1. I am grateful to Professor A. M. Snodgrass and others for valuable comments on a version of this paper which was read to the Cambridge Philological Society in October 1983.

2. TAPA 70 (1939) 139–57Google Scholar; and, with slightly different arguments, Aristotle's Poetics: the argument (1957) 164–79Google Scholar.

3. The stagecraft of Aeschylus (1977) 457–8 n. 4Google Scholar.

4. One argument that he uses to justify both deletions is that τε in 16 () should be coordinated with τε in 21 (). But, even if both deletions were right, the words could not naturally be so taken, since the two instances of τε would still be separated by two of καί.

5. The trouble with this deletion is that it involves separating σατυρικοῦ in 20 from σατυρικήν in 22; it is difficult to believe, with Else, that in this context the latter word might mean little more than ὀρχηστικωτέραν. It might be thought preferable to bracket both (inserted by someone who wrongly took , to mean intentionally comic diction?) and ; but Aristotle did regard trochaic rhythm as κορδακιώτερος; (Rhet. 1408b36).

6. I shall similarly be ignoring the view (ill-founded, I think) of de Montmollin, D., La Poétique d'Aristote (1951) 42–5Google Scholar, that everything from (15) to Σοφοκλῆς (19) is a later addition by Aristotle to his own text.

7. Dr James Diggle suggests to me that it may originally have contained one: ‘Aeschylus … gave dialogue the leading role <and introduced the skene>’ That would meet this particular difficulty, but others would remain.

8. Vitr. 7. praef. 11.

9. The testimony about Democritus and Anaxagoras tends to be implicitly accepted in books on Greek painting but treated with more reserve, if at all, in books on Greek philosophy (e.g. Guthrie, W. K. C., A history of Greek philosophy 11 (1965) 270–1)Google Scholar. A sceptic might wonder how perspective is to be reconciled with Democritus' theory of vision, and why, if the subject had such a respectable philosophic pedigree, it was neglected by later philosophers such as Aristotle.

10. Dem. 21.147, [Andoc.]4.17, Plut., Alc. 16Google Scholar. The anecdote need not be true, but it has as much claim to consideration as Vitruvius' testimony.

11. Among those who heard this paper delivered there was a feeling that I had been too sceptical about Agatharchus and his connection with Aeschylus. Perhaps so; but I should anyway wish to claim that Agatharchus' innovation, whatever its date, could only have been a permanent painted set, which cannot be relevant to the context in Aristotle (see below).

We should in any case discard the feeble compromise of Rumpf, A., JHS 67 (1947) 13CrossRefGoogle Scholar, who claims, on no better evidence than Ar., Ach. 10Google Scholar, that Aeschylo docente tragoediam in Vitruvius could refer to a posthumous revival of an Aeschylean play. Plainly Vitruvius' words mean what they say, and we must either accept them as they stand or reject them outright.

12. TAPA 76 (1945) 1–10Google Scholar. A supplementary article at WS 72 (1959) 75–107Google Scholar does not discuss the Aristotelian passage.

13. Pickard-Cambridge, A., The dramatic festivals of Athens ed. 2 (1968) 130–2Google Scholar; Lesky, A., Ges. Schr. (1966) I 241–2Google Scholar.

14. Cf. 1456a27-30: the practice of composing ἐμβόλιμα instead of relevant odes was begun by Agathon.

15. 1452a32-3, 1453all-12, 1453b6-7, 1455a16-18, 1456a25-7, 1460b32-5. Note, however, that most of these passages refer to a single exemplary play (O.T.), and that others by Sophocles are not immune to criticism (1453b36-54a2, 1460a30-2).

16. At 1449b22-4 the definition of tragedy is explicitly linked to what has preceded, and the reference must be partly, at least, to Chapter 4.

17. Vahlen, J., Beiträge zu Aristoteles Poetik (1914) 15, 250Google Scholar.

18. Aeschylus' Niobe is praised at 1456a17, and at 1455a4-6 his Choephori (admittedly a three-actor play) is implicitly treated as a complex tragedy with a creditable recognition scene.

19. For figures see Ziegler, K., R-E VI A 2.1956Google Scholar; Griffith, M., The authenticity of ‘Prometheus Bound’ (1977) 123Google Scholar. But of course Aristotle did not make a precise count, and in fact the figures for Pers. differ little from those for Ag., in which he would doubtless have said that dialogue played the leading role.

20. I do not regard the length as significant, however. Throughout the Poetics the space devoted to different topics bears no discernible relation to their importance.

21. As for the true origin of the third actor, I accept the view of most modern scholars that he was introduced in the period when both Aeschylus and Sophocles were active, and that ancient scholars had no evidence, beyond that of the plays, for attributing his introduction to any individual. Such an innovation would doubtless have had to be approved by the Archon, who would have ensured that it was available to all three competing poets.

22. Cf. Taplin (n. 3) 477-9.

23. Arnott, P., Greek scenic conventions in the fifth century B.C. (1962) 93–4Google Scholar; Baldry, H. C., The Greek tragic theatre (1971) 45–9Google Scholar; Melchinger, S., Das Theater der Tragödie (1974) 162–4Google Scholar; Blume, H.-D., Einführung in das antike Theaterwesen (1978) 60–3Google Scholar. In essence this view is not new, but goes back at least to Gardner, P., JHS 19 (1899) 252–64CrossRefGoogle Scholar. Note also Beare, W., Phoenix 7 (1953) 77–9CrossRefGoogle Scholar; id., The Roman stage ed. 3 (1964) 275-8.

24. The dramatists may well have taken an interest in the decoration of the theatre, and it is possible to imagine Sophocles making a suggestion about it to the authorities. But, if such an incident had occurred in the fifth century, Aristotle could not easily have known if it; and, if he had known of it, he would hardly have thought it worth mentioning here.

25. E.g. Vita Soph. 6: Sophocles invented the bent stick according to Satyrus, white boots according to Istrus.

26. Capps, E., The introduction of comedy into the City Dionysia (1903)Google Scholar; id., Hesperia 12 (1943) 10-11.

27. Cf. perhaps 1448a33-4, where Epicharmus is dated ‘much earlier’ than Chionides and Magnes.

28. I assume that the word πρόλογος is used in the technical sense of Poetics 12. If it meant a single introductory speech, as it sometimes does, it would be still less appropriate here.

29. There was certainly a period when prologues were employed in some tragedies and not others, and, for all We know, this period may have been a long one (cf. Them. Or. 26.316d, where the introduction of prologues is credited to Thespis on Aristotle's authority). Aristotle may not have known of any comedies without prologues.

30. Aristotle clearly did not share the view of Tzetzes (Kaibel, , CGF 18Google Scholar) that Cratinus limited the number of actors to three.

31. Polyb. 12.28a.l,6; Strabo 5.3.8; Plut., Arat. 15.2Google Scholar; Sext. Emp., Math. 1.88Google Scholar; Diog. Laert.2.125; Heliod. 7.7, 10.38; Vitr. 1.2.2; see also Schuhl, P.-M., Platon et l'art de son temps ed. 2 (1952) 77–9Google Scholar (cited by Hourmouziades). Note the architectural emphasis of these sources; Strabo is thinking of an extensive townscape, others of a single building, but there is no sign that the word was ever applied to the painting of natural landscapes.

32. Keuls, E. C., Plato and Greek painting (1978) 72–87Google Scholar.

33. I must mention here the so-called ‘Würzburg skenographia’, a fourth-century Tarentine vase-painting discussed (with references to earlier literature) by Simon, E. and Otto, B., AA 88 (1973) 121–31Google Scholar. I believe, however, that this represents a real, three-dimensional structure, not a two-dimensional projection (cf. the way in which real structures are depicted on the vases illustrated by Pickard-Cambridge, A. W., The Theatre of Dionysus in Athens (1946) 82–112)Google Scholar. Prima facie it must do so, since two of the columns are shown in front of open doorways, with figures passing through them. It is suggested that in reality only one door in each doorway could be opened, the column being painted on the other; but then a most odd effect would have been produced whenever an actor sidled on through a doorway and at once passed in front of a column which he ought to have passed behind. Also the supposed trompe l'oeil effect of porches projecting a long way forward from the ‘picture plane’ (which could anyway have worked properly from only one viewpoint) would be a remarkably ambitious one for a fourth-century painter to attempt.

34. For what there is, and for the limited conclusions that can be drawn from it, see Beare, , Roman stage (n. 23) 295–302Google Scholar.

35. Vitr. 5.6.8; Pollux 4.126-32; see Pickard-Cambridge (n. 33) 234-9.

36. Beare, , Roman stage (n. 23) 248–55Google Scholar.

37. Sifakis, G. M., Studies in the history of Hellenistic drama (1967) 49–51Google Scholar.

38. Vitr. 5.6.9.

39. One would not guess from Simon, E., The ancient theatre, trans. Vafopoulou-Richardson, C. E. (1982) 25–7Google Scholar, that this had ever been questioned; but it is questioned, independently and forcefully, by Pickard-Cambridge (n. 33) 227-30 and Lehmann, P. W., Roman wall paintings from Boscoreale (1953) 90–4Google Scholar. A fortiori Simon's attempt to connect these paintings with individual plays appears worthless.

40. Hourmouziades, N. C., Production and imagination in Euripides (1965) 35–57Google Scholar; Webster, T. B. L., Greek theatre production ed. 2 (1970) 13–17Google Scholar; Simon (n. 39) 20-7; Seale, D., Vision and stagecraft in Sophocles (1982) 19, 27, 114, 163Google Scholar. Newiger, H.-J. in Seeck, G. A. (ed.) Das griechische Drama (1979) 467–74Google Scholar attempts to mediate.

41. Both the points in this paragraph are made by Baldry (n. 23) 47.

42. Arnott (n. 23) 100.

43. At 654-5, in the ‘Deception Speech’, Ajax says that he will go ‘to the bathing place and the meadows by the shore’; at 805-6 Tecmessa, organising the search-party, seems to assume that in this, at least, he was truthful; and, since she succeeds in finding his body, we may suppose that he was. But this takes some working out, and we are given no further indication of the setting.

44. E.g. Seale (n. 40) 163.

45. HSCP 76 (1972) 57Google Scholar.

46. Aesch., Eum. 231Google Scholar, Eur., Alc. 746Google Scholar, Hel. 385, Rhes. 564.

47. Cf. Taplin (n. 3) 338-40, 390-1.

48. So e.g. Webster (n. 40) 17-18 (except that he inevitably drags in the ekkyklema).

49. Taplin (n. 3) 455 claims that almost every play written after the skene was built (shortly before 458 in his opinion) makes positive use of it. For the purpose of this argument, however, he gives P.V. an earlier date than he gives it elsewhere. It is in fact probable that the skene did exist when P.V. was produced, and is ignored (the only thing it could represent is a rock, and there is no trace in the text of a rock rising higher than the place where Prometheus is bound). There are also Euripidean plays (Supp., Tro.) in which the skene is used very little and it is unclear what it represents.

50. Blume (n. 23) 55, 62.

51. So Webster, T. B. L., Sophocles: Philoctetes (1970) 8Google Scholar.

52. So Robinson, D. B., CQ 19 (1969) 36CrossRefGoogle Scholar. I am attracted to Robinson's account, which disposes of the artificiality of Philoctetes' supposed entrance through the cave at 219 (cf. Dawe, R. D., Studies on the text of Sophocles III (1978) 123–5)Google Scholar. But I hesitate; Dale, A. M., Collected papers (1969) 128Google Scholar, was justified in stressing the fact that Philoctetes' approach is heard, not seen, at 210-18, and Robinson's explanation of Eur., Cycl. 707Google Scholar (p. 38) seems forced.

53. E.g. now Kamerbeek, J. C., The plays of Sophocles: Commentaries VII, the Oedipus Coloneus (1984) 24Google Scholar.

54. So Diggle, J., Studies on the text of Euripides (1981) 33–4Google Scholar.

55. UCPCP 14 (1950–1952) 94 n. 14Google Scholar.

56. Seale (n. 40) 143 objects that ‘an exit along either of the parodoi would appear to be in the direction of either Athens or Thebes’. But this is too rigid; departures by the two parodoi are not restricted, in this play or elsewhere, to only two destinations (both the Athenian Stranger and Polynices have arrived from the near neighbourhood). If we think rather of ‘the side of danger’ and ‘the side of security’, with Taplin in Sophocle (Entret. Fond. Hardt 29, 1983) 160, then the latter can fittingly include thesite of Oedipus' death, as well as Athens.

57. So Jebb (p. xxxvii of his edition); Dale (n. 52) 125-6.