questa critica a ukasiewicz cfr. ora
Severino 2005, pp. 39-40. Il passo aristotelico sar discusso tematicamente nel 4.
5 ukasiewicz 1910a/2003, p. 24.
sui principi aristotelici della contraddizione 143
T24 (a) a b e j , r e j , d
(b) e j , b q j d (c) c , ,
d (d) , []
(a) Se infatti vero dire che bianco o non bianco, allora necessariamente bianco o non
bianco, e (b) se bianco o non bianco, allora era [?] vero afermare o negare; e (c) se non
sussiste, allora si dice il falso, e (d) se si dice il falso, allora non sussiste []
Arist. Int. 9, 18a39-b3
Si danno pertanto quattro casi:
(a) Se vero (dire) che S (non) P, allora necessariamente S (non) P;
(b) Se S (non) P, allora necessariamente vero (dire) che S (non) P;
(c) Se S non P, allora necessariamente falso (dire) che S P;
(d) Se falso (dire) che S P, allora necessariamente S non P.
Il punto di ukasiewicz , a mio avviso, assolutamente corretto: il passo citato corri-
sponde a quella che i medievali (Giovanni Buridano) hanno chiamato regula Aristotelis1
e i moderni (Michael Dummett) chiamano Tesi di Equivalenza.2 Tale regola o tesi si
applica ai rapporti tra essere ed essere vero e tra non essere ed essere falso, e si compendia in
due tesi di equivalenza: la tesi di equivalenza fra essere ed essere vero:
(ab) vero (dire) che S P sse S P
e la tesi di equivalenza fra non essere e essere falso:
(cd) falso (dire) che S P sse S non P.
ukasiewicz prende in considerazione, a proposito del pnc, solo la tesi di equivalenza
fra essere ed essere vero, e correttamente fa derivare lequivalenza tra la formulazione on-
tologica e la formulazione logica del pnc dallapplicazione della regula Aristotelis. Tale
regola, per quanto riguarda i rapporti fra essere ed essere vero, stabilisce due principi: la
clausola (a) stabilisce il principio della discesa semantica da essere vero a essere:
(a) Se vero (dire) che S P, allora S P;
la clausola (b) stabilisce la conversa, cio il principio della ascesa semantica da essere a
essere vero:
(b) Se S P, allora vero (dire) che S P.
Secondo ukasiewicz la formulazione logica deriva dalla formulazione ontologica in
base al principio della discesa semantica: infatti,
se vero il giudizio che assegna a un dato oggetto un attributo, allora quelloggetto lo possiede;
se vero il giudizio che nega a un oggetto un attributo, allora quelloggetto non lo possiede. Se
quindi i due giudizi contraddittori fossero veri entrambi, in tal caso lo stesso oggetto avrebbe un
attributo e nello stesso tempo non lo avrebbe. Questo per non possibile in virt del principio
ontologico di contraddizione; quindi giudizi contraddittori non possono essere veri nello stesso
Viceversa, la formulazione ontologica deriverebbe dalla formulazione logica in base al
principio dellascesa semantica:
1 Giovanni Buridano, Sophismata, viii 2, 45, 47 (cfr. Knne 2003, p. 151 e n. 183).
2 Dummett 1973, p. 445. 3 ukasiewicz 1910a/2003, p. 24.
144 walter cavini
se un oggetto ha un attributo [] vero il giudizio che glielo assegna; se non lo possiede []
vero il giudizio che glielo nega. Quindi se lo stesso oggetto avesse un attributo e insieme non
lavesse, dovrebbero essere veri due giudizi contraddittori. Questo per non possibile in virt
del principio logico di contraddizione; nessun oggetto pu possedere e non possedere nel con-
tempo lo stesso attributo.1
Mi sembra tuttavia pi intuitivo pensare che per Aristotele la formulazione logica deri-
vi dalla formulazione ontologica in base al principio dellascesa semantica e la formu-
lazione ontologica derivi dalla formulazione logica in base al principio della discesa se-
mantica. Infatti, per il principio dellascesa semantica, valgono le seguenti implicazioni:
Se S P, allora vero che S P, e se S non P, allora vero che S non P; dunque se non (S P e
S non P), allora non ( vero che S P ed vero che S non P),
in base alle quali deriviamo dalla formulazione ontologica la formulazione logica del
pnc. E viceversa, per il principio della discesa semantica, valgono le seguenti impli-
Se vero che S P, allora S P, e se vero che S non P, allora S non P; dunque se non ( ve-
ro che S P ed vero che S non P), allora non (S P e S non P),
in base alle quali deriviamo dalla formulazione logica la formulazione ontologica del
In ogni caso, ukasiewicz ha avuto senzaltro il merito di applicare nel suo libro la
regula Aristotelis per derivare lequivalenza tra le due formulazioni, anche se, subito
dopo, sembra ritenere che tale regola coincida con la defnizione aristotelica di giu-
dizio vero:
Considero esatta questa conclusione [cio lequivalenza delle due formulazioni, logica e onto-
logica]; essa consegue dallidea secondo cui, giustamente, lente e i giudizi veri si corrispondo-
no reciprocamente. Questo concetto si basa sulla defnizione di giudizio vero; vero il giudizio af-
fermativo che attribuisce a un ogetto quellattributo che esso possiede; vero il giudizio negativo che nega
a un ogetto un attributo che esso non possiede. E viceversa: Ogni ogetto possiede lattributo che gli vie-
ne attribuito dal giudizio vero; e nessun ogetto possiede lattributo che gli viene negato dal giudizio vero.
Aristotele avrebbe accettato una simile defnizione, poich dice:
Metaph. 7, 1011 b 26-27: e [] a [], e k r d e c k c r [].
Dire di ci che , che , e di ci che non , che non , la verit.
Da queste defnizioni risulta necessariamente lequivalenza tra il principo ontologico e il princi-
pio logico di contraddizione.2
Che la regula Aristotelis, cio la tesi di equivalenza, coincida con la defnizione aristote-
lica di verit, come ukasiewicz sembra ritenere, e che la defnizione aristotelica di ve-
rit sia una teoria della verit come corrispondenza (lente e i giudizi veri si corrispon-
dono reciprocamente), sono tesi, a mio avviso, tuttaltro che ovvie e richiederebbero
un esame e una critica approfonditi, cosa che non possibile fare qui.3 ukasiewicz, nel
1 ukasiewicz 1910a/2003, p. 25. 2 ukasiewicz 1910a/2003, p. 25.
3 Per la tesi di equivalenza rimando a Cavini 1993b, p. 87; per la defnizione aristotelica di (dire il) vero e (di-
re il) falso rimando a Cavini 1998. Per una critica della tesi di equivalenza come defnizione della verit cfr.
McGinn 2000, cap. 5. Per una interpretazione della teoria aristotelica della verit come teoria della corrispon-
sui principi aristotelici della contraddizione 145
saggio in tedesco dello stesso anno, in cui riassume le tesi del suo libro, fonda senzal-
tro lequivalenza tra le due formulazioni sulla defnizione aristotelica di verit di Me-
taph. 7, intesa come una teoria della verit come corrispondenza tra enunciati e fatti,
senza pi evocare la regula Aristotelis;1 e, com noto, nel suo libro (cap. viii) considera
principio ultimo non il pnc ma la defnizione di giudizio vero. Vedremo in seguito ( 4)
come per Aristotele la defnizione di verit non si possa considerare un principio dimo-
strativo, a diferenza del pnc, e in che senso questultimo e non la defnizione di verit
sia il principio di tutti i principi dimostrativi.
Tornando allequivalenza tra la formulazione ontologica e la formulazione logica del
pnc, si tratta, secondo ukasiewicz, di una equivalenza che Aristotele avrebbe []
chiamato [] soltanto logica e non reale.2 A sostegno di questa tesi, in apparenza crip-
tica, ukasiewicz cita un celebre passo di Metafsica :
T25 a a e e r r f , a a e b r e
Infatti non perch noi pensiamo in modo vero che tu sei bianco, che tu sei bianco, ma
perch tu sei bianco, che noi, dicendo questo, diciamo il vero.
Arist. Metaph. 10, 1051b6-9
e commenta: Pertanto: lente tanto una ragione logica della verit dei giudizi, quan-
to una causa reale della loro enunciazione; la verit dei giudizi invece solo una ragio-
ne logica e non una causa reale dellente. Suppongo che Aristotele avrebbe cos formu-
lato questa diferenza, se solo se ne fosse reso conto chiaramente.3 In realt, Aristotele
sembra essersi reso perfettamente conto che la tesi di equivalenza fra essere ed essere
vero non stabilisce una simmetria completa fra i due termini e non va interpretata in
tal senso, perch lessere causa dellessere vero e non viceversa, cio fra i due termini
intercorre una relazione asimmetrica di priorit causale. In un passo delle Categorie (non
citato da ukasiewicz) troviamo di seguito enunciate la tesi di equivalenza fra essere ed
essere vero e quello che si convenuto di chiamare Principio c,4 cio la tesi della prio-
rit causale di essere rispetto a essere vero:
T26 (a) e a r a c r e e d
(b) a , c > (c)
d , (d) a c > ,
(e) b b c r e , e
r e (f ) a r e j c
c j c .
denza isomorfa cfr. ora Crivelli 2004a, pp. 23-25 e cap. 4. Per la diferenza fra la tesi di equivalenza e il Princi-
pio c vedi infra, n. 4.
1 ukasiewicz 1910b, p. 18/2000, p. 392.
2 ukasiewicz 1910a/2003, p. 25. 3 ukasiewicz 1910a/2003, p. 26.
4 Dummett 1976/1993, p. 52: If a statement is true, there must be something in virtue of which it is true (c
sta per Corrispondenza in quanto tale principio underlies the philosophical attempts to explain truth as a cor-
respondence between a statement and some component of reality). Dummett (1959/1978, p. 14) chiaramente di-
stingue questo principio dalla tesi di equivalenza: the correspondence theory expresses one important feature of
the concept of truth which is not expressed by the law It is true that p if and only if p [] that a statement is
true only if there is something in the world in virtue of which it is true. Il Principio c , a mio avviso, tutto quan-
to si pu trovare in Aristotele di una teoria della verit come corrispondenza, e una prova che la regula Aristotelis
di Cat. 12 non equivale alla defnizione semantica di verit di Metaph. 7. Per la discussione del Principio c cfr.
ora in particolare Knne 2003, pp. 148-154.
146 walter cavini
(a) infatti essere un uomo1 si converte secondo la consecuzione di essere nellenunciato
vero al riguardo, (b) perch se x un uomo, allora vero lenunciato con cui diciamo che
x un uomo; (c) e certo vale la conversa: (d) se infatti vero lenunciato con cui diciamo
che x un uomo, allora x un uomo; (e) ma lenunciato vero non in alcun modo causa
del fatto che la cosa sia <cos>, invece la cosa che appare in qualche modo causa del
fatto che lenunciato sia vero: (f ) infatti perch la cosa o non <cos> che lenuncia-
to si dice vero o falso.
Arist. Cat. 12, 14b14-22
Le clausole (a)-(b) e (c)-(d) enunciano rispettivamente il principio di ascesa e di discesa
semantica, cio la regula Aristotelis o tesi di equivalenza:
vero dire che S P sse S P,
ma le due clausole successive, con lenunciazione del Principio c, chiariscono il fatto che
limplicazione reciproca fra essere ed essere vero non elimina lasimmetria fra i due ter-
mini, cio la priorit causale di essere rispetto a essere vero:
Che S P ci in virt del quale vero dire che S P,
mentre non vale linverso.
Ma una volta stabilita lequivalenza logica delle due formulazioni del pnc tramite
lapplicazione del principio aristotelico di ascesa e discesa semantiche, qual il signif-
cato di tale equivalenza, che Aristotele per altro assume tacitamente senza sentire mai
il bisogno di dimostrarla in tutta la sua trattazione del pnc in Metafsica ? Come si vi-
sto nelle Categorie e nel De Interpretatione, le propriet logiche di cui gode lopposizione
contraddittoria sono, da un lato, la negazione della co-falsit dei contraddittori, sottoli-
neata in particolare nelle Categorie (cfr. T6 e T7), dallaltro, la negazione della loro co-
verit, evocata due volte nel De Interpretatione (cfr. T8 e T9). Metafsica assume sen-
zaltro la negazione della co-verit dei contraddittori come logicamente equivalente alla
formulazione ontologica del pnc. Il pnc corrisponde dunque per Aristotele alla pro-
priet logica dellopposizione contraddittoria per cui i contraddittori non possono es-
sere veri insieme o allo stesso tempo.
3.2. A diferenza del pnc, per il quale possibile distinguere in Aristotele una formula-
zione ontologica e una formulazione logica non sinonime ma logicamente equivalen-
ti, laltro principio dimostrativo discusso in Metafsica , cio il pte, non ha una formu-
lazione ontologica ma solo una formulazione logica in due varianti sinonimiche, come
si evince dalla sua enunciazione allinizio di Metaph. 7:
T27 (a) \Aa c b f r , (b) \ j j
\ e .
(a) Ma daltra parte non neanche possibile che vi sia un intermedio fra i contraddittori,
(b) ma necessario o afermare o negare un predicato qualsiasi di un soggetto.
Arist. Metaph. 7, 1011b23-24
1 Non c ragione di tradurre r con esiste un uomo dando a r il valore di esistenza (de se-
cundo adiacente) e non di copula (de tertio adiacente) come fanno tutti gli interpreti recenti del passo (Colli 1955,
Ackrill 1963, Oehler 1984, Zanatta 1989, Bods 2002, Ildefonse e Lallot 2002), che sono perci costretti a
tradurre a c r con as to implication of existence (Ackrill) o quand on envisage la
consquence de lexistence (Bods) et similia: (il fatto) che S P non causa dellesistenza ma della verit
dellenunciato corrispondente S P. Per il valore predicativo ellittico di r cfr. per esempio r
e di Int. 9, 18b1 (T24).
sui principi aristotelici della contraddizione 147
La formulazione aristotelica ricorrente del pte in realt una variante pi concisa di
(b), cio o / j o j j , con
/ al posto di \ e .1 Cos nella formulazione standard del princi-
pio non vi alcun riferimento esplicito a un terzo escluso, alla esclusione della possi-
bilit di un intermedio fra i contraddittori, cio di un enunciato vero equivalente alla
loro negazione congiunta, come avviene invece in (a). Ma chiaramente le due formu-
lazioni, negativa (a) e afermativa (b), sono logicamente equivalenti:2
(a) Px (Px Px) (b) Px (Px Px).
Che per Aristotele la formulazione (a), cio lesclusione della possibilit di un interme-
dio fra i contraddittori, sia equivalente alla versione semantica del principio, cio alla
negazione della co-falsit dei contraddittori, formalmente:
Px (Px Px) Px (FPx FPx),
risulta chiaro per due ragioni. In primo luogo, dal co-testo adiacente: l\Aa c inci-
pitario chiaramente continua quanto detto alla fne del capitolo precedente, che si con-
cludeva appunto negando la co-verit dei contraddittori e (quindi) dei contrari. Dunque
non solo i contraddittori non possono essere entrambi veri allo stesso tempo (per il
pnc), ma non possono essere neanche entrambi falsi allo stesso tempo (per il pte). In
secondo luogo, Aristotele equipara alla fne di Metaph. 7 la negazione del pnc alla dot-
trina di Eraclito, che, col sostenere che tutte le cose sono e non sono, rende tutto quan-
to vero, e quella del pte alla dottrina di Anassagora, <che, col sostenere> che vi un
intermedio fra i contraddittori, rende tutto falso;3 e nel capitolo seguente, esaminan-
do la tesi anassagorea nulla vero (o tutto falso) e la tesi eraclitea tutto ve-
ro,4 cos conclude a proposito della tesi anassagorea:
T28 (a) b b e b j n ,
r a r . (b) j j
, r a
(a) Ora, se ci che vero afermare non altro che ci che falso negare, impossibile che
tutto sia falso: necessario infatti che una <sola> delle due parti della contraddizione sia
vera. (b) Inoltre, se necessario o afermare o negare ogni cosa, impossibile che entram-
be <le parti della contraddizione> siano false: una <sola> infatti delle due parti della con-
traddizione falsa.
Arist. Metaph. 8, 1012b8-13
1 Cfr. Frede 1970, pp. 77-78; Frede 1985, pp. 79-80; Cavini 1998, pp. 6-7; Crivelli 2004a, p. 229 n. 92 (ma Cri-
velli considera erroneamente Metaph. 8, 1012b10-11 [ a r ],
una versione semantica del pte e non una variante ellittica della rcc).
2 Cfr. Mignucci 1975, pp. 237-238. Una variante platonica della formulazione negativa in Prm. 156c6-7: X
d , > x , dove i predicati contrari ed
sono evidentemente da intendersi come contraddittori ( = c ).
3 Metaph. 7, 1012a24-27. Ovviamente, la tesi eraclitea non la negazione contraddittoria del pnc, ma la te-
si contraria, e la tesi anassagorea non la negazione contraddittoria del pte, ma la tesi contraria. Sul fatto che
Aristotele assuma una negazione forte (cio la contraria e non la contraddittoria) del pnc, cfr. Mignucci 1996,
pp. 51-52 e n. 23. Lo stesso vale per il pte.
4 Su queste tesi e la loro autoconfutazione cfr. ora in particolare Castagnoli 2005, pp. 58-66; 2007, pp. 45-47.
148 walter cavini
Si tratta di due argomenti contro la tesi anassagorea e quindi a favore della versione se-
mantica del pte. La premessa del primo argomento data dal signifcato di vero e fal-
so: vero che S P sse falso che S non P. Pertanto, se entrambi i contraddittori fos-
sero falsi, allora sarebbero veri i rispettivi contraddittori: se falso che S P, allora
vero che S non P, e se falso che S non P, allora vero che S P; cio sarebbe vio-
lato il pnc: necessario infatti che una <sola> delle due parti della contraddizione sia
vera. Pertanto la negazione del pte implica quella del pnc.1 La premessa del secondo
argomento invece lo stesso pte nella sua versione predicativa: se o S P o S non P,
allora impossibile che S P e S non P siano entrambi falsi, perch per il pte una
<sola> delle due parti della contraddizione falsa.2
Si pu dunque concludere che, da un punto di vista logico, il signifcato aristotelico
dei due principi il seguente: per il pnc uno solo dei contraddittori vero, cio non pos-
sono essere entrambi veri allo stesso tempo; per il pte uno solo dei contraddittori fal-
so, cio non possono essere entrambi falsi allo stesso tempo. Ma queste sono appunto
le propriet logiche che contraddistinguono lopposizione contraddittoria rispetto agli
altri tipi di opposizione, come risulta dai passi esaminati in precedenza delle Categorie e
del De Interpretatione: le versioni semantiche del pnc e del pte defniscono congiunta-
mente la natura dell aristotelica. Pertanto, quello che nelle Categorie (T7)
detto essere l dellopposizione contraddittoria, cio la propriet per cui di una cop-
pia di contraddittori necessariamente uno vero e laltro falso (rcc), risulta essere una
conseguenza immediata (un corollario) della congiunzione di pnc e pte: se i contrad-
dittori non possono essere n insieme veri (pnc) n insieme falsi (pte), allora necessa-
riamente uno vero e laltro falso (rcc).3
1 A questo argomento (la negazione del pte implica quella del pnc) fa da pendant largomento di Metaph. 4,
1008a2-7 ( s e , d j j . a
b d , d \ \ a
, b , d i ), che deriva la conversa: la negazione
del pnc implica quella del pte, a riprova del fatto che Aristotele, contrariamente a quanto sostenuto da Wolf (ve-
di infra, p. 161) e da alcuni interpreti recenti (per esempio Berti 2002, p. 30), non deriva il pte dal pnc pi di quan-
to non derivi il pnc dal pte. Di diversa natura, a mio avviso, come cercher di mostrare in seguito, la priorit
che Aristotele attribuisce al pnc rispetto agli altri principi dimostrativi e quindi anche al pte.
2 Cfr. Alex. Aphr. in Metaph. 340.9-18 h. e Longo 2005, p. 132 n. 83.
3 Jean-Baptiste Gourinat (2001, p. 74) osserva in proposito: La conjonction du principe de contradiction et du
principe du tiers-exclu tels que les formule Aristote implique donc que deux propositions contradictoires ne sont
ni toutes les deux vraies ni toutes les deux fausses. Que lune soit vraie et lautre fausse nen rsulte que si lon admet le
principe de bivalence. Sinon, il reste la possibilit que les deux propositions ne soient ni vraies ni fausses (corsivo mio), e
conclude (p. 75): Aristote conoit donc le principe de contradiction et le principe du tiers-exclu comme lis lun
lautre. Mais, comme il ny conjoint pas la thse de la bivalence, il nadjoint ni lun ni lautre la thse qui leur est
habituellement lie et selon laquelle, de deux propositions contradictoires, lune est vraie et lautre fausse (cor-
sivo mio). La possibilit che le due proposizioni non siano n vere n false ovviamente esclusa dal fatto che la
contraddizione per Aristotele lopposizione di afermazione e negazione (cfr. T4), cio di enunciati dichiarativi
e quindi, per il principio di bivalenza che li defnisce (cfr. T1), necessariamente o veri o falsi. Anche le eccezioni al-
la rcc esaminate nei capitoli centrali del De Interpretatione non rimettono in discussione questo assunto, contra-
riamente a ci che sostiene Gourinat (p. 64) a proposito di Int. 9. In realt, Gourinat ritiene (contro ukasiewicz)
che il passo di Int. 4 (T1) non sia senzaltro da intendersi come lenunciazione aristotelica del principio di bivalen-
za per gli enunciati dichiarativi, ma ammetta uninterpretazione alternativa pi debole: lessere vero o falso sa-
rebbe solo condizione sufciente ma non necessaria per essere un enunciato dichiarativo. Cfr. Gourinat 2006, p.
51: Mais ce que dit exactement Aristote nest pas que tout discours assertif (apophantique) est vrai ou faux, mais
que cest dans le discours assertif quexistent le vrai et le faux. Cela nexclut pas a priori que certaines propositions
puissent tre autre chose que vraies ou fausses, ou ntre ni vraies ni fausses. La phrase nest donc pas parfaite-
ment explicite.
sui principi aristotelici della contraddizione 149
Come scrive lucidamente Peter Suber,1
The Principle of Non-Contradiction (PNC) and Principle of Excluded Middle (PEM) are fre-
quently mistaken for one another and for a third principle which asserts their conjunction.
Given a statement and its negation, p and ~p, the PNC asserts that at most one is true. The PEM
asserts that at least one is true. The PNC says not both and the PEM not neither. Together,
and only together, they assert that exactly one is true.
Let us call the principle that asserts the conjunction of the PNC and PEM, the Principle of
Exclusive Disjunction for Contradictories (PEDC). Surprisingly, this important principle has
acquired no particular name in the history of logic.
PNC at most one is true; both can be false
PEM at least one is true; both can be true
PEDC exactly one is true, exactly one is false
Clearly the PEDC is not identical to either the PNC or the PEM, and the latter two are not iden-
tical to one another.
The PEM is simple inclusive disjunction for p and ~p. The PNC is the denial of their conjunc-
tion. Conjoining these gives us exclusive disjunction: at least one of the contradictories is true
(PEM) and not both are true (PNC).
Queste considerazioni (al di l dellapparato della logica classica, che ovviamente non
pu essere ascritto ad Aristotele) mi sembrano del tutto in linea con la logica aristote-
lica della contraddizione, come risulta dai testi fn qui esaminati. Suber trova sorpren-
dente che un principio cos importante come quello che asserisce la congiunzione del
pnc e del pte non abbia trovato un nome particolare nella storia della logica, e per par-
te sua lo intitola Principio della Disgiunzione Esclusiva per i Contraddittori. In real-
t, almeno per quanto riguarda la storia della logica antica, ma, come vedremo, non sen-
za efetti sulla storia della logica moderna, il principio in questione non rimasto n
inosservato n anonimo. A partire almeno dal commento di Alessandro di Afrodisia al-
la Metafsica, i commentatori greci di Aristotele mostrano senzaltro di conoscere tale
principio, che intitolano, pi succintamente, assioma della contraddizione o dei con-
traddittori ( ).
La prima testimonianza al riguardo quella del commento di Alessandro di Afrodi-
sia a Metaph. B 1, 995b6:
T29 b e c c d r, d e a
d r , d e i e , d a
r []. d e b a d
E un assioma quello per cui i contraddittori non possono essere veri e falsi insieme, come
anche quello per cui cose uguali alla stessa cosa sono anche uguali fra loro, e quello per cui,
se da cose uguali si sottraggono cose uguali, anche quelle che restano sono uguali []. E il
pi comune di tutti quanti gli assiomi certo quello della contraddizione [o dei contrad-
dittori]: infatti vero in tutti i casi.
Alex. Aphr. in Metaph. 175.7-12 h.2
1 Suber 2002, p. 1.
2 Maddalena Bonelli (2001, p. 62) non rileva questo primo esempio di uso inclusivo della formula
in Alessandro di Afrodisia e scambia in questo passo lassioma della contraddizione per il principio di
non contraddizione tout court (p. 63). Luso inclusivo rilevato invece correttamente da Angela Longo (2005, p. 134
n. 86).
150 walter cavini
cui si pu aggiungere quella del commento di Asclepio al passo di Metaph. B 2 relativo
alla seconda aporia:
T30 d e , c
1 j []. a x
j j c
, a b j d r d c r
c .
E per prima cosa introduce lassioma della contraddizione [o dei contraddittori], e aferma
che impossibile che i contraddittori siano mai entrambi veri o entrambi falsi []. Dicen-
do dunque come per esempio che ogni cosa necessario o afermarla o negarla, ha mo-
strato che non possibile che i contraddittori siano entrambi falsi; e dicendo invece o an-
che impossibile insieme essere e non essere, ha mostrato che non possibile che i
contraddittori siano insieme entrambi veri.
Ascl. in Metaph. 158.11-32 h.
Questuso collettivo o inclusivo testimoniato anche in due passi del commento di Am-
monio al De Interpretatione: il primo il gi citato testo T15, il secondo fa parte del com-
mento a Int. 12:
T31 a e
quelle <proposizioni> invece che premettono le negazioni dei modi, salvano lassioma del-
la contraddizione, non risultando mai n insieme vere n insieme false.
Ammon. in Int. 223.2-4 b.
Invece nel commento di Siriano al passo di Metafsica B 2 relativo alla seconda aporia si
parla di una nozione comune riguardante la contraddizione (d ),
che comprende, secondo i predecessori di Aristotele (a ), due as-
siomi,2 cio quello del terzo escluso e quello di non contraddizione:
T32 a b s a a a 3 r, e
j d [], a \ a ,
b a e a , b
z d d d
b a d , e b c b ,
a j j , d e
a , \ r c b
a , b e .
Pertanto [Aristotele] dichiara che le nozioni comuni sono principi dimostrativi, non gi in
quanto procurano una disposizione in noi o ci spingono a delle azioni [], ma in quanto
ce ne serviamo per le dimostrazioni, chiamandole noi assiomi per il fatto di essere state
assunte cos presso tutti, e avendo di esse una conoscenza pi evidente che <la conoscen-
za> delle cose dimostrate. Una di queste <nozioni comuni> anche quella riguardante la
contraddizione, ma presso i pi antichi due sono le cose che vengono assunte riguardo ad
1 Il composto gi in Aristotele: cfr. Int. 10, 19b36; Metaph. K 6, 1063a21.
2 Cfr. Syrian. in Metaph. 18.28-29 k. e Longo 2004, p. 88.
3 Il termine aristotelico non d ma d (Metaph. B 2, 996b28, 997a20-21) o d
(APo. i 32, 88a36). Per la derivazione delle d di Euclide dalle d aristoteliche, cfr. Ross 1949,
pp. 56-57.
sui principi aristotelici della contraddizione 151
essa [scl. riguardo alla contraddizione]: una quella per cui niente le sfugge, bens neces-
sario che ogni cosa sia signifcata o in modo afermativo o in modo negativo, cosicch in
nessun caso entrambe le proposizioni falliscano il bersaglio [scl. siano false], ma luna delle
due sia vera; e laltra che impossibile che queste <proposizioni> siano vere insieme. Egli
[scl. Aristotele] ora fa uso di entrambe le cose [].1
Syrian. in Metaph. 18.9-22 k.
Angela Longo, commentando in due occasioni questo passo,2 ritiene che Siriano di-
stingua due assiomi della contraddizione, a diferenza di Alessandro di Afrodisia e di
Asclepio, che parlano invece di un solo assioma. I due assiomi distinti da Siriano sono
chiaramente quello del terzo escluso ( necessario che ogni cosa sia signifcata o in mo-
do afermativo o in modo negativo, cosicch in nessun caso entrambe le proposizioni
falliscano il bersaglio [scl. siano false], ma luna delle due sia vera) e quello di non con-
traddizione ( impossibile che queste <proposizioni> siano vere insieme): il primo sia
nella sua formulazione sintattica ( necessario che ogni cosa sia signifcata o in modo
afermativo o in modo negativo) sia in quella semantica che ne consegue, in quanto le
due formulazioni sono logicamente equivalenti (cosicch in nessun caso entrambe le pro-
posizioni falliscano il bersaglio [scl. siano false], ma luna delle due sia vera); il secon-
do nella sola formulazione semantica. Ma a proposito del primo dei due assiomi della
contraddizione cos distinti, Longo sostiene sorprendentemente che nel commento di
Siriano a Metafsica tale assioma non lassioma del terzo escluso, ma ci che ne con-
segue, ovvero limpossibilit della co-falsit di due proposizioni contraddittorie, e il fat-
to che una di esse sia vera.3 Siriano dunque presenterebbe il principio per cui ne-
cessario o afermare o negare ogni cosa (ovvero il principio del terzo escluso) come il
preludio allassioma della contraddizione circa limpossibilit della co-falsit di due pro-
posizioni contraddittorie.4 In altri termini, Siriano riterrebbe che dal principio del ter-
zo escluso consegua lassioma della contraddizione, ma non viceversa, cio che i due
principi non siano logicamente equivalenti. Il che ovviamente falso. Quello che Lon-
go chiama il principio del terzo escluso in realt solo la variante sintattica del prin-
cipio, in simboli:
(1) p (p p);
mentre quello che Longo chiama assioma della contraddizione circa limpossibilit del-
la co-falsit di due proposizioni contraddittorie la sua variante semantica, in simboli:
(2) p (Fp Fp),
e le due formule, come facilmente dimostrabile,5 sono logicamente equivalenti.
Attribuire a Siriano la tesi della non equivalenza fra principio del terzo escluso (cio
la variante sintattica del principio) e assioma della contraddizione (la variante semanti-
ca), e quindi la distinzione fra principio e assioma, sarebbe senzaltro attribuirgli un er-
rore non banale. Ma Longo ritiene che in un passo della parte fnale del commento a
1 La traduzione italiana quella di Longo 2004, pp. 85-86, lievemente modifcata. Cfr. anche Longo 2005, pp.
133-134 n. 86, e p. 189. 2 Cfr. Longo 2004, pp 84-90; 2005, pp. 133-135 n. 86.
3 Longo 2004, p. 89. 4 Ivi, p. 90 (corsivo mio).
5 Per la terza legge di De Morgan, p (Fp Fp) p (Fp Fp); ma, per lequivalenza tra verit e non-
falsit della negazione classica (cfr. Arist. Metaph. 7, 1017a31-32: e r d e , e
b c r b a ; Int. 11, 21a22-23), Fp Vp e Fp Vp; dunque, per la tesi ari-
stotelica di equivalenza: Vp p e Vp p, p (Fp Fp) p (Vp Vp) p (p p).
152 walter cavini
Metafsica , Siriano chiaramente distingua gli assiomi riguardanti la contraddizione da
quello del terzo escluso.1 Il passo merita in efetti di essere riportato per intero ed esa-
minato con cura:
T33 T e a g a d
, a d d , x
a e d , d f a ,
j d a \ b
j c d . b y
; j e k r j e c k c r b
d . b d ; j e c k r j e k c
r. b e b , e b b
f d []
Dicendo cos tante cose contro tali posizioni, il flosofo [scl. Aristotele] non solo conferma
gli <assiomi> riguardanti la contraddizione, ma discute anche di altri assiomi, come per
esempio che impossibile che i contrari siano presenti nella stessa cosa sotto lo stesso ri-
spetto e nello stesso modo, e che non vi alcun intermedio fra i contraddittori: se infatti vi
qualcosa, sar senzaltro vero o falso, perch tale propriet appartiene a ogni enunciato
dichiarativo; ma nientaltro soggetto per natura a essere vero o falso tranne lafermazio-
ne e la negazione; dunque non vi qualcosa <di intermedio fra i contraddittori>. E in che
modo solo queste [scl. lafermazione e la negazione] sono vere? Chiunque dica il vero o di-
ce che ci che o <dice> che ci che non non , cio unafermazione e una negazio-
ne. E in che modo anche sono false? Chiunque dica il falso o dice che ci che non o <di-
ce> che ci che non : la prima unafermazione falsa, la seconda una negazione falsa.
Non vi dunque alcun intermedio tra afermazione e negazione []
Syrian. in Metaph. 78.22-32 k.
In questo passo, secondo Longo, Siriano distinguerebbe a chiare lettere gli assiomi ri-
guardanti la contraddizione (a d ) dallassioma del terzo escluso
(f ), cio i due principi semantici per cui i contraddittori non
possono essere n veri insieme n falsi insieme dal principio sintattico per cui non vi
alcun intermedio fra i contraddittori, il principio del terzo escluso appunto. In realt,
in questo passo, Siriano commenta cursoriamente sia la fne di Metaph. 6 sia linizio
di Metaph. 7. Alla fne di 6, e quindi alla conclusione della trattazione aristotelica del
principio di non contraddizione ( 3-6), si riferiscono sia laccenno ai numerosi argo-
menti aristotelici contro chi nega tale principio (Dicendo cos tante cose contro tali po-
sizioni) sviluppati in 4-6, sia quello alla derivazione del principio di non contrariet
( impossibile che i contrari siano presenti nella stessa cosa sotto lo stesso rispetto e
nello stesso modo) dal principio di non contraddizione ( 6, 1011b15-22), che chiude
6. Allinizio di 7 si riferiscono sia la menzione dellassioma secondo cui non vi al-
cun intermedio fra i contraddittori ( 7, 1011b23) sia quella della defnizione aristoteli-
ca di (dire il) vero e (dire il) falso ( 7, 1011b25-29).
Tenendo conto di questa collocazione del passo, sorprende a prima vista che Siriano
usi il plurale a d in riferimento a quanto detto in precedenza da Ari-
stotele in difesa del pnc, dal momento che lo stesso Siriano, in quattro casi precedenti,
si riferito a tale principio chiamandolo (al singolare) e (66.14,
68.29k.), e (69.7-8k.) e e a c (74.17-
1 Cfr. Longo 2004, p. 90 e n. 26; 2005, pp. 132-135.
sui principi aristotelici della contraddizione 153
18k.). Ma il plurale (unanimemente attestato dai codici) si giustifca alla luce del com-
mento di Siriano a un passo di Metaph. 4, in cui Aristotele aferma che tra le conse-
guenze della negazione del pnc vi anche la negazione del pte ( s
e , d j j , 1008a2-4). Siria-
no commenta riferendosi al pte come a laltro dei due assiomi della contraddizione
(e [] ).1 Non vi dubbio pertanto che anche
il pte nella sua variante sintattica ( j j ) sia per Siriano uno dei
due assiomi della contraddizione; e quanto al plurale, che in efetti ricorre solo in Siria-
no,2 in realt corrisponde perfettamente alluso sia collettivo sia distributivo che del ter-
mine assioma della contraddizione fanno anche gli altri commentatori greci di Ari-
stotele: Alessandro di Afrodisia, per esempio, dopo aver chiamato, collettivamente,
assioma della contraddizione la congiunzione del pnc e del pte (cfr. T29), chiama con
lo stesso nome, distributivamente, il solo pnc,3 seguito in questo da Siriano4 e da Ascle-
pio;5 mentre Filopono nel suo commento agli Analitici Secondi e Simplicio in quello alla
Fisica chiamano, distributivamente, assioma della contraddizione sia il pnc sia il pte.6
Resta infne da considerare la menzione da parte di Siriano dellassioma secondo cui
non vi alcun intermedio fra i contraddittori, assioma che sarebbe altro rispetto a
quelli della contraddizione. Dal momento che anche Siriano come gli altri commenta-
tori greci di Aristotele considera il principio aristotelico del terzo escluso nella sua for-
mulazione afermativa (ogni cosa necessario o afermarla o negarla) come uno dei
due assiomi della contraddizione, non resta che concludere che per lui lassioma se-
condo cui non vi alcun intermedio fra i contraddittori non la formulazione negati-
va del principio aristotelico del terzo escluso, e quindi non un assioma della contrad-
dizione, ma un altro assioma. Il che ovviamente falso: come aveva gi visto
correttamente Alessandro di Afrodisia, lassioma secondo cui non vi alcun interme-
dio fra i contraddittori equivale alla versione semantica del principio aristotelico del ter-
zo escluso, cio al principio per cui i contraddittori non possono essere falsi insieme, e
quindi non un assioma diverso dal principio aristotelico del terzo escluso, ma solo la
sua variante negativa (non vi alcun intermedio), logicamente equivalente alla sua
variante afermativa (ogni cosa necessario o afermarla o negarla):
T34 c x , a
f . \ i a e b c
> i e d e c e d e . a
b , , ( a i f -
), e d e e b r, e b .
Avendo mostrato che non possibile che i contrari appartengano insieme <allo stesso sog-
getto>, mostra in seguito, mediante numerosi argomenti, che non vi alcun intermedio
fra i contraddittori. Con ci pu stabilire altres che i contraddittori non sono neanche fal-
si insieme, cui seguirebbe che essi dividono in ogni caso il vero e il falso. Se infatti non so-
no n veri insieme, come stato mostrato, n falsi insieme (ci che potrebbe essere detto
1 Syrian. in Metaph. 71.14-15 k. 2 Cfr. anche Syrian. in Metaph. 79.16 k.
3 Cfr. Alex. Aphr. in Metaph. 271.24, 272.10 h. In due occasioni Alessandro di Afrodisia chiama il pnc anche
c : cfr. in Metaph. 269.22, 272.7 h., e Prantl 1855-1870, i, p. 622. Lo stesso fa Asclepio, in Me-
taph. 258.10-13, 259.35 H. 4 Cfr. Syrian. in Metaph. 66.14, 68.29, 69.7-8, 74.17-18 k.
5 Cfr. Ascl. in Metaph. 251.6-7, 251.19-21, 258.12-13, 259.27-28, 291.21-25, 292.36-38, 293.22-23 h.
6 Per il pnc, cfr. Phlp. in APo. 135.6-9, 136.12, 137.18-19, 138.17, 138.31-139.2, 140.3-6, 141.7 w.; Simp. in Ph. 240.15-16,
985.19-20 d.; vedi anche Elias in Cat. 110.9-11 b. Per il pte, cfr. Phlp. in APo. 137.33-138.1, 140.11-13, 181.2-3 w.; Simp.
in Ph. 21.26-27, 1021.3-4 d.; vedi anche Ammon. in Int. 146.8-12, 222.8-10 b.
154 walter cavini
intermedio fra i contraddittori), allora rimane che in ogni caso una delle due parti della
contraddizione sia vera e laltra falsa.
Alex. Aphr. in Metaph. 328.7-13 h.
In questo testo Alessandro di Afrodisia riassume con tutta chiarezza quanto si venuto
sostenendo fn qui: (a) al principio aristotelico di non contraddizione, nelle sue varie for-
mulazioni ontologiche e logiche, corrisponde il principio semantico per cui i contrad-
dittori non possono essere veri insieme (Se infatti non sono n veri insieme, come
stato mostrato); (b) al principio aristotelico del terzo escluso, nelle sue due formula-
zioni logiche (sintattiche) afermativa e negativa, corrisponde il principio semantico per
cui i contraddittori non possono essere neanche falsi insieme (Con ci pu stabilire al-
tres che i contraddittori non sono neanche falsi insieme); infne (c) dalla congiunzio-
ne dei due principi deriva come corollario il principio semantico per cui di due con-
traddittori uno vero e laltro falso, cio i contraddittori dividono in ogni caso il vero
e il falso,1 dal momento che lopposizione contraddittoria si ha solo tra afermazioni e
negazioni, cio tra enunciati dichiarativi veri o falsi. Alessandro di Afrodisia, seguito so-
stanzialmente dagli altri commentatori greci di Aristotele, chiama assioma della con-
traddizione, sia collettivamente sia distributivamente, i principi semantici corrispon-
denti ai principi aristotelici di non contraddizione e del terzo escluso. Tali principi
semantici, insieme al loro corollario (l dellopposizione contraddittoria o rcc), co-
stituiscono i principi aristotelici della contraddizione.
Non sappiamo chi sia stato il primo a chiamare il principio
per cui i contraddittori non possono essere n veri insieme n falsi insieme, e quindi, di-
stributivamente, sia il principio aristotelico di non contraddizione sia quello del terzo
escluso, cio gli assiomi di cui tratta Aristotele nel libro della Metafsica. Come si vi-
sto, la locuzione testimoniata solo a partire dal commento di Alessandro di Afrodisia
a Metafsica B (T29); e luso che ne fa il commentatore peripatetico quello di una for-
mula ormai acquisita. Tuttavia il signifcato flosofco di tale formula merita fn dora al-
cune considerazioni. In primo luogo, Aristotele aveva enunciato ma non nominato con
una formula ad hoc i principi di non contraddizione e del terzo escluso: la formula scel-
ta dai commentatori non solo colma (almeno in parte, come vedremo) tale lacuna, ma
ha un signifcato flosofco non banale. Per Aristotele, i principi di non contraddizione
e del terzo escluso sono assiomi flosofci, cio principi dimostrativi, principi della
nostra conoscenza scientifca del mondo, che appunto per Aristotele il sapere dimo-
strativo. Come tali sono oggetto, insieme alla sostanza, della flosofa prima: al floso-
fo primo che spetta indagare se siano veri o no,2 ci che fa Aristotele in Metafsica
confutando chi nega tali principi. Ora, averli intitolati , se da un
lato corrisponde esattamente al modo in cui Aristotele li presenta e ne stabilisce elen-
cticamente la verit, dallaltro tuttavia sottolinea con tutta evidenza il rilievo particola-
re che la contraddizione assume in Aristotele per il sapere dimostrativo e la flosofa pri-
ma. Il genitivo non pu essere inteso come epesegetico o dichiarativo,
dal momento che la contraddizione o i contraddittori non sono un assioma in senso ari-
1 Per la locuzione dividere sempre il vero e il falso come sinonima della rcc cfr. Barnes 2007, p. 76: Con-
tradictory opposition, which holds (in principle) between an afrmation and a corresponding negation has this
special characteristic: if two items are contradictory opposites, then (in an ancient jargon) they divide truth and
falsity at any given time, exactly one of them is true and exactly one of them is false.
2 Arist. Metaph. 3, 1005a30.
sui principi aristotelici della contraddizione 155
stotelico, cio un principio dimostrativo; il suo valore sar dunque quello di genitivo di
argomento: lassioma della contraddizione il principio dimostrativo riguardante la con-
traddizione o i contraddittori. Ma questo signifca, come vedremo nella quarta e ultima
parte di questo lavoro, che la contraddizione e i principi che la regolano (non contrad-
dizione, terzo escluso e regola delle coppie contraddittorie) sono intimamente associa-
ti in Aristotele alla dimostrazione e al sapere dimostrativo, e come tali riguardano la f-
losofa prima.
3.3. Per concludere questa parte, mi sembra opportuno rievocare, sia pure sommaria-
mente, le peripezie che legano l dei commentatori greci di Ari-
stotele alla storia del principio di non contraddizione e del terzo escluso nella logica mo-
derna. Il punto di partenza, a mio avviso, pi signifcativo rappresentato dalla Logica
Hamburgensis di Joachim Jung o Jungius (1638), la cui opera fu molto stimata da Leibniz
ed considerata la logica pi importante del Seicento.1 Nel capitolo sulla dimostrazio-
ne,2 Jungius distingue dapprima in generale due principia contradictionis, cio gli assio-
mi Quidvis est vel non est e Idem non simul est, et non est; quindi articola il principium con-
tradictionis o Axioma contradictionis3 in due principi: uno afermativo, corrispondente al
principio aristotelico del terzo escluso, e uno negativo, corrispondente al principio ari-
stotelico di non contraddizione; infne coniuga logicamente i due principi in quello che
chiama Axioma contradictionis completum:4
T35 12. Ex axiomatibus duo primo loco posita [scl. Quidvis est vel non est e Idem non simul est, et
non est] principia contradictionis dicuntur, item principia principiorum, et axiomata axiomatum,
quippe quorum veritati reliquorum principiorum veritas innitatur.
14. Principium contradictionis, sive Axioma contradictionis, aliud est afrmativum, aliud negati-
vum: utrumque vel direct vel refexiv propositum.
15. Afrmativum idemque direct propositum est, Quidvis vel est vel non est, sive Quidvis vel est
hoc, vel non hoc, Item Quidvis est vel tale, vel non tale.
16. Afrmativum idque refex propositum ita se habet, Quodvis de quovis vel afrmandum vel
negandum est, sive De quvis notione quaevis notio vel ipsa vel contradictoria ejus afrmanda est.
1 Cfr. Kneale 1962, pp. 313-314/1972, p. 359; Risse 1964-1970, i, pp. 521-522.
2 Jungius 1638, pp. 322-331 (Meyer 1957, pp. 224-228): Logicae Liber Quartus, Cap. x. De Demonstratione, ejusque
materi, et in specie de Principijs.
3 Il curatore della ristampa della Logica Hamburgensis (1957), Rudolf W. Meyer, non segnala in apparato tra le
fonti la derivazione dal greco dei commentatori greci di Aristotele, che invece Jungius
aveva evidentemente presenti, in particolare il commento di Alessandro di Afrodisia alla Metafsica, che poteva leg-
gere nella traduzione latina cinquecentesca di Juan Gins de Seplveda (1527). Disponibile era anche una tradu-
zione latina cinquecentesca del commento di Siriano alla Metafsica a cura di Girolamo Bagolino (1558). Per le tra-
duzioni latine rinascimentali di Alessandro di Afrodisia, cfr. Cranz 1958.
4 Ancora alla fne del Cinquecento, Giulio Pace (Pacius) sembra ignorare, nel suo commento analitico alla Isa-
goge di Porfrio e allOrganon aristotelico (1597), il termine principium contradictionis. Tuttavia, commentando il
passo degli Analitici Secondi in cui Aristotele aferma che nessuna dimostrazione assume <come premessa> il prin-
cipio secondo cui non possibile insieme afermare e negare, cio il principio di non contraddizione (APo. i 11,
77a10-11), Pace osserva: Est autem principium omnibus scientiis commune, Contradictionis alteram partem esse
veram, alteram falsam, e aggiunge: Hoc principium, quamvis vnum videatur, tamen duabus partibus constat, et
in duo principia diuiditur: nempe afrmatio & negatio non sunt simul verae: &, non est vtraque simul falsa. Prior
pars declaratur in hac particula: posterior in particula sequenti (p. 303a). Cfr. anche Francisco Surez, Disputa-
tiones Metaphysicae (1597), iii, 3.5 (Esposito 2007, p. 612): quae aliis verbis a dialecticis dici solent: Impossibile est duas
contradictorias esse simul veras, et, Impossibile est esse simul falsas; quae duo constat esse valde diversa; et primum fun-
dari in illo principio, Impossibile est idem simul esse et non esse, secundum vero in alio, Necesse est esse vel non esse.
156 walter cavini
17. Negativum idemque direct enuntiatum tale est, Idem non simul et est, et non est, sive Idem
non simul est et hoc et non hoc, sive Idem non simul est et tale, et non tale.
18. Negativum refex propositum ita habet [sic], Idem de eodem non licet simul afrmare et nega-
re, sive De edem notione non licet simul contradictorias notiones afrmare.
19. Si disjunctiva particula completo singnifcatu [sic] sumatur, Negativum in Afrmativo in-
cluditur, et tunc Axioma contradictionis completum appellatur.
Jungius, Logica Hamburgensis, pp. 324-325
da notare anzitutto come entrambi i principia contradictionis il cui nome, come in Ari-
stotele, ancora coincide con la loro enunciazione siano considerati da Jungius princi-
pia principiorume axiomata axiomatum, cio come Jungius estenda anche al principio del
terzo escluso quel tratto di assoluta priorit rispetto agli altri principi, che Aristotele ave-
va assegnato solo al principio di non contraddizione.1
Per quanto riguarda poi larticolazione del principium contradictionis, sive Axioma con-
tradictionis, in un principio afermativo, il terzo escluso, e in un principio negativo, la
non contraddizione, da notare in entrambi i casi la distinzione tra formulazione di-
retta (direct propositum) e formulazione rifessiva o inversa (refex propositum),
che, come si visto, riproduce unanaloga distinzione presente anche nelle varie for-
mulazioni aristoteliche dei due principi. In particolare, la formulazione diretta del prin-
cipio afermativo, cio del terzo escluso: Quidvis vel est <hoc>, vel non est <hoc>, intro-
duce una novit rispetto ad Aristotele, essendo chiaramente una formulazione
ontologica del principio, quella divenuta poi canonica (p (p p)), mentre in Aristo-
tele, come si visto, abbiamo solo formulazioni logiche, fra cui quella che Jungius chia-
ma rifessiva (refex propositum): Quodvis de quovis vel afrmandum vel negandum est.
Infne, merita senzaltro sottolineare il signifcato di quello che Jungius chiama Axio-
ma contradictionis completum: se, argomenta Jungius, riformuliamo il principio Quidvis
vel est vel non est, in cui la disgiunzione inclusiva (o non-esclusiva)2 vel esclude solo la
co-falsit dei disgiunti, e diamo alla particella disgiuntiva un valore completo,3 cio
esclusivo: Quidvis aut est aut non est, in questo modo includiamo il principio negativo, la
non contraddizione, nel principio afermativo, il terzo escluso, in quanto la disgiun-
zione esclusiva aut esclude non solo la co-falsit ma anche la co-verit dei disgiunti, e
la proposizione disgiuntiva vera solo nel caso in cui esattamente uno dei disgiunti sia
vero.4 Cos otteniamo lassioma completo, cio esclusivo, della contraddizione, a ri-
1 Aristotele descrive il principio di non contraddizione, da un lato, come il principio pi certo di tutti <i prin-
cipi> (Metaph. 3, 1005b11-12, 17-18, 22-23; cfr. 6, 1011b13), dallaltro, come per natura principio anche di tutti
gli altri assiomi (Metaph. 3, 1005b33-34); ed solo al principio di non contraddizione che applica enfaticamente
lattributo platonico di ( 3, 1005b14).
2 Cfr. Quine 1950, pp. 10-11/1960, pp. 19-20.
3 Il signifcato completo della particella disgiuntiva, di cui parla Jungius, richiama la , il confitto
perfetto o completo, della logica stoica secondo la testimonianza di Galeno (Inst. Log. iv 1-3 e xiv 5), quello in
cui i configgenti non possono n coesistere () n perire insieme (: in termini di pro-
posizioni configgenti, il coesistere equivale allessere entrambe vere, la metafora del perire insieme allessere en-
trambe false) (Nasti De Vincentis 2002, p. 49).
4 LAxioma contradictionis completum di Jungius coincide dunque con quello che Peter Suber (cfr. supra, p. 150)
chiama the Principle of Exclusive Disjunction for Contradictories, cio formalmente p w p, sorprendendosi
del fatto che questo importante principio non abbia ricevuto un nome particolare nella storia della logica. Jungius
in realt ha avuto il merito non solo di dare un nome a tale principio, ma anche di enunciarlo correttamente da
un punto di vista logico. Ma in efetti, almeno su questo punto, il suo acume logico, cos apprezzato da Leibniz,
non sembra avere trovato seguaci.
sui principi aristotelici della contraddizione 157
prova del fatto che anche per Jungius assolutamente chiaro che ai due principi
sintattici, afermativo e negativo, della contraddizione corrispondono i due principi se-
mantici aristotelici, cio rispettivamente la negazione della co-falsit e della co-verit
dei contraddittori.
Se Jungius d una formulazione sintattica, afermativa e negativa, diretta e inversa,
del principium contradictionis, Leibniz ne d invece almeno due distinte formulazioni se-
mantiche. La prima in ordine di tempo quella che troviamo nei Nouveaux Essais sur
lEntendement Humain (1705, ma pubblicati postumi nel 1765):
T36 Le principe de contradiction est en gnral: une proposition est ou vraie ou fausse; ce qui ren-
ferme deux nonciations vraies, lune que le vrai et le faux ne sont point compatibles dans
une mme proposition, ou quune proposition ne saurait tre vraie et fausse la fois; lautre que
loppos ou la ngation du vrai et du faux ne sont pas compatibles, ou quil ny a point de
milieu entre le vrai et le faux, ou bien: il ne se peut pas quune proposition soit ni vraie ni fausse.
Leibniz, Nouveaux Essais, iv ii 1
Quello che Leibniz chiama qui principio di contraddizione, cio il principio semantico:
p (Vp Fp),
in realt il principio di bivalenza, e come tale non riguarda la contraddizione, cio una
coppia di enunciati contraddittori, ma ogni singolo enunciato dichiarativo vero o falso.
Pu sembrare quindi sorprendente che Leibniz chiami principio di contraddizione un
principio che apparentemente non riguarda la contraddizione, e che egli verosimil-
mente desume non da Aristotele ma dalla logica stoica (in particolare dalla testimo-
nianza del De Fato di Cicerone),1 per la quale il fondamento della dialettica non il prin-
cipio di non contraddizione bens quello di bivalenza.2 In realt, secondo Leibniz,
questo principio semantico generale (en gnral) contiene (renferme) due principi
semantici particolari, entrambi negativi: il primo stabilisce quune proposition ne saurait
tre vraie et fausse la fois, cio:
p (Vp Fp),
ed chiaramente una formulazione semantica del principio di non contraddizione; il se-
condo stabilisce invece che il ne se peut pas quune proposition soit ni vraie ni fausse, cio:
p (Vp Fp),
ed equivale, a sua volta, a una formunazione semantica del principio del terzo escluso.
In che senso allora il principio di contraddizione di cui parla Leibniz, cio il principio
di bivalenza, conterrebbe i due principi di non contraddizione e del terzo escluso? E per-
ch chiamarlo principio di contraddizione?
Il principio di contraddizione di Leibniz contiene i due principi di non contraddizione
e del terzo escluso verosimilmente nel senso che soddisfa entrambe le condizioni di ve-
1 Leibniz (Essais de Thodic [1710], 169) parafrasa e cita in particolare Cic. Fat. 21: Epicure se laissa aller nier
le premier et le plus grand principe des vrits de raison; il niait que toute nonciation ft ou vraie ou fausse. []
Aprs cela, il na point besoin dtre rfut; et Chrysippe se pouvait dispenser de la peine quil prenait de confr-
mer le grand principe des contradictoires, suivant le rapport de Cicron dans son livre De fato: Contendit omnes
nervos Chrysippus ut persuadeat omne \A aut verum esse, aut falsum. [] (corsivo mio).
2 Cfr. Cic. Luc. 95: Nempe fundamentum dialecticae est, quidquid enuntietur (id autem appellant ,
quod est quasi ecfatum) aut verum esse aut falsum.
158 walter cavini
rit da essi stabilite. Ma allora occorre che, come per lAxioma contradictionis completum
di Jungius, la particella disgiuntiva sia presa in senso esclusivo:
p (Vp w Fp),
cos da escludere sia la co-verit dei disgiunti (quune proposition ne saurait tre vraie et
fausse la fois) sia la loro co-falsit (quune proposition soit ni vraie ni fausse). In questo
modo il principe de contradiction di Leibniz coinciderebbe con lAxioma contradictio-
nis completum di Jungius, e questo spiegherebbe anche la scelta di Leibniz di conservare
il nome di principio di contraddizione a un principio che apparentemente non riguar-
da la contraddizione.
Come l dei commentatori greci di Aristotele e il principium
contradictionis di Jungius, cos anche il principe de contradiction leibniziano compren-
de sia il principio di non contraddizione sia quello del terzo escluso, cio soddisfa le con-
dizioni di verit di entrambi. Ma mentre nei commentatori greci di Aristotele tale prin-
cipio generale designa sia i due principi particolari sia la loro congiunzione, e in Jungius
si articola nelle due formulazioni sintattiche afermativa e negativa, in Leibniz il princi-
pio di contraddizione un principio semantico equivalente alla congiunzione di due
principi semantici particolari, che restano anonimi, cio coincide con lAxioma contra-
dictionis completum di Jungius, ed ha a fondamento non i principi aristotelici della con-
traddizione, ma il principio stoico di bivalenza come legge fondamentale della logica.1
Tale interpretazione corroborata, a mio avviso, dalla defnizione, in apparenza
diversa, che Leibniz d del principe de la contradiction negli Essais de Thodice (1710):
T37 [] il y a deux grands principes de nos raisonnements: lun est le principe de la contradiction,
qui porte que de deux propositions contradictoires, lune est vraie, lautre fausse; lautre
principe est celui de la raison dterminante [].
Leibniz, Essais de Thodice, 442
Anche in questo caso si tratta di un principio semantico, ma riguarda espressamente due
proposizioni contraddittorie e non una singola proposizione vera o falsa. Tale principio
non altro che l aristotelico dellopposizione contraddittoria (cfr. T7), cio la rcc,
che per Aristotele, come si visto, un corollario dei due principi semantici della con-
traddizione: se i contraddittori non possono essere n insieme veri n insieme falsi, al-
lora necessariamente saranno uno vero e laltro falso. mileBoutroux(1881/1940, p. 143),
commentando il 31 della Monadologie sul principio della contraddizione in riferi-
mento al 44 della Thodice, rinvia a Cat. 10, 13b2-3 (T6), ma sospetto che Leibniz in T37
tenesse conto piuttosto di un passo del De Fato ciceroniano, opera a lui senzaltro ben
presente nella Thodice (cfr. 169 e supra, p. 158 n. 1), cio Fat. 37 (T10): Necesse est
enim in rebus contrariis duabus (contraria autem hoc loco ea dico, quorum alterum ait
quid, alterum negat), ex iis igitur necesse est invito Epicuro alterum verum esse, alte-
rum falsum. Epicuro accusato di aver negato, per salvare la libert darbitrio nel caso
1 La tesi che il principio di bivalenza o il suo rifuto siano le leggi pi fondamentali della logica sar sostenuta
in seguito da ukasiewicz, per il quale il principio di non contraddizione e del terzo escluso non sono che teore-
mi di un sistema formale di logica bivalente, la cui validit dipende appunto dal principio di bivalenza: cfr. Gou-
rinat 2006, p. 55.
2 Cfr. Risse 1964-1970, ii, pp. 187-188; Liske 2007, pp. 52-53. Lo stesso principio in Pacius 1597, p. 303a: Est autem
principium omnibus scientiis commune, Contradictionis alteram partem esse veram, alteram falsam: vedi supra,
p. 156 n. 4.
sui principi aristotelici della contraddizione 159
degli enunciati contingenti al futuro, le premier et le plus grand principe des vrits de
raison, cio quello secondo cui ogni enunciato o vero o falso (cfr. Thodice, 169, ad
Cic. Fat. 21); linvito Epicuro di Fat. 37 evidentemente richiama la sua negazione del
principio stoico di bivalenza, e questo porta a concludere che per Cicerone la negazio-
ne del principio di bivalenza equivalga a quella della rcc (alterum verum esse alterum
falsum). Leibniz seguirebbe Cicerone nel considerare equivalenti il principio stoico di
bivalenza (omne \A aut verum esse, aut falsum, aut disgiunzione esclusiva) e
la rcc, e dal momento che per Leibniz il principio stoico di bivalenza le plus grand
principe des vrits de raison e le grand principe des contradictoires, cio equivale al
principio di (o della) contraddizione, anche la rcc risulta essere, come in T37, una for-
mulazione possibile di tale principio. Il che, da un punto di vista logico, senzaltro ve-
ro: il principio leibniziano di contraddizione, cio la disgiunzione esclusiva:
p (p w p),
escludendo congiuntamente la co-verit e la co-falsit dei disgiunti, asserisce che esat-
tamente uno vero ed esattamente uno falso.1
Con Leibniz avviene perci una frattura nella venerabile e millenaria tradizione ari-
stotelica dei principi della contraddizione. Il principio leibniziano di contraddizione
contiene s i principi aristotelici dell, ma chiaramente un principio stoico
dell, cio dell vero o falso. Leibniz coniuga in questo modo la tradi-
zione aristotelica dell con quella stoica del principio forte di bi-
valenza, cio della disgiunzione esclusiva di vero o falso, ma a questultimo che asse-
gna il titolo di plus grand principe des vrits de raison, cos come Crisippo lo aveva
considerato il fondamento stesso della dialettica. E se per Aristotele assiomi della flo-
sofa prima e principi della dimostrazione sono i principi che regolano la contraddizio-
ne garantendo in questo modo la nostra conoscenza scientifca del mondo, per Leibniz
tale conoscenza garantita invece dal principio stoico di bivalenza per le verit di ra-
gione e dal principio di ragione sufciente o determinante per le verit di fatto.2
Il passo successivo, che render insanabile la frattura con la tradizione aristotelica dei
principi della contraddizione e ne decreter in qualche modo la fne, sar compiuto di l
a poco dagli eredi di Leibniz dellIlluminismo tedesco, in particolare da Christian Wolf
e dal suo allievo Alexander Baumgarten. A essi si deve lintroduzione e la difusione dei
nomi, divenuti poi canonici, del principio di non contraddizione e di quello del terzo
escluso. Scrive per esempio Wolf nella sua Philosophia Prima, sive, Ontologia (1729):
T38 Propositio haec; Fieri non potest, ut idem simul sit & non sit, dicitur Principium Contradi -
ctionis, ob rationem mox adducendam. Principium autem Contradictionis jam olim adhibuit
Aristoteles eodemque usi sunt Scholastici in philosophia prima instar axiomatis maxime generalis.
Wolff, Philosophia Prima, 293
1 Cfr. Christian Wolff, Philosophia Rationalis sive Logica (1728), 532: Propositionum contradictoriarum altera
necessario vera; altera necessario falsa; Philosophia Prima sive Ontologia (1729), 53: Coincidit haec propositio [scl.
pte] cum altera in Logicis demonstrata [], quod propositionum contradictoriarum altera necessario vera, altera ne-
cessario falsa. Communiter etiam dicitur, inter contradictoria non dari medium.
2 Cfr. Leibniz, Nouveaux Essais, iv ii 1; Monadologie, 31-32.
3 Gi nella cosiddetta Metafsica Tedesca (cio nei Vernnftige Gedanken von Gott, der Welt und der Seele des
Menschen, auch allen Dingen berhaupt) del 1719, Wolf scriveva ( 10): Und solchergestalt rumen wir berhaupt
ohne einiges Bedenken diesen allgemeinen Satz ein: Es kan etwas ni cht zugl ei ch seyn und auch ni cht
seyn. Diesen Satz nennen wir den Gr und des Wi derspr uchs []. Rafaele Ciafardone, nella sua traduzione
160 walter cavini
T39 Quodsi quis demonstrationem propositionis, quod propositionum contradictoriarum altera sit
necessario vera, altera necessario falsa, quam dedimus (. 532 Log.), perpendere voluerit; is non
minus animadverteret, eandem in principium contradictionis resolvi. Quoniam vero pro-
positio, Quodlibet est vel non est, aut, si mavis, principium exclusi medii inter duo contradictoria
ex principio contradictionis deduci potest, ideo illo posito, hoc una ponitur.
Wolff, Philosophia Prima, 54
Pi concisamente e formaliter Baumgarten nella sua Metaphysica (1739):
T40 seu, praedicatorum contradictoriorum nullum est subiectum, seu, nihil est, et non est.
0=A+non A. Haec propositio dicitur principium contradictionis & absolute primum.
Baumgarten, Metaphysica, 7
T41 Ergo omne possibile aut est A, aut non A, seu, omni subiecto ex omnibus praedicatis contra-
dictoriis alterutrum convenit. Haec propositio dicitur principium exclusi tertii, seu medii, inter
duo contradictoria.
Baumgarten, Metaphysica, 10
ormai evidente che i principi che ancora Jungius considerava principia contradictionis,
ovvero due formulazioni distinte, afermativa e negativa, del principium contradictionis,
si sono in qualche modo separati e hanno assunto ciascuno per la prima volta un nome
proprio. Wolf chiama principium contradictionis quella che in Jungius era la formu-
lazione negativa diretta del principium contradictionis, cio Fieri non potest, ut idem si-
mul sit & non sit; mentre chiama principium exclusi medii inter duo contradictoria
quella che in Jungius era la formulazione afermativa diretta del principium contradictio-
nis, cio Quodlibet est vel non est. Il titolo onorifco di principium contradictionis spet-
ta ormai soltanto al principio di non contraddizione; e, mentre in Aristotele, come si
visto,1 i due principi della contraddizione risultavano interderivabili e la loro congiun-
zione aveva come corollario l dellopposizione contraddittoria, cio la rcc, in
Wolf si sottolinea invece lassoluta priorit logica del principium contradictionis (cio del
principio di non contraddizione) sia rispetto al principio del terzo escluso sia rispetto al-
la rcc. Questultima si risolve nel principio di non contraddizione (quod propositionum
contradictoriarum altera sit necessario vera, altera necessario falsa [] in principium con-
tradictionis resolvi), poich dal principio di non contraddizione si pu dedurre il prin-
cipio del terzo escluso (propositio, Quodlibet est vel non est, aut, si mavis, principium ex-
clusi medii inter duo contradictoria ex principio contradictionis deduci potest), ma non
viceversa. Infatti per Wolf i due principi non sono interderivabili, in quanto la deriva-
zione del principio di non contraddizione dal principio del terzo escluso presupporreb-
be inevitabilmente il principio di non contraddizione e quindi si risolverebbe in una pe-
tizione di principio.2
italiana della Metafsica Tedesca (1999), traduce Diesen Satz nennen wir den Gr und des Wi derspr uchs con
Chiamiamo questo principio principio di non contraddizione, ma Diesen Satz si riferisce alla proposizione
universale enunciata in precedenza, e quindi va tradotto con questa proposizione e non con questo principio;
e soprattutto Gr und des Wi derspr uchs chiaramente la traduzione tedesca di principium contradictionis e
quindi non va tradotto con principio di non contraddizione (locuzione ancora ignota a Wolf e che, come vedremo,
attestata solo a partire dai primi dellOttocento), ma con principio di (o della) contraddizione.
1 Cfr. supra, p. 149 n. 1.
2 Cfr. Wolff, Ontologia, 54, p. 37: Enimvero si vim consecutionis, qua principium contradictionis ex princi-
pio exclusi medii inter duo contradictoria infertur, ad vivum reseces; attendenti constabit, committi utique circu-
lum in demonstrando, hoc est, principium contradictionis, quod inferri debebat, revera supponi, ut inferri possit.
sui principi aristotelici della contraddizione 161
chiaro dunque perch i principi aristotelici della contraddizione si risolvano per
Wolf in un unico principio, quello di non contraddizione, e perch solo a tale princi-
pio spetti il titolo onorifco di principium contradictionis (contradictionis genitivo di ar-
gomento), mentre per il principio del terzo escluso si debba coniare un nome diverso e
ad hoc. Lenfasi posta da Aristotele in Metaph. 3 sulla priorit epistemica del principio
di non contraddizione rispetto agli altri principi dimostrativi (il suo essere il principio
pi certo di tutti) e sulla sua priorit apodittica (il suo essere la credenza ultima pre-
supposta da ogni dimostrazione in quanto per natura principio anche di tutti gli altri
assiomi), diviene per Wolf anzitutto una priorit logica: dal principio di non contrad-
dizione si possono dedurre anche gli altri principi, ma non viceversa, se non si vuole ca-
dere in una petizione di principio.
Ma lepilogo della storia dei principi aristotelici della contraddizione non dato sol-
tanto dalla loro separazione nella prima met del Settecento in due distinti principi, di
(o della) contraddizione e del terzo escluso, ma si prolunga almeno fno alla prima me-
t dellOttocento con una innovazione lessicale, minima ma importante, di cui siamo
divenuti inconsapevoli eredi e che quindi ci riguarda direttamente.
Nella prima met dellOttocento, il flosofo scozzese William Hamilton (1788-1856)
scriveva nelle sue Lectures on Metaphysics, pubblicate postume fra il 1858 e il 1860:
T42 The highest of all logical laws, in other words, the supreme law of thought, is what is cal-
led the principle of Contradiction, or more correctly the principle of Non-Contradiction [corsi-
vo mio]. It is this: A thing cannot be and not be at the same time, Alpha est, Alpha non
est, are propositions which cannot both be true at once. A second fundamental law of
thought, or rather the principle of Contradiction viewed in a certain aspect [corsivo mio], is cal-
led the principle of Excluded Middle between two Contradictories. A thing either is or it is
not, Aut est Alpha aut non est; there is no medium; one must be true, both cannot.
Hamilton, Lectures on Metaphysics, xxxviii (vol ii, p. 368)
Linteresse e la novit di questo testo risiedono, a mio avviso, nella precisazione or mo-
re correctly the principle of Non-Contradiction, che per la prima volta, fra i testi fn
qui citati, introduce il termine principio di non-contraddizione (con o senza trattino),
termine che per noi, a diferenza di Hamilton, curiosamente ormai sinonimo di prin-
cipio di contraddizione. Curiosamente,1 perch chiaro che fra i due termini esiste una
profonda diferenza grammaticale, che anche una profonda diferenza di signifcato.
1 Anche senza conoscere il retroscena storico che si cercato fn qui di ricostruire, curioso che si accetti
comunemente di chiamare lo stesso principio logico con due nomi che risultano in realt contraddittori (di
contraddizione vs di non contraddizione), come se, soprattutto in logica, la presenza o meno di una negazione
fosse solo una variante stilistica: cfr. Cavini 2007, p. 10. Francesco Berto (2006, p. 42 n. 2) ha ora il merito di se-
gnalare almeno in una nota questa curiosa e contraddittoria polinomia: Curiosamente, il (pnc) viene chiamato
Principio di Contraddizione, oltre che di Non-Contraddizione.
Una correzione analoga a quella di Hamilton anche in Kneale 1962, p. 357 n. 2: Law of non-contradiction
would be a better name, malgrado in precedenza gli autori abbiano usato regolarmente Law of Contradiction
(cfr. pp. 11 e 46, ma a p. 168 principle of non-contradiction). Anche Longo sottolinea la novit della denominazio -
ne recente del principio: nelle denominazioni antiche non c la negazione (principio di non contraddizione). Lag-
giunta di essa marca il carattere normativo negativo assunto da tale principio [], rispetto a un carattere iniziale
piuttosto descrittivo, che prende atto di unimpossibilit pi che esprimere un divieto (Longo 2005, p. 89; cfr.
Lemaire 2005, p. 288 n. 7). Ma se la denominazione moderna ha un carattere normativo negativo (nel senso di desi-
gnare una legge che vieta la contraddizione), quella antica non ha un carattere descrittivo (lassioma o il principio
della contraddizione non descrive la contraddizione, ma ne declina le propriet). Correttamente invece Longo2004,
p. 88, parla di assiomi che in generale riguardano la contraddizione ( genitivo di argomento).
162 walter cavini
Nella espressione principio di (o della) contraddizione il sintagma di (o della)
contraddizione non pu essere inteso come un complemento di specifcazione epese-
getica o dichiarativa, come invece, per esempio, il sintagma della temperanza
nellespressione la virt della temperanza, perch la temperanza una virt, ma la
contraddizione non un principio; deve essere inteso invece, come gi i genitivi
e contradictionis in e principium contradictionis,
come complemento di argomento, analogamente, per esempio, allespressione assio-
ma delle parallele: il principio di (o della) contraddizione il principio che riguarda la
contraddizione, cos come lassioma delle parallele lassioma che riguarda le parallele.
Al contrario, nellespressione principio di (o della) non contraddizione il sintagma
di (o della) non contraddizione va inteso come complemento di specifcazione epese-
getica o dichiarativa, perch la non contraddizione appunto un principio, analoga-
mente al sintagma del terzo escluso nellespressione principio del terzo escluso, per-
ch il terzo escluso anchesso un principio. La precisazione di Hamilton si pu
considerare un esempio di quella che i flologi chiamano interferenza mentale:1 la
possibile ambiguit grammaticale e di signifcato insita nel termine tradizionale prin-
cipio di contraddizione lo porta a correggerlo disambiguandolo con una negazione, in
analogia con la grammatica e il signifcato dellaltro termine ormai tradizionale prin-
cipio del terzo escluso.
Ma vi un altro motivo di interesse per noi nel passo di Hamilton, che merita di es-
sere segnalato, e cio il fatto che Hamilton considera il principio del terzo escluso non
come una seconda legge fondamentale del pensiero, ma piuttosto come il principio
di Contraddizione visto sotto un certo aspetto. Questo spiega forse anche il suo erro-
re di attribuire al principio del terzo escluso la stessa valenza logica del principio di non
contraddizione, cio: Una cosa o o non , []; non vi alcun medio; una deve esse-
re vera, entrambe non possono <essere vere> [corsivo mio], mentre, come si visto, avreb-
be dovuto dire che per il principio del terzo escluso entrambe non possono essere false.
Tuttavia, al di l di questo forse non banale logical slip, signifcativo il fatto che Ha-
milton da un lato assegni la priorit fra le leggi del pensiero al principio di non con-
traddizione (come gi aveva fatto Wolf seguendo Aristotele), dallaltro consideri il
principio del terzo escluso solo una variante del primo (cos come Jungius aveva consi-
derato i due principi due varianti, rispettivamente afermativa e negativa, dello stesso
principium contradictionis o ).
Ma il legame di Hamilton con la tradizione dei principi aristotelici della contraddi-
zione emerge ancora pi chiaramente in una appendice alla Lecture sopra citata, scritta
nel 1855, poco prima della morte. In essa Hamilton cos compendia il suo pensiero sul-
la dottrina della contraddizione:
T43 The doctrine of Contradiction, or of Contradictories, ( ), that
Afrmation or Negation is a necessity of thought, whilst Afrmation and Negation are
incompatible, is developed into three sides or phases, each of which implies both the others,
phases which may obtain, and actually have received, severally, the name of Law, Principle,
or Axiom. Neglecting the historical order in which these were scientifcally named and
articulately developed, they are:
1, The Law, Principle, or Axiom, of Identity, which, in regard to the same thing, imme-
diately or directly enjoins the afrmation of it with itself, and mediately or indirectly pro-
hibits its negation: (A is A).
1 Devo questo suggerimento a Simonetta Nannini, che ringrazio.
sui principi aristotelici della contraddizione 163
2, The Law, &c. of Contradiction, (properly Non-contradiction), which, in regard to con-
tradictories, explicitly enjoining their reciprocal negation, implicitly prohibits their recip-
rocal afrmation: (A is not Not-A.) [].
3, The Law, &c. of Excluded Middle or Third, which declares that, whilst contradictories
are only two, everything, if explicitly thought, must be thought as of these either the one
or the other: (A is either B or Not-B.)
A parte il riferimento obbligato per un logico dellOttocento al principio di identit,1 e
la formulazione kantiana del principio di non contraddizione: A non non-A,2 questo
testo pu considerarsi un compendio della tradizione dei principi aristotelici della con-
traddizione, cos come siamo venuti ricostruendo fn qui: da un lato, la dottrina della
contraddizione o dei contraddittori, per la quale Hamilton evoca esplicitamente
l dei commentatori greci di Aristotele (citazione che poteva ora
trarre direttamente dal commento di Alessandro di Afrodisia alla Metafsica edito da
Hermann Bonitz nel 1847); dallaltro, i due principi, di non contraddizione (anche in
questo caso correggendo il nome tradizionale) e del terzo escluso, come lati o fasi di
tale dottrina, insieme al principio di identit, che tuttavia ne entra a far parte solo con
Wolf e meriterebbe una storia a s.
Cos si pu dire conclusa la tradizione dell o dei principi
aristotelici della contraddizione, e si pu, e forse si deve, senzaltro accettare, come si
fatto tacitamente fn qui, la correzione di Hamilton e considerare come non sinonimi
principio di (della) contraddizione e principio di non contraddizione. La contraddi-
zione non un principio, almeno per Aristotele e per la logica (e la metafsica) classica,
mentre la non contraddizione come il terzo escluso sono insieme principi della con-
traddizione, cio assiomi della opposizione contraddittoria, e principi della dimostra-
zione, cio assiomi della flosofa prima. Fin qui ci siamo occupati solo del primo aspet-
to. tempo ora di considerare i principi della contraddizione come principi della
dimostrazione e assiomi della flosofa prima.3
Universit di Bologna
1 Sui principi di identit, non contraddizione e terzo escluso in Hamilton, cfr. Raspa 1999, pp. 84-89, che tutta-
via non nota la correzione proposta da Hamilton di principio di contraddizione in principio di non-contraddi-
2 Sul principio di non contraddizione in Kant cfr. in particolare Berti 1987, pp. 145-175; Raspa 1999, pp. 62-79;
Capozzi 2002, pp. 454-459. Per un precedente aristotelico della formulazione kantiana del pnc (A non non-A)
vedi APo. i 11, 77a18.
3 Gli incunaboli di questo lavoro risalgono al xiii Symposium Aristotelicum (Pontignano 1993) sul De Interpre-
tatione, purtroppo tuttora inedito, e alla mia discussione del paper di Jonathan Barnes sui capitoli 4-6 dellopera
aristotelica. In seguito ho avuto modo di presentare e discutere parti di questo lavoro in vari convegni e seminari,
in particolare alle Universit di Cambridge (1997), Salonicco (1997), Venezia (1998), Macerata (2001), San Paolo del
Brasile (2002), Liegi (2004), Venezia (2005), Lille iii (2005) e allIstituto Veritatis Splendor di Bologna (2005). La
discussione coi partecipanti a questi incontri stata uno stimolo importante a chiarire e precisare i temi trattati.
Ringrazio soprattutto Jonathan Barnes, David Charles, Mary Louise Gill, Marco Zingano, Franco Ferrari, Carlo
Natali, Andr Laks e Fabienne Blaise per le domande, i dubbi, gli utili suggerimenti; Giuseppe Cambiano, Carlotta
Capuccino, Luca Castagnoli, Francesco Fronterotta, Walter Leszl, Simonetta Nannini, Andrea Piatesi, Luciana
Repici, Mario Vegetti e un anonimo referee per aver letto con attenzione e abnegazione alcune stesure intermedie
di questo lavoro, suggerendomi vari approfondimenti e salvandomi da non poche sviste e oscurit; infne Vassilis
Karasmanis, Juliette Lemaire e Ccile Wartelle per avermi dato la possibilit di leggere le loro tesi di dottorato, e
Bob Sharples per avermi permesso di citare un suo inedito.
164 walter cavini
Ackrill 1963 J. L. Ackrill (a cura di), Aristotles Categories and De Interpretatione,
Oxford, Clarendon Press, 1963.
Antonini 1994 Francesca Antonini (a cura di), Marco Tullio Cicerone: Il fato, Milano, bur,
Barnes 1975 Jonathan Barnes (a cura di), Aristotles Posterior Analytics, Oxford, Cla-
rendon Press, 1975.
Barnes 1993 Jonathan Barnes (a cura di), Aristotle: Posterior Analytics, Second Edition,
Oxford, Clarendon Press, 1993.
Barnes 2007 Jonathan Barnes, Truth, etc.: Six Lectures on Ancient Logic, Oxford, Cla-
rendon Press, 2007.
Baumgarten 1739 Alexandri Gottlieb Baumgarten Metaphysica (1739), Editio vii,
Halae Magdeburgicae, Carol. Herman. Hemmerde, 1779 (rist. anast.
Hildesheim-New York, Olms, 1982).
Berti 1987 Enrico Berti, Contraddizione e dialettica negli antichi e nei moderni,
Palermo, LEpos, 1987.
Berti 2002 Enrico Berti, Il principio di non contraddizione: Storia e signifcato, in
Pietro Bria e Fiorangela Oneroso (a cura di), Bi-logica e sogno: Sviluppi
matteblanchiani sul pensiero onirico, Milano, Franco Angeli, 2002, pp.
Berto 2006 Francesco Berto, Teorie dellassurdo: I rivali del Principio di Non-Con-
traddizione, Prefazione di Graham Priest, Roma, Carocci, 2006.
Bobzien 1998 Susanne Bobzien, Determinism and Freedom in Stoic Philosophy, Oxford,
Clarendon Press, 1998.
Bods 2002 Richard Bods (a cura di), Aristote: Catgories (2001), Paris, Les Belles
Lettres, 20022.
Bonelli 2001 Maddalena Bonelli, Alessandro di Afrodisia e la metafsica come scienza
dimostrativa, Napoli, Bibliopolis, 2001.
Bonitz 1870 Hermann Bonitz, Index Aristotelicus, Aristotelis Opera, vol. v, Berolini,
Academia Regia Borussica, 1870.
Boutroux 1881/1940 mile Boutroux (a cura di), Leibnitz [sic]: La Monadologie, Paris, Librairie
Delagrave, 1881 / traduzione italiana di Ernesto Codignola, G. G. Leib-
niz: La Monadologia, Firenze, La Nuova Italia, 1940.
Calcante 1992 Cesare Marco Calcante (a cura di), Cicerone: La natura divina, Milano,
bur, 1992.
Capozzi 2002 Mirella Capozzi, Kant e la logica, vol. i, Napoli, Bibliopolis, 2002.
Cassin e Narcy 1989 Barbara Cassin e Michel Narcy (a cura di), La dcision du sens: Le livre
Gamma de la Mtaphysique dAristote, Paris, Vrin, 1989.
Castagnoli 2005 Luca Castagnoli, The Logic of Ancient Self-Refutation: From Democritus
to Augustine, Ph.D. Thesis, Cambridge, 2005.
Castagnoli 2007 Luca Castagnoli, Everything is True, Everything is False: Self-Refuta-
tion Arguments from Democritus to Augustine, Antiqvorvm Philosophia,
1 (2007), pp. 11-74.
Cavini 1991 Walter Cavini, Semantica fondamentale: Note sul 44 di Essere e Tem-
po, in Gabriele Usberti (a cura di), Problemi fondazionali nella teoria del
signifcato, Firenze, Olschki, 1991, pp. 35-48.
Cavini 1993a Walter Cavini, Aristotele e i modi del paradosso, in A. M. Battegazzore
(a cura di), Dimostrazione, argomentazione dialettica e argomentazione
retorica nel pensiero antico, Genova, Sagep, 1993, pp. 61-81.
sui principi aristotelici della contraddizione 165
Cavini 1993b Walter Cavini, Chrysippus on Speaking Truly and the Liar, in Klaus
Dring e Theodor Ebert (a cura di), Dialektiker und Stoiker: Zur Logik der
Stoa und ihrer Vorlufer, Stuttgart, Steiner, 1993, pp. 85-109.
Cavini 1998 Walter Cavini, Arguing from a Defnition: Aristotle on Truth and the Ex-
cluded Middle, in Nikolaos Avgelis e Filimon Peonidis (a cura di), Aristo-
tle on Logic, Language and Science, Thessaloniki, Sakkoulas, 1998, pp. 3-15.
Cavini 2007 Walter Cavini, Il principio di non contraddizione in Aristotele, in Alberto
Strumia (a cura di), Il problema dei fondamenti da Aristotele a Tommaso
dAquino allontologia formale, Siena, Cantagalli, pp. 9-20.
Chiaradonna 2005 Riccardo Chiaradonna, I contrari e i termini intermedi, in Bruno Centro-
ne (a cura di), Il libro Iota (x) della Metafsica di Aristotele, Sankt Augustin,
Academia, 2005, pp. 157-169.
Ciafardone 1999 Rafaele Ciafardone (a cura di), Christian Wolf: Metafsica Tedesca, Mila-
no, Rusconi, 1999.
Code 1987 Alan Code, Metaphysics and Logic, in Monan Matthen (a cura di),
Aristotle Today: Essays on Aristotles Ideal of Science, Edmonton, Alberta,
Academic Printing and Publishing, pp. 127-149.
Colli 1955 Giorgio Colli (a cura di), Aristotele: Organon, Torino, Einaudi, 1955 (rist.
anast. Milano, Adelphi, 2003).
Cranz 1958 F. Edward Cranz, The Prefaces to the Greek Editions and Latin Transla-
tions of Alexander of Aphrodisias, 1450 to 1575, Proceedings of the Ameri-
can Philosophical Society, 102 (1958), pp. 510-546.
Crivelli 2004a Paolo Crivelli, Aristotle on Truth, Cambridge, Cambridge u.p., 2004.
Crivelli 2004b Paolo Crivelli, Aristotle on the Liar, Topoi, 23 (2004), pp. 61-70.
Del Re 1976 Rafaello Del Re (a cura di), Marco Tullio Cicerone: Le Dispute Accademi-
che, Milano, Mondadori, 1976.
Dufour 2004 Richard Dufour (a cura di), Chrysippe: uvre Philosophique, 2 tomi, Paris,
Les Belles Lettres, 2004.
Dummett 1959/1978 Michael Dummett, Truth (1959), in Id., Truth and Other Enigmas, Lon-
don, Duckworth, 1978, pp. 1-24.
Dummett 1973 Michael Dummett, Frege: Philosophy of Language, London, Duckworth,
Dummett 1976/1993 Michael Dummett, What is a Theory of Meaning? (ii) (1976), in Id., The
Seas of Language, Oxford, Clarendon Press, 1993, pp. 34-93.
Esposito 2007 Costantino Esposito (a cura di), Francisco Surez: Disputazioni Metafsiche
(1996), Nuova edizione riveduta e ampliata, Milano, Bompiani, 2007.
Fait 2007 Paolo Fait (a cura di), Aristotele: Le confutazioni sofstiche, Roma-Bari, La-
terza, 2007.
fds Karlheinz Hlser, Die Fragmente zur Dialektik der Stoiker: Neue Samm-
lung der Texte mit deutscher bersetzung und Kommentaren, 4 voll., Stutt-
gart-Bad Cannstatt, Frommann-Holzboog, 1987.
Frede 1970 Dorothea Frede, Aristoteles und die Seeschlacht: Das Problem der Con-
tingentia Futura in De Interpretatione 9, Gttingen, Vandenhoeck & Ru-
precht, 1970.
Frede 1985 Dorothea Frede, The Sea-Battle Reconsidered: A Defence of the Traditio-
nal Interpretation, Oxford Studies in Ancient Philosophy, 3 (1985), pp.
Gaskin 1995 Richard Gaskin, The Sea Battle and the Master Argument: Aristotle and
Diodorus Cronus on the Metaphysics of the Future, Berlin-New York, de
Gruyter, 1995.
Gobbo 1997 Elena Gobbo, La concezione diretta dell negli Analitici primi e
nelle Confutazioni sofstiche, Elenchos, xviii (1997), pp. 311-357.
166 walter cavini
Gourinat 2001 Jean-Baptiste Gourinat, Principe de contradiction, principe du tiers-
exclu et principe de bivalence: Philosophie premire ou organon?, in M.
Bastit e J. Follon (a cura di), Logique et mtaphysique dans lOrganon
dAristote, Louvain-La-Neuve ecc., Peeters, 2001, pp. 63-91.
Gourinat 2006 Jean-Baptiste Gourinat, Lhistoire du principe de bivalence selon
ukasiewicz, in Roger Pouivet e Manuel Rebuschi (a cura di), La
philosophie en Pologne: 1919-1939, Paris, Vrin, 2006, pp. 37-66.
Hamilton 1861-1866 William Hamilton, Lectures on Metaphysics and Logic, a cura di H.
L. Mansel e John Veitch, 4 voll., Edinburgh-London, Blackwood,
1861-18662 (rist. anast. Stuttgart-Bad Cannstatt, Frommann-Holzbo-
og, 1969-1970).
Heidegger 1925/26 Martin Heidegger, Logik: Die Frage nach der Wahrheit (Marburger
Vorlesung, Wintersemester 1925/26), Gesamtausgabe, vol. 21, a cura
di Walter Biemel, Frankfurt am Main, Klostermann, 1977.
Ildefonse e Lallot 2002 Frdrique Ildefonse e Jean Lallot (a cura di), Aristote: Catgories,
Paris, Seuil, 2002.
Jungius 1638 Logica Hamburgensis [] Autore Joachimo Jungio, Hamburgi, Literis
Rebenlinianis [Jacob Rebenlin], m.dc.xxxviii.
Kneale 1962/1972 William Kneale e Martha Kneale, The Development of Logic,
Oxford, Clarendon Press, 1962 / traduzione italiana di Amedeo G.
Conte e Luca Cafero, Storia della logica, Torino, Einaudi, 1972.
Knne 2003 Wolfgang Knne, Conceptions of Truth, Oxford, Clarendon Press,
Lemaire 2005 Juliette Lemaire, La contradiction chez Aristote: Analyse et problmes,
Thse de Doctorat, Universit Paris x-Nanterre, 2005.
Liske 2007 Michael-Thomas Liske, Gottfried Wilhelm Leibniz, Mnchen, Beck,
2000 (traduzione italiana di Paolo Rubini, Leibniz, Bologna, il Muli-
no, 2007).
Longo 2004 Angela Longo, Siriano e i precedenti pre-aristotelici del principio della
contraddizione, Documenti e studi sulla tradizione flosofca medie-
vale, xv (2004), pp. 81-97.
Longo 2005 Angela Longo, Siriano e i principi della scienza, Prefazione di J. Bar-
nes, Napoli, Bibliopolis, 2005.
lsj Henry George Liddell, Robert Scott, Henry Stuart Jones, A
Greek-English Lexicon (1843), Oxford, Clarendon Press, 19409.
ukasiewicz 1910a/2000 Jan ukasiewicz, O zasadzie sprzecznosci u Arystotelesa (1910) / Du
principe de contradiction chez Aristote, traduit du polonais par Dorota
Sikora, prface de Roger Pouivet, Paris, Lclat, 2000.
ukasiewicz 1910a/2003 Jan ukasiewicz, O zasadzie sprzecznosci u Arystotelesa (1910) / Del
principio di contraddizione in Aristotele (1910), traduzione dal polacco
di Grazyna Maszkowska, a cura di Gabriele Franci e Claudio Anto-
nio Testi, presentazione di Maurizio Matteuzzi, Macerata, Quodli-
bet, 2003.
ukasiewicz 1910b/2000 Jan ukasiewicz, ber den Satz von Widerspruch bei Aristoteles, Bul-
letin international de lAcadmie des Sciences de Cracovie, Classe
de Philosophie (1910), pp. 15-38 / traduzione italiana di Venanzio Ra-
spa, Sul principio di contraddizione in Aristotele, Paradigmi, xviii
(2000), pp. 389-410.
ukasiewicz 1921/1970 Jan ukasiewicz, Logika dwuwartos ciowa, Przeglad Filozofczny,
13 (1921), pp. 189-205 / traduzione inglese di O. Wojtasiewicz, Two-va-
lued Logic, in Jan ukasiewicz, Selected Works, a cura di L. Borkow-
ski, Amsterdam-Varsavia, North Holland-pwn, 1970, pp. 89-109.
sui principi aristotelici della contraddizione 167
ukasiewicz 1930/1970 Jan ukasiewicz, Philosophische Bemerkungen zu mehrwertigen Syste-
men des Aussagenkalkls, Comptes Rendus des Scances de la Soci-
t des Sciences et des Lettres de Varsovie, Cl. iii, 23 (1930), pp. 51-77
/ traduzione inglese di H. Weber, Philosophical Remarks on Many-Va-
lued Systems of Propositional Logic (1967), ripresa in Jan ukasiewicz,
Selected Works, a cura di L. Borkowski, Amsterdam-Varsavia, North
Holland-pwn, 1970, pp. 153-178.
Magris 1994 Aldo Magris (a cura di), Marco Tullio Cicerone: De fato / Sul destino,
Milano, Mursia, 1994.
Maier 1896-1900 Heinrich Maier, Die Syllogistik des Aristoteles, 3 voll., Tubingen,
Laupp, 1896-1900.
McGinn 2000 Colin McGinn, Logical Properties: Identity, Existence, Predication,
Necessity, Truth, Oxford, Clarendon Press, 2000.
Meyer 1957 Rudolf W. Meyer (a cura di), Joachimi Jungii Logica Hamburgensis,
Hamburg, Augustin, 1957.
Mignucci 1975 Mario Mignucci, Largomentazione dimostrativa in Aristotele: Com-
mento agli Analitici secondi, vol. i, Padova, Antenore, 1975.
Mignucci 1996 Mario Mignucci, Consistenza e contraddizione in Aristotele, Diano-
ia, 1 (1996), pp. 53-60.
Mignucci 2003a Mario Mignucci, I principi delle dimostrazioni: Unanalisi della secon-
da aporia di Metafsica B, in Vincenza Celluprica (a cura di), Il libro B
della Metafsica di Aristotele, Napoli, Bibliopolis, 2003, pp. 75-101.
Nasti De Vincentis 2002 Mauro Nasti De Vincentis, Logiche della connessivit: Fra logica
moderna e storia della logica antica, Bern ecc., Haupt, 2002.
Nuchelmans 1973 Gabriel Nuchelmans, Theories of the Proposition: Ancient and
Medieval Conceptions of the Bearers of Truth and Falsity, Amsterdam-
London, North Holland, 1973.
Oehler 1984 Klaus Oehler (a cura di), Aristoteles: Kategorien, Berlin: Akademie-
Verlag, 1984.
Pacius 1597 Julius Pacius, In Porphyrii Isagogen et Aristotelis Organum commenta-
rius, Francofurti, Apud Heredes Andreae Wecheli, mdxcvii (rist.
anast. Hildesheim: Olms, 1966).
Paolillo 1966 M. Paolillo (a cura di), Cicerone: De fato, Firenze, Le Monnier, 1966.
Pesce 1970 Domenico Pesce (a cura di), Cicerone: Il fato, Padova, Liviana, 1970.
Prantl 1855-1870 Carl Prantl, Geschichte der Logik im Abendlande, 4 voll., Leipzig,
Hirzl, 1855-1870 (rist. anast. Graz, Akademische Druck- u. Verlagsan-
stalt, 1955).
Priest 2004 Graham Priest et alii (a cura di), The Law of Non-Contradiction: New
Philosophical Essays, Oxford, Clarendon Press, 2004.
Quine 1950/1960 W. V. Quine, Methods of Logic, New York, Holt, 1950 / traduzione
italiana di Michele Pacifco, Manuale di logica, Milano, Feltrinelli,
Quine 1976 W. V. Quine, The Ways of Paradox and Other Essays (1966), Revised
and enlarged edition, Cambridge, ma, Harvard u.p., 1976.
Raspa 1999 Venanzio Raspa, In-contraddizione: Il principio di contraddizione alle
origini della nuova logica, Trieste, Parnaso, 1999.
Reale 2004 Giovanni Reale (a cura di), Aristotele: Metafsica, Milano, Bompiani,
Risse 1964-1970 Wilhelm Risse, Die Logik der Neuzeit, 2 voll., Stuttgart-Bad Can-
nstatt, Frommann, 1964-1970.
Ross 1908 W. D. Ross (a cura di), The Works of Aristotle Translated into English,
vol. viii: Metaphysica, Oxford, Clarendon Press, 1908 (19282).
168 walter cavini
Ross 1949 W. D. Ross (a cura di), Aristotles Prior and Posterior Analytics, Oxford, Cla-
rendon Press, 1949.
Russo 1973 Antonio Russo (a cura di), Aristotele: Metafsica, Bari, Laterza, 1973.
Seel 2001 Gerhard Seel (a cura di), Ammonius and the Seabattle: Texts, Commentary
and Essays, Berlin-New York, de Gruyter, 2001.
Severino 2005 Emanuele Severino, Fondamento della contraddizione, Milano, Adelphi,
Sharples 2003 R. W. Sharples, Ammonius and the Sea-Battle: Standard and Non-Standard
Interpretations (2003), inedito.
Suber 2002 Peter Suber, Non-Contradiction and Excluded Middle, www.earlham.
svf Ioannes ab Arnim, Stoicorum Veterum Fragmenta, 3 voll., Lipsiae, Teub-
ner, 1903-1905; vol. 4: Indices, a cura di Max Adler, Lipsiae, Teubner, 1924.
Talanga 1896 Josip Talanga, Zukunftsurteile und Fatum: Eine Untersuchung ber Aristo-
teles De interpretatione 9 und Ciceros De fato, mit einem berblick ber die
sptantiken Heimarmene-Lehren, Bonn, Habelt, 1986.
Varzi 2004 Achille C. Varzi, Conjunction and Contradiction, in Priest 2004, pp. 93-
Viano 1974 Carlo Augusto Viano (a cura di), La Metafsica di Aristotele, Torino, utet,
Wedin 2004 Michael V. Wedin, Aristotle on the Firmness of the Principle of Non-Con-
tradiction, Phronesis, xlix (2004), pp. 225-265.
Weidemann 2002 Hermann Weidemann (a cura di), Aristoteles: Peri Hermeneias (1994),
Berlin, Akademie Verlag, 20022.
Whitaker 1996 C. W. A. Whitaker, Aristotles De Interpretatione: Contradiction and
Dialectic, Oxford, Clarendon Press, 1996.
Wolff 1729 Christiani Wolffii Philosophia prima, sive, Ontologia(1729), Editio nova,
priori emendatior, Francofurti, Ofcina Libraria Rengeriana, 1736 (rist.
anast. Hildesheim, G. Olms, 1962).
Zanatta 1989 Marcello Zanatta (a cura di), Aristotele: Le categorie, Milano, bur, 1989.
Zanatta 1992 Marcello Zanatta (a cura di), Aristotele: Dellinterpretazione, Milano, bur,
sui principi aristotelici della contraddizione 169
the arguments APA TO XHMA TH EE
in aristotle

s sophistical refutations*
Ermelinda Valentina Di Lascio
1. Introduction

s treatment of the fallacy of (form of expression) in

the Sophistical Refutations (Sophistici Elenchi, hereafter se) is especially intriguing
and has been thought to occupy a particular place in his broader discussion of the
linguistic fallacies.1 Various qualms about Aristotles treatment seem to emerge in the
scholarly literature:
1. The fallacy of form of expression confates what are actually quite diferent kinds
of arguments;2
2. Although catalogued by Aristotle as a linguistic fallacy, it deals at least as much with
ontology as with language;3
3. When it does deal with language, actually it often amounts to a form of ambigui-
ty, i.e. something very similar to the fallacies of homonymy and amphiboly.4
These diferent but related views seem to suggest the same general conclusion:
Aristotles classifcation, at least as far as linguistic fallacies are concerned, is signif-
cantly defective. Nonetheless, scholars also claim that the fallacy of form of expression
is of particular interest for Aristotle and for us.
In this article I shall illustrate Aristotles treatment of this fallacy in se (se 4, 166b10-
19 and se 22) and analyse how the arguments classifed under it work. This analysis will
lead us to reject the three abovementioned criticisms: the fallacy includes just one kind
of argument, is correctly classifed as linguistic, and occupies its own unique position
in the group of linguistic fallacies (it does not overlap with any of the others).
I shall also consider one further thesis which seems to be advanced, or at least im-
plied, in all the studies I am aware of:
* The content of this article is part of my Ph.D. research in the Faculty of Classics, Cambridge University. My
deepest thanks go to Nick Denyer for supervising my work with great skill and patience. I owe many thanks also
to Luca Castagnoli and Geofrey Lloyd for reading and discussing a version of this article with close attention, and
to David Sedley for very useful comments. I am also grateful to Paolo Fait, who kindly allowed me to read his
commentary on the Sophistical Refutations before it was published. I read it when most of this article was already
written, and we are in agreement on several specifc issues; the main point on which I disagree is signalled and
discussed in n. 1 on p. 182 of this article. Finally, I thank Kelli Rudolph for help with the English and St. Johns Col-
lege Cambridge, the Arts and Humanities Research Council, and the Newton Trust for supporting my research
in Cambridge in the last three years.
1 Cf., e.g., Schreiber [24], p. 37: The examples of this fallacy are of somewhat more philosophical interest
than the previous examples of homonymy and amphiboly.
2 Cf. Dorion [7], pp. 347-348 n. 307, 353-354 n. 319, 358 n. 331; Schreiber [24], e.g. pp. 37, 44f., 47, 51, 92-93, 170
(Although Aristotle cites distinctions between Categories as being necessary for resolving these errors, some of
his examples involve confusions among diferent individuals of the same Category, or confusions between uni-
versals and particulars in the same Category.); Tabarroni [27], p. 189; Tabarroni [28], p. 17.
3 Cf. Schreiber [24], pp. 37-38, 169-171; Tabarroni [27], p. 196.
4 Cf. Dorion [7], pp. 353-354 n. 319, 358 n. 331; Edlow [9], pp. 23, 28.
4. The answerer undergoing a sophistical refutation depending on the form of the
expression is persuaded of the soundness of the sophists argument or, more generally,
is deceived by it because he is tricked into falsely believing (or perhaps believes in the
frst place) that similarity of linguistic form implies similarity of the things signifed.1
I shall argue that this idea is incorrect: the answerer is never led to that false belief
nor, more generally, is he ever thinking in ontological terms while facing the sophism.
It is Aristotles analysis and solution that relies on ontological distinctions.
2. Presentation of the Fallacy (se 4, 166b10-19)
In se 3 Aristotle has presented refutation as the primary goal that an eristical arguer
will try to achieve. In the opening of se 4 he informs us that actually it is possible to
refute someone in two ways, and introduces his fundamental distinction between the
arguments depending on linguistic fallacies and those depending on non-linguistic
ones. He deals frst with the linguistic fallacies in se 4 and focuses on the non-linguis-
tic ones in se 5. Before coming to the arguments depending on the form of the ex-
pression, he focuses on the other fve linguistic fallacies: homonymy, amphiboly, com-
position, division, and accent. The arguments depending on these turn out to follow
usually the same general pattern. First the sophist submits to the answerer a dilem-
matic question to which the answerer must reply yes or no, thus selecting the the-
sis he is going to defend in the dialectical exchange. Then the sophist begins asking
several questions to make the answerer accept premisses from which he will fnally in-
fer the contradictory of the answerers chosen thesis, i.e. refute it.2 Of course, his refu-
tation will not be sound, either because the conclusion does not really follow from the
premisses granted (the deduction is invalid) or because the conclusion is only appar-
ently the contradictory of the answerers thesis. The answerer, however, will fail to re-
alise this because of some trick the sophist adopts. Generally speaking, in the case of
homonymy, amphiboly, composition, and division some of the propositions involved
will be liable to be interpreted in more than one sense, which allows the sophist to
surreptitiously take the answerers reply in a diferent way from the intended one and
use it accordingly. Similarly, in the accent fallacy some word will have two possible
pronunciations, each with a diferent meaning, and the sophist will use one in his
question but then derive his conclusion as if the premiss granted by the answerer had
contained the other.
1 Cf. Dorion [7], e.g. pp. 348 n. 308, 349 n. 310 (Dorion seems to think that even the sophist himself is deceived
by the similarity of linguistic expression: le questionneur se laisse abuser par la forme de la ), 351 n. 311 (Du
fait que voir [] a une terminaison caractristique des verbes lactif, le questionneur croit pouvoir le ranger
parmi le catgorie de lactivit), 353 n. 318; Ebbesen [8], p. 200; Edlow [9], p. 23; Fait [10], e.g. pp. xiv, xv, xvi, xix-
xxi, xxii-xxv (le cause dellapparenza dei paralogismi sono credenze errate che determinano un errore di valu-
tazione delle confutazioni; si induce linterlocutore a credere di aver concesso che vedere (horan) appartiene alla
categoria del fare, come il tagliare e il bruciare; lanalisi psicologica dei paralogismi [] postula che il rispon-
dente ragioni correttamente e sia confuso dal ritenere soddisfatta una condizione necessaria alla validit del
ragionamento che invece non soddisfatta); Huelsen [13], pp. 176-177; Kirwan [14], p. 42 ([] his inexpert re-
spondent [] infers identity of meaning-type from identity of expression-type); schreiber [24], e.g. pp. 37 (The
linguistic source of the persuasiveness of these errors steams from the false belief that similarities of word end-
ings [] refect analogous similarities in the things signifed), 38-39, 41, 43 (The reason for the false belief that see-
ing is an activity is that it has the same morphological terminations as do the other verbs signifying activities),
44, 51, 92-93, 169-171 (on p. 169 Schreiber lists the false presuppositions which he thinks Aristotle believes are
jointly necessary and sufcient to account for all examples of false reasoning); Tabarroni [27], e.g. p. 186.
2 Alternatively, the conclusion can be absurd or otherwise unacceptable.
172 ermelinda valentina di lascio
Aristotle fnally introduces the sophisms depending on the form of expression:
T1 The <arguments> depending on the form of expression (a e ) occur
when what is not the same is said in the same way (e c e ),1 i.e.
what is masculine () <is said as> feminine () or what is feminine as masculine or
what is neuter () as one of these two, or again (j ) something qualifed as
quantifed or something quantifed as qualifed or something active as afected or something
disposed in a certain way as active, and the other things as distinguished before;2 for by
means of the expression it is possible to signify as an action what is not among actions. For
example, to be healthy () is said similarly to to cut () or to build
() because of the form of expression; nonetheless, the former signifes something
qualifed and disposed in a certain way, the latter an action. In the same way also for the other
cases. (166b10-19)3
Aristotle labels the kind of sophistical arguments he is introducing as a e
,4 while in se 22, the chapter devoted to the solution to this fallacy, the same
arguments are described as a e a c (178a4-5), which in
T1 indicates the cause of the fallacy. The two descriptions appear to be connected in so
far as diferent things are said similarly in the sense that the linguistic form of the
terms in which they are expressed is similar. The phrase might actu-
ally seem to have a narrower sense than , i.e. to refer specifcally to
the external form or shape of the word,5 and the example in T1 might appear to con-
frm that: , and are similar not only in that they are all verbs
in the same voice, mood and tense (active present infnitives), but more specifcally be-
cause they share something in their surface aspect, their ending . However, al-
though undeniably this makes these verbs more similar to one another than if they did
not have in common their ending, mood and tense (and Aristotle chooses a most ef-
fective example to illustrate the fallacy), this is only one possible (and not necessary) in-
gredient of what counts as similarity of expression, and in various cases we will not en-
counter any analogous resemblance in external form. For example, let us consider this
Categories passage in which Aristotle mentions the false expectations which the noun
man can generate:
T2 Every substance seems to signify an individual ( ). Concerning primary substances, it
is indisputably true that they signify an individual: for, what they denote is indivisible ()
and one in number ( ). As for the secondary substances, it appears likewise that they
signify an individual because of the form of the noun ( ), when
one says man or animal, but this is not true; rather they signify something qualifed (
), because the subject (e ) is not one as the primary substance <is>, but man
and animal are said of many things. However, they do not signify simply () something
qualifed ( ), as white (e ) does: for white signifes nothing but something qual-
ifed (), whereas the species and the genus mark of something qualifed in relation to
the substance (d e e ). For they signify what a substance is (a
). (Cat. 5, 3b10-21)
1 For the translation of cf. below p. 183 n. 1.
2 Although this reference is usually thought to be Top. i 9, I do not agree that Aristotle intends to refer specif-
cally to the list of categories mentioned there (cf. below 3. 1).
3 All translations are mine. I have tried to give literal translations.
4 This label is also used in the list of the fallacies depending on language at se 4, 165b27, and again at se 7, 169a29-
30 and similarly at se 8, 170a15. 5 Cf. Poste [21], p. 11; Ebbesen [8], p. 212.
the arguments APA TO XHMA TH EE 173
The nouns denoting secondary substances, like man or animal, can appear to de-
note an individual . In commenting on this passage Am-
monius and Philoponus speak of e as responsible for the assumption
that and signify :1 the singular character of the word could make
one imagine that the corresponding is one in number. And, Simplicius
comments, it is easy to be deceived and believe that man is similar to Socrates, i.e. is
numerically one.2 What two words like and have in common is
their being singular nouns, but there is no strict similarity of the words (their endings
- and - are not identical); nonetheless, Aristotle is happy to speak of
as responsible for the perceived similarity. Analogously, similarity of
in se will refer not only to identity of ending, but also to some other linguistic
aspect words may have in common; the boundaries of this notion, which of course is
a key issue, will emerge more clearly as we proceed, and I shall reconsider the matter
in 4.
Having introduced the general meaning of form of expression, let us analyse the
arguments depending on it. T1s example (to be healthy is said similarly to to
cutan action) suggests that these sophisms occur when a word which shares cer-
tain similarities with other words is thereby treated as if it also signifed the same kind
of thing. It also suggests that these sophisms can be solved by clarifying the diferent
kinds of things to which the terms involved really refer (to what category, in the Aris-
totelian technical sense of category, they belong).3 But before we can proceed to ex-
amine this solution in more detail, we need to discuss a problematic point in Aristotles
presentation of the fallacy.
In T1 Aristotle appears to distinguish two groups of arguments, separated by j ;
only for the second group is the solution clearly formulated in terms of his categories
(the other things as distinguished before refers to remaining categories). Categories
are also presented as a fundamental defensive tool against this class of sophisms at the
very beginning of se 22: we have the kinds of predications (a )
(178a5-6), and only the second group of arguments is discussed there. What is the solu-
tion to the arguments belonging to the frst group, then? If it is a diferent one, why does
Aristotle not give them as much attention as to the others? Is their solution only too ob-
vious? Or was Aristotles intention to deal with them too in se 22, but then he forgot,
or to revise se 4 and make some additions there? Before we can even attempt an answer,
we should understand Aristotles list of examples in the frst group, because this itself
presents some problems. It is not clear whether by , and Aristotle is
referring to the gender of things or of the words signifying them. One way to under-
stand the text is to take , and as consistently referring to things, as
suggested by the previous what is not the same is said in the same way: a male thing
said as a female thing (along the lines of what follows: a qualifed thing said as a quan-
tifed one). Although Aristotle does not provide us with any exaplanation of what this
can mean, I suggest we can fnd some example in the discussion of solecism in se 14:
is a masculine word and a feminine one but they refer to things of neuter
1 Ammon. in Cat. 48, 12-49, 11; Phlp. in Cat. 71, 14-74, 3. 2 Simp. in Cat. 102, 23-27.
3 I do not intend to suggest that the categories to which Aristotle appeals here are exactly the same as those
listed in the Categories or in the Topics, but only that they are analogous kinds of distinctions.
174 ermelinda valentina di lascio
gender (wineskin and bed respectively). A diferent way to interpret the text is to take
, and to refer to word genders and not to things (a feminine term
said as a masculine one). When a thing of feminine gender is named by a feminine
word belonging to the second declension but ending in - in the nominative, an end-
ing which characterises almost exclusively masculine nouns, one could thereby be in-
duced to believe that word (and therefore the thing it signifes) to be masculine, exact-
ly because it ends in -, rather than, e.g., in - or -. I suggest that Aristotle in T1 has
primarily in mind cases of the former kind; however, we cannot exclude that he con-
sidered also cases of the latter kind (provided of course that the presence of the article
or of an adjective does not reveal the real gender of the word).
We can now conjecture how the two groups of arguments distinguished in T1 re-
semble each other and how they difer. In both cases we deal with mistaking the kind
of thing to which a word refers because of some linguistic similarity, but we do not have
a category mistake in both cases, not at least in the same sense. In the frst group the
confusion concerns the gender of the thing (and, perhaps, sometimes also of the word),
not its Aristotelian category (for example, whether we think that the bed belongs to the
female gender, as its name would indicate, or, correctly, to the neuter one, we shall cer-
tainly agree that it is the same kind of thing),1 whereas in the second one there is a cat-
egory mistake in the Aristotelian technical sense. Therefore, one might conclude that
what the sophisms belonging to the two groups share is the cause of the fallacy (what
is not the same is said in a similar way), and what they difer in is their solution (the
ones can be solved by using Aristotles categories, the others can not). But this presents
us with a problem, since in se more than once Aristotle maintains that arguments de-
pending on the same fallacy will necessarily have the same solution, or the solution will
not be the real one (cf. e.g. T10 below). Perhaps then we have to consider the matter
diferently: the solution of all the arguments a e consists in
recognising that although x and y are expressed similarly, they are not the same kind of
things; for a large number of these arguments, then, Aristotles categories are the best
way of expressing the diference, although this is not the case with sophisms based on
gender confusion. We shall see later (4) a further reason why it made perfect sense for
Aristotle to treat the case of gender confusion along with diferent cases of similarity
of form of expression.
3. The Sophisms and their Solution (se 22)
3. 1. Categorial Distinctions (178a4-8)
Let us now look at se 22, Aristotles most extensive treatment of the sophistic argu-
ments depending on the form of expression. The chapter opens with a reminder of
something we already know from T1:
T3 It is also clear how one should oppose the <arguments> depending on the fact that things
which are not identical are said similarly, if we have the kinds of predications (a
). For () one man, being questioned, has granted that there is () not
1 Pace Schreiber ([24], pp. 47-48), who speaks of confusion between two diferent primary substances (a sub-
stance of one sex is confused with another substance of a diferent sex), thus describing this as another case of
category mistake, in the technical sense of category.
the arguments APA TO XHMA TH EE 175
something which signifes what-it-is; the other has shown that there is () some rela-
tive or quantifed thing, but these appear to signify what-it-is because of the expression.
Dorion suggests that the term here must carry a broad meaning, since not
all the sophisms in se 22 involve category mistakes in the Aristotelian technical sense of
category (cf. point 1 in 1): his examples are the case of the confusion between gram-
matical genders, which we have analysed in the previous section, and se 22s frst
sophism (A), which we shall analyse shortly. Schreiber makes a similar complaint and
cites, together with the case of gender confusion, the sophism at 178b33-34 (argument
(L) below) and the Third Man sophism at 178b36-179a10 (argument (N) below) as ex-
amples of sophisms which are not based on category mistakes, but respectively on a
confusion between diferent items in the same category (two substances) and a con-
fusion of a particular with a universal.1
Dorions suggestion is correct only in that Aristotle is not using exactly the same
categories he lists in Top. i 9 or in Cat. 4; however it fails to notice the more important
point that even the diferent categories to which Aristotle refers in se 22 embody the
same kind of ontological distinctions as those of the Topics and the Categories. I think
that we are entitled to consider all the arguments mentioned in se 22 as based on cat-
egory mistakes in a fairly strict sense (we shall verify whether this is correct case by
case). For the present, we should notice that Aristotle explains () why we can solve
this kind of fallacy by using a , by referring to the case of ex-
changing substances for things in the categories of relation or quantity (178a6-9).2 Al-
so the way he presents his solution (if we have the kinds of predications) might be a
hint that he is referring to some quite defnite classifcation of his own which one
would need to know and employ (although, again, not necessarily identical to the Top-
ics or Categories ones).
3. 2. The Sophists Manoeuvres (178a9-178b7)
The frst two sophisms Aristotle presents involve the categories of action and afection
(178a9-28). I reconstruct the frst (178a9-11) as follows:
<(1) To cut () is to do something? Yes.> P
<(2) To have cut () is to have done something? Yes.> P
<(3) Is it possible to cut and to have cut the same thing at the same
time? No.> P
<(4) Then, is it possible to do () and to have done ()
the same thing at the same time? No. Answerers Thesis (T),
from (1), (2), (3)
1 Dorion [7], pp. 347-348 n. 307; Schreiber [24], pp. 44-45.
2 The details of this reference, however, are quite obscure, but will become more clear when we will have con-
sidered an argument which fts this description (argument (D) below; cf. p. 180 n. 1). The translation of the verb
also presents some problem: some interpreters take it in its existential sense (e.g. Poste [21]: The thesis
denies the existence of a substance, and the questioner proves the existence of a relation or quantity), others as
meaning to belong <as an attribute> (e.g. Pickard-Cambridge [20]: For the one man, say, has granted, when
asked, that a term denoting a substance does not belong as an attribute, while the other has shown that some at-
tribute belongs which is in the Category of Relation or of Quantity). Neither proposal makes the text very in-
telligible, but since there is no better alternative, we have to adopt the rather clumsy translation T3.
176 ermelinda valentina di lascio
<(5) But it is possible to see () and have seen () the
same thing at the same time. P
<(6) Therefore, it is possible to do and to have done the same thing
at the same time.> T, from (5)
The sophist produces a merely apparent refutation of the answerer by making him con-
cede that it is impossible to do and to have done the same thing at the same time (4),1
and by subsequently objecting that it is possible to see and to have seen the same thing
at the same time (5). The apparent refutation2 relies on the sophists tacit assumption,
which the answerer has never conceded, that , being an expression similar to verbs like
(and itself ), signifes an action. According to Aristotle the argument is
paralogistic, because actually seeing is not an action, but an afection (as the answerer
should point out in his solution), and therefore cannot constitute a genuine counterex-
ample to the answerers thesis that it is not possible to do and to have done the same thing
at the same time. Here the similarity of expression consists in the fact that the verbs in-
volved have the same mood, voice and tense (infnitive, active and present in the frst
set; infnitive, active and perfect in the second one).
According to Dorion, this argument does not involve a category mistake, but exploits
the contraposition between actualisation () and process (): each mo-
ment of a vision is in itself a complete seeing and thus is both to be seeing and to have
seen, which is not true of a process (each moment of the process is not a complete
process, but only a part of it).3 This would be an example of a sophism which cannot
be solved by categorial distinctions, and therefore it would require us to understand a
at the beginning of se 22 in a broad non-technical sense. Although
the diference between building something and seeing something could be described,
in Aristotles own terms, as a distinction between and respectively, I
suggest that this is not what Aristotle had in mind here. Dorions interpretation cannot
make any sense of Aristotles remarks on the argument:
T4 If someone there, while granting that it is not possible to do and to have done the same thing
at the same time, were to say that it is possible to see and to have seen <the same thing at the
same time>, he would not have been refuted yet, if he says that to see is not an action but an
afection (, c e a); for this question must
be added. (178a16-19)
If Dorion were right, the missing premiss necessary to make the sophists refutation
genuine should be Seeing is a process, and not Seeing is an action. Moreover, we
would have two diferent solutions for two arguments, (A) and the immediately
following one, (B), which Aristotle evidently associates because of their close resem-
blance, and the second of which is clearly solved by saying that to see does not signi-
fy an action, but an afection, as we shall see shortly. We could say, of course, that build-
ing is not only an action but also a and seeing not only an afection but also an
, but this does not seem to be applicable to every case in which the categories
of action and afection are involved, and therefore would not represent a real solution
(arguments based on the same fallacy require identical solutions).
1 (1), (2), (3) are not necessary for the argument, but provide a basis for (4) to be more likely granted by the
2 It is worth noticing that this is not a refutation with the usual Aristotelian structure, with an initial dilem-
matic question. 3 Dorion [7], pp. 349-350 n. 310.
the arguments APA TO XHMA TH EE 177
The second argument runs as follows (178a11-16):
(1) To do something is to be afected? No. T
(2) Is cut (), is burnt (), perceives () are
said similarly and all signify an afection? Yes. P
(3) And to speak (), to run (), to see () are said
similarly? Yes. P
(4) But to see () is to perceive (). P
(5) Therefore, to see is both to do something and to be afected at the
same time. T, from (2), (3), (4)1
That the verbs in (3) are all similar in form does not imply that they all signify the same
kind of thing, namely some action, which is instead again the tacit assumption which
the sophist would need to draw the conclusion (5) validly. The solution is that, while
similar to and , does not signify to do something, but to be afected
by something (notice that not even in (2) similarity of reference was implied by similar-
ity of expression).2
But in what sense is said similarly to and ? Here there is no surface
similarity ( does not end in , although a Greek speaker might have sensed that
its ending results from the same termination which is applied to the stem of the other
two verbs), but the three verbs are all in the active (the similarity between ,
, and is instead stricter). The similarity of here consists in be-
longing to the same morphological category, in our terms. But did Aristotle possess the
same concept of morphological category? The fact that he is not more specifc could
incline us to think that the answer is negative; however one might object that he is
generic purposely, in order to include several diferent cases in the same class. I shall
come back to this question in 4 after examining the whole variety of arguments dis-
cussed in se 22.
Although the underlying sophistical principle is the same, the arguments we en-
counter next are more sophisticated and puzzling. Let us reconstruct the frst (178a29-36):
(1) Did you lose what () you had at one stage but later do not have? Yes. T
(2) If you have ten () dice and you lose only one ( ), do you
still have ten dice? No. P
(3) But you have not lost the ten dice you had before. P
(4) Therefore, it is not the case that you lost what you frst had but later
do not have. T, from (2), (3)
1 Strictly speaking (5) is not the contradictory of (1), but is inconsistent with it.
2 The reconstruction of argument (B) is not fully convincing, since the fact that certain verbs are said similarly
would work better as an implicit assumption on which the sophist relies (and as such was described by Aristotle
in the case of the previous argument (A)). An alternative version of the sophism might then be:
(1) Is cut () and is burnt () are to be afected? Yes. P
(2) And to speak () and to run () are to do something? Yes. P
(3) To do something is to be afected? No. T
(4) But to see () is to perceive (). P
(5) Therefore, to see is both to do something and to be afected at the same time. from (1), (2), (4)
(4) can work as a counterexample to (3) only on the basis of the sophists tacit assumption that , because
similar to and , must signify an action, and , because similar to and ,
must signify an afection.
178 ermelinda valentina di lascio
It is not easy to understand how this sophism works according to Aristotle. Dorion iden-
tifes an ambiguity in the relative pronoun , which would be responsible for the con-
fusion between the category of substance and that of quantity.1 This suggestion pres-
ents the problem of making the fallacy collapse onto that of homonymy (cf. point 3 in
1): the sophist would rely on the fact that can mean both la chose que (substance)
and le nombre de choses quels (quantity). This would make (1) ambiguous, and suit-
able to be understood by the answerer in one way (with referring to a substance) but
taken by the sophist himself in another (with referring to a quantity), which is the hall-
mark of the fallacies of homonymy and amphiboly (cf. se 19, 177a9-11). Moreover, it
would be difcult to interpret the previous arguments (A) and (B) along similar lines:
for we would never say that is ambiguous and that its meanings are to be an ac-
tion and to be an afection.2
The fundamental starting point for understanding correctly how argument (C)
works is that (1) is not ambiguous at all in Aristotles view; the pronoun is clearly
meant to range over items that are 3 abstracting from any quantifcation, and in this
way it is employed by the sophist and understood by the answerer, who in fact gives a
correct afrmative answer. It is only if we substituted with , a pronoun ranging
over quantities, that we would get a question which cannot be answered by a simple
yes or no: while it is necessary that if I had a single pound and I do not have it any-
more, I lost it, it is not necessary that if I had ten pounds, but I do not have them any-
more, I lost all of them.
But if (1) is not ambiguous, what allows the sophist to replace the pronoun , which
is meant to be substituted by in (1), with (which is instead ) without the
answerer noticing this? I suggest that it is not that can appear to signify a quantifed
thing, but rather it is that, while signifying something quantifed, might be taken
to signify an individual because of its form of expression. This is confrmed by 178a6-8
(the second part of T3) in which Aristotle outlined the typical structure of the argu-
ments depending on form of expression: it is what signifes something quantifed that
was said to appear to signify an individual, not the other way around. Further confr-
mation comes from the following sophism (178a36-178b1):
(1) Could one give what () one does not have? No. T
(2) But I do not have only one ( ) die, for I have ten (). P
(3) I can give only one die. P
(4) Therefore one could give what one does not have. T, from (2), (3)
Here, again, no premiss is ambiguous. The problem lies in the fact that in (2) and (3), as
Aristotle says explicitly, does not signify , but a type of relation (namely, not
1 Dorion [7], pp. 352-353 n. 316.
2 One might ofer a partial answer to this objection by saying that the diference between the present fallacy
and that of homonymy is that in the former the ambiguity of reference is between two diferent categories under
which the same thing can be described, whereas in the latter it is between two things; however, we would still be
diagnosing the problem as a form of ambiguity.
3 T occurs three times in se 22 (178a39, 178b5, 28) and occurs six times (178b38, 179a2, 179a4 [bis], 179a6,
179a8). I suggest that here they have the same meaning, i.e. an individual (which includes the primary substances
of the Categories). For the seminal study of the meaning on in Aristotle cf. Smith [26].
the arguments APA TO XHMA TH EE 179
with something else [ ]), and thus the sophist is not entitled to draw the
conclusion (4) as if it signifed some for which the pronoun stands in (1).1
Next Aristotle ofers an illustrative example in which the same kind of fallacy is much
more evident, so that we can understand it better (178b1-7):
(1) Could one give what () one does not have? No. T
(2) Could one give something quickly () even if one does not have
it quickly? Yes. P
(3) Therefore, one can give what one does not have. T, from (2)
Aristotle explains that the deduction from (2) to (3) is invalid for the usual reason: since
to give quickly is not to give something (), but to give in some way (z),
cannot replace the pronoun in (1). But the faw is so evident here that presumably no
sophist would employ such an argument with hope of success:2 does not re-
semble in its linguistic form terms which signify individuals, and thus
cannot be taken to be . Similarly, in (C) and (D)
is not simply to have something (only one die), but to have something (die) in some
way (only one), and that is why to consider as the object of , although
correct from a grammatical point of view, is unwarranted from an ontological point of
view (the fact that it is used as if it meant does not imply that it does mean ).
Thus, the fundamental linguistic resemblance which and share (unlike
) with the relative pronoun and which favours the sophists manoeuvre to use
them surreptitiously as if they could replace it is, in our terms, that they all play the
same syntactical role: () and () are the objects
of both and , just like . Moreover, they qualify a term which itself
signifes individuals, and also has a more exterior feature in common with
, namely the ending (if the adverb were used, the sophism would
lose most of its strength). Things being so, that of similarity of expression appears now
to be an extremely broad notion which confates aspects we would distinguish (identi-
ty of mood, voice and tense, syntactical role, ending).
Aristotles analysis of the sophisms depending on the form of expression is quite sub-
tle and fascinating, but not immediately straightforward for us, who might be tempted
to diagnose them in diferent terms. Although we have already considered and discard-
ed an alternative reading of argument (C), the impression that (D) should be analysed
diferently is maybe even stronger. Dorion in fact suggests two possible diagnoses, dif-
ferent from Aristotles. His frst proposal is to understand a
as ambiguous: does it mean that one has no die at all or that one has more
than one die? His second proposal consists instead in taking the sophism to depend on
combination and division: if we construe with the sentence means
that one has more than one die (il na pas un osselet seul), whereas if we construe it
with the verb it means that one has no die at all (il na seulement pas un osselet).3 On
1 We can now understand the second half of T3. There is not something which signifes what-it-is describes
the negative answer to (1) in (D), in which what () refers to an individual substance. There is some relative
thing in T3 describes (3) I can give only one die in (D), in which only () signifes a relation, but its
is similar to that of die ().
2 Pace Dorion ([7], p. 354 n. 320), who believes this to be a genuine example of another sophism.
3 Dorion [7], pp. 353-354 n. 319.
180 ermelinda valentina di lascio
both interpretations, then, the sentence a would be
ambiguous between whether one has more or less than one die (cf. point 3 in 1): but
what shows that this can not be the real problem is that even if we disambiguate the
second premiss in (D), as I have chosen to do on p. 179 (But I do not have only one die,
for I have ten), the sophism still works.
3. 3. The of the Answerer
Aristotle ofers us an analysis of the sophisms depending on the form of the expression,
but he does not explain how the answerer, once trained in spotting the fallacy, should
behave practically. Presumably the answerer should reafrm his thesis, e.g. confrm that
one cannot give what one does not have in (D), and clarify that what is possible to do
is something quite diferent, to give what one has but not in the way one has it (e.g.
someone could give one die only, and not with other nine, which is the way in which
he possesses that die). But when is he supposed to clarify this? I suppose that just as in
the case of homonymy and amphiboly (cf. se 19), the answerer can either block the
sophism when the initial question is asked, if he can foresee what kind of argument the
sophist is going to run (cf. se 23, 179a21-22), or denounce the fallacy at the end.
But if he is not trained yet in spotting such a fallacy and consequently blocking this
kind of sophisms, how is he tricked by them? I have referred to the linguistic simi-
larity between the words involved in the sophisms as responsible for the fallacy. One
might think, then, that the sophism works by tricking the answerer into believing, for
example in (C) and (D), that , because of its linguistic form, refers to the same
kind of thing (an individual) over which the pronoun ranges, and therefore that the
substitution is warranted and the conclusion follows.1 However, this is not how the
sophism works. The answerer need not share, even for one second, that false belief,
and actually must not share it if the argument is to be sophistical; it is the questioner
who infers the conclusion as if the answerer had made that concession. The problem
with the answerer is rather that, because of the similarity between expressions, he
will fail to readily identify where the fallacy lies (and not, again, that he positively be-
lieves from the very beginning, or is led to believe, that words having a similar form
of expression must signify the same kind of things, and that therefore the sophists in-
ference is valid). Similarly, in (A) and (B) not only has the answerer never conceded
that to see signifes an action,2 but, if asked, probably he would not concede this
(or, more generically, that words linguistically similar thereby also share the same
kind of referent).
The scenario of an answerer who does hold the false belief that similarity of lin-
guistic form implies similarity of ontological referent (or also some more specifc be-
lief like the one that signifes ) is theoretically possible, but not one which
Aristotle seems to have primarily in mind in se. Sophisms are supposed to aim at de-
feating and fooling the respondent regardless of his actual beliefs. My suggestion would
require a broader discussion of the nature of sophisms in se which I cannot undertake
here. However, I shall at least try to tackle what would probably be the main objection
against my suggestion: in se 7 Aristotle seems to identify and list exactly the answerers
false presuppositions (in Schreibers terms) or incorrect beliefs (in Faits terms) in-
1 Cf. p. 172 n. 1. 2 The sophist is in fact careful to avoid such a question: cf. T4 above.
the arguments APA TO XHMA TH EE 181
volved in each single fallacy.1 To begin challenging this interpretation, let us consider,
for example, the case of homonymy and amphiboly:
T5 The trick () of the <arguments> depending on homonymy and amphiboly occurs be-
cause of the inability to distinguish (c ) what is said in many senses (for
some things are not easy to distinguish, e.g. being and the same). (se 7, 169a22-25)
Aristotle is not saying that the answerer holds the positive false belief that the term in-
volved in the sophism has just one meaning, but only, more weakly, that he fails to
recognise the two diferent meanings with which it is used. He is not ready to notice
that the term is employed with two diferent meanings in the argument, but treated by
the sophist as if it had only one, and this lack of awareness and preparation is what the
sophist exploits. In se 7 is not an error to which the answerer himself is com-
mitted, i.e. a false belief which makes him accept that the sophism is sound, and it is
not even deception in the strong sense that the answerer is tricked into believing that
the argument is valid: is where the sophists stratagem and astuteness lie, i.e. the
trick the sophist exploits without the answerer being able to identify it (i.e. it is at most
an omissive error of the answerer). The same is true of our case:
T6 <The trick () of the arguments> depending on the form <of expression> occurs be-
cause of the similarity of the language. For it is difcult to distinguish what things are said
similarly and what are said diferently [], since we assume () that each thing
predicated of something is individual ( ), and we perceive it as one in number (); for
being an individual and being seem to apply especially to what is one in number and what is
a substance (0). (se 7, 169a29-36)
It could be tempting to interpret the verb as indicating a positive belief of
the answerer. However, the assumption of the answerer need not be any kind of con-
scious and positive belief, but can be taken to amount to no more than an acceptance
1 Fait [10] and Schreiber [24] explicitly suggest such an interpretation. As I understand it, Faits general idea
is that Aristotles answerer in se is someone who knows the correct defnition of refutation, and therefore rea-
sons well, but holds an incorrect belief which makes him think that the sophistical argument he is presented with
fulfls all the conditions for being a real refutation (when it actually does not) and therefore makes him accept the
argument as valid. For example, in a refutation depending on homonymy there will be a word which has more
than one meaning and therefore the refutation will not meet all the conditions for being a real refutation, but the
answerer will not realise this because he holds the incorrect belief that the word in question has only one mean-
ing. This incorrect belief is actually embodied in a premiss which the questioner has not asked and therefore the
answerer has not granted but thinks to have granted. Should the sophist ask such a question, the answerer will
probably concede it and the refutation will turn out to be valid. (Fait [10], pp. xvii-xxv.) Schreibers interpreta-
tion has various points in common with Faits. What we are mainly interested in here is his thesis that explaining
why the sophistical refutations appear true to the answerer consists in isolating that false presupposition that is
accepted by the victim and would render the reasoning sound if it were true (Schreiber [24], p. 168). These false
presuppositions can concern the nature of the language or the nature of the world and determine how an argu-
ment is classifed (He [sc. Aristotle] places diferent false arguments under the same type of fallacy whenever the
same false presupposition can account for the appearance of proper reasoning in each example [p. 168]). Schreiber
identifes nine false presuppositions, fve concerning language and four concerning the world (he lists them on p.
169) and explains that if a false presupposition about language is involved in a sophism, then this will be classifed
as linguistic, otherwise it will be classifed as non-linguistic. My main general objection to Faits and Schreibers
approach is that if the answerer holds a false belief, then the argument turns out to be valid and the refutation to
be a genuine refutation of him, i.e. not sophistical! Why shouldnt the sophist, then, ask to grant a premiss which
expresses the answerers false belief ? The point in fact is rather that the sophist avoids asking to concede that pre-
miss since if he asked this, the answerer would probably recognise where the sophist is heading to and where the
fallacy lies. Moreover, I doubt that Aristotle has in mind someone who knows the correct defnitions of syllo-
gism and refutation: as we shall see shortly, Aristotle makes clear that the answerer is inexpert of dialectic.
182 ermelinda valentina di lascio
(passive and unconscious) consisting simply in the failure to denounce that the sophist
is treating as something that actually is not (we assume in the sense of we let
the sophist assume without protesting, and in this sense we can be considered re-
sponsible for the error). This failure is prompted by the inexperience of the answerer
and by the objective similarity of linguistic expressions which helps to conceal the
sophists stratagem.1
One might object, more generally, that Aristotles own accounts of apparent syllo-
gism and refutation goes against my suggestion:
T7 A syllogism is made of certain things posed in such a way that it is necessary to say something
diferent from what has been posed through what has been posed. A refutation is a syllogism
with the contradictory of the conclusion. Some <syllogisms and refutations> do not achieve
this, but appear <to achieve it> for many causes. (se 1, 164b27-165a4)
The fnal sentence explains that an apparent syllogism fails to do what a real syllogism
is supposed to do, i.e. what is laid down in the defnition of syllogism (frst part of T7),
but it also adds that at the same time it appears to deliver what in fact it does not deliv-
er. One might then suppose that for the answerer to be fooled apparent arguments must
appear to him to be made of certain premisses posed in such a way that it is necessary
to assert the conclusion through what has been posed, and might wonder how this can
happen without the answerer holding some false beliefs. My reply is that in Aristotles
view there is no genuine persuasion of the answerer that these sophisms are syllogistic in
the technical sense. At se 1, 164b25-27 Aristotle gives us essential information: the argu-
ment falsely appears syllogistic to the lay person ( ). Just as somebody can adjust
and embellish himself to look as if he were healthy when he is not, and just as a yellow
metal can look like gold and a silvery metal can look like silver, similarly some argu-
ments only look like syllogisms (164a23-26). I take the point to be that just as in the for-
mer cases one lacks the skills and the knowledge of medicine and metals necessary to
identify counterfeit samples, in the latter case one lacks the skills and knowledge of di-
alectic to distinguish real from counterfeit syllogisms (and these skills are exactly what
Aristotle intends to provide in se). I also take it that the resemblance the layperson per-
ceives between a merely apparent syllogism and a real one is a very vague one, or at
most one with as low a degree of precision as low is the level of his experience.
Of course there can be various degrees of similarity between someone who is
healthy and someone who only looks healthy, and between gold and a yellow metal,
and depending on these and on the relevant expertise one possesses it will be more or
less difcult to spot the fake healthy person and the fake piece of gold. Similarly, there
1 In T1 I translated as is said (taking it as a synonym of ): The <arguments> depend-
ing on the form of the expression occur when what is not the same is said in the same way. I chose not to trans-
late as is understood exactly because that translation would have given the impression that Aris-
totle is committed to the stronger claim that for the fallacy to occur it is necessary that the answerer holds the
false belief that similar words always have similar referents. However, a more literal translation might be defend-
ed now on the basis of my present suggestion. One could suppose that in T1 Aristotle is referring to the deepest
cause of the success of these arguments, i.e. what the sophist exploits to his advantage; in other words, the rea-
son why we are puzzled and at loss when we are presented with this kind of sophism. And this is nothing but the
fact that the answerer will have trouble in identifying where the fallacy lies, because he perceives the words as sim-
ilar. We might then translate in T1 as is understood in the same way or is interpreted in
the same way, but only if we keep in mind that the sense is not that the answerer believes that e.g.
and both refer to particulars. Alternatively, we might supply an implicit <by the sophist> after is inter-
preted in the same way, so that it is clear that this is not a mistake of the answerer.
the arguments APA TO XHMA TH EE 183
can be diferent degrees in which an unsound refutation appears sound and accordingly
it will be more or less difcult to realise that it is not a real refutation: Aristotle makes
this clear at se 33, 182b6-31, in which he states that some arguments are more difcult to
solve than others, even if they all depend on the same fallacy. But he also indirectly
makes clear that the degree of experience and skill of the answerer are crucial, although
even the most expert can encounter difculties in solving some very challenging so-
phistical arguments. We can think at least of three possible scenarios:
A completely inexperienced answerer who is easy to deceive and who, when sub-
mitted to an apparent argument (even a ridiculous one), thinks he has really been re-
A less inexperienced answerer who realises that he has been tricked (especially if the
argument is ridiculous), but is not able to spot the fallacy;
An experienced answerer who will be able to solve virtually any argument (although
it is possible that he will struggle with the most difcult ones).
Of course Aristotle is not considering case (3): that is the ideal scenario when some-
one has already studied his dialectic. As for case (1), when a layman is presented with a
sophism, he perceives this as a real syllogism not in the sense that he is convinced of
the conclusiveness of the argument (i.e. believes that it possesses all the features a syl-
logism must have); his problem is that he is not even able to check whether it is a real
syllogism because he does not know the relevant criteria. He is judging the argument
on the basis of its loose resemblance with genuine syllogisms, just like someone who
cannot see the details of an image because he is looking at it from a distance (se 1,
164b26-27).1 It is a sort of sensation () the layman is relying on, not some pre-
cise account, which he does not possess, of what a syllogism is (otherwise that account
would be incorrect or he would really be deceived by the sophist into falsely believing
that the sophism fts that account). Therefore, there is no false belief involved. Similar-
ly, there is none in case (2): this answerer too will have no precise idea of what a real
refutation is. However, he will probably have the feeling that something is wrong with
the argument and that he has not been treated fairly by the questioner, but has been
the victim of a puzzling sophism; yet, he will be completely incapable of spotting
where the fallacy lies and unmasking the questioners strategy ( just like someone who
witnesses a magicians trick is aware that he has been tricked but is incapable of ex-
plaining the trick). This is the aim of the sophist, and not persuading the answerer that
the argument is valid and its conclusion true.
In 3. 3 we discarded an interpretation which makes some of these sophisms de-
pendent on some sort of homonymy, and thus rejected, at least partially, the scholarly
charges illustrated under points 1 and 3 in the Introduction. The present discussion has
also refuted point 4, and thrown some light on the problem arising in point 2: the fact
that the answerer does not hold any false ontological belief as the result of certain
1 For example, certain questions can be asked as if to establish the premisses of a syllogism and then some-
thing else can be asserted as if it were the conclusion; moreover, some feeling that the conclusion follows from
the premisses can be promoted by the sophist, e.g. by stressing the therefore which introduces the conclusion.
It seems, then, that this answerers basic mistake could be described as a fallacy of the consequent (e.g. This
metal is yellow, but gold is yellow, therefore this metal is gold [cf. se 5, 167b1-20]; somebody who accepts this ar-
gument is committing the mistake of taking gold is yellow as convertible, i.e. as implying what is yellow is
gold). I am not claiming that the fallacy of the consequent is more fundamental than the others: the answerers
mistake can be still reduced to the fallacy of ignorance of what a refutation is (cf. se 6).
184 ermelinda valentina di lascio
features of language already tells us that ontology is not a main focus of this fallacy.
What promotes the sophists trick is the linguistic form of some words used: since they
are similar the answerer does not notice the illegitimate substitutions the sophist is
making; therefore to classify this fallacy among the linguistic ones is without a doubt
3. 4. Alternative Solutions (178b8-36)
Let us now go back to the text of se 22 and discuss the next passage, in which Aristotle
considers and criticises three alternative solutions to the sophisms depending on the
form of expression:
T8 Also all the following <arguments> are similar <to the previous ones>: Could one strike
with the hand one does not have?, or Could one see with the eye one does not have?, for
one does not have only one <eye>. (a) Some people, then, solve <this kind of argument> by
saying that the person who has more than one thing, whether eye or anything else, also has
a single one; (b) others <solve it by saying> that one also acquired what one has: For this
man gave only one pebble; and certainly this other man, they say, has only one pebble from
him. (c) Others, again, <solve it> by rejecting straight away the initial question, <and say>
that it is possible to have what one never acquired, e.g. having acquired sweet wine, because
it went bad in the course of receipt, <it is possible> to have it sour. (178b8-16)
Solution (a) appears to be specifcally conceived for sophisms like the following:
<(1) Could one see with the eye one does not have? No. T
<(2) One does not have only one ( ) eye. P
<(3) One could see with only one eye.> P
<(4) Therefore one could see with the eye one does not have. T, from (2), (3)1
The proponents of solution (a) seem to suggest that the answerer should not accept
premiss (2):2 it is not true that one does not have only one eye; since one has two eyes,
one also has a single eye. In other words, they understand (2) as amounting to One has
no eye at all, not even one and they reject it. If the answerer rejects (2), the sophist of
course cannot proceed on with his argument.
Both solutions (b) and (c) refer to arguments which are not fully quoted by Aristot-
le, but at least in the case of solution (c) it seems clear that the initial question of the
sophism is Is it possible to have what one never acquired? This information is suf-
cient to understand the proposed solution: for the sophism to work, the answerer is ex-
pected to reply No to the initial question, but the proponents of this solution suggest
that he can give a positive answer, since something which one acquires can change its
properties over time (e.g. wine can change from sweet to sour) and thus become some-
thing which one has without ever acquiring it. In this way, the sophist can not run his
argument, which I suggest might have been as follows:
1 Alternatively, and more neatly, the following sophism, which is not a refutation, but leads to a paradoxical
conclusion (the third most important sophistical aim: cf. se 3):
(1) Could one see with the eye one does not have? No. P
(2) But one does not have only one eye. P
(3) Therefore one cannot see with only one eye, which is absurd.

, from (1), (2)
2 Pace Dorion ([7], p. 355 n. 322), who understands that they are rejecting the conclusion (4).
the arguments APA TO XHMA TH EE 185
(1) Is it possible to have what () one never acquired? No. T
(2) One has ten () pebbles (sour wine). P
(3) One never acquired ten pebbles (sour wine), but ffty (sweet wine). P
(4) Therefore it is possible to have what one never acquired. T, from (2), (3)1
To understand solution (b) we now need to reconstruct the whole sample argument.
From the few clues Aristotle gives us, we might think he is alluding to the same
sophism, (G), for which solution (c) is formulated (in both cases we deal with having
and acquiring things), but I suggest that the two arguments must be diferent (albeit
similar in their topic) for the following reasons. First, I cannot see any way of recon-
structing an argument which satisfes all the requirements we can extrapolate from T8
(b) and T8 (c): for example, we have seen that in (c) Aristotle is quite clear that the con-
tradictory of the answerers thesis must be It is possible to have what one never ac-
quired, but this would be difcult to square with the positive form One never acquired
what one has which we fnd in (b). Second, presenting three diferent solutions to three
diferent arguments serves Aristotles purpose better, by making it more evident that
diferent arguments based on the same fallacy would have diferent solutions, which can
not be the case according to him (cf. e.g. T10 below). Thus, let us try to reconstruct the
argument to which solution (b) is meant to apply:
(1) Did one acquire what () one has (and did not lose)? Yes. T
(2) He has ten () pebbles, but he never acquired ten pebbles (for frst
his father gave him fve pebbles, then his mother gave him four pebbles
and fnally his brother gave him only one pebble). P
(3) Therefore he never acquired what he has. T, from (2)
On this reconstruction alternative solution (b) would consist in rejecting premiss (2): for
we should not say that someone has ten pebbles if he did not acquire them all togeth-
er, but we should say that he has only one pebble from the brother from whom he
acquired only one pebble, that he has four pebbles from the mother from whom he
acquired four pebbles, and that he has fve pebbles from the father from whom he ac-
quired fve pebbles.
What T8s three solutions have in common is that they all suggest that there is some
proposition which the answerer is expected to concede for the sophist to be able to con-
struct his argument, but which the answerer should not concede. Let us see why this
approach does not work for Aristotle:
T9 But, as was said also before, all these people direct their solutions not against the argument,
but against the person ( e e a e e ). For if this were the solu-
tion [sc. rejecting one of the questions], then, one could not solve by doing the opposite <of
that on which the argument depends> ( e x ), just as in the
other cases (e.g. if the solution is So-and-so is true in a way and not in another, then, if one
grants that what is said is true without any qualifcation, the conclusion follows; but if the
conclusion does not follow, then that [sc. distinguishing] could not be the solution). But what
1 Cf. also Schreibers reconstruction ([24], p. 41):
(1) Is it not the case that what one person gives to another, that second person has? Yes.
(2) But one person gave another only one vote, yet the second person has ten votes, not only one.
186 ermelinda valentina di lascio
we say with regard to the examples discussed before is that, even if all the premisses are grant-
ed, still there is no syllogism. (178b16-23)
This passage is quite obscure. I suggest the following structure: the frst sentence
presents Aristotles criticism of the inadequate solutions, the second and fnal sentences
ofer an argument for that criticism, while the long parenthesis gives an example of those
arguments to which Aristotle has referred immediately before it (as in the other cases).
Aristotle does not discuss each single solution of T8,1 but presents one general
criticism: these solutions are not e e , but e e . We fnd a very
similar remark at se 20, 177b31-34, when Aristotle considers solutions diferent from his
for some sophisms based on the fallacy of combination (presumably he refers to this
passage at the beginning of T9 by as was said also before):
T10 But it is clear that they do not solve well: for the solution to the arguments depending on
the same thing is the same, but this <solution> does not ft <all the arguments of the same
kind>, nor all the ways of formulating the questions, but is valid against the person who
questions, not against the argument ( \ d b
, \ e e , e e ).2
In T10 the reason why some inadequate solutions are rejected as merely ad hominem
seems to be that it would be impossible to use them for all the possible alternative for-
mulations of the same argument and for all the arguments depending on the same fal-
lacy, i.e. combination. In T9 the reason ofered seems to be the following: For if this
were the solution, then, one could not solve by doing the opposite, just as in the other
cases []. But what we say with regard to the examples discussed before is that, even if
all the premisses are granted, still there is no syllogism. I understand e
as the opposite <of that on which the argument depends>, and not as the opposite
<of the solution> (If this were the solution [sc. rejecting one of the questions], if one
did its opposite [sc. did concede all the questions], one could fnd no solution, as in the
other cases). I suggest that the frst reading makes more sense, and that the passage
must be understood as follows. From se 23 we know that the general strategy for solv-
ing the sophisms depending on language consists in doing the opposite of that on which
they depend, i.e. in making some distinction (in the case of the form of expression, for
example, saying that although one cannot give what one does not have, one could give
what one has but not in the way one has it). From se 18 we also learn that making dis-
tinctions is the way to solve an apparent argument, whereas eliminating one of the
questions is the way to solve a valid argument which has one or more false premisses.3
On the basis of this, I suggest that Aristotles argument in T9 is the following. If we ac-
cept that eliminating one of the premisses is the solution, then this solution will not
consist in making some distinctions, i.e. in doing the opposite of that on which the ar-
gument depends. But we know that non-syllogistic arguments must be solved by means
of distinctions (from se 18) and that the arguments depending on the form of expres-
sion are non-syllogistic and remain such even if all the premisses are granted (last sen-
1 For example, he could have criticised solution (c) by saying that even if something changes its properties over
time it still remains the same thing, and therefore it is not true that the (sour) wine is something one has but has
not received.
2 I agree with Dorion ([7], pp. 346-347 n. 303) that refers to the fact that there can be
various ways of formulating the premisses of the same argument.
3 Cf., in particular, se 18, 176b35-36, 177a2-6.
the arguments APA TO XHMA TH EE 187
tence of T9; with what we say with regard to the examples discussed before Aristot-
le might refer to 178b4-5, in which he explained with the help of a fake sophism
[argument (E) above] why these arguments are invalid); therefore we must solve them
by making some distinction, and not by eliminating one of the questions (implicit
Whether or not the implicit steps of Aristotles arguments are exactly those I just
spelled out, it seems clear that his qualm about the alternative solutions in T9 is that
they consist in rejecting some proposition which the sophist is proposing to the
answerer, and not in contesting the validity of the argument itself. This is why the
alternative solutions are unsatisfactory, because they misidentify the problem in some
As for the elliptical parenthetical remark of T9 (e.g. if the solution is So-and-so is
true in a way and not in another, then, if one grants that what is said is true without
any qualifcation, the conclusion follows; but if the conclusion does not follow, then
that [sc. distinguishing] could not be the solution), this supports the claim that the
solution to the rest of apparent arguments depending on language is by means of
distinctions. Aristotle has just said that in the other cases the solution consists in the
opposite of that on which the argument depends, i.e. in making distinctions, and the
parenthesis ofers an example of one of these other cases. When the solution consists
in distinguishing between diferent meanings of a question (as in the case of the
sophisms based on homonymy and amphiboly), if the extra premiss that the relevant
question has only one meaning were assumed, then the argument would be valid: this
shows that distinguishing the meanings of the question is the correct solution (other-
wise the argument would not be valid once the fact that the question has only one
meaning has been assumed).
Ultimately, then, the solutions criticised in T9 are e e and not e
e for the same reasons displayed in T10. They are not e e because
they neither show that the argument is invalid nor why (i.e. what is the fundamental
fallacy). And they are e e because they too apply only to specifc argu-
ments (and for this reason they appear to work): for example, it is difcult to see how
one could solve the sophism (F) by answering yes to the initial question, as suggest-
ed by solution (c) to sophism (G). But arguments of the same kind should not have dif-
ferent solutions. Thus if an answerer proposed such a solution, he would be criticising
the questioners chosen argument or specifc formulation of the premisses, and maybe
the audience would have the impression that the solution works, but he would not be
really solving the sophism, because he would not be revealing the fundamental fallacy on
which it is constructed and which is in common to the arguments of the same kind and
ultimately to all the arguments depending on linguistic fallacies.
1 Aristotles argument can be schematised as follows:
<(1) If one solves a sophism depending on the form of expression by eliminating one of the
questions, one is not solving it by doing the opposite of the cause of the argument, i.e. by
making distinctions. P (T9)
<(2) Non-syllogistic arguments must be solved by making distinctions.> P (SE 18)
<(3) Arguments depending on the form of expression are non-syllogistic and remain such even
if all the premisses are granted. P (T9)
<(4) Therefore one must solve the sophisms depending on the form of expression by making
distinctions.> from (2), (3)
<(5) Therefore one can not solve them by eliminating one of the questions.> from (1), (4)
188 ermelinda valentina di lascio
Having criticised the unsatisfactory alternative solutions of T8, Aristotle goes on to
present other sophisms, which I shall explain briefy to show that they work in the same
way as those analysed in 3. 2. Let us consider the frst two (178b24-31):
(1) Has someone written what () is written? Yes. P
(2) The false () sentence () you are sitting is now written. P
(3) But it was a true () sentence when it was being written. P
(4) Therefore someone has written a false and true sentence at the same

, from (1), (2), (3)
( J)
<(1) Does the one who learns learn what () he learns? Yes. P
<(2) Someone learns what is slow fast (e f ). P
<(3) Therefore what he learns is slow fast.>

, from (1), (2)
These arguments, which are not in the form of refutations but lead to absurdity,1 must
be solved according to Aristotles usual analysis: and in (I) and in
( J) are not fair substitutes for the pronoun in (1), since the latter ranges over individ-
uals whereas the former signify diferent kinds of things, i.e. qualities in (I) and a man-
ner (in which something is learnt) in ( J). The reason why one might fail to notice that
these expressions do not signify the same kind of things which the pronoun signifes
is the usual one: they stand in the same syntactical relation with respect to the verb as
, and accompany substantival forms with which they share the case ( also has the
same ending as ).
Let us now look at the following arguments (178b31-34):
<(1) Does one tread upon () what () one walks through? Yes. T
<(2) One walks through the whole day (c ). P
<(3) But one does not tread upon the whole day.> P
<(4) Therefore one does not tread upon what one walks through.> T, from (2), (3)
<(1) Does one swallow what () one drinks? Yes.> P
<(2) But one drinks the cup (c ). P
<(3) Therefore one swallows the cup>.

, from (1), (2)
The usual trick is at work here. The pronoun in (1) ranges over individuals, whereas
the whole day and the cup in (2) signify, respectively, the time interval during which
one walks (in the category of time) and the receptacle from which one drinks (pre-
sumably in the category of place);2 therefore the conclusion can not be validly derived
from the premisses. I do not agree with Schreiber who takes (L) as a case in which the
confusion is not intercategorial, but between diferent items in the same category:
certainly that from which liquid is drunk is as much a substance as the liquid that is
1 We can easily turn these sophisms into refutations by adding initial questions like: Could one write a sen-
tence both false and true at the same time?, and Can one learn something which has opposite properties at the
same time?. However, they would be less efective since the initial question would give away what the sophist is
aiming at. Aristotle deals with the sophistic aim of leading to absurdity or falsity in se 12, and his remarks on the
relation between leading someone to absurdity or falsity and refuting someone can be found at se 15, 174b12-18.
2 I attempt to locate these items in the Categories categories, but, as I said above, this is not necessary in order
to save Aristotles consistency.
the arguments APA TO XHMA TH EE 189
drunk. It is simply a diferent substance ([24], p. 45). If this were true, there would be a
confation of diferent kinds of fallacy under the same heading form of expression;
however, we have just seen that (L) can be easily analysed like the other sophisms so far,
and therefore no inconsistency arises.
One minor diference from previous sophisms is that the terms involved in (K) and
(L) do not qualify other terms which signify individuals (as, e.g., in (I) and ( J)). Here the
actual term referring to a is left in both cases usefully implicit (One walks through
<something> the whole day; One drinks <something from the> the cup); should it
be expressed, these sophisms would not deceive anyone.
Dorion ([7], p. 358 n. 331) gives a diferent interpretation of (K), suggesting that the
sophism exploits the confusion between the transitive use of (
= suivre une route) and the intransitive one ( c = marcher
toute la journe). But this description seems somehow to make premiss (1) in (K) am-
biguous, which, once again, is not Aristotles own analysis.
The next argument is the following (178b34-36):
<(1) Is it either by learning or by discovery that one knows what () one knows?
Yes. T
<(2) But if of a pair of things (a ) someone has discovered one and learnt
the other, the pair is not known to him by either method.1 P
<(3) Therefore it is not either by learning or by discovery that one knows
what one knows.> T, from (2)
In (1) signifes (the thing known), whereas in (2) a means a quantity (al-
though one might fail to spot the diference, because of its syntactical role and of the
presence of the article). In fact to the frst question the answerer should reply: Yes, but
not the things which one knows (\ L ; cf. se 23, 179a23-24). This, how-
ever, does not mean that (1) is ambiguous, as Dorion suggests: le pronom relative
(178b34), qui est manifestement le pivot du sophism, est ambigu dans la mesure o il
peut tre compris comme un singulier individuel (ce que je sais = le chose que je sais)
ou comme un singulier collectif qui est lquivalent dun pluriel (ce que je sais =
lensemble des choses que je sais).2 This is not Aristotles analysis: as I have pointed out
earlier, it would not make sense for him to say that the relative pronoun has more than
one meaning.
Let us now take stock of some interim results. All the sophisms we have analysed in
this section
1) follow the same pattern: the sophist exploits some linguistic similarity between the
words, assuming tacitly that because they are similar in this way they must also signify
the same kind of thing and drawing his conclusion accordingly (contra charge 1 in the
2) do not rely on any form of ambiguity, and therefore do not overlap with other
kinds of linguistic fallacies (contra charge 3 in the Introduction).
We need now to consider only the fnal and most famous argument of se 22, to ver-
ify whether it fts and corroborates our analysis.
1 For the kind of reasoning which could support the truth of this premiss cf. s.e. m xi, 11-14.
2 Dorion [7], p. 358 n. 331.
190 ermelinda valentina di lascio
3. 5. The Third Man Sophism (178b36-179a10)1
The Third Man is an excellent example of Aristotles obscura brevitas in presenting ex-
amples of sophisms in se:
T11 <There is> also <an argument which concludes> the existence of a third man besides
man and the particular men (d \ e d f \
); for man and every common term do not signify an individual, but something
qualifed or quantifed or a relative or something of this kind (e a d e
e a j e j j ). Likewise,
also in the case of Coriscus and Coriscus the musician, are they the same or diferent? For,
the frst <term> signifes an individual, the second one something qualifed, so that it is
not possible to distinguish the latter from the former (\ e ). But
what leads to the conclusion that there is a third man is not distinguishing <man from par-
ticular men> ( e b e ), rather conceding that
<man> is essentially an individual; for it is not possible that also the essence of man, like
that of Callias, is being an individual. And if someone says that what is distinguished is not
essentially an individual, but something qualifed, nothing changes: for there will be a uni-
tary entity besides () the many <particular men>, i.e. man. It is clear, hence, that one
must not concede that what is similarly predicated of many things signifes an individual,
but that it signifes something qualifed or a relative or something quantifed or something
of this kind. (178b36-179a10)2
Aristotle informs us only of the conclusion of this sophism, namely that there is a third
man. Is it possible to reconstruct the premisses through which this sophistical refuta-
tion proceeds? It is not easy to answer this question, as shown by the fact that the
sophism Aristotle is referring to has been identifed in two radically diferent ways by
commentators. Let us turn to them.
At least twice in the Metaphysics (A 9, 990b15-17; Z 13, 1038b34-1039a3) Aristotle uses
to label an argument which he employs against the Platonic theory of
Forms. While no formulation of this argument can be found in that work, it is usually
supposed that a version of it is preserved by Alexander in his commentary on Meta-
physics A 9:
T12 One can prove the Third Man also in the following way. If what is predicated truly of sev-
eral things is also diferent from ( ) the things of which it is predicated, and it is
separated () from them [], there will be a Third Man. For, if what is predi-
cated is diferent from the things of which it is predicated, and has its own existence (\
), and man is predicated both of the particular men and of the Form, there
will be a Third Man besides () the particular men and the Form. Thus, there will be al-
so a Fourth Man which is predicated of that Third Man, of the Form and of the particular
men; similarly, there will be also a Fifth Man, and so on ad infnitum. (Alex. Aphr. in Metaph.
84, 21-85, 3)
This is not the only argument quoted by Alexander under the label of Third Man, but
one of the four versions he reports. Alexander claims that the frst was employed by
1 This section is an abridged and revised version of the frst half of Di Lascio [6].
2 I shall discuss here only the frst part of the passage (178b36-39). For the analysis of the rest of the passage see
Di Lascio [6].
the arguments APA TO XHMA TH EE 191
Eudemus of Rhodes in his d , and attributes the second to anonymous
sophists, the third to Polyxenus and, fnally, the fourth (T12s one) to Aristotle.
While almost all the modern commentators agree on the fact that the Third Man ar-
gument in question in T11 is the one Alexander attributes to Aristotle in T12,1 Pseudo-
Alexander (in se 158, 20-26) and the anonymous commentator (in se 54, 11-16) believe
that Aristotle is alluding to the version (the second one) which Alexander attributes to
the sophists:
T13 An argument presented by the sophists which led into the Third Man was the following. If
when we say Man walks we are referring neither to the Form of Man (for it is incapable
of motion) nor to any particular man (for how could we refer to someone we do not know?
For we know that man walks, but we do not know of which particular man we are saying
that he walks), then we are saying that some other Third Man besides () these walks;
therefore, there will be a Third Man of which we say that he walks. (Alex. Aphr. in Metaph.
84, 7-14)
Which argument is Aristotle discussing in se 22? In order to provide an answer, it is nec-
essary to consider carefully the context into which these arguments should ft, to veri-
fy whether the choice of one or the other could guarantee a sufcient internal coher-
ence to the text and a satisfactory understanding of it. Let us, then, continue the analysis
of T11 to discover the of the sophism proposed by Aristotle, and check whether
the arguments of T12 and T13 are compatible with it.
Aristotle explains that the Third Man sophism springs from treating something as
, when it is not (178b37-39: for man and every common term do not signify an
individual, but something qualifed or quantifed or a relative or something of this
kind).2 We have already said that () in se 22 refers to individuals (cf. p. 179 n. 3),
but the six occurrences of this expression in T11 (a term predicated of many things does
not signify (178b38); Coriscus is (179a1-2) and Callias is (179a4-5))
suggest that in this case indicates more specifcally what Aristotle calls
in the Categories. Man, then, unlike Callias, is not (179a4-5).
What is Aristotles solution? First we read that man and every common term
() do not signify , namely a particular entity. Te at 178b38 is equiva-
lent to e d at 179a8-9. A common term is, therefore, a
term which is predicated of many things. There are various places in the Aristotelian
corpus which show that Aristotle can use e interchangeably with e
(e.g. in Metaph. Z13, 1038b34-1039a3, in which Aristotle speaks of the Third Man similarly
to se, frst and then e d are used to express the
same idea) and the well-known defnition of in the De Interpretatione (7, 17a39-
b1) confrms this identity.
Thus, the trick of the Third Man sophism lies in the erroneous assumption that a uni-
versal term signifes an individual. As always, it is the sophist who tacitly deploys this
1 Cherniss [3], Dorion [7], Fait [10], Mariani [18], Poste [21], Ross [23], Schreiber [24], Sharma [25],
White [31]. Colli [4] and Taylor [29] disagree.
2 Bekker [2] reads: ead ee, a j jj
. Ross [22] adopts a diferent punctuation (without comma after ) and does not read j after
, but j before j . Ross reading might be preferable because it conforms these lines to the last
ones of the passage in which the same idea is expressed (179a8-10). However, would account for the addition of
a case like that of in sophism ( J) (which is not covered by the lists of categories at Cat. 4 and Top. i 9).
192 ermelinda valentina di lascio
false assumption to refute his interlocutor. Since the respondent did not concede
explicitly that man signifes , he has not really been refuted (cf. T4). But the
audience, if inexperienced, will believe the contrary. Here Aristotle does not explain
what deceives them, but a manoeuvre of the same kind as we have seen in the previous
arguments of se 22 should be at stake. A term denoting another kind of entity (sec-
ondary substance) is surreptitiously treated by the sophist as if it denoted a substantial
individual, namely a primary substance.1 The trick must be facilitated, here too, by
similarity: is a singular noun like K (in an important sense, it is
said as K) and this can help hide the fact that it should not be treated as if it de-
noted a primary substance, one in number. As Wittgenstein remarked, one of the
greatest sources of philosophical bewilderment is that a substantive makes us look for
a thing that corresponds to it.2 The false expectations which the term man can gen-
erate are mentioned also in the Categories (cf. T2), where Aristotle says that when we
say man or animal these nouns of secondary substances seemto denote a particular
entity .
But if a universal term does not signify , what does it signify? In T11 Aristotle
says that it signifes or or or something else of this kind (it is
clear that with Aristotle refers to the remaining kinds of things). The
same idea is repeated in the fnal lines (179a8-10), in which we fnd used inter-
changeably instead of (we fnd also at 179a6).3
But, more specifcally, what does man signify? What category does man belong to?
We fnd the answer in the second part of T2, where Aristotle claims that a secondary
substance signifes : for the subject is not one as the primary substance <is>,
but man and animal are said of many things. And what is said of many things, ac-
cording to the defnition at Int. 7, 17a39-b1, is a universal. We can presume that also in
T11 Aristotle intends man to signify (or ). Unfortunately, to say that
man signifes is quite obscure, because in the Categories Aristotle dedicates
the whole eighth chapter to the category of quality, distinguishing its various senses,
but he does not mention any sense in which man and animal could be identifed as
. Various scholars have discussed this problem, but the best solution seems to
be the one ofered by Ackrill,4 which is endorsed and developed by Devereux,5 and
which we fnd already in Simplicius (in Cat. 103, 8-31). This solution consists in recog-
nising that the term has various senses and, consequently, can be used in two dif-
ferent kinds of questions: the frst one is What kind (of man) is Socrates?, and re-
quests a quality as the answer; the second one is What kind (of substance) is
Socrates?, and requests the species or the genus to which Socrates belongs. When Ar-
istotle at 3b19 in T2 says that white signifes nothing else than , the term is em-
1 It must be noticed that Aristotle does not say that man signifes a secondary substance (in se there is no dis-
tinction between primary and secondary substances). However, it is plausible that he has in mind an analogous
distinction here too (or one which includes it).
2 Wittgenstein [32], p. 1. One could also think that the mistake is induced by the fact that the term man can
be used both as a predicate and as a defnite description in a specifc context (in which it would have the function
of denoting an individual). However, the fact that our sophism is presented as depending on the form of expres-
sion and not as depending on something like double meaning suggests that this is not what Aristotle has in mind
3 se 7, 166b10-19 shows that is used as , which is true also in the Categories (cf. T2). This seems to
be true also in the Topics: cf., e.g., vi 6, 144a17-19.
4 Ackrill [1], pp. 88-89. 5 Devereux [5], pp. 349-352.
the arguments APA TO XHMA TH EE 193
ployed in the frst sense; but when, at 3b15-16, he says that man signifes , the
term is used in the second sense.1
Thus, we can conclude that in T11 man also signifes or, interchange-
ably, , in the sense I have clarifed for T2. Therefore, the sophism discussed in
T11 will be solved by realising that man does not signify an individual; rather, it sig-
nifes what kind of substance a thing is, namely it signifes a species of a substance. In
other words, the sophism will be solved by recognising that the nature of man is to
be a universal.
Having seen how the sophism must be solved, we can go back to the thorny prob-
lem of its identifcation. Let us start by analysing the version of the Third Man argu-
ment which Alexander attributes to anonymous sophists in his commentary on
Metaphysics A 9 (T13).2 The sophism works as follows: if we say Man walks we are
saying neither that the Form of Man walks (for Forms are incapable of motion), nor
that some particular man walks (for we do not know of which man in particular
walks is predicated); we should then conclude that there is a Third Man, of which
we say that he walks.
Pseudo-Alexander, the anonymous commentator and the scholia all refer to this
sophism while commenting on T11, using terms almost identical to those used by
Alexander.3 This fact, together with Alexanders explicit attribution of the Walking Man
sophism to the sophists,4 could support the idea that this is the sophism discussed in
se 22. Nevertheless, the majority of modern scholars, following Cherniss, believe that
this is not the case.5 One reason is that this sophism does not conclude the existence of
a Third Man on the basis of the alleged similarity between the Form of Man and the
particular men, i.e. their being individual things, which instead seems to be required by
the overall logic of the passage. Consequently, Cherniss suggests that the Walking Man
sophism was quoted by the commentators of se just because they were deceived by the
fact that Alexander attributes it to the sophists: since se is a work on sophistical falla-
cies, they thought that the version of the Third Man discussed at the end of se 22 had
to be sophistical, and not the valid Aristotelian Third Man argument.6
1 T2 considered together with T11 might show that Aristotle had in mind two diferent classifcations of cate-
gories at two diferent times. One classifcation (in T11) considered things like man as (or ) and
therefore grouped them together with qualifed things in the sense of Cat. 8, although Aristotle was well aware
of their diferent nature (which he tries to explain, I believe, in the second part of T11). A second classifcation (in
T2) grouped secondary substances together with primary ones, although Aristotle stresses, here too, their im-
portant diferences. The fact that T2 explains away a primary substance as and a secondary substance as
might be a clue that T2 is later than T11. 2 Cf. p. 192 above.
3 [Alex. Aphr.] in se 158, 20-26; Anon. in se 110, 11-17; sch. 46, 54, 124, 164, 229 ad se 22, 179a2-3; sch. 116, 150, 327
ad se 22, 178b36; sch. 35 ad se 22, 178b39 (for the relations between the two commentaries and the various scholia
cf. Ebbesen [8], vol. i, pp. 259-261, 268-285, 290-301, 333-340).
4 One could also conjecture the sophisms paternity, given its similarity to the following argument of Stilpo
quoted by Diogenes Laertius (d.l. ii 119 = ii o 27 ssr): whoever speaks of man does not speak of anyone, since
he speaks neither of this man nor of that one (for, why should he be speaking of this man rather than that one?
Therefore, he is not speaking of this man).
5 Cherniss [3], pp. 290-291 n. 194 and Appendix iv (pp. 500-505).
6 Cherniss also suggests that Alexanders passage citing the various Third Man arguments (Alex. Aphr. in
Metaph. 83, 34-85, 12) was interpolated, with the addition of the section containing the sophistical one and the one
attributed to Polyxenus. Ebbesen suggests that the interpolator might be Pseudo-Alexander, who seems to have
revised Alexanders commentary on the Metaphysics after writing his own commentary on se. If this is true, it is
reasonable to suppose that the other commentators have simply followed Pseudo-Alexanders choice of the
sophism: therefore, we would have no explicit link between the SE Third Man and the Walking Man sophism.
194 ermelinda valentina di lascio
I shall now propose some arguments based on an examination of the logical compat-
ibility of the Walking Man sophism with what is said in T11 and undermine the idea that
this is the sophism to which Aristotle refers. To begin with, for this idea to work, it would
be necessary to understand \ as besides the Form of Man in Aristotles for-
mulation of the sophism (d \ e d f \
). This is the way in which many commentators have understood \ ,1
but this translation is neither the only possible one nor the most obvious, and actually it
is itself the consequence of a precise interpretation of the sophism. A further difculty
consists in the fact that if the sophism referred to in T11 were the Walking Man sophism,
it could be employed exclusively against a supporter of the theory of Forms.2 This
sounds strange: as far as I know, this would be the only example in se of a sophism so
specifcally aimed, and also the only place in which the Forms are alluded to. The latter
difculty could be partially overcome if one thinks that in the Topics there are places in
which Aristotle ofers strategies both for demolishing theses formulated by supporters
of the theory of Forms and for formulating theses in accordance with it; moreover, there
are two occurrences of the term , the meaning of which must be Man
himself or Form of Man (Top. v 7, 137b6; vi 10, 148a17).3 But could Aristotles solution
to the sophism in T11 be accepted by a supporter of the theory of Forms? It is evident
that if this sophism is addressed only to supporters of Forms its solution should be ac-
ceptable for them; but things seem to be diferent. The sentence Man walks makes one
believe that there is a Third Man because we can identify neither the Form of Man nor
any particular man as reference of the term man. Aristotles solution (for man and
every common term do not signify an individual, but something qualifed or quantifed
or a relative or something of this kind) could be applied to the Walking Man sophism
only in the following way: when we say Man walks the term man does not denote
an individual because, instead, it denotes a universal (in the sense that the human kind
has the capacity of walking). But this solution would be problematic: could a friend of
the Forms accept it, when it leads one to admit the existence both of the Form of Man
and of the universal man? This is unlikely: Aristotle formulates his notion of universal
just in opposition to the Platonic one of Form. Could Aristotle, at least, accept this solu-
tion (the solutions Aristotle proposes for the other sophisms he reports are in accordance
with his own doctrinal beliefs, based in primis on correct categorial distinctions)? This is
not possible on the present hypothesis: Aristotle could never postulate the existence both
of the universal man and of the Form of Man. At this stage, the only possibility left
would be to say that Aristotles solution allows one to somehow escape the sophism, but
it is merely an apparent solution since it commits one to a triple distinction (the partic-
ular men, the Form of Man and the universal man) that is unacceptable both for Aris-
totle and for a supporter of the theory of Forms.
1 Colli translates oltre alluomo in s ([4], p. 704), Dorion distinct de home ([7], p. 178), Mansion ct
de lHomme en soi ([17], p. 194), Poste besides the ideal man ([21], p. 71), Schreiber besides [man] itself ([24],
p. 44), Zanatta oltre quello in s ([33], p. 221).
2 The question When we say Man walks are we referring to the Form of man? will have the negative an-
swer which the sophist needs to proceed only if the answerer accepts that Forms exist and knows that they do not
move. It is not possible to think of a situation where the answerer just assumes (even if he does not believe), for
the sake of argument, that there are Forms and that they do not move: he will never accept the conclusion, but
immediately protest that Man walks refer to the human kind who has the capacity of walking.
3 Cf. Owen [19].
the arguments APA TO XHMA TH EE 195
In conclusion, I believe that it is implausible that the Third Man sophism discussed in
T11 is the one proposed by Pseudo-Alexander and the anonymous commentator,
namely the version attributed to the sophists in Alexanders commentary on Meta-
physics A 9 (T13).
The second argument to be examined is the which Aristotle ex-
posed, according to Alexander, in his d (T12). I shall not address the problem
of establishing the exact logical form of the Aristotelian or under-
standing its position in the work on Forms:1 these questions would require a long dis-
cussion and would take us too far from the passage we are interested in. What I shall
verify is, again, whether the hypothesis that the Aristotelian Third Man is the one dis-
cussed in T11 is compatible with what is said there.
Let us start by reconstructing the Aristotelian argument on the basis of Alexanders
(1) What is predicated truly of several things is diferent () from the things of which it is
predicated and has its own existence (\ ).
(2) Man, being predicated of several particular men, is diferent from them and has its own
existence (Form of Man).
(3) Man is predicated not only of particular men, but also of the Form of Man.
(4) Therefore, there will be a Third Man besides () the particular men and the Form of Man,
diferent from them and having its own existence.
(5) Man is predicated not only of particular men and of the Form of Man, but also of the Third
(6) Therefore, there will be a Fourth Man besides the particular men, the Form of Man and the
Third Man, diferent from them and having its own existence; and so on ad infnitum.
Alexander suggests that this argument is virtually identical to the one Eudemus em-
ployed in his d , in which it is plausible that also fallacies due to the similarity
of the form of expression were dealt with:2 if this is true, it is probable that Eudemus
discussed the Third Man argument among these fallacies. This could support the hy-
pothesis that Aristotle included among the sophistical refutations a e
the Third Man argument which he used in the d , which was virtually
identical to Eudemus.3 And, in fact, the idea that the Third Man referred to in T11 is
the Aristotelian argument has been accepted by most modern scholars. Cherniss sug-
gests that it is not a genuine problem that Aristotle has classifed his own objection to
the theory of Forms as a sophistical argument: it is only to reveal a real fallacy present
in the theory of Forms that the Third Man regress relies on the fallacious substantivi-
sation of the predicate man, typical of that theory.4
However, I believe that this idea faces insurmountable difculties which, again, can
be fully understood only when we try to square the Aristotelian Third Man with what
1 For this cf., e.g., Fine [11], Leszl [15].
2 In his d a c Galen, when dealing very briefy with the sophisms depending
on the form of expression (xiv 593.4-12 K.), seems to refer to Eudemus, even if some manuscripts are corrupted
and not all the scholars agree in reading E at 593.9 (ebbesen [8], vol. i, pp. 15-16 reads E<>). How-
ever, the idea suggested by Spengel and Prantl that this passage alludes to Eudemus d seems plausible.
In support of this, we should notice that Galen wrote a commentary on Eudemus work (Libr. Propr. 19.47, 10-11 K.).
3 Cf. cherniss [3], p. 290 n. 194.
4 Aristotle himself considers it [sc. the Third Man argument] to be a fallacy a e but
valid all the same against the Ideas because the demonstration of their existence from the d is itself
just this fallacy (Cherniss [3], pp. 290-291).
196 ermelinda valentina di lascio
is said in T11. First there would be a paradoxical inversion of roles: the sophist with his
argument would open the answerers eyes to a real fallacy that undermines his theory.
Hence, Aristotles solution would be the solution not only to the sophism but also to
the error underlying the theory supported by the answerer, who would necessarily be
a Platonist (it becomes necessary, again, to understand at 178b37 in the strong
sense of besides the Formof Man). Even more important is that if we accept Cherniss
suggestion, the Third Man argument discussed in se should not be considered a
sophism. The fact that it uses an assumption, the substantivisation of the predicate
man in the Form of Man, which is false (at least from an Aristotelian perspective) does
not mean that it refutes the interlocutor in a merely apparent or sophistical way, be-
cause the interlocutor is ready to accept that assumption, and the Third Man validly fol-
lows from it. The Third Man would then be a non-sophistical refutation which shows
that the (Platonic) interlocutor has contradictory beliefs, since he believes both in the
uniqueness of the Forms and in a theory from which a regress derives forcing him into
admitting the existence of infnitely many Forms of Man. Moreover, we have seen that
the fallacy at stake in T11 consists in treating man as . However, it is important
to remember that this sophistic trick should be in some way facilitated by the linguistic
form of (man), a singular noun: but where does this deception lie in the
d argument?
Now that both the initial candidates playing the role of the Third Man sophism in
se have been discarded, we should attempt to formulate a diferent hypothesis. I sug-
gest that the Third Man of T11 should be an argument with a structure analogous to
that of the Aristotelian Third Man of the d , but formulated in an interroga-
tive and elenctic form, not specifcally levelled against the Platonic theory of Forms,
and, obviously, sophistical, based on a category mistake somehow exploiting the lin-
guistic form of the term . The following is a possible reconstruction of such
a refutation:
(1) Is there a third man besides () man and the particular men? No. T
(2) Man ( ) is man ( ). P
(3) Socrates ( ) is (a) man ( ). P
(4) Callias ( K) is (a) man ( ). P
(5) Man is diferent () from Socrates. P
(6) Man is also diferent from Callias, and from all the other men. P
(7) Therefore man is something else besides () all men.1 from (5), (6)
(8) Therefore there is a third man besides man and the particular men. T, from (2), (7)
As one can see from my reconstruction, I do not take the of the initial ques-
tion (178b37) in the Platonic sense of besides the Man himself, but simply, and more
generically, as besides man. This is the most natural reading of the Greek text, since
refers epanalectically to . It is not by chance, then, that Cherniss, al-
though suggesting that T11s sophism is the Aristotelian Third Man, translates by
man, even if this is not compatible with his own interpretation of the passage.2 My
interpretation of makes the sophism general enough, starting from its initial
1 Cf. Alex. Aphr. in Metaph. 80, 8-15 and [Alex. Aphr.] in Metaph. 813, 22-29.
2 Also Pickard-Cambridge translates by distinct from man ([20], p. 304).
the arguments APA TO XHMA TH EE 197
question, to be employed against any interlocutor, and not conceived ad hoc for a par-
ticular, Platonic adversary.1
How does the sophism (N) work? (2) is accepted by the answerer as a trivial tauto-
logical assertion of self-identity (if he denied it, he would appear to be saying some-
thing paradoxical), and (3) and (4) are clearly accepted as straightforward predications.
(5), (6), and (7) are also easily accepted by the answerer because he takes them as stat-
ing the trivial non-identity () of man with any particular man (their denial would
lead him to assert, for example, that Plato is not a man, since he is diferent from
Socrates and Callias, who would be identical to man): this also confrms that he has
taken (3) and (4) in their most obvious predicative sense, and not as stating identities.2
Once the sophist has managed to have the answerer agree that man is diferent and
distinct from all men in (7), he infers (8), that there must be a third man besides man
and the particular men. But how does he get that conclusion? What he tacitly assumes,
but the answerer has never conceded, is that since the term in (2) is lin-
guistically similar to and K in (3) and (4) (they all play the same syn-
tactical role of subject, and they all are singular masculine nouns in the nominative
case), then it must also signify (the specifc point that they are particular men
is then established by the fact that man is predicated alike of all of them: (2), (3) and
(4)).3 By surreptitiously treating man as an individual man, the sophist draws his con-
clusion as if man in (7) were included in the group of all men (and then the man
which is distinct from the members of that group should be a further one, a third
man). Aristotles solution (man does not signify ) aims at unmasking that tacit
assumption not shared by the answerer (which might remain unnoticed thanks to the
inexperience of the answerer and also to the quickness of formulation typical of the
agonistic context).4
1 Of course, it is theoretically possible that a sophist employs this argument against a Platonic answerer who
understands man in (1) as the Form of man (and who, therefore, will be really refuted), but this is not what
Aristotle primarily has in mind. For this would be the case of someone holding a false belief which makes the
refutation real, but we have already discarded this scenario as a general interpretation of Aristotles sophistical
refutations depending on the form of expression (cf. above 3. 4). Actually that the Third Man Sophism is not the
Aristotelian Third Man because of the unwelcome consequences seen above confrms my general thesis that no
false beliefs of the answerer are involved in the sophisms.
2 It might appear odd that the sophist asks both whether Socrates is a man and whether man is diferent from
Socrates, but that this was possible is shown by the examples Aristotle gives us at se 5, 166b28-36, when he deals
with the arguments due to the fallacy of accident. Presumably, the answerer would never take Socrates is (a) man
as an identity. Therefore, the next question whether man is diferent from Socrates, which he takes as a denial of
an identity, would not seem odd to him.
3 This is, I think, also one reason why Aristotle does not solve this sophism by saying that in (3) Socrates is (a)
man and (2) man is man the predication is (a) man has diferent senses (in the former it would mean that
Socrates belongs to the human kind, whereas in the latter it could only mean that the human kind is identical to
itself ). For, it is the fact that man is treated as referring to something particular that decides the sense in which
to be (a) man is predicated of it, and it is not the predication itself which already carries a specifc meaning. If
man is taken to signify a universal, to be (a) man cannot apply to it in the sense that man is a member of hu-
man kind!
4 My reconstruction here is diferent in some aspects from the one I proposed in Di Lascio [6]. Changes have
been made to make this argument more similar to the previous examples in se 22. One advantage of this new ver-
sion is that two crucial questions (i.e. But isnt it the case that man is diferent from each of the men of which it
is predicated? and But could it be the case that Socrates, Plato and Callias are men, and this man is not?) are
now asked in a diferent and better way. Whether man is man is asked at the beginning of the argument, and sep-
arately from the question whether Socrates and Callias are men, so that less room is left for the answerer to doubt
whether he should clarify that man is not man in the same sense in which Socrates and Callias are.
198 ermelinda valentina di lascio
Reconstruction (N) satisfes those minimal requirements I had indicated and squares
well with the context of T11. Moreover, it confrms that 1) all the arguments in se 22 in-
volve, broadly speaking, categorial distinctions; 2) it is some of the linguistic features of
a word that allows the fallacy to take place and the sophist to trick the answerer; 3) there
is no ambiguity involved in any of these arguments; 4) when the answerer undergoes
these sophisms, he does not entertain any false belief about the ontological counter-
parts of the terms involved.
4. Form of Expression Revisited
We can now return to the general issue of elucidating the concept of similarity of
. The analysis of the sophisms in se 22 has revealed that this concept
is quite broad, encompassing various distinct items: identity of mood, voice and tense
(verbs), of case and number (nouns), of syntactical role, of ending (clearly, these are not
exclusive). Of course we cannot presume that Aristotle would have spelled out his no-
tion of linguistic similarity in precisely these terms. The few grammatical concepts he
seems to possess, or at least to which he refers in his work, are presented at Poet. 20 and
at Int. 2-3; would these be sufcient to fesh out satisfactorily his notion of similarity of
(form of ) language? I suggest not only that they are, but that Aristotle might have had
them in mind when speaking, broadly, of similarity of . Let us frst
quote his accounts of noun, verb, and variation:
T14 A noun () is a compound signifcant sound without time, no part of which is signif-
cant in itself (Poet. 20, 1457a10-12)1
T15 A verb () is a compound signifcant sound with time, no part of which is signifcant in
itself, as also in the case of nouns. For man and white do not signify when, but walks
or walked additionally signify the present time (the former) and the past time (the latter).
(Poet. 20, 1457a14-18)
T16 A variation () is of a noun or of a verb; on one hand it signifes of this (), or
to this (), or such things; on the other hand one (d) or many (), e.g. men or
man ( j ),2 or aspects of delivery (a ), e.g. question or com-
mand. For walked () and used to walk () are variations of a verb ac-
cording to these forms. (Poet. 20, 1457a18-23)
T17 Of Philo () and to Philo () and the like are not nouns but variations of
a noun. (Int. 2, 16a32-b1)
T18 Similarly, was healthy and will be healthy are not verbs but variations of a verb. They
difer from the verb in that it additionally signifes the present time, while they <addition-
ally signify> all times except the present. (Int. 3, 16b16-18)
I suggest that whenever in se Aristotle speaks of similarity of or the words
or phrases are similar in that either (a) they are both nouns, or (b) they are both verbs,
or (c) they are both the same variation of nouns or verbs. Let us see how this is consis-
tent with the examples in se 22:
1 Int. 1, 16a15 and Poet. 20, 1457a16 show that not only nouns but also adjectives are . Lucas suggests that
also pronouns and probably adverbs are included ([16], p. 202). Whitaker adds also participles ([30], p. 59).
2 Although here Aristotle mentions the nominative singular, I take it that this is not a variation of a noun, but
a noun. Probably he meant to refer to some singular variation of .
the arguments APA TO XHMA TH EE 199
In argument (A), , are the same variation of .
In argument (A), , are the same variation of .
In argument (B), , , are the same variation of .
In argument (B), , , are all .
In arguments (C), (D), (F), (G), (H), , , , and are the same varia-
tion of .
In argument (I), , , , are all .
In argument ( J), e , , are the same variation of .
In arguments (K), (L), c , c , are the same variation of .
In argument (M), a , are the same variation of .
In argument (N), , K, are all .
From T17 it seems that the nominative singular of a word is an (cf. also T20 be-
low), while the other infections are .1 Similarly, only the present indicative of
a verb qualifes as , while the other tenses and moods are (T16 and T18).
T15 might seem to contradict this, but I take it that the two examples walks and
walked are not both of (one is of a , the other of a ): Aristotles
point is that they both additionally signify time, while nouns and of nouns do
not.2 I also assumed that infnitives are variations of a 3 (it seems quite obvious that
we can exclude that they are , since we have said that only the present indicative
counts as ), although this is highly hypothetical since Aristotle never discusses in-
fnitives and there might be another possible interpretation.4
Nowhere does Aristotle explicitly claim that the relative pronoun -- and numer-
als like x-- are , but this seems to be a plausible conjecture, since they sat-
isfy T14s requirements and there is no other grammatical class recognized by Aristot-
1 However, it is possible that both the nominative singular and the nominative plural must be considered
: at Int. 2, 16b1-5 the criterion for being an is to form a true or false proposition when is, was, or
will be is added, and this is satisfed both by the nominative singular and by the nominative plural of a noun (of
course, for the plural we will need to add are, were or will be).
2 The problem could be solved by supposing that if something is a of a noun/verb, then it is also a
noun/verb (whereas the reverse is not true), but T17 and T18 do not seem to allow this.
3 What tense infnitives signify will be determined by the main verb of the sentence in which they occur.
4 To my knowledge there is only one passage in which Aristotle may be making an indirect comment on an
A is a compound signifcant sound of which some parts signify something by themselves (\ a) (for
not each consists of nouns and verbs, for example the defnition of man, but it is possible to have a
without verbs, certainly there will always be a part which signifes something) for example Cleon in Cleon walks
(x K K). (Poet. 20, 1457a23-28)
This is how Lucas ([16], pp. 33, 203) reads and punctuates the text. Gallavotti ([12], pp. 74-75, 178-179), instead,
proposes to put certainly there will always be a part which signifes something (
) out of the parenthesis and to read K[] ( is in the oldest mss) in place of
K. According to Gallavotti, this reading would explain why only K is signifcant: does
not signify any tense and does not express any predication since it is an infnitive. I think that Gallavottis reading
K is very appealing from a textual point of view and eliminates the problem of why Aristotle
would not consider the present as signifcant in itself as well; however, it is puzzling that Aristotle would
write K and not e K at the end of the sentence. On the supposition that Gallavottis reading of the
text is correct, we can understand that while K is signifcant in itself, is not, because it is just like a
signifcant part of a compound name which is not signifcant in itself. Just like is not signifcant in itself when
extracted from the noun K, but only tends to signify (Int. 2, 16a19-26; cf. Whitaker [30]), in the same
way is not signifcant when separated from the logos K, but only tends to signify some-
thing. Thus, at least when infnitives depend on another verb, it is possible that they should not be considered vari-
ations of verbs.
200 ermelinda valentina di lascio
le (verbs, variations, connectives, elements) which could include them.1 Moreover, their
(y-w-y, >-->, etc.) signify , , etc. and or ( just like the
of nouns: cf. T16 and T17), and we know that can only be of
or . The same is true of the indeclinable noun : I take it that all its
are identical. Given the existence of such a wide and poorly defned class of
(and therefore of their ), it is not surprising that so many things turn
out to be said similarly according to Aristotles criterion (cf. se 6, 168a25-26). This might
sound like a negative note; however, it is the generality of this classifcation that enables
Aristotle to distinguish very lucidly words said similarly and words not said similar-
ly and therefore allows him to analyze the sophisms depending on the linguistic forms
as he does. Moreover, I believe, this does not mean that Aristotle would not be aware of
the fact that e.g. and , albeit both nouns, somehow difer, but his way
to express the diference would be to look at the kind of things they signify.
Of course, the fact that some words are all of nouns or verbs will not be
sufcient to call them similar in form of expression. For example, in argument (E)
does not resemble a word which signifes an individual: for, although it is prob-
ably a of a noun,2 just like , it is not the same , i.e., to borrow Aristotles
jargon, signifes , while does not (possibly it signifes ). Aristotle
does not clarify why diferent instances of of diferent nouns (e.g. and
) are the same (he simply says that they both mean ), but the
fact that he possesses a concept of identity of allows us to use his own concepts
and terminology, and not anachronistic ones, to account for the broad phenomenon of
similarity of .3 Not only would it be anachronistic to refer to our
more refned grammatical categories to understand Aristotles solutions, but some of
these categories would have been of no use to Aristotle. For example, by labelling
as an adjective, as a relative pronoun, and as a substantive, one would miss
what they have in common: from this point of view Aristotles less refned grammar,
as I have already said, serves his distinction between words having similar forms and
words which have diferent forms much better.
When the fallacy of form of expression was frst introduced in se 4, Aristotle men-
tioned a distinct group of sophisms which he does not discuss further in se 22. In 2 I
suggested that he was hinting at cases in which, for example, a masculine noun is used
to refer to a neuter object. Does my present account of what counts as similarity of
expression suit them too? While the same broad categories of noun, verb, and varia-
tion would be insufcient, at Poet. 21, 1458a8-17 Aristotle distinguishes nouns on the ba-
sis of their gender: so we can say that is said similarly to not only because
they are both , but also because they are both feminine (however, names
something female, while does not). This narrows the gap between this species of
similarity of expression and that discussed in se 22: both can be spelled out in terms of
Aristotles own linguistic categories.
1 I am assuming here that Aristotle intended this grammatical classifcation to be exhaustive (cf. Poet. 20,
1456b20-21). This, of course, does not imply that it is really exhaustive, and, in fact, was not taken to be so by later
readers (cf. the possible interpolation of at Poet. 20, 1456b21 [Gallavotti [12], p. 169; Lucas [16], pp. 201-202]).
2 Cf. p. 199 n. 1.
3 Of course, there are important features of , , and which Aristotle did not (and could
not) mention explicitly (e.g. all the various possible syntactical roles which each could play), but we can suppose
that they were already somehow implicitly embodied in these notions.
the arguments APA TO XHMA TH EE 201
There seems to be no trace of this same notion of outside se. At
Poet. 19, 1456b9 we fnd the plural a , but this refers to types of dis-
course or speech acts, i.e. commands, narrations, prayers, questions, answers, threats
(it is difcult to be more precise because Aristotle does not deal with them further on
the grounds that they are not subject-matters of the art of poetics, but of the art of de-
livery). At Rh. ii 24, 1401a8 the meaning of is also very broad, i.e. the
way to arrange an argument (e.g. in such a way as to give the impression that it is an
enthymeme), while at Rh. iii 8, 1408b21 and 28 refers to the style of
discourse (rhythm and meter). Aristotles concept of similarity of in
se is, then, an unicum. It seems likely that Aristotle narrowed and adapted the looser no-
tion of () of which we fnd examples in the Poetics and Rhetoric in
order to use it with reference to single words or phrases when in se he had to deal with
a certain kind of sophism, which were otherwise difcult to classify.
5. Conclusion
Our analysis of Aristotles treatment of the fallacy of the form of expression in se con-
frms what I had anticipated in the Introduction: certain objections frequently advanced
in the literature do not survive careful scrutiny, and Aristotles exposition is more co-
herent and convincing than usually admitted.
Let us look briefy at a fnal interesting point which Aristotle makes when dealing
with the sophistical arguments which lead the answerer to solecism:
T19 And in a certain way solecism is similar to those refutations which are said to depend on
the fact that things which are not the same <are said> similarly. For, just as in the latter
() it happens to solecise about things (d ), so in the former
() it happens to solecise about words (d ): for
man is both a thing and a word, and so is white. (se 14, 174a5-9)
There are both a certain similarity and an important diference between the fallacy of
form of expression and solecism: just as to solecise is to speak ungrammatically with
words, similarly to treat certain expressions as if they referred to items in the same on-
tological category only on the basis of their superfcial linguistic similarity is speaking
ungrammatically of things (violating the philosophical grammar of those expressions,
we might say, as opposed to their linguistic grammar).1 To verify Aristotles claim we
need to analyse briefy an example of sophistical argument leading to solecism (se 32,
(1) Does that () which () you say truly exist truly? Yes. P
(2) You truly say something () to be (r) a stone (). P
(3) Therefore something () is () truly a stone (). Solecistic C, from (1), (2)
What we can extrapolate from Aristotles analysis of this sophism is that in (3) is
solecistically used in the accusative instead of the correct nominative case, which is the
way it should occur since it replaces the nominative of premiss (1) (to say is
not to say [] but [se 32, 182a11-12]). The answerer is invalidly led to the
1 Pace Dorion ([7], p. 317 n. 234), who understands T19 as saying that the solecism concerns things, whereas
the fallacy of form of expression concerns words.
202 ermelinda valentina di lascio
solecistic conclusion (its grammatically correct version being
), and the reason why the sophist might convey the impression that such a move
is warranted is that for other terms (like and ) diferent cases have identical forms:
T20 It appears <that the solecistic conclusion follows> because the variation of the noun,
which is diferent, appears similar <to the noun> (a e c
). (se 32, 182a26-27)
It is now clear what the sophisms depending on the form of the expression and those
leading to solecistic conclusions have in common: in the former the sophist exploits the
fact that the same linguistic form is used for items in diferent categories, while in the
latter he exploits the fact that the same linguistic form is used for diferent cases (in our
terms, and are used both for the nominative and the accusative, while Aristot-
le would say that they are used both for the nouns themselves and for one of their vari-
ations). T20 shows very well how the latter linguistic fact can be described in terms very
similar to those used for the fallacy of the form of expression. But while in the case of
solecisms it is the conclusion that is ungrammatical, in the case of the sophisms de-
pending on the form of the expression the premisses which the answerer accepts are
perfectly sound: it is only if taken too literally and according to their surface grammar
that they speak incorrectly of reality (the sophist tacitly adopts this unnatural reading
to infer some absurdity or refute the answerer).
To highlight these similarities and diferences between the two fallacies makes us ap-
preciate how refned Aristotles analysis was. Even if he had possessed a more precise
theory of grammar and syntax, he probably would still have formulated his account in
analogous terms, because he would have considered such an approach as the only one
capable of denouncing the real fundamental problem of all the sophisms involved, al-
lowing him to group them all together. The fact that Aristotles diagnosis and solution
require us to adopt a certain set of ontological beliefs did not escape his notice, but was
by no means perceived as a drawback by him. Rather, Aristotle seems to be almost
proud of this feature, at least if we judge by the way he introduces his solutions: we can
solve these sophisms correctly thanks to his formulation of the appropriate ontologi-
cal distinctions (T3). The identifcation and analysis of the group of sophisms depend-
ing on the form of expression and the formulation of their solution might be one of
Aristotles most important innovations in the ancient debate on linguistic fallacies,1 and
still remains an original and thought-provoking contribution which should be taken
into serious account in modern discussions of fallacies.
St. Johns College Cambridge
[1] Ackrill J. L., Aristotles Categories and De Interpretatione, Oxford, Clarendon Press, 1963.
[2] Bekker I., Aristotelis Opera, 5 vols., Berlin, Reimer, 1831-1870.
[3] Cherniss H., Aristotles Criticism of Plato and the Academy, Baltimore, John Hopkins Press, 1944.
[4] Colli G., Aristotele: Organon, Torino, Einaudi, 1955.
[5] Devereux D. T., Aristotles Categories 3b10-21: A Reply to Sharma, Ancient Philosophy, 16
(1998), pp. 341-352.
1 For example, in Platos Euthydemus, unlike what happens for the other fallacies, there seems to be no trace
of the sophisms which Aristotle classifes as depending on the form of expression and of his analysis.
the arguments APA TO XHMA TH EE 203
[6] Di Lascio E. V., Third Men: The Logic of the Sophisms at se 22, 178b36-179a10, Topoi, xxiii, 1
(2004), pp. 33-59.
[7] Dorion L.-A., Aristote. Les rfutations sophistiques: Introduction, traduction et commentaire,
Paris, Vrin, 1995.
[8] Ebbesen S., Commentators and Commentaries on Aristotles Sophistici Elenchi: A Study of Post-
Aristotelian Ancient and Medieval Writings on Fallacies, 3 vols., Leiden, Brill, 1981.
[9] Edlow R. B., Galen on Language and Ambiguity: An English Translation of Galens De Caption-
ibus (On Fallacies) with Introduction, Text, and Commentary, Leiden, Brill, 1977.
[10] Fait P., Aristotele. Le Confutazioni Sofstiche (Organon vi), introduzione, traduzione e com-
mento, Bari-Roma, Laterza, 2007.
[11] Fine G., On Ideas: Aristotles Criticism of Platos Theory of Forms, Oxford, Clarendon Press, 1993.
[12] Gallavotti C., Aristotele. DellArte Poetica, Milano, Mondadori, 1974.
[13] Huelsen R., Concrete Accidental Terms and the Fallacy of Figure of Speech, in Kretzmann N.
(ed.), Meaning and Inference in Medieval Philosophy, Dordrecht-Boston-London, Kluwer, pp.
[14] Kirwan C., Aristotle and the So-Called Fallacy of Equivocation, The Philosophical Quarterly,
29 (1979), pp. 35-46.
[15] Leszl W., Il De Ideis di Aristotele e la teoria platonica delle Idee, Firenze, Olschki, 1975.
[16] Lucas D. W., Aristotle. Poetics, Oxford, Clarendon Press, 1968.
[17] Mansion S., Note sur la doctrine des catgories dans les Topiques, in Owen G. E. L. (ed.), Ari-
stotle on Dialectic: The Topics. Proceedings of the Third Symposium Aristotelicum, Oxford,
Clarendon Press, 1968, pp. 189-201.
[18] Mariani M., Il Terzo Uomo nelle Confutazioni Sofstiche, in Cellucci C., Di Maio M. C.,
Roncaglia G. (eds.), Logica e flosofa della scienza: Problemi e prospettive. Atti del Convegno
della silfs (Lucca, 7-10 gennaio 1993), Pisa, ets, 1994, pp. 3-13.
[19] Owen G. E. L., Dialectic and Eristic in the Treatment of the Forms, in Owen G. E. L (ed.), Ari-
stotle on Dialectic: The Topics. Proceedings of the Third Symposium Aristotelicum, Oxford,
Clarendon Press, 1968, pp. 103-125.
[20] Pickard-Cambridge W. A., Sophistical Refutations, in Barnes J. (ed.), The Complete Works of
Aristotle, 2 vols., Princeton, Princeton University Press, 1984, vol. i.
[21] Poste E., Aristotle on Fallacies or the Sophistici Elenchi, London, MacMillan&Co., 1866.
[22] Ross W. D., Aristotelis Topica et Sophistici Elenchi, Oxford, Oxford University Press, 1958.
[23] Ross W. D., Aristotle. Metaphysics, 2 vols., Oxford, Clarendon Press, 1924.
[24] Schreiber S. G., Aristotle on False Reasoning: Language and World in the Sophistical Refuta-
tions, Albany, State University of New York Press, 2003.
[25] Sharma R. K., What is Aristotles Third Man Argument against the Forms?, Oxford Studies
in Ancient Philosophy, 28 (2005), pp. 123-160.
[26] Smith J. A., T in Aristotle, The Classical Review, 35 (1921), p. 19.
[27] Tabarroni A., Figura dictionis e predicazione nel commento ai Sophistici Elenchi di Egidio
Romano, Documenti e studi sulla tradizione flosofca medievale, 2 (1991), pp. 183-215.
[28] Tabarroni A., Figure of Speech and Aristotles Division of Fallacies, in Cellucci C., Di Maio
M. C., Roncaglia G. (eds.), Logica e flosofa della scienza: Problemi e prospettive. Atti del
Convegno della silfs (Lucca, 7-10 gennaio 1993), Pisa, ets, 1994, pp. 15-24.
[29] Taylor A. E., Parmenides, Zeno and Socrates, Proceedings of the Aristotelian Society, 16
(1915-1916), pp. 234-289.
[30] Whitaker C. W. A., Aristotles De Interpretatione: Contradiction and Dialectic, Oxford,
Clarendon Press, 1996.
[31] White N. P., A Note on E, Phronesis, 16 (1971), pp. 164-168.
[32] Wittgenstein L., The Blue and Brown Books, Oxford, Blackwell, 1958.
[33] Zanatta M., Aristotele. Le Confutazioni Sofstiche, Milano, bur, 1995.
204 ermelinda valentina di lascio
Jean Levi
Lchange de pomes
comme forme suprme de la persuasion
est par ces mots que Marcel Granet caractrise la langue chinoise dans les pre-
mires pages de son ouvrage magistral, la Pense Chinoise: Langage rude et fn la
fois, tout concret de puissance et daction, on sent quil sest form dans des palabres
o safrontaient des volonts ruses. Ces palabres o princes et ambassadeurs cher-
chent faire prvaloir leurs ambitions sans heurter la susceptibilit de la partie adverse
demandent perspicacit, prudence, modration, mais aussi esprit de dcision, got du
commandement et sens de la repartie. Elles sont prtexte banquets et ripailles o lon
se divertit en coutant les purs accents des lithophones, des cloches, des luths et des
cymbales aux sons desquels voluent les danseuses, lgres et gracieuses. Elles sont
aussi et surtout le lieu o schangent des propos graves, des paroles empreintes de sa-
gesse, o se profrent des leons de haute politique et de profonde vertu. Cest l, au-
tour des nattes disposes savamment selon le rang de chacun, en haut des degrs de la
grande salle dapparat que la fne feur de laristocratie faisait assaut de courtoisie et
dloquence afn dillustrer, ministre ou Grand ofcier, son prince, prince ses anctres
et sa ligne. Lors de ces banquets, il tait dusage dofrir ses htes un pome pour
tourner un habile compliment, en les comparant tel ou tel personnage des temps pas-
ss glorif par ces vers. Bien souvent, sous couvert dhommages courtois, les diplo-
mates les plus habiles faisaient connatre indirectement, au prix dun dtournement de
sens, leurs sentiments profonds et leurs vritables intentions. Si bien que ces discus-
sions diplomatiques o se mesurent les volonts ruses se bornent des changes de
chansons. Il ne sagit pas de pomes improviss pour loccasion mais des pices tires
dun compendium de chants qui forment comme le fonds de la sagesse immmoriale
des vieux pays chinois. On y trouve des hymnes dynastiques, des odes excuts loc-
casion des sacrifces aux anctres ou entonns lors des runions solennelles; y fgurent
aussi des chants damour changs par les garons et les flles dans les ftes paysannes
du printemps. Cest ce rpertoire folklorique que les hommes dEtat vont solliciter le
plus volontiers des fns persuasives, exprimant travers le langage du badinage
amoureux aussi bien les rapports de force entre Etats que les ambitions de ceux qui les
servent. Composs dans des circonstances tout autres que des pourparlers diploma-
tiques, le sens premier des pomes est subverti par un travail mtaphorique qui vise
transmuer lexpression spontane et directe du sentiment en un langage cod et allusif,
tout en lui insufant la force incitative de ce verbe o demeure une essence de nces-
sit qui est la vertu premire de tout rite et lme de la spontanit qui est le moteur de
tous les jeux. Lart rhtorique dalors consiste dans le choix du pome le mieux ap-
propri aux circonstances. Lhomme dEtat expert est celui qui sait, au cours des ban-
quets qui suivent lchange protocolaire de compliments et de cadeaux, slectionner,
dans le vaste corpus des chants, lode dont les paroles, dtournes de leur signifcation

concrte, aura la plus forte charge allusive dans le contexte de leur nonciation. Lin-
tention est signife de faon dautant plus persuasive que le sens nest pas dans les
mots, mais ct, dans ce qui est insinu, dans le jeu des substitutions paradigma-
tiques qui permet de renvoyer un lot de notations fgures de scnes images qui
parlent aux sentiments et produit grce au recours des emblmes concrets un efet
frappant sur limagination en sorte que les dsirs et la volont de linterlocuteur en
sont afects.
Lors dune assemble seigneuriale, le prince de Jin reproche un chef barbare de la
tribu Rong davoir pactis avec lennemi et lui rappelle devant les autres seigneurs les
bienfaits de ses anctres vis vis des Rong. Le prince Rong rfute point par point le r-
quisitoire de son accusateur par un discours parfaitement argument. Mais surtout il
prend soin de conclure sa proraison par le pome La mouche verte qui tout en tant
une mise en garde contre les calomniateurs laisse la porte ouverte la conciliation:
La mouche verte va bourdonnant se poser sur la haie,
prince aimable et avenant ne croyez pas le calomniateur!
la mouche verte va bourdonnant se poser sur les jujubiers.
Le calomniateur ne connat pas de bornes,
il sme la zizanie entre tous les pays
Tout est dans ces trois mots prince aimable et avenant; le prince comprend les inten-
tions du chef barbare, saisit la perche qui lui est tendue; revenu de meilleurs senti-
ments, il fait amende honorable et traite avec afabilit son alli.
Faut-il presser ses interlocuteurs de conclure un trait en prenant un engagement
ferme, faute de quoi on se tournera ailleurs on chantera Jupes trousse:
Si tu as pour moi des pense damour
je trousse ma jupe et passe la Zhen
Mais si tu nas point de penses pour moi
Est-ce quil ny a pas dautres hommes?
le plus fou des jeunes fous!
Mais de mme que lhabile diplomate est celui qui sait choisir le pome dont les conno-
tations sajusteront le mieux la situation en raison de ltincelle motive que le choc
entre le halo de signifcations vagues contenues dans la gangue des mots et les circons-
tances de sa dclamation russira faire jaillir, de mme on ne saurait sacquitter de ses
missions dambassadeur que si lon se montre capable de comprendre mi-mots les in-
tentions qui se cachent derrire les formules toutes faites des chants damour, car le
message vritable nest pas dans les mots eux-mmes. La signifcation vritable nest ac-
cessible qu des auditeurs dont loreille aiguise par une longue pratique sait dchifrer
les indices tnus, cachs dans ces formules anodines, dans ces poncifs dcors de motifs
ternellement repris, lintimit de la personne, son for intrieur.
Cest ainsi que, partir de cet usage de lchange de pomes crypts o le sens du
message mane de laurole incitative de leurs connotations, sest dveloppe la tho-
rie des propos subtils wei yan. Le Lshi chunqiu, une encyclopdie philosophique com-
pile vers 240 av. J. C., en fournit lillustration canonique sous forme dun dialogue en-
tre Confucius et un prtendant au trne vinc, le prince Bai, qui tout en ayant pour
thme la question de savoir en quoi prcisment consiste les propos subtils en est la
mise en uvre exemplaire, si bien que le dialogue est tout la fois question et rponse
la question, et, en se donnant lui-mme comme propos subtil, en fournit comme la
206 jean levi
mise en abyme. Le prince Bai, qui veut fomenter un coup dEtat, va trouver Confucius,
pour avoir son avis et demande:
Peut-on se faire comprendre mots couverts?
Confucius ne rpond pas. Le prince insiste:
Et si on jette une pierre dans leau, quen est-il?
Il suft de plonger pour la rcuprer, rpond Confucius.
Et si lon jette de leau dans de leau?
Le cuisinier Yiya tait capable de distinguer, au got, les eaux de la Zhe et celles de la Sheng
leur confuent.
Alors il ne sert rien davoir recours la parole allusive?
Qui a dit cela? Mais il faut connatre dabord le sens vritable de la parole.
Lencyclopdie philosophique en tire la leon suivante:
Le prince Bai ne comprit pas ce que Confucius avait voulu dire. Celui qui a saisi la signifcation
de la parole na pas besoin de recourir la parole qui nen est jamais que lappendice. Si on se
mouille pour attraper le poisson et court pour prendre le gibier ce nest pas par plaisir mais par
ncessit. Cest pourquoi la parole suprme abolit la parole, lagir suprme est non-agir. Hlas,
une intelligence superfcielle ne sattache qu laccessoire. Cest la raison pour laquelle le prince
Bai prit sous les coups de la loi.
Lauteur entend dire que le vritable homme dtat est capable de se faire compren-
dre mi-mots ou mme sans mots du tout, grce un simple silence. La parole ne
sera jamais quun pis aller, on ny recourt que faute de mieux; cest dune certaine fa-
on un aveu dimpuissance. Ainsi celui qui a compris cela, ayant le verbe rare et nig-
matique, ne se laisse ni percer jour ni deviner. Il peut ds lors agir linsu de tous et
surprendre ses adversaires. De fait, le prince Bai se dcouvrit beaucoup trop tt, lais-
sant ses ennemis le temps de sorganiser, de contre-attaquer et de le dfaire. Car cest
bien de secret dont il sagit, ou plutt de la science dun change de paroles qui qui-
vaudrait du mutisme; dun accord tacite qui permettrait des tres dlite de se com-
prendre sans recourir au langage ou bien qui transformerait le langage en le contraire
de ce quil est. Le dialogue du prince Bai et de Confucius est lui-mme dj explicatif.
Il vient clairer, dans le Lshi chunqiu, le clbre apologue de lenfant et des libellules:
Un jeune garon passionn de libellules, avait su si bien les amadouer quil pouvait
samuser avec elles des heures durant sans quelles fssent mine de scarter. Son pre
lui ayant demand un jour de lui en attraper quelques unes, pas une libellule ne vint
foltrer ses cts le lendemain: elles avaient subodor ses intentions. Ou plutt ds
lors quil ntait plus question dactivit dsintresse mais que le sujet agissant se
trouvait habit dune intentionnalit, celle-ci ne pouvait manquer de transparatre au
dehors et dalerter ceux qui en taient la cible. Le Liezi, un ouvrage taoste, reprend
la fable en mettant le garon aux prises avec des mouettes et en supprimant lchange
entre Confucius et le comploteur pour ne conclure que par la formule nigmatique:
Cest pourquoi il est dit: la parole suprme est dans la non-parole, lagir suprme est
dans le non-agir. Ce qui est accessible lintellect ne saurait tre que superfciel. si-
gnifant par l que si dj, alors que pas un mot na t profr, des mouettes devinent
les intentions amicales ou hostiles, que dire lorsquon sexprime en phrases articules
devant un auditoire lesprit en veil; mais surtout est prononc un arrt sans appel
contre la parole. Elle ne saurait vhiculer quun sens trivial, sans profondeur, que tout
un chacun est en mesure de capter, car il nest pas lexpression du libre jeu du dyna-
de la dispute sophiste au dialogue philosophique 207
misme ontologique. Un arrt quun autre auteur taoste, Zhuangzi, exprimera sous la
forme suivante:
Lcriture nexcdant jamais la parole, cest donc cette dernire qui est prcieuse; ce quil y a de
prcieux en elle nest rien dautre que lide exprime. Lide sattache toujours quelque chose
dinefable. Ainsi le monde apprcie les mots et les transmet par les livres. Bien quils soient hau-
tement rvrs, je les juge indignes destime. Car ce que lon vnre en eux ne saurait me les ren-
dre prcieux Hlas, tous mes contemporains demeurent persuads que les formes et les cou-
leurs, les sons et les vocables sufsent rendre compte de la ralit extrieure. Mais comment le
pourraient-ils? Cest pourquoi celui qui sait ne parle pas et celui qui parle ne sait pas. Comment
le commun pourrait-il comprendre cette vrit?
Cela implique quil est des moyens de se faire comprendre par dautres truchements que
la parole, ou bien en recourant une parole o ce qui est signif ne serait pas dans les
mots mais au del ou en de des mots, dans les silences que cachent les vocables. Ainsi
la vritable loquence serait dans une parole qui sabolirait dans lextnuation de son
sens. Cest pourquoi Confucius, qui se voulait le continuateur de la tradition de
lchange subtil sest exclam un jour: Jaimerai ne plus parler! Il savait plus quaucun
autre que Celui qui sait ne parle pas, celui qui parle ne sait pas, cest pourquoi le sage
pratique lenseignement sans paroles. Dans la pure ligne des banquets nobles, le Ma-
tre vise transformer le discours en geste efcace, faire de la parole la continuation
de laction grce la transmutation rituelle. Il sagit dlaborer des procdures qui sub-
vertissent la fonction langagire du langage afn que le sens des mots soit toujours en
dehors deux, quil acquiert par ce transfert de la signifcation des paroles au contexte
de leur locution une efcacit presque divine. La parole vraie, la parole qui ne sanan-
tit pas elle mme dans son propre bavardage rhtorique porte un message qui ne se
trouve jamais dans ce que les vocables dsignent. Ce qui est dit ne concide jamais avec
ce qui est exprim par les mots. Le sens est ailleurs, fottant dans cet indicible excdant
toute parole, parce quil est lexpression mme de lintriorit du sujet, dont la vrit
nest jamais accessible que par un dtour. Cest dire que le chant et les paroles quils v-
hiculent valent avant tout par leur fonction rituelle. Ainsi donc pour la tradition confu-
cenne le discours porteur de son seul sens est second par rapport lacte total quest
le geste formellement correct. Le geste se substitue au langage articul comme moyen
de persuasion pour produire un efet de contrainte plus efcace que le simple discours
argument. Les cas dactes persuasifs sont nombreux. Lun des plus beaux exemples est
fourni par Hou Chenzi, un Grand ofcier du Lu. Passant par le Wi, la faveur dune
ambassade au Jin, il est convi un festin par son ami, le grand trsorier de droite Ke
Chen. Celui-ci appelle sa troupe de musiciens sans la faire jouer. Et la fn du banquet
lui remet un jade.
Hou Chenzi, sans mme prendre cong de son hte quitte la principaut en toute hte.
Ses suivants stonnent Hou Chenzi explique alors:
Il ma donn boire pour rjouir, il a convi les musiciens sans les faire jouer pour me ma-
nifester ses soucis; il ma ofert un jade pour me dire quil comptait sur moi. Tout ceci montre
que le Wi va connatre des troubles graves.
De fait. Il venait peine de passer la frontire quune rvolution de palais clatait Ke
Chen trouvait la mort dans les trubles. Hou retournait au Wi emmener la femme et le
fls de son ami chez lui et pourvoir leur entretien. Il rendit le jade au fls de Ke Chen
sa majorit.
208 jean levi
Mobilit rhtorique et disputes sophistes
La rhtorique comme mouvement
Cependant, par un de ces paradoxes dont lhistoire a le secret, au moment o elle sla-
bore, la conception dun discours qui vaudrait plus comme chant ou chorgraphie que
comme argumentation joue un rle trs marginal dans la pratique politique; au cours
des trois sicles suivant elle connut une clipse presque totale avant de ressurgir sous
les Han et de simposer comme orthodoxie. Du ve au iiie sicle avant notre re les ma-
tres du discours tiennent le haut du pav. Sophistes et rhteurs exercent une tyrannie
telle que mme les tenants de la suprmatie du rite sur la parole se laissent eux aussi en-
traner par les sirnes du discours. Sans doute est-ce que lhistoire tait passe par l et
que les transformations socio-politiques qui avaient boulevers les relations entre Etats
et provoqu la dislocation de laristocratie Zhou avaient disqualif la ritualit et le lan-
gage crmoniel qui lui tait consubstantiel pour laisser la place la seule efcacit pra-
tique. Que de chemin parcouru en efet depuis lpoque de Confucius! En ces temps l
toute conduite tait attendue parce que circonscrite dans un temps et un code rituels,
chacun avait sa place assigne, tant dans lespace que dans la hirarchie: le paysan cul-
tivait la terre dans la campagne, lartisan fabriquait des produits la ville, le marchand
faisait circuler les biens sur les marchs, tandis que le noble, par la parfaite matrise des
gestes protocolaires, assurait lharmonie des difrents ordres.
La Chine a connu entre les vie - iiie sicles les bouleversements les plus rapides et les
plus profonds de son histoire. En lespace de quelques sicles, le rgime rituel de la
royaut Zhou, dj srieusement mis mal, achve de se dsagrger pour faire place
un ordre neuf dont lempire centralis et absolutiste sera laboutissement ultime.
Les mutations dans lorganisation matrielle entrane lapparition dune classe de sp-
cialistes verss dans les arts conjecturaux: devins, sophistes ou stratges. Louverture du
temps sur linfni des possibles est source de mobilit, mobilit sociale et spatiale, mais
aussi intellectuelle. Le sophiste, lambassadeur, ou le stratge, hommes ns du change-
ment, le suscitent et le devancent; tout au moins se montrent-ils habiles sy plier. Avec
lui se dveloppe une forme dintelligence indite, caractrise par la promptitude du
coup dil et de lexcution, par lagilit mentale et dont le domaine comprend tout ce
qui est instable, fuctuant.
Cette mobilit va se traduire en premier chef par le caractre itinrant de lensei-
gnement. Lenseignement noble se droulait dans des lieux spcifques dont les formes
et la disposition avaient une valeur symbolique, en mme temps que leur schma ar-
chitectural sajustait au contenu des matires et au type de disciplines qui y taient pro-
digus. Cependant, paradoxalement, cette activit, enracine dans un lieu au point que
cest de celui-ci que drive le mot servant la dsigner la salle de tir larc , sera iti-
nrante au moment o se constitue le modle de lenseignement lettr. Ltude se pas-
sera de tout lieu, ou mieux son lieu fut le non-lieu des routes et des chemins. Les tri-
bulations de Confucius, premier matre, qui se dplace travers les provinces des
principauts du milieu, entranant sa suite une foule de disciples, en qute dun prince
servir ofriront le paradigme de toute activit denseignement durant au moins trois
sicles. Lenseignement, de sdentaire, se fait transhumant. Il nest savoir que dans la
mobilit, dune part parce quil faut voyager pour sinstruire et que dautre part les
disciplines qui sont prises rclament de la ductilit desprit et enfn, ces savoirs et ces
de la dispute sophiste au dialogue philosophique 209
comptences sont le moyen de la mobilit sociale. La politique ouvre un ambitieux
disposant de quelque talent la meilleure des voies pour parvenir aux honneurs. Cest
ainsi que beaucoup se dtournent du mtier de leurs pres pour sengager dans cette
carrire et se gagner du renom. Ce groupe de professionnels de la chose publique, en-
glob sous le terme trs gnral de shi, se distingue des classes traditionnelles aussi
bien des nobles que des paysans par son instabilit, une instabilit la fois gogra-
phique, sociale et intellectuelle. Gographique dabord: les politiciens ne cessent dal-
ler dune principaut lautre proposer leurs services aux dirigeants,et les coles o les
disciples reoivent des matres une doctrine, ou mieux une recette (dao) sont itinrantes;
sociale ensuite: ds lors quil a trouv un prince servir, le lettr est propuls de la boue
du ruisseau aux ors ministriels, un va-nu-pieds peut tre promu, lissue dune ou de
deux entrevues, de la plus haute dignit; intellectuelle enfn et surtout: lintelligence se
dfnit comme la capacit danticiper sur les vnements et de se plier aux alas des cir-
constances. Le spcialiste de la politique ne se caractrise pas seulement par sa mobilit
physique et mentale. Homme de parcours, il est aussi et surtout homme de discours.
En se faisant vagabond, lenseignement va librer la parole. Le matre de sagesse de-
vient matre du langage. De ritualiste il se fait rhteur. Tous les lettrs, quelle que soit
leur doctrine doivent consacrer une large place lart de la persuasion, car partir du
moment o lenseignement se donne pour la rponse opportune aux sollicitations de
linstant et des retournements de circonstances, il tait invitable que se dveloppe un
art de la repartie, art dautant plus ncessaire que le but ultime de lenseignement vise
former un personnel politique. Mme les mules de Confucius se convertissent la
rhtorique. Xunzi reconnat quil ne saurait y avoir de vritable sagesse sans loquence.
Mencius est rompu toutes les subtilits de lart oratoire. Lart de faire suppose donc
un art de dire. Et dao, le chemin renvoie, certes, une doctrine ou des recettes (dao),
mais encore la parole (dao) qui la transmet. Ces hommes des recettes (dao) sont donc
des hommes du chemin (dao) qui est indispensable la matrise de la parole (dao). On
comprend ds lors le rle fondamental que devait jouer lloquence durant la priode
des Royaumes combattants. La rhtorique y devient la reine des sciences. Elle obit en
efet une triple ncessit: elle est ncessaire pour obtenir un emploi auprs dun
prince; ncessaire pour avoir le dessus dans des joutes oratoires contradictoires; nces-
saire aussi et surtout dans les tractations diplomatiques pour infuencer et convaincre.
Le verbe sophiste, se modelant sur son contenu, se caractrise par son extrme ducti-
lit. Il ne prtend aucune vrit autre que celle de lefcacit pratique. Lart oratoire
chinois sintresse en premier lieu la dimension performative du langage. La parole y
est action. Tant Hui Shi que Gongsun Long qui sillustrrent par des paradoxes fameux
furent chargs de missions importantes. Rien ne les distingue sinon leur degr de
moindre russite des diplomates et des agents doubles tels Zhang Yi, Su Qin, Su Dai
ou Gan Mao. Aussi la dfnition courante des sophistes grecs, donne par les diction-
naires, tels le Grand Robert, pourrait sappliquer mot pour mot leurs homologues de
lEmpire du Milieu: ctaient eux aussi des matre de rhtorique et de philosophie, des
sortes de confrenciers qui allaient de ville en ville et enseignaient lart de parler en pu-
blic, les moyens de lemporter sur ses adversaires dans une discussion, de dfendre par
des raisonnements subtils ou captieux nimporte quelle thse. Ce qui fait mentir las-
sertion de Taine, qui afrme, dans sa Philosophie de lArt: Nulle part ailleurs (quen
Grce) on vit un groupe dhommes minents et populaires enseigner avec succs et avec
gloire, comme le faisaient les Gorgias, les Protagoras et les Polus, lart de faire paratre
210 jean levi
bonne une mauvaise cause et de soutenir avec vraisemblamce une position absurde si
choquante quelle ft. La seule difrence peut-tre entre les deux espces de sophistes
de lHellade et de la Chine cest que les seconds apprenaient tout autant lart de parler
en priv quen public et avaient dvlopp un art de la persuasion dans le tte tte ou
dans le secret des cabinets du Prince. En un mot, ils cumulaient les talents de diplomates
et dorateurs.
Largumentation des rhteurs joue sur la prminence accorde au temps sous la
forme des circonstances de son nonciation; mais les sophistes ne peuvent pervertir le
langage que parce quils sautorisent la mtaphore et la comparaison qui elles sont
dordre synchronique. Par les glissements smantiques quelles favorisent, elles four-
nissent un instrument puissant dans la manipulation frauduleuse des catgories lo-
giques. On raconte une anecdote amusante propos de Gongsun Long. Elle montre
limportance de lenjeu. Comme le roi Min de Qi, protecteur des arts et des lettres
sextasiait de lhabilet oratoire de Gongsun Long devant un de ses rivaux, celui-ci ft
remarquer quil devait ses succs nullement la rigueur logique mais son art de la
mtaphore et il lui recommanda de lui en interdire lemploi. Le roi du Qi, branl,
convoqua le rhteur et lui signifa que dsormais il devait sabstenir en sa prsence de
recourir lanalogie, la comparaison ou la mtaphore.
Bien, ft Gongsun Long, si quelquun demande ce quest une arbalte et que je lui rponde une
arbalte est une arbalte estimeriez vous que jai rpondu de faon satisfaisante, et que mon in-
terlocuteur pourra savoir ce quest une arbalte?
Certes non, dit le roi.
Si je lui dis une arbalte est une sorte darc muni dun mcanisme spcial, qui permet de
dcupler la tension de la corde est-ce que cela serait plus clair et permettrait de comprendre ce
quest cet objet?
Ainsi vous voudriez que je renonce au seul moyen de faire toucher du doigt mon interlo-
cuteur des notions qui lui sont totalement trangres dans le domaine des ides alors que vous
venez den admettre la validit pour de simples objets matriels? Linconnu ne se saisit qu par-
tir dapproximations connues.
Les sophistes tirent leur force de leur science du discours et de la parfaite matrise des
fgures de style; ils ne peuvent en efet pervertir le langage que dans la mesure o ils
en ont dmont les mcanismes. Ils ont compris tout le proft quon peut tirer de
lusage abusif de la mtaphore et de la comparaison. Certes, la comparaison, ainsi que
ne se fait pas faute de le rappeler Gongsun Long est indispensable dans toutes les op-
rations de lesprit. Elle seule permet, par approximations, de comprendre linconnu
partir du connu. Ainsi on ne pourra faire comprendre un interlocuteur qui ignore ce
quest une arbalte quen la rapprochant dun arc. Toutefois, ouvrant la porte tous
les amalgames, elle peut conduire aux pires abus, toutes les dsignations devenant
mensongres. La pratique de la langue des sophistes, loin de servir llaboration
dune science objective des catgories du discours, va tre utilise des fns subversives
et nihilistes.
Le tour de galop du cheval blanc
Le sophiste est un dlinquant. Il violente les catgories logiques et sintroduit en tratre
dans les sentiments ou les penses de ses interlocuteurs.
de la dispute sophiste au dialogue philosophique 211
Pour la plupart des discoureurs des Royaumes combattants, le langage est avant tout
un moyen dmouvoir lautre, de le forcer sortir de sa rserve. Ce qui ne veut pas dire
faire appel au pathtique; cela reviendrait sen remettre au bon vouloir de son inter-
locuteur; il sagit dexposer la situation de telle faon que lautre, pouss par la nces-
sit incoercible des vnements ou plutt de leur prsentation tendancieuse , se
trouve oblig dagir comme souhait son corps dfendant. Largumentation doit pos-
sder la puissance contraignante dun canal de drivation ou dune pente escarpe do
dgringolent pierres et billes de bois. Une dispute entre sophistes se joue comme une
partie dchecs: un change de bons coups lissue duquel lun des adversaires ne trouve
plus rien rpondre.
Cet art de la rplique fond sur le renversement dassociations fges nest efcace
que dans linstant; il requiert une science de lopportunit. Il est indissociable de cette
agilit desprit, de cette ductilit mentale acquise au prix de mille rencontres faites au
dtour du chemin et qui contraignent avoir la cervelle toujours en alerte pour tre
immdiatement capable de rpondre au bon coup de ladversaire par une parade su-
prieure. Lenseignement dispens par les rhteurs, par laccent mis sur les circons-
tances et les dispositions de lauditeur, ainsi que par la mobilit confre, grce au re-
cours massif lanalogie et la mtaphore voire la fable ou lanecdote, fait du discours
un instrument de manipulation: il est usage tactique. En ce sens la rhtorique est trs
proche de lart de la guerre chinois tel quil est expos dans le Sunzi. La stratgie sup-
pose linvestissement du temps par lespace; la tactique,elle, se meut dans linstant. En
raison de son non-lieu, elle est tributaire du temps; il lui faut guetter le moment pro-
pice afn de saisir au vol les possibilits de proft. Elle joue avec les vnements pour en
faire des occasions. La stratgie postule un lieu susceptible dtre circonscrit comme
un propre. Tandis que la tactique, dpourvue de lieu autre que celui de lautre,
cherche profter de ce lieu et de jouer de la force trangre. Sunzi parle de construire
sa victoire sur les mouvements de lennemi ce qui est une faon de reconnatre que
lon se repose sur lautre pour conqurir lespace partir dune pratique intelligente du
temps. Il lui faut utiliser les failles que les conjonctures particulires ouvrent dans les-
pace dif par la stratgie. Cest ce que le Sunzi appelle dans son langage imag: Vous
vous prsentez dabord comme une vierge timide; lennemi laisse ber louverture,
alors, rapide comme le livre vous vous engoufrer dans louverture sans lui donner le
temps de la fermer. Ce sont les mmes qualits dopportunisme et dutilisation des
mouvements de ladversaire qui sont mises en uvre par la rhtorique chinoise.
Comme le militaire, le sophiste pratique lefraction. Comme le dit avec force un trait
de rhtorique, le Guiguzi, le Livre de Matre Val des Dmons, mis sous le patronage du
professeur de rhtorique des deux plus clbres diplomates et hommes dEtat de la fn
du ive sicle av. J. C, Zhang Yi et Su Qin: Lessentiel cest douvrir la brche et une fois
quon a devin les dsirs de celui quon veut circonvenir, on lance sa ligne avec la bonne
amorce, et, lui faisant miroiter lobjet de ses secrets espoirs, on le tient alors sa
Le caractre frauduleux du personnage se manifeste de faon exemplaire dans le pa-
radoxe un cheval blanc nest pas un cheval de Gongsun Long. La formule non seule-
ment dstabilise les rapports logiques impliqus dans le systme des dsignations mais
encore mine les fondements de la socit. Les sophistes jourent en virtuoses des rso-
nances inconscientes des images; elles leur permirent, travers la mise en cause des ca-
tgories grammaticales, de bousculer les catgories sociales. Lexpression un cheval
212 jean levi
blanc nest pas un cheval (qui vise dnoncer larbitraire de la distinction opre par
la langue entre lobjet et ses qualits) fait mouche en raison de la charge mtaphorique
du cheval. Larrire-fond symbolique qui sattache lanimal fait que la phrase renvoie
immdiatement des noncs tels que un fls pieux nest pas un fls un loyal sujet
nest pas un sujet, etc., formules qui remettent en cause le rle de catgorisation so-
ciale prts aux noms et au langage dans le systme du ritualisme confucen. Le para-
doxe va faire des ravages. La contamination de la pratique sophiste stend tous les
courants. Mme les moralistes confucens vont y avoir recours pour faire prvaloir
leurs vues et tenter de restaurer lantique systme de valeurs. Le cas de Mencius est
exemplaire. Lui qui ne cesse de pourfendre le cynisme des sophistes ne peut se passer
de combattre avec leurs armes dans les joutes dialectiques qui lopposent aux tenants
des coles adverses. Cet change entre Mencius et un contradicteur, Elvation, pro-
pos du caractre spontan des vertus sociales, est difant:
Je vois quelquun qui est mon an, et pour cela je lhonore en tant quan. Mais cette qualit
dan ne se trouve pas en moi mais est extrieure moi. Cest comme la qualit dun objet qui
nest pas en moi: je considre une chose blanche comme blanche, ce blanc ne mest pas propre
mais se trouve dans lobjet. De ce que la qualit de blanc nous est extrieure jen dduis de mme
que le devoir est extrieur nous.
Sil ny a aucune difrence implique par la blancheur entre un cheval et un homme, en
va-t-il de mme pour lge chez un cheval et un homme? Fais-tu dpendre le devoir de lge de
lobjet ou bien du respect que tu dois lge?
En sorte quil semblerait quil ne puisse se drouler de discussion sur quelque sujet que
ce soit sans que le cheval blanc ne viennent y faire son petit tour de galop pour semer
la pagaille!
Discours logiques du lieu, discours captieux du moment
Les Chinois opposent le ci, le discours rhtorique et feuri et le discours logique et ar-
gument li; lun, discours raisonnable, permanent ressortit lespace, lautre le discours
spcieux, qui ne vaut que par les circonstances de son nonciation est de lordre du
temps. Le discours raisonn est de lordre de lespace en ce quil nest pas li un temps
particulier. Au contraire, il est dot dune validit universelle et transhistorique. Sap-
puyant sur les lois de la raison confondues il est vrai chez les lettrs traditionalistes
avec la Parole des Anciens il se donne pour lexpression de rgles qui sont les mmes
que celles qui commandent aux mouvements clestes. Il est rgi par les mmes lois qui
sont luvre dans la nature et qui structurent le relief li et commandent la rpar-
tition des populations sur le sol en groupements territoriaux li. Discours sans ornement
qui ne table que sur sa propre logique et sa propre rigueur de dduction pour empor-
ter ladhsion, il est aride. Il se refuse aux grces de la stylistique et aux rgles de la rh-
torique, il dit le vrai et donc ne chatoie pas. Le lettr qui y recourt se met en position
dinfriorit car il se prive des moyens de sduction du langage, mais ce dsavantage mo-
mentan se rachte par une victoire long terme: avec le passage du temps, sa validit
et sa vrit se rvlent aux interlocuteurs profonds et rfchis. Ceux qui recourent au
li, au discours logique, sont toujours aplatis dans la discussion; ils sont pris dans les an-
neaux mouvants dune dialectique chatoyante et ondoyante comme les anneaux lisses
de la dispute sophiste au dialogue philosophique 213
dun grand serpent. Le ci quant lui blouit, il tincelle et produit un nuage de paillettes
qui une fois dissipes laissent voir le vide de largumentation linterlocuteur abus.
Cest pourquoi il est de lordre de linstant. Les changes entre le sophiste Gonsun Long
et le confucen Kong Zigao, un descendant en ligne directe du Matre en personne, tels
que les ont rapports le Kong cong zi, Le Livre des matres de la famille Kong sont
exemplaires de ces deux types darguments. Dans les disputes contradictoires qui lop-
posent son adversaire devant le prince de Pingyuan de Qi, Zigao se fait chaque fois
clore le bec; il ne parvient rfuter les thses du sophiste qu tte repose, une fois quil
se trouve seul seul avec le prince. En un mot il ne peut laborer une dfense et d-
monter la faute logique de lautre que lorsque le facteur temps nintervient pas, quand
il na pas rpondre du tac au tac un coup imprvu port par un adversaire qui sait
profter de tous les ressorts de la rhtorique et est habile dstabiliser linterlocuteur
par des arguments inattendus, des formules incongrues, des raisons spcieuses mais in-
gnieusement travesties en enchanements rigoureux. Aussi ce nest quune fois que
Gongsun Long a quitt la place quil peut dcrter: Ses propos sont faux mais trs ru-
dits; ils sont ingnieux mais contraires la logique, voil pourquoi jai prfr me taire.
Mais le mieux est den fournir un extrait, on comprendra immdiatement de quoi il
Kong Zigao se rendit au Zhao et rencontra Long la cour du prince de Pingyuan. Une fois en
prsence du matre, il lui dit:
Ce qui mest venu aux oreilles de votre conduite et de vos discours ma donn envie de vous
connatre. Toutefois en dpit de lestime que minspire lensemble de votre enseignement, il est
pourtant un point qui me chifonne: cest votre paradoxe cheval blanc nest pas cheval. Si vous
en dbarrassiez votre doctrine, je serais prt devenir votre disciple et vous servir avec la fd-
lit du chien.
Vous venez de profrer une normit. Mon enseignement consiste justement rectifer les
dnominations grce ce paradoxe. Si je le supprime de mon enseignement, autant renoncer
ma carrire de pdagogue. Alors nest-ce pas aller contre le plus lmentaire bon sens que de
vouloir se mettre sous ma frule juste quand je naurai plus rien enseigner? En principe on prend
un matre parce quon estime quil vous est suprieur dans une discipline, et quil a donc des choses
vous apprendre, que ce soit dans le domaine de la connaissance, de la raison ou de la sagesse; or
vous me demandez dabandonner ce qui fait ma supriorit pour me suivre comme disciple.
() Mais pour en revenir cette question du paradoxe cheval blanc nest pas chevalqui vous
tarabuste tant, je vous rappellerai que la mme thse a t soutenue par votre anctre Confu-
En efet le roi de Chu, sen alla chasser le dragon et le rhinocros dans le parc de Yunmeng.
Au retour, il perdit son arc prcieux, son entourage voulut organiser une battue pour le rcup-
rer. Le roi, fataliste, les arrta:
A quoi bon? un homme du Chu la perdu, un homme du Chu le retrouvera.
Confucius layant appris dit:
Le roi de Chu ne pratique pas la charit jusquau bout. Il aurait d dire un homme la perdu
un autre le retrouvera, quoi bon ajouter du Chu?
Dites-moi, Messire nest-ce pas illogique dapprouver Confucius qui difrencie un homme de
Chu dun homme tout court, et de me contredire quand je distingue un cheval blanc dun che-
val tout court?
() Sur le coup, Zigao ne trouva rien rpondre; mais une fois que le sophiste se fut retir, il
Je nai pas voulu rpondre quelquun qui profte de sa vaste rudition pour assner des
contrevrits et qui se sert de son habilet rhtorique pour habiller des paradoxes.
214 jean levi
Un autre jour le prince de Pingyuan invita tous ses htes et clients un banquet en lhonneur
de Zigao et lapostropha en ces termes:
Vous navez pas hsit faire une longue route pour venir me trouver, et maintenant vous
vous apprtez nous quitter! Est-il possible que vous partiez sans quon ait pu dcider de la va-
lidit des thses de matre Long?
Quand les arguments sont rellement convaincants la vrit apparat delle-mme; elle ne
saurait tre lie mon dpart.
Puis-je entendre vos raisons?
Je ne donnerai pas mon opinion personnelle, mais je me contenterai de citer les classiques
qui contiennent la rfutation de ses thses. On lit par exemple dans les Annales des Printemps et
des Automnes: Six hrons ont vol reculons: en les observant de loin, on a vu quils taient six,
en les examinant plus attentivement on a distingu quil sagissait de hrons. Dans cette expres-
sion les hrons font pendant au cheval et le nombre six la couleur blanche de la formule che-
val blanc. Dans la perception dun objet, la couleur le blanc en loccurrence , apparat au pre-
mier regard; puis quand on scrute plus en dtail on se rend compte de la forme, on dcouvre
alors quil sagit dun cheval. Les couleurs se distinguent par difrents noms, linstar des qua-
lits internes qui ont leurs propres manifestations externes. Cest pourquoi quand je dis bai ma
blanc cheval,1 il y a correspondance entre le nom et la chose. De la mme faon, de la soie ou
des fbres de chanvre peuvent, suivant la faon dont elles sont traites par les tisserandes, don-
ner de la toile noire ou crue; mais noir ou beige, mme si le nom des couleurs difre, le mat-
riau,lui, est le mme. Cest pourquoi les pomes du Shijing parlent de su si crue soie, et non
pas de si su soie crue; le Livre du rituel parle, quant lui de zhi bu noire toile et non linverse,
de noir buf etc on pourrait citer des exemples linfni. On nomme dabord la couleur puis
lobjet ou le matriau. Et il en va de mme de toute chose. Le dterminant prcde le dtermin;
les sages, depuis lantiquit jusqu nos jours, ont toujours procd de cette faon.
Dans ses discours le sage prise la conformit lordre des choses (wuli) et non pas lhabilet
rhtorique (ci)
Le prince se tournant vers ses invits dit:
Pensez vous que Long soit capable de rfuter ces arguments?
Un des invits dit:
Il aurait certainement des arguments spcieux opposer, mais il ne le pourrait pas sur le ter-
rain de la logique.
Lart du retournement de dispositif chez Mencius
Le discours dfnit un espace signifant et peut valoir ce titre comme terrain de mani-
pulations ou si lon prfre de champ de manuvres, la faon des damiers, des chi-
quiers, ou des tables du jeu de loie. Il est propice au dploiement tactique. Face un
adversaire, il se prsente comme un parcours hriss de piges, de chausse-trappes, de
faux-semblants et de trompe-lil. Si le discours sophiste est tortueux, sil recourt des
stratgies biaises, il sapparente ces arts de la ruse cette mtis des Grecs remar-
quablement analyse par Jean Pierre Vernant et Marcel Dtienne. Aussi beaucoup de
dialogues ou de disputes entre sophistes et matres doctrinaux doivent se lire comme
une srie de mouvements de joueurs dchecs. On propose ladversaire des gambits;
on le force par un coup rpondre par un coup oblig son corps dfendant. Mme
lorsque lautre croit avoir trouv la parade, celle-ci tait prvue dans le calcul de lhabile
rhteur, et elle fournit le moyen par lequel il sera rduit au silence. Toutefois pour
1 En chinois le dterminant prcde le dtermin, ladjectif qualifcatif se place donc ncessairement avant le
nom quil qualife.
de la dispute sophiste au dialogue philosophique 215
contraindre lautre agir en fonction de ses propres vues, il faut quil ne puisse exami-
ner loisir les alternatives qui se prsentent lui; la rapidit est essentielle. Les dbats
entre Mencius et ses contradicteurs sont clairants de cet usage retors du verbe so-
phiste. Ils sont dautant plus intressants que Mencius est tenu pour un dispensateur de
sagesse, certains allant jusqu considrer que sa parole vaut comme un ferment susci-
tant la rfexion et lveil la vrit. Il nen est rien. La parole de Mencius est une parole
de pouvoir. Elle est l pour imposer un dire devant une assemble en un dbat contra-
dictoire dont les ds sont pips. Cest dans une arne quelle se profre et elle est une
arme pour un enjeu qui nest pas la vrit ou le salut de lme dautrui mais pour lob-
tention de gloire et de prbendes. Lart rhtorique de Mencius se dfnit comme la
science des moyens les plus propres clore le bec lautre en dehors de toutes proc-
cupations logiques, mme si la logique peut tre mise contribution comme technique
efcace. Mencius se rvle un adversaire de premire force. Son art oratoire vise pro-
duire une altration dans le dispositif discursif et par ce dsquilibre inopin emporter
lavantage. Cet art de la repartie nest oprant que parce quil est apte saisir loccasion
et subvertir le dploiement mtaphorique de ladversaire procd formalis par cer-
tains auteurs en termes de translation de structures (mapping). Un change entre le
philosophe confucen et le sophiste Shunyu le Tonsur, gloire de lacadmie Jixia, o
safrontaient les esprits les plus alertes de lempire, permettra de dmonter le procd:
Les rites, ce quon dit, veulent quune femme et un homme nentrent jamais en contact.
Est-ce vrai?
Mais si votre belle-sur se noie, lui tendez-vous la main?
Naturellement, moins davoir un cur de loup! Si la rgle est que les personnes de sexe
oppos ne peuvent avoir un contact direct de la main la main, les circonstances commandent
que la rgle soit transgresse.
Dans ce cas dites-moi pourquoi, alors que lempire se noie ne lui tendez-vous pas la main?
Quand lempire se noie on le secourt avec le Tao. Quand sa belle-sur se noie on la secourt
avec la main. Voudriez vous que je secoure lempire avec la main?
Tonsur, pris de court, resta coi.
Il sagit dune pirouette et non dun argument. Tonsur croit marquer un point en met-
tant Mencius en contradiction avec lui-mme. Pour cela il utilise habilement largument
dont se sert Mencius pour prouver la validit de sa thorie dune nature humaine fon-
cirement bonne: si un enfant tombe dans un puits on lui tend spontanment la main
pour le retenir sans aucune autre considration quun lan de piti irrpressible surgi
du cur, pour lopposer la stricte sparation des sexes prne par la morale rigoriste
des confucens laquelle Mencius souscrit pleinement. Dans le but de faire clater la
contradiction de la faon la plus saisissante, il recourt une mtaphore; il compare lem-
pire en proie aux convulsions un homme qui se noie et auquel Mencius refuserait de
tendre la main comparaison la vrit lgitime si elle tait explicite. Mencius laisse
ladversaire pousser lavantage en dveloppant le paralllisme pour mieux le piger en
le retournant contre lui; il dmonte labsurdit de la fgure par le retour au sens littral:
on ne sauve pas un empire avec sa main, mais par le Tao, cest dire en inculquant aux
hommes des principes moraux.
Un autre exemple de cette facult de Mencius de retourner son proft les compa-
raisons sophistes labores par la partie adverse est fourni par une passe darmes entre
216 jean levi
Matre Elvation et le philosophe propos de la nature humaine. Tandis que Matre El-
vation soutient que la nature humaine nest ni bonne ni mauvaise mais quelle est mal-
lable et dpend de lducation, Mencius soutient quelle est foncirement bonne.
La nature humaine nest-elle pas losier et le devoir le panier quil sert tresser?
Non. Peut-on tresser des paniers sans contrarier la nature de losier? On doit couper et at-
tendrir losier pour le rendre souple et pouvoir le forcer. Sil en tait ainsi cela voudrait dire quil
faut torturer lhomme pour le plier aux vertus sociales. Une telle ide ne peut que lloigner des
vertus de Bont et de Justice.
Mencius ne rpond pas sur le fond; il ne rfute pas la validit de la thse de son contra-
dicteur, mais se place sur le seul terrain du bien-fond de la comparaison. Elvation pro-
pose une image qui dpeint un aspect de lassertion quil veut mettre en valeur: la fexi-
bilit de la nature humaine que lon peut plier dans un sens ou dans un autre, vers le
bien comme vers le mal. Mencius profte des connotations ngatives de limage pour
embarrasser son adversaire: lhomme ne pourrait acqurir des qualits morales et de-
venir un animal social quau prix dune torture.
Elvation, contr propos de la premire mtaphore de losier, va recourir une au-
tre image conventionnelle pour faire valoir sa thse:
Alors, la nature ne serait-elle pas comme une eau qui tourbillonne derrire ses digues? Si on lui
ouvre un passage gauche, elle ira gauche, si on lui ouvre une voie droite elle ira droite.
Elle est incapable de discerner le bien du mal, de mme que leau ne fait pas de difrence entre
la droite et la gauche.
Certes, leau ne distingue pas entre la droite et la gauche, mais elle distingue entre le haut
et le bas. La nature humaine tend au bien comme leau tend vers le bas. Tout homme est bon
comme leau suit la ligne de pente.Il est vrai que si vous la frappez violemment, leau jaillira vers
le haut; si vous dressez une digue, vous pouvez arrter son cours; mais obit-elle sa tendance
naturelle? non, elle subit une contrarit. Il en est de mme de lhomme que les circonstances
peuvent dtourner du bien.
Ce morceau a t analys de bien des faons, mais quels que soient linterprtation four-
nie et le jugement port sur le niveau des comptences logiques ou des capacits dia-
lectiques des Chinois quil manifeste, les exgtes se sont toujours placs dans la pers-
pective de lhistoire de la pense. En sorte que les sinophiles dfendent la mthode
analogique qui y est mise en uvre, en dbusquant les chanons rests implicites pour
dmontrer que les dveloppements de Mencius supposent la parfaite matrise des syl-
logismes, les ethnocentristes stigmatisent le manque de rigueur dialectique de Mencius
en soulignant les imprcisions de ses catgories et le fou de largumentation, les tenants
de la radicale altrit de la pense chinoise arguent quil nest nullement question de d-
monstration mais dincitation, il sagit damorcer un processus de maturation qui u-
vrant en quelque sorte linsu du contradicteur lui-mme, par une sorte de persuasion
insidieuse, oprerait sa conversion au terme dun cheminement souterrain. Rien de
tout cela. Il faut recourir une autre grille de lecture: ni celle de la logique ni celle de la
sagesse. Mencius est un sophiste ou disons pour tre plus prcis quil est rompu
toutes les ruses de la rhtorique. A lpoque trouble o il professe sa doctrine, dfen-
dre que lhomme est bon par nature et quen faisant fond sur cette bont on peut assu-
de la dispute sophiste au dialogue philosophique 217
rer le fonctionnement harmonieux de la socit est aussi paradoxal que de soutenir
quun cheval blanc nest pas un cheval. Seule difrence, lun se situe sur le plan des qua-
lits morales lautre sur le plan des qualits physiques. Lanalyse qui permet, ce me sem-
ble, de rendre compte de faon adquate des procdures dployes par Mencius est celle
dveloppe par Michel de Certeau dans Les Arts de faire. Mencius joue sur largument
dautorit pour le retourner contre lui-mme. Lanalogie de la nature humaine docile
comme leau, ou dun individu mallable comme losier, linstar de tous les lieux com-
muns, a la force persuasive de lvidence: ils coulent de source si je puis dire. Toute las-
tuce de Mencius est de jouer avec ces formules toutes faites et de les retourner comme
des gants. Il produit une altration dans le dispositif du savoir convenu et, par ce ds-
quilibre inopin, remporte lavantage. Michel de Certeau parle dun art de la mmoire
propos de cette pratique retorse de la langue; cen est un en efet dans la mesure o
lon puise dans le stock des vrits existantes et des confgurations connues pour leur
infiger linfme altration qui les transforme en confgurations gagnantes. Il sagit
dtre dans le lieu de lautre sans en tre toutefois le dtenteur; il faut savoir tirer partie
de la distorsion sans rien y perdre en ne se trouvant toujours en dehors du systme des
associations conventionnelles. Cet art de la rponse opportune laide de dtourne-
ment dimages fges na de valeur que parce quil sadapte aux circonstances, quil est
science de loccasion. Leau dans la comparaison propose par Elvation est utilise en
raison de ses connotations din-difrence ping et de ductilit. Confucius lui-mme dit
que leau est carre dans un rcipient carr, ronde dans un vase rond. Mencius va lui
opposer dautres associations toutes aussi prgnantes: celle de leau coulant vers le bas
suivant la ligne de plus grande pente, ces nouvelles images bousculent les prcdentes,
elles font vaciller le premier rseau signifant et, pour fnir, se substituent lui. Il va sans
dire quelle ne renvoie aucune vrit. Toutefois les associations paradigmatiques pro-
poses par Mencius seront dautant plus fortes que le philosophe dtourne une autre
image, inverse de celle quil va imposer. Les lgistes, thoricien cyniques du pouvoir to-
talitaire, en efet, font reposer le systme absolutiste sur le postulat dune mchancet
foncire de lhomme ou plutt ils soutiennent que seules les passions rendent les
hommes gouvernables en sorte quil est bon que lhomme soit mchant, cest la soif du
proft qui en fait un animal sociable, et pour cela ils recourent eux aussi la mtaphore
de leau: lhomme disent-ils court aprs le proft comme leau coule suivant la ligne de
plus grande pente.
Les dialogues philosophiques du Zhuangzi
La subversion zhuangzienne du verbe sophiste
par lexcs sophistique mme
Les complots de tous contre chacun, la plthore dergoteurs dbouchent sur une totale
impuissance matriser le cours des vnements. Les tenants de lordre, tel Xunzi ne
cessent de fulminer contre les Gongsun Long, vouant les sophistes et leur cheval blanc
aux gmonies. Ces gens l, temptent-ils, sont des faussaires qui mettent sur le mar-
ch la mauvaise monnaie des dsignations mensongres rendant le langage dans son
ensemble sujet caution. A partir du moment o lon ne peut plus qualifer une chose
sans oblitrer sa nature, il devient impossible de sanctionner quelque conduite que ce
soit, sous peine dentacher dinexistence lobjet mme de la sanction. La justice ne pour-
rait plus sexercer et tous les criminels courraient en libert! Cest par ce biais que la pra-
218 jean levi
tique confucenne des dsignations a t branle. Dautant que lefort des lettrs tra-
ditionalistes pour rfuter les paradoxes sophistes sur le plan de la logique est loin dtre
couronn de succs. La plupart du temps, ils se font clouer le bec par leurs adversaires
dans les joutes oratoires et ce dautant plus facilement quils cherchent les combattre
en se plaant sur le terrain de ladversaire.
Pour avoir raison des arguments sophistes, il suft donc de dplacer le terrain de la
lutte des brumes du discours lvidence du rel. Ainsi Han Fei, le thoricien de lab-
solutisme imprial, a recours, comme il convient pour anantir dun trait ce genre
dapories, largument massif et massue de lvidence du rel. De mme quon prouve
le mouvement en marchant, et quAristote balaie ddaigneusement le trait Du non-tre
de Gorgias par le sentiment dvidence absolue de lexistence du rel et par le consen-
tement gnral quil provoque chez tous, de mme il rfute le fameux, le lancinant che-
val blanc nest pas un cheval, qui sape toute la pratique confucenne en faisant appel
la dure loi de la ralit. Il se contente de raconter cette anecdote:
Le sophiste Ni Yue aprs avoir triomph de tous ses adversaires de lacadmie Jixia en soutenant
victorieusement quun cheval blanc ntait pas un cheval, fut oblig dacquitter les droits sur les
chevaux aux douaniers, quand mont sur son cheval blanc, il voulut passer la frontire.
La dmonstration est imparable. Nul sophiste, aussi loquent soit-il, ne pourra faire va-
loir son argument un agent du fsc.
Il existe une autre rfutation de Gongsun Long. Elle est luvre dun philosophe mar-
ginal, Zhuangzi. Elle est la plus corrosive, car elle se situe sur un autre plan; ni sur ce-
lui de la logique, ni sur celui de la dure loi du rel, vue sous les espces de ladministra-
tion des douanes.
Pour dsamorcer les excs de la dmarche analytique des sophistes, laquelle aboutit
la disparition aussi bien du rel que des moyens de sa reprsentation et pour conduire
son travail de sape contre le verbe comme instrument privilgi de la pense du spar,
Zhuangzi retourne le langage contre lui-mme et discrdite la dmarche sophiste par
le comique. Ainsi parodiant Gongsun Long, il fournira sa propre variante loufoque du
paradoxe du cheval blanc. Elle est digne de Lewis Caroll:
En efet, vouloir dmontrer partir de lide en soi que lide pour soi nest point lide-en-elle-
mme, vaut moins que de dmontrer, en partant de la non-ide que lide-dans-la-chose nest
pas lide-en-elle-mme. De la mme faon dmontrer partir de lide de cheval-en-gnral
quun cheval blanc nest pas un cheval vaut moins que de dmontrer en partant de lide de non-
cheval en particulier que tout cheval est un cheval blanc en gnral. Car en vrit du point de
vue de lunit suprme, lunivers nest quune ide et lempire quun cheval!
Mais Zhuangzi ne se borne pas ridiculiser la phrasologie des sophistes. Il met en
scne tout au long de ses essais des personnages qui safrontent dans des circonstances
les plus diverses et sous les travestissements les plus varis. Ce peut tre des fgures al-
lgoriques tels Enonc du Non-agir, Evanescence des Images, Grande puret, Egar,
etc., des fgures lgendaires comme lEmpereur Jaune, des personnages historiques ou
des sages de lantiquit comme Laozi et Confucius, ce peut tre encore des rhteurs
contemporains comme le sophiste Huizi quil met toujours aux prises avec lui-mme
et dont il fnit toujours par avoir raison, parce quil ne se place jamais sur le terrain de
la raison mais de la draison cest dire de lvidence sensible. Les personnages saf-
frontent soit loccasion dune rencontre, soit dans une conversation amicale, ou bien
ce peut tre un sage entreprenant le matre des hommes et produisant sur lui par la seule
de la dispute sophiste au dialogue philosophique 219
vertu de son verbe une conversion salutaire. En efet beaucoup de dialogues mettent en
scne une conversion, ou la transformation dun personnage un dirigeant lissue
dun face face. Zhuangzi y parodie la situation archtypale telle quelle est dcrite par
de nombreux textes imprgns par la mentalit sophiste. Il joue sur le caractre conven-
tionnel des situations. Il sagit pour le sophiste ou le diplomate de capter lattention et
de provoquer un choc. Su Qin et Su Dai, mais aussi Front de Bufe, Huizi et Gongsun
Long taient pass matres dans cet art. L o Zhuangzi se distingue de la pratique com-
mune, cest que les interventions de ses sages, loin de gagner linterlocuteur telle ou
telle option politique, plonge linterlocuteur dans les afres de la perplexit. Ils produi-
sent une rvolution qui se traduit essentiellement par lbranlement de toutes les certi-
tudes. Les discours des excentriques de Zhuangzi sont de vritable lectrochocs, car sils
miment le discours sophiste, ils sont en ralit des anti-discours conduisant au rejet vio-
lent de la politique, comme ils sont des antidotes contre la rhtorique sophiste et contre
le langage lui-mme, tout au moins dans son usage ordinaire. En un mot Zhuangzi se
sert de lart du discours comme dune machine de guerre dirige contre le discours
mme. Avec une gale maestria, il a recours aux procds rhtoriques des diplomates,
des sophistes et des lettrs itinrants, ces Huizi, ces Mencius, ces Su Dai et ces Front de
Bufe,mais il poursuit des fns opposes. Ils contribuent au travail de sape contre la pen-
se du spar dont les catgories de la langue sont tout la fois les causes et les mani-
festations. Ainsi en est-il de ce dialogue innarrable digne de Rabelais entre un souve-
rain et divers conseillers sophistes qui ridiculise et les princes et la politique tout en
fournissant un modle de rhtorique sophiste:
Le roi Ying de Wei avait conclu un trait avec le prince Tianmu de Qi. Celui-ci, lencre peine
sche, lavait dj viol. Le roi de Wei, furieux, mditait de le faire assassiner par un de ses si-
caires. Layant appris, le gnral Tte de Bufe sentit le rouge de la honte lui empourprer le front.
Il se rendit auprs du roi et lui dit:
Peuh! cest un scandale! Vous alignez dix mille chars de guerre et vous voulez confer votre
vengeance un tueur gages! Cest vous conduire en homme du commun. Donnez-moi cent
mille hommes. Jattaquerai le Qi. Je ravagerai le pays, je le mettrai feu et sang, jgorgerai les
hommes, je violerai les femmes, je ferai main basse sur les chvres, les vaches et les chevaux. Le
roi de Qi en fera une jaunisse et une tumeur lui poussera sur le dos; il aura peur, il senfuira. Alors
je le fouetterai, je lui briserai les reins et je le rduirai en hachis
Le Grand intendant Jizi semporta contre Tte de Bufe:
Nous venons juste dachever les dfenses de notre territoire; notre force vient de ce que nous
nous sommes abstenus de faire la guerre durant dix ans. Et voil ce gnral qui veut tout re-
mettre en question. Cest un fauteur de troubles. Ne lcoutez pas!
Cest alors que Matre Hua ayant eu vent de toute lafaire intervint:
Ncoutez ni lun ni lautre! Ce sont des fauteurs de troubles! Et celui qui vous conseille de
ncouter ni lun ni lautre est aussi un perturbateur!
Alors que faire? scria le roi ne sachant plus quel saint se vouer.
Agir selon le Tao.
Le roi tait dsempar. Il ft appel Matre Hui le sophiste. Matre Hui ladressa un certain
Dai Jinren.
Dai Jinren lui dit:
Vous savez ce quest un escargot?
a a deux cornes. Sur lune des cornes il y a le royaume dAgression. Sur lautre le royaume
de Barbarie. Ils se font constamment la guerre, en sorte que les cadavres jonchent le sol par cen-
taines de milliers. Quand lun est battu, il ne faut pas plus de quinze jours pour quil remette a.
220 jean levi
Cest une blague, ft le roi indign. a na aucun sens!
Alors je vais tout de suite lui en donner un: pensez-vous que lespace qui nous entoure ait
une limite?
Non, lespace est illimit.
Dans un espace illimit, notre terre nest-elle pas un petit atome de boue?
Cest vrai.
Sur ce point imperceptible de la nature il y a la Chine, et au milieu de la Chine, insignifants,
il y a le Qi et le Wei, avec lun sa capitale Linze et lautre sa capitale de Grandpont. Est-ce si dif-
frent des deux royaumes de Barbarie et dAgression?
Non, il ny a aucune difrence.
Le visiteur ayant pris cong, le roi resta longtemps perdu dans ses penses. Lorsque Huizi vint
le trouver aprs le dpart de Dai, il lui dclara:
Ah, ce visiteur est un grand homme! Mme les saints rois ne lui arrivent pas la cheville!
Cest vrai. Quand on soufe dans une trompette, elle met un son clatant; quand on soufe
dans le pommeau dune pe, cest tout juste sil en sort un pft!. Tout le monde sextasie sur la
vertu de Yao et de Shun mais compar Dai Jinren, ils sont comme le pft! mis par le pommeau
dune pe.
Mais on aurait tort de voir dans lutilisation des dialogues un simple pastiche du langage
sophiste. Le travail du ngatif zhuangzien creuse beaucoup plus profond. Cest ce que
nous allons essayer de montrer par lanalyse de deux passages essentiels: un dialogue
entre le patriarche du taosme, Laozi et le pre du Confucianisme, Confucius en per-
sonne; et une discussion entre le sophiste Huizi et Zhuangzi.
Laozi, Confucius et la copulation des insectes
Le chapitre xiv du Zhuangzi intitul Tianyun, du Mouvement cleste, se termine par
une discussion entre Laozi et Confucius. Elle est si absurde quon la croirait tire dAlice
au pays des merveilles.
Confucius dit Laozi:
Depuis longtemps jtudie les Six Canons. Livre des Odes, Livre des Documents, Livre des Rites, Li-
vre de la Musique, Livre des Mutations et Annales des Printemps et Automnes nont plus de secret pour
moi; alors allez savoir pourquoi jai eu beau expliquer aux soixante-douze princes qui jai rendu
visite ce qutait la Voie des anciens rois et les inciter suivre les traces des ducs de Zhou et de
Shao, il ny en pas eu un seul pour mcouter. Est-ce que cela tient aux hommes quil est difcile
de convaincre ou au Tao quil est difcile de mettre en lumire?
Mon pauvre ami, rtorqua Laozi, cest une chance que vous nayez pas rencontr un prince
qui sappliqut faire rgner lordre en son sicle. Car les Six Canons ne sont que les empreintes
laisses par les pas des anciens souverains! comment pourraient-ils tre la cause de ces traces?
Vos discours sont leur image: lempreinte a beau provenir de la chaussure, elle ne saurait se
confondre avec elle.
Ainsi il suft aux hrons de se regarder dans le blanc de lil sans ciller, aux insectes de chan-
ter lun au dessous de lautre pour se fconder; en un mot sitt que mle et femelle saccou-
plent suivant leur espce, la transformation vitale saccomplit sans entraves. La nature inne ne
peut tre modife; on ne change pas le cours du destin, on narrte pas le mouvement des
saisons, on nemprisonne pas le Tao. Tout est possible qui la trouv, rien nest possible qui
la perdu.
Confucius resta clotr chez lui trois mois durant. Puis il retourna auprs de Laozi:
a y est, jai trouv, dit-il au Matre, corbeaux et pies couvent, les poissons crachent leur frai,
les insectes taille fne muent, et la naissance dun petit frre fait pleurer son an. Oui, voici trop
longtemps que je nassume plus ma nature dhomme en participant aux transformations. Et in-
de la dispute sophiste au dialogue philosophique 221
capable dtre un homme en participant aux transformations, jaurais encore la prtention de
transformer les hommes?
Tu y es! dit Laozi.
Bon nombre de sayntes du Zhuangzi doivent leur efet corrosif la parodie des Entre-
tiens de Confucius. Zhuangzi se sert de la terminologie et de la problmatique du ritua-
lisme formel, pour les retourner contre elles-mmes en en poussant les implications
jusquau non-sens. Que vient faire la copulation chez les oiseaux dans une discussion
dont on pressent quelle a trait aux rapports entre laction et sa reprsentation? Le point
central ou tout au moins prsent comme tel de la discussion entre les deux hommes
concerne la validit dun discours qui nayant pas la ralit de laction, se rsume un
pur savoir livresque, simple refet dun refet. Lcriture nest-elle pas en efet ersatz
dune parole, elle mme trace sans paisseur dune intention ou dune action? Nan-
moins les deux articulations discussion sur la chaussure et sa trace, envole lyrique sur
la fcondation sont toutes deux lexpression dun mme principe, celui du transfor-
misme universel, impliqu par le systme des correspondances qui sous-tend le ritua-
lisme confucen. Tout au long du texte courent en fligrane des motifs, qui sans jamais
tre explicitement dsigns, forment la trame mme de la discussion.
Le mot donn en ouverture est jing, les Classiques. Confucius, il le confesse lui-
mme, na pas russi convaincre son sicle en dpit de sa science des liujing, des Six
Classiques. Par le dploiement du champ smantique du mot sur la chane syntagma-
tique du discours, cest toute la problmatique du rapport entre geste et parole, entre
criture fge et soufe transformateur de laction qui va tre mise en scne. Jing, ca-
non, signife aussi chemin: cest la voie droite de la Vertu. Le droit chemin trouve sa
rplique, ou plutt sa concrtisation morale, dans les Six Canons. On comprend com-
ment les deux termes peuvent sappeler et entrer en rsonance. De canon on passe
chemin, puis de chemin trace. Un chemin nest rien dautre que la voie fraye
par la trace des pas. Et de chemin et trace on drive sur marche (li), terme qui en
chinois dsigne la fois laction (la marche) et lobjet qui laccompli (la chaussure).
Il est donc lgitime de passer du livre (de la thorie abstraite) laction (qui en est la
source). Cest pourquoi Laozi peut dire Confucius que ses fameux classiques ne sont
que les empreintes laisses par les pas des anciens souverains et non la cause de ces
traces. Si jing est le chemin sur lequel marche lhomme la conduite droite oprant les
justes distinctions, il nest pas le pas lui-mme, lacte efectif, mais sa trace, ji. Et de fait
Confucius avoue traquer dans les livres la trace ji des hauts faits des Anciens. Zhuangzi
a encore dautres motifs de sarrter sur le terme. La trace est fondamentale encore dans
lconomie de ce passage, parce quelle permet, en outre, par le recours dun seul
mot,de faire appel au mythe de lcriture comme trace. Une tradition relativement an-
cienne veut que lcriture ait t cre sur le modle des empreintes de pas laisses sur
le sol par les oiseaux et les animaux sauvages. Zhuangzi joue en matre avec les repr-
sentations traditionnelles. Elles lui permettent de tlescoper la conception lettre du Li-
vre canonique comme chemin, dans la thorie de lcriture comme trace. A partir du
moment o le canon en tant que corps crit se rduit au simple reliquat dune action,
il ne saurait avoir le prestige du geste efcace, vivant; il est au rite ce que lempreinte
est la chaussure. Et comme le terme qui la dsigne li, renvoie dans son acception fac-
titive au mouvement crmoniel, le texte accomplit un cycle complet, lissue duquel
la dimension rituelle du canon se trouve entirement vacue.
222 jean levi
Le canon est la marque scripturale (et pour Laozi fge) de la vertu (dao). Cette vertu,
la marche de qui sengage sur la voie droite la manifeste. Le modle de cette marche
parfaite, dans limagerie confucenne, est la dambulation royale accomplie dans le Pa-
lais des lumires. Se conformant au cycle saisonnier, elle manifeste le mouvement c-
leste (tianyun titre du chapitre o fgure notre passage) tout en lassurant. Ce parcours
rituel convertit en activit humaine le mouvement des forces naturelles; il est propre-
ment lacte civilisateur par excellence. Grce lui sopre la transformation spontane
des tres. Cest le wanghua, linfuence royale difusant la culture depuis le centre
jusquaux confns de lunivers. Cette action est shen, divine, en ce quelle possde la
mme efcience invisible que le vent (feng). Confucius a pu dire: La vertu du prince est
le vent; celle du peuple est lherbe; lherbe se courbe sous laction du vent. Les murs
(fengsu) se rsument aux habitudes (su) faonnes par lair (feng) de vertu manant de
lattitude royale et que trahissent les airs de musique, les chants (feng) et les danses en
vogue parmi le peuple; les mlodies donnent le diapason des coutumes, renseignent sur
les penchants dune communaut. Par une musique rgle on modre la luxure, et par
laction spontane du vent de la civilisation, on ramne le peuple la correction des ma-
nires. Ces chants, feng ou guofeng, Confucius les a dailleurs compils en un canon ri-
tuel, le Livre des Odes afn de modifer (hua) les comportements de ses contemporains,
autrement dit de les civiliser. Ce sont ces fameuses chansons que les seigneurs dcla-
maient loccasion des banquets diplomatiques en leur donnant des connotations hau-
tement morales et politiques, tout en les chargeant de leur propre intriorit.
Mais feng a dautres rsonances, elles renvoient celles-l la sexualit et la licence.
Fengliu, ce sont les attachements amoureux. De mme que les histoires de brise et de
lune, fengyue, dsignent coucheries et aventures galantes, les chants de pays, les Guo-
feng consigns par Confucius dans le vnrable Livre des Odes ntaient rien dautre que
des chants damour ces chants alterns quentonnent les villageois au printemps, sous
lefet du soufe fcondant du yang montant, avant de sapparier. De la mme faon, si
hua peut dsigner la transformation accomplie par lacte civilisateur, il renvoie aussi au
cycle de la vie et la gestation: le ftus passe par neuf tapes, hua avant de venir au
monde. La procration est assimile une mue du vivant. Zhuangzi reprend une ex-
pression dissyllabique qui dans le discours lettr dsigne laction rituelle accomplie par
le geste royal pour la retourner contre elle mme en lidentifant au cours naturel de la
vie, dans lequel tres et choses se transforment deux-mmes et agissent sans efort les
uns sur les autres grce la copulation. Ce qui lui permet en outre de se prvaloir se-
crtement des thories du Yijing, lesquelles forment le cadre indpassable de la pense
chinoise. On peut y lire en efet dans un des plus importants commentaires: hommes
et femmes saccouplent et tous les tres sont engendrs par transformation. Et voici
rveill par le mot hua, les transformations, un lot de formules calendriques quiva-
lentes: un temps de yin, un temps de yang cest le Tao; Vantail ferm, cest le Passif,
vantail ouvert cest lActif, un temps douverture, un temps de fermeture, ce sont les
Transformations, emblmes vivaces, clatants de puissance vocatrice, ces dictons
possdent lentire efcacit du rite, parce quils reclent la fracheur et la spontanit
de la posie populaire qui, aux dires mmes des lettrs, est lme dune nation. En
voquant la sexualit, Zhuangzi ne fait que revenir la source de lacte transformateur
et y saisir cette essence de ncessit naturelle qui est la vertu premire de tous les rites,
se conformant lesprit du Confucius compilateur des posie champtres des Guofeng
et exgte du Livre des Mutations.
de la dispute sophiste au dialogue philosophique 223
Ossife par le culte des livres, ptrife par une pratique, lcriture, qui fge les choses,
la doctrine de Confucius a perdu toute force persuasive. Elle ne peut infuencer per-
sonne. Si le Matre, dit Zhuangzi, veut retrouver le soufe spontan qui, limage du
ciel,meut les tres en silence, il faut quil renonce aux classiques, quil abandonne la
trace pour retrouver la chaussure, responsable de la trace, quil se dtache en un mot
de lcriture pour porter son regard sur le monde qui lentoure, et o sefectue sans ef-
fort et sans peine lacte transformateur.
Nous nen avons pas encore fni avec lanalyse lexicale du passage. Il est un autre mot
qui revient constamment et dont nous navons pas relev toutes les connotations. Cest
le terme fondamental de Tao (dao). Il apparat au dbut du texte dans la bouche de
Confucius, comme une sorte de dveloppement explicatif du contenu des canons. Il
voque la vertu (dao) des anciens rois, dans une phrase forme de deux propositions
symtriques o le terme est mis en parallle avec la trace, mais aussi,au sens fgur, les
exploits. Ayant tudi les canons, Confucius voulait en faire connatre aux autres le
contenu. Cest ainsi qu il exposa la vertu (dao) des anciens rois et (les traces laisses
par) les exploits (ji) des ducs de Zhou et de Shao. Ce faisant, le rapprochement opr
par le balancement rhtorique de la phrase ressuscite le sens premier du mot Tao qui
est voie, chemin. Cest bien la route trace par les actes des sages antiques que le
Matre veut retrouver, une route qui se reconnat des traces.
Plus largement, dao dsigne un agir, un faire; il peut vouloir dire mthode ou
technique, mal distingue dailleurs de la vertu, en tant quefcace. Le texte va uti-
liser une fois encore le recoupement des sens de dao, la vertu-voie et de jing, le che-
min-canon, pour ravaler la voie des confucens une trace. Les rgles de conduite at-
testes par les canons, la Voie confucenne des sages de lantiquit, sopposent
radicalement, en tant quobjet inerte, produit et non cause agissante, cet autre Tao,
cette autre voie, qui est celle des taostes. Ce Tao l est vertu et chemin, mais en tant
que pure activit, il est non pas le chemin trac par la marche, il est le mouvement
mme qui trace le chemin tout en faisant progresser sur la Voie. Retournant lexpres-
sion la fn de sa diatribe, Laozi peut conclure sur le Tao, mais un Tao rinvesti de
toute son efcience et de toute sa vitalit pratique. Ce Tao du Yijing dfni comme le
processus de procration universelle. Tao de la ncessit irrpressible que rien ne sau-
rait emprisonner. Emprisonner dans quoi? dans une trace crite, dans un savoir li-
vresque certes,mais plus gnralement dans un discours. Car dao a aussi le sens de
dire, de parole. La formule du Laozi La Voie que lon peut exprimer par les mots nest
pas la voie constante (Dao ke dao fei chang dao) est lexemple le plus clbre et le plus
troublant du jeu sur le double sens de dao. Un sens o se dvoile toute la problma-
tique du geste et de la parole. Et ce, tous les niveaux dinterprtation. Une premire
lecture, qui privilgierait le caractre trs pragmatique du Laozi, donnerait dao, dans
ses deux occurrences, le mme sens de prendre un chemin, dagir, de mettre en
pratique; le texte pourrait tre traduit de la faon suivante: La Voie que lon doit sui-
vre nest pas une voie constante. Cest dire que dans la vie relle, concrte, il faut
savoir sadapter aux circonstances et saisir lopportunit du moment, sans jamais se f-
ger dans une attitude rigide. Cest cette interprtation que font leur tant les stratges
que les lgistes. Mais ce premier sens sen superpose un second qui prend les deux
dao dans une acception difrente: parole et Voie, Principe la fois immanent et
transcendant do lAction puise son origine. Le sens de la phrase serait alors: La Voie
Voix nest pas la vraie Voie. Toute parole, parce quelle est parole, ne peut atteindre
224 jean levi
qu des aspects particuliers et partiels du rel, quelle ne restitue jamais dans sa pl-
nitude. On ne saurait trouver une rgle immuable pour laction dans aucun discours,
dans aucune doctrine. Les deux sens, loin de sexclure, sont complmentaires. En efet
le Tao ternel se retrouve dans laction et par laction, une action qui est toujours
changeante, fuctuant en fonction des sollicitations du moment. Nous atteignons l
lultime niveau de signifcation du passage, un sens juste esquiss, plus suggr que
rellement exprim. La scne commence par un aveu dchec. Confucius confesse
quil a t incapable de convaincre les hommes de les transformer au moyen du lan-
gage. Car Confucius, parle, disserte, explique, en un mot il se dpense dans cette vaine
activit du discours, que le mot dao qualife, mais que cette activit mme de parole
disqualife en tant que dao agissant, en tant quactivit irrpressible et spontane, ce
mouvement naturel que rien ne peut ni ne doit enfermer, et surtout pas lcriture,
sous peine de rduire la doctrine (dao) des sages des restes fossiles, den faire des
routes dj balises,et de suivre des ornires. En dernire analyse, cest bien dans la
mesure o lcriture est refet du discours quelle sannihile comme porteuse dun
message dot dun pouvoir efcace.
Zhuangzi nous dit ceci: non seulement lenseignement de Confucius narrive pas se
priver de la parole, mais encore il prend pour base les textes du pass. Croyant retrou-
ver dans les crits des sages un peu de leur esprit; il ne fait que pervertir encore davan-
tage le sens de sa propre pdagogie et oblitre ce que gardait de spontan la relation de
matre disciple.
Les livres dispensent encore moins de savoir que les mots.Ils ne sont que paroles
mortes, lettres mortes, les djections des anciens comme il est dit ailleurs. Ils ne ren-
ferment aucune vrit, aucune exprience vraie, puisque celle-ci se situe au del en
de plus exactement du langage, dans le geste parfait.
En ce sens le dialogue entre Laozi et Confucius doit tre rapproch dun autre trs
clbre dialogue du Zhuangzi qui met cette fois-ci aux prises le duc Huan de Qi, lisant
les paroles des saints avec un charron. Zhuangzi dmontre, par ouvrier interpos,
quil y a, dans toute activit, quelque chose que les mots chouent exprimer; plus
forte raison les signes crits, qui ne sont que de la parole morte cristallise.
Au cur de la critique taoste de lducation confucenne, se niche le constat de cette
contradiction, qui traverse toute la tradition lettre: lenseignement du rite par le rite
ne peut supprimer la transmission par le langage. La porte de la critique du langage
tient ce que dans la hirarchie des valeurs traditionnelles, il nest pas premier mais se-
cond par rapport au geste dans son efectivit rituelle. A partir du moment o le rite
saccompagne du langage et devient prtexte un discours, sa nature se pervertit; il
nest plus expression spontane du monde, action irrpressible et invisible du ciel, mais
forme vide et accessoire, ce que Laozi appelle lcorce de la foi.
Zhuangzi, Huizi et les courges du roi de Wei
Le chapitre douverture du Zhuangzi, lun des plus importants du livre, intitul
Xiaoyao you, Randonnes extatiques se clt sur un double dialogue entre Zhuangzi
et Huizi:
Le roi de Wei, dclara Huizi, ma laiss des graines de courges gantes. Je les ai plantes et elles
ont donn des fruits tellement normes quils ne pouvaient ni servir de jarre, les parois ntant
pas assez solides, ni tre dbites pour faire des coupes, car elles auraient t trop plates pour
de la dispute sophiste au dialogue philosophique 225
rien contenir. Pour tre normes, a, elles taient normes, mais nen ayant aucun usage, je les
ai rduites en miettes!
Cela prouve, rtorqua Zhuangzi, que tu es incapable de voir grand. Un homme de Song avait
dcouvert un baume contre les engelures. Sa descendance lutilisait depuis lors de pre en fls
pour laver en toutes saisons la bourrette de soie. Un tranger en entendit parler et en ofrit cent
pices dor. La famille se runit pour examiner sa proposition: Voil des gnrations que nous
nous chinons faire mtier de trempeurs de bourrette et nous navons pu amasser grand peine
que quelques pices dor. Maintenant que nous avons loccasion de nous trouver la tte de cent
pices dor en un seul jour en vendant notre secret de fabrication, ce serait folie de tergiverser!.
Une fois en possession de la formule, lhomme se rendit auprs du roi de Wu. Or il se trouvait
que le Yue venait douvrir les hostilits contre son voisin. Le roi de Wu accorda ltranger le
commandement de sa fotte. Pendant lhiver, il y eut un engagement naval et la fotte du Yue fut
dfaite grce au baume dont les marins avaient enduit leurs mains pour les protger du froid. Le
Yue cda une portion de son territoire au roi de Wu qui lofrit en apanage son gnral pour le
rcompenser de ses services. Ainsi le mme baume permit dans un cas dacqurir un fef tandis
que dans lautre il ne servit jamais qu laver la bourrette de soie.
Et Zhuangzi conclut:
Tu avais une courge de cinquante boisseaux. Pourquoi ne pas avoir song en faire un na-
vire pour voguer sur les feuves et les lacs au lieu de te lamenter quelle tait trop plate pour rien
contenir? Cela prouve que tu as lesprit triqu!
Un autre jour Huizi dclara Zhuangzi:
Il existe de grands arbres appels ailantes dont le tronc bossel ne suit pas le trac du cor-
deau et les branches tordues le compas ou lquerre. Ils poussent sur le bord du chemin mais au-
cun charpentier ne leur accorde un regard. Il en va de mme de tes discours grandiloquents mais
inutiles dont la foule unanime se dtourne.
Tu nas jamais vu une civette? rtorqua Zhuangzi. Elle se fait toute petite et se tapit, guet-
tant le rat qui musarde. Elle bondit de ci de l, elle saventure tantt en haut tantt en bas, et
pour fnir elle est prise au pige et meurt dans les flets. Le yack, lui, est aussi vaste que les nuages
qui fottent dans le frmament, et pourtant en dpit de sa taille, il serait bien incapable dattra-
per un souriceau. Tu as un grand arbre et tu te lamentes de son inutilit. Pourquoi ne pas le plan-
ter dans les contres du non-tre ou dans les steppes de la solitude infnie. On dambulerait tout
autour dans linsouciance, et on stendrait, oublieux de tout, son ombrage, assur quil ne
mourra pas prmaturment sous les coups de la hache, ni ne subira latteinte daucun tre.
Nayant aucune utilit, rien de fcheux ne saurait lui advenir.
Les deux squences, qui mettent en scne les mmes protagonistes se rpondent terme
terme; elles sont construites sur le mme patron: Huizi ouvre les hostilits en recou-
rant une parabole et Zhuangzi lui rpond sur le mme ton en lui renvoyant une autre
Dans lun et lautre change, le dbat se conclut sur lvocation lyrique dun voyage:
navigation au long cours sur les feuves et les lacs dans le premier cas, vagabondage mys-
tique dans linconditionn hors des frontires de lici-bas, dans le second, amplifant les
thmes de la premire passe darmes, la faon dune partition musicale o les motifs
esquisss en ouverture se trouvent pleinement dvelopps dans la suite du morceau.
Lenjeu de la dispute dans les deux cas est le mme. Il tourne autour de la question
du langage. Huizi reproche Zhuangzi ses thories grandioses sans utilit pratique.
Cest dire en ralit, pour un pragmatique et un ambitieux comme Huizi, sans appli-
cation dans le domaine politique. Huizi en efet est un sophiste. Habile dans le manie-
ment des catgories du langage, il fait passer le blanc pour le noir et les vessies pour des
lanternes En un mot cest un coupeur de cheveux en quatre qui cherche par des dis-
226 jean levi
tinctions artifcielles et spcieuses avoir toujours le dernier mot dans des disputes. Cet
art rhtorique, il la mis au service de ses ambitions. Diplomate de premier ordre, poli-
ticien retors aux combinaisons machiavliques, Huizi intrigue, complote, voyage dune
principaut lautre afn de conclure des alliances, ngocier des traits et convaincre les
princes de lui confer les rnes de lEtat. A linstar de tous les habiles ngociateurs de
son temps, il sait dtecter le dfaut de la cuirasse et possde de ce fait une puissance de
persuasion souveraine. Pourtant Huizi et ses pareils, si experts prvoir les ractions
dautrui et jouer des passions, se laissent surprendre un jour ou lautre par les ruses
de lhistoire, et fnissent comme la civette, la trappe ou pris dans les flets de la loi. Cest
contre ces dangers que Zhuangzi met en garde son ami, en mme temps quil dnonce
la vanit de son art oratoire qui lempche de comprendre la porte de ses thories quil
taxe dextravagantes. Zhuangzi, au rebours de son ami qui, en bon sophiste, semploie
tout distinguer, veut tout faire fusionner dans une obscure identit. A travers des
histoires baroques qui semblent premire vue absurdes voire grotesques, il seforce
de dmontrer la fausset des catgories du discours sur lesquelles sappuie un Huizi. Il
entend sattaquer aux distinctions opres par le langage dans le tissu de la totalit v-
cue, en retournant le verbe contre lui-mme, afn de briser la cangue des dtermina-
tions et des jugements de valeur.
Mais le Zhuangzi est tout la fois proclamation dune thorie du langage et mise en
uvre efective de cette thorie. Aussi, en mme temps quil met en scne deux atti-
tudes antagonistes vis vis des deux domaines solidaires du langage et de la domina-
tion, par le truchement de paraboles et dallgories, il prend soin de les lester dasso-
ciations symboliques demi caches, diaprant son verbe de rsonances polyphoniques
qui le soustraient la linarit du discours commun. En efet, on ne peut rellement
comprendre la signifcation du passage si lon ne peroit pas larrire fond lgendaire sur
lequel il se construit. Le discours philosophique se double, la faon de ces tofes de
damas double trame, dun autre discours, dordre mythique celui-l, qui, contrariant
le discours premier, lui confre sa vritable signifcation par lcart qui se creuse entre
ces deux voix. Lindice de la prsence de ce quon pourrait appeler une anamorphose
conceptuelle dans le dialogue entre Zhuangzi et Huizi se manifeste dans la mtaphore
nautique, engendre par limage de la gourde, qui permet denter le dbat entre les deux
penseurs sur la problmatique de la fragmentation de la totalit, travers les mythes du
dluge, et dinsrer la pratique usuelle de la langue et ses rpercussions politiques dans
le cadre plus vaste dune catastrophe cosmique. La squence qui ferme le premier cha-
pitre douverture des sections intrieures du Zhuangzi, doit se lire la lumire du
mythe de la mort de Chaos qui vient en clore le dernier chapitre. Non seulement plac
dans un rapport de symtrie celui-ci entretient avec louverture des rapports dinter-
textualit, mais en outre il en dveloppe et en amplife les motifs, faisant apparatre en
pleine lumire les connexions caches.
Le chapitre vii du Zhuangzi se termine par un mythe de cration dont le caractre
la fois lamentable et comique possde dans son indigence mme quelque chose de fas-
cinant. Bien que succincte, lhistoire cumule les strates. On peut y voir aussi bien une
fable cosmogonique, quun merveilleux conte philosophique sur les malheurs de la psy-
ch ds lors quelle se laisse prendre au pige de la ralit extrieure, on peut linter-
prter tout aussi lgitimement comme une charge contre la religion froce de la piti
dun Mencius, dont la compassion sme partout la mort et la dsolation, ou encore
comme une parodie des runions seigneuriales, fournissant le paradigme de la fausset
de la dispute sophiste au dialogue philosophique 227
des rapports humains. De toute manire nous pressentons que, par del le dsarroi que
suscitent en nous les ractions absurdes des protagonistes, cest la question mme de
lEtre qui est en jeu. Mais en voici dabord une traduction littrale:
Lempereur de la mer du Sud tait Shu, lempereur de la mer du Nord tait Hu, lempereur du
milieu tait Hundun. Comme chaque fois quils staient retrouvs chez Hundun celui-ci les
avait reus avec la plus grande amnit, Shu et Hu se concertrent sur la meilleure faon de le
remercier de ses bonts: Les hommes, dclarrent-ils, ont sept ouvertures pour voir, entendre,
manger, respirer. Lui seul nen a aucune. Et si on les lui perait? Chaque jour ils lui perforrent
un orifce. Au septime jour Hundun avait rendu lme.
La mort de Hundun a tout dun drame boufon. Mais le ton comique ne doit pas mas-
quer le substrat cosmique sur lequel la fable est construite. Le rcit de la mort de Chaos
fait un clin dil un vieux fond mythique dont laire de difusion est trs vaste. Mythe
de Fuxi et Ngua, li au cycle du dluge des Miaos, mythe du chien Panhu des Mans du
Sichuan, enfn et surtout mythe de Bangu, probablement dorigine tibtaine. Issu de
luf cosmique primordial form par le ciel et la terre alors troitement uni en un tout,
le dmiurge Bangu semploie les sparer, et une fois sa tche acheve, il meurt, se
transforme et son cadavre, en se dcomposant, donne naissance notre monde, avec
ses feuves, sa vgtation, ses phnomnes atmosphriques et ses luminaires qui
constellent la vote cleste. Ces rcits visent dabord afrmer ltroite parent entre
le corps de lhomme et lunivers, entre le macrocosme et le microcosme. Cette prsence
du mythe se manifeste aussi bien dans les noms des protagonistes: Hundun (Chaos),
Shu (Illico) et Hu (Presto), que dans le thme du percement des ouvertures. Bangu est
apparu au centre de luf primordial, form par ltroite union de deux principes anti-
thtiques qui vont peu peu se dissocier pour former le Ciel, en haut, form des l-
ments purs et lgers et de la Terre, en bas, constitue de composants plus lourds et plus
grossiers. Chaos occupe lui aussi le centre, fanqu de deux acolytes, occupant chacun
un point de lespace oppos, symbolisant le nord et le sud, mais encore la lumire et
lobscurit et les emblmes qui y rpondent: yin et yang qui sont aussi, sur laxe verti-
cal, Terre et Ciel en haut et en bas, mais qui associs donnent Shuhu, dissyllabe qui sert
dsigner lclair, vecteur de la fcondation cosmique. Bangu est lagent dune opra-
tion douverture il fend luf primordial, sparant le Ciel et la Terre au moyen de sa
hache , de mme que Chaos est le patient dune opration douverture, puisquon m-
nage des orifces dans une face qui, dpourvue de toute cavit et de toute protubrance,
doit se prsenter comme luf cosmique. A la suite de ces interventions les deux per-
sonnages meurent. Mais tandis que Bangu devient le monde tel que nous le voyons,
Chaos se voit doter des moyens de voir le monde tel quil est. Le mythe de Fuxi et N-
gua est certes inscrit en fligrane dans la trame du meurtre de Chaos, qui en serait la va-
riante cocasse, mais il commande aussi larmature structurelle du premier volet du
dialogue de Zhuangzi et de Huizi qui sans cela demeure peu clair. On comprend mal
en efet la signifcation de la fable du baume, qui serait parfaitement gratuite si elle ne
reproduisait pas, en sous-main, pourrait-on dire, la trame de lanecdote des courges. Le
premier dialogue fait surgir des thmes propres au dluge. Grce la courge intacte le
rhteur aurait pu voguer sur les fots, la faon de Fuxi et Ngua qui surent chapper
linondation en prenant place dans lnorme cucurbitace ne des graines miracu-
leuses du roi dragon. On notera que Huizi sme les graines de cucurbitaces, gantes
elles aussi, quil tient dun personnage dont la dignit lgale un dragon. A lintgrit
de la courge primordiale qui sert de vhicule cleste au couple incestueux et indivis,
228 jean levi
soppose le dcoupage de la courge ne de leur union dont rsulte lhumanit mortelle,
marque par ses divisions en peuples, en clans et en groupes exogames, de mme que
Huizi ne peut concevoir dautres usages de la plante magique que dans son morcelle-
ment de la mme faon quil dcoupe, avec les catgories du langage analytique, la to-
talit mouvante et confuse du rel en objets et en units discrtes, tablissant ainsi en-
tre les tres des relations toujours prouves sur le mode des rapports hirarchiques.
On notera en sus que chez certaines ethnies, le mythe de Fuxi-Ngua sert de rcit tio-
logique la varit des langues humaines. Dans un rcit Nahsi lindistinction symboli-
se par la courge ou la boule de chair ne des rapports incestueux entre le frre et la
sur est remplace par le silence: le couple primordial met au monde trois enfants nor-
maux, ceci prs quils sont muets, mais ils retrouvent chacun lusage de la parole
quand un cheval fait intrusion dans le champ de navets familial. Pour prvenir leurs pa-
rents, chacun poussera un cri difrent, donnant ainsi naissance aux grandes langues de
la rgion: tibtain, bai et nahsi.
Lanecdote du baume, quant elle, afrme sa solidarit avec la geste du dluge tout
dabord en flant la mtaphore aquatique. Elle nabandonne pas llment liquide en ef-
fet, au premier chef par le lavage des dchets de soie, et en second lieu par la rfrence
un combat naval qui permet une lvation au sommet de la hirarchie, de mme que
la navigation du couple des jeunes gens les lve littralement jusquau ciel avant de les
conduire au sommet dune montagne. Le paralllisme est conserv ensuite travers le
dtail tnu mais essentiel de lactivit laquelle se livre la famille du Song: ils sont des
laveurs de bourrette le mot employ est guang, glos par xu. Le mot xu dsigne ces d-
chets qui entourent le cocon de soie et qui ne pouvant tre traits comme le fl de soie,
sont travaills et cards la faon de la laine ou du coton. Le terme, mtonymiquement,
renvoie donc au cocon de ver soie, qui fgure dans le rcit de fondation des Man du
Sichuan en tant que signe de la totalit et de lindistinction. En efet le chien Panhu, an-
ctre fondateur de la tribu des Man, est n de la srosit auriculaire dune concubine
dun empereur de Chine lgendaire, en tout point semblable un cocon de ver soie.
En mme temps il manifeste le rapport entre la fragmentation de la totalit et le lan-
gage humain ordinaire, lauxiliaire descriptif xuxu, dsignant des propos insipides qui
se droulent comme lcheveau de la bourre de soie, un bavardage vain et strile qui ne
saurait avoir de prise efective sur le rel.
Le motif nautique va se doubler dun motif politique travers la nouvelle applica-
tion que lui fait subir le hros du rcit. Toutefois jusque dans sa critique de la pense
utilitariste le texte nabandonne pas le terrain du mythe, car travers la rfrence au
pays de Yue et sa lutte contre Wu, on voit se profler la fgure de son hros civilisateur:
Yu le Grand. Or celui-ci sest signal par ses exploits dingnieur hydraulique dans les l-
gendes chinoises qui historicisent le cycle lgendaire du dluge labor par les socits
tradition tatique par opposition au cycle de Fuxi et Ngua caractristique dethnies
galitaires et communautaires qui ignorent les lignes dynastiques. Le personnage de
Yu le grand, fondateur de la dynastie des Xia dont lanimal ponyme est le dragon, em-
blme de la Souverainet, constitue lalternative absolutiste des socits qui, bien que
marques par les stigmates de la fragmentation, portent encore, travers linceste fon-
dateur, le souvenir de la totalit perdue.
Mais le rseau souterrain des correspondances ne sarrte pas l. Aucun dtail na t
choisi au hasard. Le baume, dit le texte, sert primitivement protger des engelures. Le
caractre dont se sert Zhuangzi est le mot gui qui signife au sens propre tortue deau.
de la dispute sophiste au dialogue philosophique 229
Les glosateurs dissertent doctement pour savoir sil faut lentendre au sens mtapho-
rique que ce sens driv signife vein comme la carapace de la tortue ou plutt se
rtractant comme la tortue qui rentre sa tte ou ses pattes dans sa carapace ou bien
sil ne sagirait pas plutt dun emprunt phontique, pour le mot zou, rid, frip, le
caractre gui pouvant lui aussi se