1 Evans, J. D. G., Aristotle's Concept of Dialectic (Cambridge, 1977);Google ScholarOwen, G. E. L., ed., Aristotle on Dialectic: The Topics (Proceedings of the Third Symposium Aris-totelicum) (Oxford, 1968);Google ScholarBarnes, J., Schofield, M., and Sorabji, R., eds., Articles on Aristotle (London, 1975).Google Scholar See the works mentioned in notes 2 to 9, below.

2 Solmsen, F., “Dialectic without the Forms,” in Aristotle on Dialectic (above, note 1), 55 (dialectic's “extraordinary and almost appalling loss of status”)Google Scholar, and Huby, P., “The Date of Aristotle's Topics and its Treatment of the Theory of Ideas,” Classical Quarterly, N.S. 12 (1962), 72, note 1,76, 79–80.CrossRefGoogle Scholar This view of dialectic's lossof status is usually connected to the belief that the Topics is one of Aristotle's early works (see Huby, “Date of Aristotle's Topics,” and Ross, W. D., Aristotle (London, 1923), 56–57, 59). Solmsen, however, maintains that Aristotle had “a very clear idea” of the difference between dialectical and demonstrative arguments when he wrote the Topics, even though the Posterior Analytics was probably written later (“Dialectic without the Forms,” 55)Google Scholar.

3 Brunschwig, J., Aristote: Topiques (Livres I–IV) (Paris, 1967), xi–xii;Google ScholarSolmsen, “Dialectic without the Forms,” 53–54Google Scholar.

4 Moreau, J., “Aristote et la dialectique platonicienne,” in Aristotle on Dialectic (above, note 1), 83–85;Google ScholarSolmsen, “Dialectic without the Forms,” 54;Google ScholarBrunschwig, Aristote: Topiques, xiGoogle Scholar.

5 Aubenque, P., “Sur la notion aristotélicienne d'aporie,” Aristote et les problèmes de méthode (Communicationspresentées au Symposium Aristotelicum tenu à Louvain du 24 août au 1^er septembre 1960) (Louvain and Paris, 1961), 10–18;Google ScholarWeil, E., “Laplace de la logique dans la pensée aristotélicienne,” Revue de Métaphysique et de morale 56 (1957), 283–315.Google Scholar Weil's article is translated into English in Articles on Aristotle (above, note 1).

6 Pater, W. A. De, S.C.J., Les Topiques d'Aristote et la dialectique platonicienne: La Méthodologie de la définition (Fribourg, 1965), 75, 82–83.Google Scholar

7 Aubenque, P., “Sur la notion aristotélicienne d'aporie,” 12–14.Google Scholar

8 For example, see the most thorough study of dialectical question and answer to date, that of Moraux, Paul, “La joute dialectique d'après le huitième livre des Topiques,” in Aristotle on Dialectic (above, note 1), 277–311.Google ScholarMoraux attributes to the interlocutor the goal of demonstrating, establishing, and proving a thesis (277, 279, 283).Google ScholarPater, De (Les Topiques d'Aristote, 81–82) andGoogle ScholarBrunschwig (Aristote: Topiques, xxix) do note the destructive character of the interlocutor's activity. Solmsen observes “the impression that with Aristotle (in the Topics) the desire to refute definitions is somewhat stronger than the desire to establish or defend them” (“Dialectic without the Forms,” 57).Google Scholar According to Brunschwig, when the thesis advanced by the responder is positive in form, the syllogism composed by the interlocutor constitutes a refutation (anaskeua-zein), whereas it is called établissement (kataskeuazein) when the initial thesis is negative (xxx). In this paper I use “refutation” more loosely so as to designate an interlocutor's attack on a responder's thesis, regardless of whether the thesis is positive or negative in form. For this general use of “refutation,” see Topics 8. 1 155b5, 8. 3 158a31, and 81–82Google Scholar.

9 See, for example, Blond, J.-M. Le, Logique et methode chez Aristote (Paris, 1939), 43.Google Scholar Le Blond adds that the nature of exercise in dialectic in no way illuminates the deepest nature of dialectic and its relationship with science. This paper seeks to show that a strong connection between these two does exist. To date, De Pater and Evans have tried to spell out the relationship between dialectical exercise and dialectic's utility for philosophic investigation. The present analysis agrees with Evans that what is crucial is the dialectician's ability to argue both sides of a question and “see the consequences of alternative hypotheses” (Aristotle's Concept of Dialectic, 31–32, 34) and with Pater, De that refutation is central to this process (Les Topiques d'Aristote et la dialectique platonicienne, 81–82)Google Scholar. The bulk of the paper (Section II) is devoted to showing how this process works and how—contrary to the claims of some scholars (see especially Moreau, “Aristote et la dialectique platonicienne,” 83–84)—dialectic is able to break out of the sphere of the probableGoogle Scholar.

10 For logos as “claim,” see the argument which follows. Aristotle uses to thesis hupechein to refer to the same activity as that characterized by to logon hupechein in Topics 8. 3 158a31, 8. 9 160b14.

11 Soph. Elen. 1 165a2–3, 6 168a36–37, 11 172b5–8. Of course, in one sense refutation is itself a proof (a syllogism with a conclusion), albeit a proof in order to disprove. “Proof” will be used with the positive meaning ofproof for the purpose of showing that something is the case.

12 The problem is complicated by Soph. Elen. 1 165a24–28, where to logon labein is contrasted with to logon dounai. Since Soph. Elen. 1 is ambiguous—in that to apseudein and ton pseudomenon emphanizein could each be construed as referring to the interlocutor or to the responder—it is better to use Soph. Elen. 34 as the standard. In Soph. Elen. 34, to logon labein clearly refers to constructing syllogisms and to logon hupechein to preventing their construction. Pickard-Cambridge's translation “standing up to an argument” would, on this basis, seem to be best, were it not for the analogy with to thesis hupechein (see above, note 10), which means defending a thesis, not standing up to one.

13 The mere fact that the interlocutor puts together a dialectical syllogism does not insure that the conclusion is true or that the refutation is real. It is the interaction between the interlocutor and the responder acting as a watchman (and, in investigation, this together with thejuxtaposition of two such proofs and reflection on them) that makes the perception of truth possible (see p. 85 below).

14 The responder must, however, conform to the ground rules Aristotle lays down; and Aristotle makes a sharp distinction between a responder who discerns grounds for rejecting a proposition legitimately and he who objects capriciously and without explanation (see Top. 8. 8).

15 Aristotle consistently puts the blame on the responder for allowing an interlocutor to construct a false refutation (Top. 8. 11 161a16–17, 161b16–18). This is in keeping with his approach throughout, namely, making the responder responsible for the level of truth or falseness a discussion achieves (see p. 86 below).

16 The situation changes when the responder does not believe everything he concedes. The present argument does not deal with this complication, because it focuses on that part ofthe responder's art that is common to argumentative and investigative dialectic.

17 Top. 8. 14 163b29–33 makes this claim for generally accepted beliefs. It seems it could be extended to rigorously scrutinized opinions and arguments that will accumulate over time.

18 Top. 8. 4 159a20–24, 8.6 160a1 1–14, 8. 7 160a27–29. Even someone with knowledge can be refuted by a sophist (Soph. Elen. 8 169b 25–29).

19 Aristotle draws a sharp distinction between testing that stays entirely within the horizons of a particular science and testing that has recourse to principles outside a science (Soph. Elen. 9 170a33–39).

The first philosophic use of dialectic mentioned in Top. 1. 2—discerning truth and falsehood—is the subject of the rest of this paper. Whether the second use is a subdivision of the first or a separate use entirely, dialectic's ability to reach truth will have a bearing on dialectic's role in connection with principles.

20 Because all people know the common principles (ta koina). However, Aristotle distinguishes between the intelligent nonexpert and the man in the street (Soph. Elen. 9 170b7–9).

21 It is possible to construe all of the Topics as well as On Sophistical Refutations as the referent. The usual interpretation of Aristotle's claim is that he first systematized an activity that others already engaged in. The question is, did the systematization itself change the character of the activity? Aristotle says in this passage that Socrates knew how to put questions, but not how to respond (Soph. Elen. 183b6–8). Aristotle's innovation is an art that enables the responder to detect falseness in the same way that someone who has knowledge can—without, however, possessing that knowledge himself.

22 The sophist pretends to a wisdom he does not have and makes money thereby. Sophistical refutations share only the former characteristic. Sophistical refutations need not, it seems, involve deliberate deception; they could be the result of “honest” mistakes.

23 If, however, a refutation is no longer deemed structurally sound when, for example, terms are used ambiguously (see p. 89 below), then dialectical refutations are those which, using true and relevant premises, reason syllogistically to the contradictory of a thesis. When premises are false, or true but irrelevant to the problem, the form would be unsound as a consequence.

24 Soph. Elen. 3–8 is on refutations by means ofthe thirteen fallacies; 16–32 on solutions to the fallacies and, thereby, to the refutations.

25 By “false dialectical proofs” I mean proofs genuinely dialectical, but which employ false (although generally accepted) premises.

26 To this should be added the need for practice (Soph. Elen. 16 175a20–30).

27 See Soph. Elen. 18 177a6–8.

28 Top. 8. 14 163a36–b4, 8. 1 155b10–11. In any case, investigators must train themselves i n both arts and must carry out whatever role they assume as honestly as if they were investigating by themselves, with no thought of looking good.

29 Top. 8. 14 163b5–33 speaks of an arsenal of generally accepted arguments; see above, note 8.

30 Top. 1. 1 100a18, 8. 5. In 8. 5 the criterion is that the premises be more generally accepted than the conclusion. The result is that all the propositions the responder concedes, or considers conceding, should be generally accepted to some degree. Aristotle seems to have in mind not how many people endorse a belief, but how few oppose it (Top. 8. 2 157b32–33).

31 See Top. 8. 6–8.

32 Those para tēn lexin.

33 See especially Soph. Elen. 7.

34 Top. 8. 10 mentions four kinds of objection (enstasis) a responder can make, only one ofwhich is a solution. See Met. 3. 1 and Aubenque, “Sur la notion aristotélicienne d'aporie” (above, note 5).

35 Top. 8. 10 160b23–39; Soph. Elen. 11 171b8–11.

36 Soph. Elen. 22 178b18–23. In practice, if two contradictories are each partly true and partly false, then there can be two solutions, since there is something true on each side. I n situations like these, the investigators will reformulate the two contradictories to take into account the mixture of truth and falseness so exposed. They will keep repeating this process until they reach contradictories, one ofwhich is entirely true, one entirely false. See Top. 8. 2 157b8–11.

37 Top. 1. 2 101a34–38, 8. 14 163b9–16.

38 This appears to be the sense of “understanding a knot” in Met. 3. 1 995a30.

39 Top. 8. 14 163b9–16.

40 There is an obvious parallel between the investigative process outlined in the Top. and Soph. Elen. and that suggested by Met. 3. 1 995a25–b4 (see above, note 38).

41 See Moreau, “Aristote et la dialectique platonicienne,” 88–89Google Scholar, on dialectic's “role gnoristique.”