¹ For the relation of this to 1449b8 see Heath, M., ‘Aristotelian Comedy’, CQ 39 (1989), 348–52.CrossRefGoogle Scholar

² Cf. Halliwell, S., Aristotle's Poetics (London, 1986), pp. 140–1Google Scholar (on πργματα as ‘events’), 144 (on the term's ‘mediating function’).

³ Halliwell (n. 2), p. 22, speaks aptly of mimesis ‘embodying’ universals; less precise is the formulation on p. 55, which speaks of a ‘representation’ of universals.

⁴ Halliwell's discussion (n. 2), pp. 231–2, is flawed in part by the assumption that Aristotle is summarising the plot of Euripides' play as such.

⁵ On the term ‘episode’ here, and its relation to the apparently different usage in the following discussion of epic, see Heath, M., Unity in Greek Poetics (Oxford, 1989), pp. 52–3.Google Scholar

⁶ The text at 1455b7–8 is disputed. Mackay, L. A., AJP 75 (1954), 300Google Scholar, seeing that the plot is not in question here, proposed the deletion of ἓξω τοû μθου, which together with a transposition, a supplement by Hermann and repunctuation gives: τ δ ὅτι νεîλεν θες λθεîν κεî ‹κα› δι τνα αἰτααν ἕξω τοû καθλου· καἱ φ' ὅτι δ. But this leaves intact a feeble doublet; it might be better therefore to remove δι τνα αἰταν also, in which case λθεîν κεî must either be transposed with Mackay or deleted (Bekker) to give: τ δ ὅτι νεîλεν θες ἕξω το καθλου· κα ø' ὅτι δ. It is easy to see how this cryptic (but not thereby un-Aristotelian) text could have attracted explanatory interpolation. On the other hand, the much simpler solution adopted by Kassel might be accepted on the assumption that the phrase ἕξω το μθου is used loosely; and that is not an extravagant assumption when dealing with Aristotle.

⁷ See e.g. Halliwell (n. 2), pp. 208–10, for a good statement of the case.

⁸ Stinton, T. C. W., ‘Hamartia in Aristotle and Greek Tragedy’, CQ 25 (1975), 221–54CrossRefGoogle Scholar = Collected Papers on Greek Tragedy (Oxford, 1990), pp. 143–85.Google Scholar

⁹ Sorabji, R., Necessity, Cause and Blame (London, 1980), pp. 295–8Google Scholar (see also pp. 279ff.); cf. Lucas, D. W., Aristotle Poetics (Oxford, 1968), pp. 301–2.Google Scholar

¹⁰ Stinton (n. 8), pp. 225–6 (with p. 225 n. 2).

¹¹ Sorabji (n. 9), p. 297. I should stress that my sole concern here is whether the events, thus interpreted, are consistent with Aristotle's theory; whether the events should be interpreted in this way (which I doubt: see section (IIc) below) is another issue.

¹² Ross, W. D., Aristotle's Physics (Oxford, 1937), p. 519.Google Scholar

¹³ Knowable, not simply known; for if it were knowable but unknown the action would be an instance, at best, of μρτημα in the narrower sense (μ παραλγως ἅνευ δ κακας EN 1135b17–18).

¹⁴ There is another incidental reason why this story cannot furnish a tragic plot: the falling of the statue is neither an action nor the result of an action (in the more precise terminology of Physics 2 it is π τοû αὐτομτου rather than π τχης) But the absence of agent and action cannot be the feature with which Aristotle is primarily concerned here, for the murderer's death would no no less objectionably fortuitous if we supposed the fall of the statue to be an act of random terrorism.

¹⁵ Strictly speaking, there is a prior condition on which the necessary or probable connection between the killing of Mitys and the fatal falling of the statue depends, but which is causally independent of the killing of Mitys: namely, the existence of an agent such as would desire and be able to effect vengeance on Mitys' behalf should he be murdered. But this does not rob the argument of its force, since the existence of such a potential avenger could be regarded as intrinsically probable, while the fact that the passer-by ‘happens to be’ the father is intrinsically improbable.

¹⁶ Sorabji (n. 9), pp. 10–11.

¹⁷ This point is anticipated by Halliwell (n. 2), p. 208 n. 11, although I would distinguish (as he does not do here) ‘outside the play’ from ‘outside the plot’. This distinction is entailed by 1454b6–8 and 1455b24–32 (the complication includes τ ἕξωθεν but starts π' ρχς this is contradictory unless the beginning [sc. of the plot] may be outside [sc. of the play]; thus τ προπεπαγμνα at 1455b30 are, as one might expect, among the πργματα of the plot); cf. also 1453b31–4, 1454b2–6. Seemingly anomalous is 1460a27–33, where ἔξω το μυθεματος is equivalent to ‘outside the play’. (Halliwell recognises the distinction at p. 211 n. 14, but fails to see that it is consistent with ch. 7, which is about the structure of a complete action, not about the structure of a play.)

¹⁸ Aristotle's principle of delimitation – the closure of the causal sequence – is admittedly problematic; see Heath (n. 5), pp. 41–2 (but p. 42 n. 7 is in error: see nn. 6 and 17 above). But this does not affect the present argument.

¹⁹ Cf. the analysis in Poetics 14, where ignorance is crucial to the most favoured types of plot (there is nothing in context to suggest that this ignorance must be avoidable, so as to exclude mischance).

²⁰ Unless there were some third event (e.g. the suicide of their mother) consequent on the two deaths together (cf. 1459a27–9 ξ ὦν ἓν οὐδν γγνεται τλος).

²¹ Cf. Lloyd-Jones, H., The Justice of Zeus (Berkeley, 1971), pp. 103–28.Google Scholar

²² Halliwell (n. 2), p. 148.

²³ This seems to be implied on, e.g., pp. 140, 143.

²⁴ God has an οἰκεα πρξις, and only ξωτερικα πρξεις cannot be attributed to him: Pol. 1325b28–30; cf. EN 1154b24–8.

²⁵ This and the following arguments are from Halliwell (n. 2), pp. 231–3.

²⁶ It is possible to marginalise the theses of ch. 25 by dismissing it as an anomaly within Aristotle's theoretical system; but this seems to me a doctrinaire procedure. Is it not more reasonable to suppose that Aristotle's practical engagement with poetry provides valuable evidence of how he himself understood his principles? If so, then ch. 25 is in a sense a key chapter for the interpretation of the Poetics. (Halliwell also raises a question, p. 233, about the bearing of ch. 25 on the ‘quasi-philosophical value’ ascribed in ch. 9; I shall take up this point in section III) below.)

²⁷ The concept of an agent with an implied role in the πρξις should not be thought difficult. The conspicuous place which Dionysus has in Euripides' play makes him a usefully clear illustration of divine agency; but it must be remembered that Aristotle's analysis is directed primarily towards plots, rather than plays (see n. 17 above). Consider, e.g., Hippolytus; Aphrodite's agency is crucial to the plot (cf. Heath, M., The Poetics of Greek Tragedy (London, 1987), p. 53Google Scholar), but it would be no less crucial even if it were conveyed obliquely or implicitly and she did not appear on stage.

²⁸ Cf. Heath, M., BICS 34 (1987), 66–8Google Scholar. As an example of how Aristotle's own practice might be corrected in this way from his principles, one may consider his objection to Aegeus' appearance in Medea (1461b20–1); is it in fact irrational? Given the evidence of divine agency in this play (cf. Heath (n. 27), p. 57), one might entertain the possibility that it is in accordance with necessity or probability.

²⁹ Stinton (n. 8), p. 232, cf. p. 226. The association of chance and compulsion is of course justified; cf. Rhet. 1369b5 βᾳ δ, ὅσα παρ' πιθυμαν ἤ τοὺς λογισμοὺς γγνεται δι' αὐτν πραττντων. ‘Wholly compulsory’ allows for those acts which, though compelled, are not κοσια without qualification, according to EN 3.1; cf. Stinton, pp. 228–9.

³⁰ Cf. Stinton (n. 8), p. 225: ‘the change from good fortune to bad must result from some action, i.e. not be the result of a mischance’.

³¹ Belfiore, E., ‘Aristotle's Concept of praxis in the Poetics’, CJ 79 (1983/1984), 110–24Google Scholar, argues that in the Poetics πρξις ‘never means “a morally (or ethically) qualified action”, that is, an action for which one may appropriately be praised or blamed’ (p. 110). This is partly, of course, because the primary use of πρξις in the Poetics is to refer to the ‘single action’, which (as we have seen) is not something which some one person does, but a product of interaction. Nevertheless, this interaction is constituted by the actions of many agents, and these actions are as such appropriate candidates for moral evaluation; so Belfiore's argument (with which I am broadly sympathetic) needs qualification. A πρξις is evaluated on the basis of προαπεσις; if the προαρεσις cannot be determined, then evaluation is not possible in practice. But Aristotle indicates that in drama προαρεσις may not be made clear (1454a16–18; in this case, the tragedy will lack ἦθος, which Aristotle regards as possible, though not desirable: 1450a24–9, cf. 1460al 1). So the actions of dramatic agents are the kind of thing that is in principle subject to moral evaluation, but may not actually be so in practice.

³² EN 1100a21; EE 1233b22–5; Rhet. 1386blO–12, 25, 27–8, 87a9, b23; Top. 109b37, 110a2.

³³ Cf. Belfiore (n. 31), 121, aptly citing Margoliouth, D. S., The Poetics of Aristotle (London, 1911), pp. 37–41.Google Scholar

³⁴ Contrast Halliwell (n. 2), pp. 144–8; the tendency to think in terms of one central agent (pp. 147–8, 217, 223–4) seems to me particularly misleading here.

³⁵ Iambic is poetry (1448b33–4) and is mimetic (1448b25–7 shows that it imitates actions); Halliwell (n. 2), p . 55 n. 15, is right here against Janko, R., Aristotle on Comedy (London, 1984), p. 61Google Scholar. In 1451b14 I believe one should read περἱ τ καθ' ἕκαοτον; cf. Heath (n. 1), p. 351 n. 28.

³⁶ Cf. Heath (n. 5), pp. 48–9, 80–1.

³⁷ Belfiore (n. 31), 119, argues that ‘serious’ in ch. 9 can be linked to the ‘seriousness’ of tragic action as denned in ch. 6 (1449b24); but this must be wrong. It is a tragic action that is by definition serious, but poetry that is more philosophical and more serious; the latter includes comedy (1449b7–9, 1451b11–15; cf. n. 1 above), and it is hard to believe that the σπονδαἱα πρξις of tragedy is to be divorced from the distinction between tragedy and comedy as imitating respectively σπουδαῖοι and ϕαûλοι set out in ch. 2 (cf. 1448a25–9, 1448b25–6). Her argument is that ‘serious and complete’ in 1449b24 becomes ‘complete and whole’ in 1450b24, which is explained in terms of necessity and probability, which is the basis of universality in ch. 9. But this is a mistake: ‘whole’ in 1450b24 is not a substitute for ‘serious’, but a gloss on ‘complete’ (cf. Phys. 207a13, 228b14, Met. 1016b17).

³⁸ Else, G., Aristotle's Poetics (Cambridge, Mass., 1963), p. 133Google Scholar, argues that learning and knowledge for Aristotle are of universals; hence ‘if you have merely recognized the resemblance of one individual … to another individual … you have not learned anything.’ He is followed by Sifakis, G. M., ‘Learning from Art and Pleasure in Learningy’, Studies in Honour of T. B. L. Webster (Bristol, 1986), i.211–22Google Scholar. But (i) this oversimplifies Aristotle's account of knowledge (see Met. 1087a10–25, with the commentaries of Ross and Annas ad loc.); and (ii) what is involved is not simply the recognition of resemblances, but an inference of identity from resemblances or other signs (cf. Sifakis, p. 219: ‘we recognize that a man with a lion-skin and club in a picture … is Herakles, because we know that Herakles is always represented with these attributes’); hence the cause is known, and the inference is not wholly individual. Sifakis' interpretation of οὗτος κεῖνος makes it the conclusion of a syllogism of which the first premise is not to be found in Aristotle (every well-imitated image is a representation of universals: but ch. 9 is concerned only with some kinds of poetry) and the second (e.g. the Athena Promachos by Pheidias is a wellimitated image of Athena as Promachos) cannot be affirmed unless the conclusion (the Athena Promachos by Pheidias is a representation of what Athena as Promachos must, necessarily or probably, be like) is already known (cf. ν μ τχῃ προεωρακώς).

³⁹ Halliwell (n. 2), p. 77 n. 42, considers some other possibilities.

⁴⁰ The cognitive process described in 1448b15–17 is of a very rudimentary kind, and Aristotle emphasises that the pleasure it affords is available even to the least philosophical: 1448b13–15. The appreciation of a tragic plot is more sophisticated, since it involves grasping the plot as instantiating universals, and the cognitive pleasure is presumably therefore greater. However, this does not necessarily imply that grasping a tragic plot is positively difficult; according to Rhet. 1410b10–11, ‘learning easily is naturally pleasant to all’.

⁴¹ Cf. Halliwell (n. 2), p. 132; but this seems to undermine his stress (pp. 101–6) on the distinction between objective and subjective probability. I note in passing that the cognitive process involved in grasping a plot as instantiating universals is the same whether or not those universals are held true of the real world; the cognitive pleasure would therefore be the same.

⁴² I discuss the nature of this pleasure in ‘Aristotle and the Pleasures of Tragedy’ (forthcoming), and attempt to show why Aristotle might have regarded this as a morally serious end.

⁴³ 1452b31–2 takes the superiority of the complex plot as established; this is intelligible if the reference to a previous discussion of περιπτεια in 1452a23 (καθπερ εἳρηται) is to 1452a3–4, since that is an argument that events παρ τν δξαν δι' ἅλληλα have greater emotional effect. (1452a23 cannot refer to 1451a12–14, which is only concerned with the change of fortune, a feature of simple plots as well as complex ones, whereas περιπτεα is one of the defining features of complex plots; cf. Glanville, I. M., CQ 41 (1947), 73.)CrossRefGoogle Scholar

⁴⁴ Cf. Heath (n. 5), pp. 43–4.

⁴⁵ Cf. Heath (n. 27), p. 12.

⁴⁶ Alternatively, there may be a progression from harm anticipated to harm on the verge of being realised; this is consistent with the cases of averted action discussed in ch. 14. For the emotional force of a merely imminent disaster see Stinton (n. 8), 253.

⁴⁷ That Aristotle envisages changes of fortune from bad to good, as well as from good to bad, as suitable tragic plots (1451a12–15, 1455b27) may seem surprising; but consider e.g. Euripides' Ion. Initially both Ion (deprived, though he does not know it, of his proper status) and Creusa are unfortunate, and both end in good fortune; what makes the play tragic is the imminence of worse disaster that besets the progress from bad to good fortune (see previous note).

⁴⁸ Thus pace Halliwell (n. 2), p. 203 n. 2, it is correct to say ‘that eutuchia is dependent on action’, and Belfiore's suggestion (n. 31), 115, that in 1450a17–20 εὐδαιηονα is used in a nontechnical sense for εὐτυχα may be right (εδαι at any rate includes ετυχα, given Aristotle's acceptance that ‘external goods’ are necessary). Kassell's athetesis is certainly incorrect: Horn, H.-J., Hermes 103 (1975), 292–9Google Scholar, has shown that it leaves Aristotle's argument incomplete; the objection to κακοδαιμονα in a17 (for which Halliwell ibid. cites a parallel from Protr. B46 Düring) casts no doubt on καἱ τ τλος…κτλ.

⁴⁹ Cf. Heath (n. 5), p. 41 with n. 6.

⁵⁰ Provided, of course, that they are not hurt: τ γρ γελοῖν στιν μρτημ τι κα αἷσχος νώδυνον καἱ οὐ øθαρτικν(1449a34–5).

⁵¹ A shortened version of this paper was read at a meeting of the Northern Association for Ancient Philosophy on 3 April 1991. I am indebted to the participants, and especially to Hans Gottschalk for his subsequent observations on the full text. The comments of CQ's anonymous referee helped clarify the presentation of the argument at a number of points.