1. Read to the Oxford Philological Society in February 1974, and to the Cambridge Philological Society in November 1975. I am grateful to Professor H.Lloyd-Jones for his valuable comments on the original draft, and to Professor E.R.Dodds, Dr M.Schofield and others for helpful criticism in the discussions.

2. This sentence was written in May 1976. Dedicated devolutionists please note that the example is purely illustrative, however tendentious it may appear.

3. I owe this example to Mr Brook Manville.

4. 1.182.1; 2.73.3; 3.3.1; 4.5.1; 4.42.4; 5.86.3; 7.214.1; 8.119. Cf. 2.123.1; 3.113.2; 3.115; 4.96.1; 6.121.1.

5. Pind. Pyth. 2.73; cf. e.g. Plat., Rep. 620cGoogle Scholar, Men. fr.333.8.

6. Callisthenes ap. Plut., Vit. Alex. 33Google Scholar = FGH 124 F 36.

7. In Callim., Hy. 6.98–9Google Scholar., Triopas begins his prayer to Poseidon by reproaching him for not acting as a true father should. Cf. Eur., HF 339Google Scholar, Hipp. 1169–70 .

8. Parry, Adam, ‘Have we Homer's Iliad?’, YCS 20 (1966) 199 fGoogle Scholar.

9. The aberration is explained by schol. Ar. l.c. in euhemerist terms: it is more probable that Lynkeus used lamps for mining, and was therefore said to see beneath the earth.

10. Mr A.S.Hollis suggests an additional reason for Aratus' expression of doubt: .

11. Heinze, R., Virgils epische Technik³ (1928) 243 nGoogle Scholar.

12. In Epod. 9.11 ff. ‘Romanus, eheu, – posteri negabitis – | emancipatus feminae | fert vallum,’ etc., Horace uses a similar phrase of a fact so scandalous as to be incredible. Cf. Virg. Eel. 10.46 ‘nec sit mihi credere tantum’: the love-lorn Gallus does not want to believe what he knows to be true. (I owe this example to Mr P.C.Millett, who also points out that in Ov. Her. 9.9 the reading ‘si creditur’ is confirmed, as against ‘sic creditur’, by analogy with the type of expression here illustrated.)

13. Ovid in fact is closest to Herodotus in the tractatio of Pythagoras, in Metamorphoses XVGoogle Scholar, namely in the at 252 ff., some of which are indeed found in Herodotus: with 273, 276 (the rivers Lycus, Erasinus), 309 (the oasis of Ammon), 392 (the Phoenix), compare Her. 7.30, 6.76, 4.181, 2.73. Here Ovid (or Pythagoras) is concerned with evidence (252 ‘vidi ego’, 282 ‘nisi vatibus omnis | eripienda fides’) and with (259–60, and esp. 359 ‘haud equidem credo’, followed by 361 ‘siqua fides rebus tamen est addenda probatis’). Such expressions, appropriate to the sober, scientific account of the world's wonders, are seldom evoked by the mythical marvels in the rest of the poem (3.311 ‘si credere dignum est’, of the birth of Dionysus, cf. 13.282; see Heinze, , Ovids elegische Erzählung [1919] 18–19)Google Scholar. Seriousness of tone is indicated here, as in the exordium, by many echoes of Lucretius. But the Lucretian manner is too marked to be entirely serious, and there is some element of irony in the sceptical rejection of the feathered Hyperboreans and Scythian women (359–60: not in Herodotus!), when so many marvellous metamorphoses have been described without comment.

14. Cf. Heinze, Ovids elegische Erzählung, l.c. (n.13).

15. Cf. Oehler, R., Mythologische Exempla in der älteren griechischen Dichtung (1925) 45, 51, 53, 62Google Scholar, al.

16. o.c. (n.11) 240 ff.; cf. Bömer l.c. (p.64).

17. By West, David, G & R 21 (1974) 28Google Scholar.

18. Melanippe the Wise fr. 484N .; see Nauck's note for similar expressions. In Sat. 2.2 f. ‘nec meus hie sermo est, sed quae praecepit Ofellus | rusticus’, etc., Horace uses the device to suggest the homespun character of his homily.

19. I owe this example to Mr W.S.Barrett, though he is not accountable for my interpretation.

20. Bowra, C.M., Pindar (1964) 55Google Scholar.

21. . This is commonly rendered ‘let all war and battle be far from the gods’, i.e. it is impious to associate gods with war. No Greek of Pindar's day thought war unworthy of gods. The impiety is to fight without gods' help, as Ajax boasted that he could (Soph., Aj. 774–5Google Scholar), and a fortiori to fight against them. For the thought, cf. Eur. fr. 1025 , fr. 391 . For ‘without the help of, cf. Soph., Aj. 165, fr. 942Google Scholar; Eur., Alc. 368Google Scholar, Hel. 1274, fr. 580.4. (One might take ‘leave afar off all war and fighting with gods’, cf. Nem. 10.72 ; but the adverbial is strange, especially in this order.)

22. Cf. Dornseiff, F., Die archaische Mythenerzählung (1933) 34 fGoogle Scholar.

23. In what follows I am indebted to discussion with Miss Georgina Lloyd.

24. See Kakridis, J.T., Hermes 63 (1928) 417Google Scholar = Calder, W.M. and Stern, J., Pindaros und Bakchylides (1970) 162Google Scholar; cf. RE xvi.1 (1933) 1152–3Google Scholar.

25. The earliest mention of the Myrtilus story in literature is in Pherekydes (FGH 3 F 37a, b). It is alluded to in Soph., El. 504 ff.Google Scholar, and more fully in Eur., Or. 988 ffGoogle Scholar. (408); it is implied in Sophocles' Oenomaus (fr. 474 P), of unknown date, and was presumably the version in Euripides' Oenomaus (cf. Hyginus, , Fab. 84Google Scholar). There is no clear evidence for it in vase-painting before c. 400: cf. Séchan, L., Etudes sur la tragédie grecque (1926) 453 f., 464Google Scholar; Squarciapino, M. Floriani, ASAA 30–32, n.s. 14–16 (1952–1954 [1955]) 135 ff.Google Scholar; EAA v.115fGoogle Scholar. I see no good reason for accepting Robert, Carl's theory (Griechische Heldensage [1920] i. 208, 214)Google Scholar that the Myrtilus version was an old one originating in Lesbos, nor the elaborate account of its relation to the other version postulated by Kakridis (o.c. [n.24] 163 ff.).

26. See Ashmole, B. and Yalouris, N., The sculptures of the temple of Zeus at Olympia (1967) 16, 175Google Scholar. Both authors believe that the story of Myrtilus' treachery is implied (12–13). Ashmole thinks it compatible with the absence of Myrtilus in his own arrangement, though he admits the difficulty (16), but not with M.'s age in that of Yalouris, who however refers (175) to his ‘heroic nudity’ (cf. e.g. Stucchi, S. in ASAA vol. cit. [n.25] 78)Google Scholar. Robertson, C.M. (A history of Greek art [1975] i.277, 279)Google Scholar holds as I do that it is the Pindaric version depicted, but regards it as a ‘Pindaric purification’ of an earlier tale of treachery. That it cannot be due to Pindar himself is however shown by the winged horses on the chest of Cypselus. In any case, one would hardly expect a version invented by Pindar to become so widely known in a few years that it could be alluded to so obliquely on a monument.

26a. So Wilamowitz, , Pindaros (1922) 414 n.Google Scholar; Studniczka, F., Abh. Akad. Leipzig (1923) iv.20Google Scholar; cf. RE l.c. (n.24), ibid. 1158.

27. Cf. R.Oehler, o.c. (n.15) 62.

28. Fraser, P.M., Ptolemaic Alexandria (1973) ii.915 n.284Google Scholar.

29. E.g. by Mair, A.W., Callimachus (Loeb, 1921) 19–20Google Scholar, endorsed by Fraser.

30. Vahlen, J., Gesammelte philologische Schriften (1923) ii.417–18Google Scholar.

31. Cf. Wilamowitz, , Hellenistische Dichtung (1924) ii.10Google Scholar.

32. Meincke, W., Unters. zu den enkomiastischen Gedicht Theokrits (1965) 176Google Scholar; q.v. for a full statement of the arguments.

33. Euripides and the judgement of Paris, JHS Suppl. No. 11 (1965) 38n. The relative αἵ in 975–6 (codd.) implies that the goddesses went to Ida, but for a frivolous reason, so that Helen's version of the story cannot be true. Naber's εἰ implying that they did not go at all, requires a dubious ellipse (‘quod fecissent, si’, Murray). Scheidweiler, F. (Philologus 97 [1948] 334–5)CrossRefGoogle Scholar also holds that Hecuba is not denying the Judgement story as such, but takes her denial to apply only to Helen's claim that Aphrodite herself escorted the lovers. This does not help to explain 975–6. (My interpretation is endorsed by Schwinge, E.R. in Gnomon 39 [1967] 651Google Scholar.)

34. Coles, R.A., ‘A new Oxyrhynchus papyrus: the hypothesis of Euripides' Alexandras’, BICS Suppl. No.32 (1974)Google Scholar.

35. Cf. Hyginus, , Fab. 91Google Scholar. Does she also appear earlier? The hypothesis does not say so, but this is not conclusive, as Coles points out (o.c. 40). My suggestion (o.c. [n.23] 69) that she might have prophesied first in frenzied lyrics, then in a later scene in sober iambics, is however wrong. The hypothesis tells us that she was when she prophesied late in the play. Lyrics and trimeters therefore came in the same scene, as in the Cassandra scene in the Agamemnon, in Eur., Tro. 308 ff.Google Scholar, and in Ennius' Alexander (see Jocelyn, H.D., Ennius [1967] 204)Google Scholar. It is thus likely that fr. 7S came at the beginning of this scene, despite the difficulty of accommodating fr. 6 there (cf. Coles, l.c.; fr. 7 might not refer to Cassandra, as he observes [o.c 54], but the reference is highly probable, cf. esp. Tro. 341).

36. As Scheidweiler (l.c. [n.23]) had already maintained, against Snell (so also Strzelecki, W., De Senecae Agamemnone [1949] 11 n.19Google Scholar). Jocelyn (o.c. 217–20) argues, if I understand him rightly, (i) that the reference to the Judgement in Ennius fr. xvii ‘iudicavit inclitum iudicium’ is in the wrong chronological order; (ii) that ‘inclitum’, ‘famous’, is an anachronism tolerable in Ennius (who elsewhere makes his actors ‘speak like a contemporary scholiast’), but not in Euripides; (iii) Euripides would not call Helen an Erinys, though Ennius (ib.) can call her a ‘Furia’, a word of wider scope (=‘pestis’); so (iv) some of Cassandra's speech in Ennius, in particular ‘iudicavit inclitum iudicium’, had no counterpart in Euripides; hence (v) these words not only do not refute Snell's contention that the Judgement was not mentioned in Euripides' Alexandros, but confirm it. But (i) Cassandra is a seer, and seers' visions need not be in logical order, even if her other visions are so in Aeschylus and Seneca; (ii) being a seer, she knows that the Judgement is to be famous; (iii) that Aeschylus calls Helen an Erinys, at Ag. 749Google Scholar has indeed been disputed, but it is still in my view very likely; certainly Euripides could have done so, as he does at Or. 1389, and Virgil does at Aen. 2.573 (Tro. 457, where Cassandra refers to herself as ‘one of three Erinyes’ obviously does not make against this). There is thus no ground for supposing (iv) that ‘iudicavit inclitum iudicium’ had no counterpart in Euripides, and even if there were, the conclusion (v) in favour of Snell would not follow.

The question whether the Judgement has already taken place or is still to come, which is unfortunately not answered by the papyrus, does not affect the argument here. I argued, after Snell and others, that ‘iudicavit’ in Ennius, as understood by Cicero, refers to the future (o.c. [n.23] 68 n.3; for a full discussion see Jocelyn, o.c. 217–20). The ‘haughty behaviour’ of which the shepherds are said in the hypothesis to accuse Paris need not imply that he was conscious of his origins and destiny, but simply that he behaved in a manner appropriate to his royal birth, like Cyrus (Her.1.108 ff.; see Euripides and the judgement of Paris, 53).

37. Coles (o.c. 33 n.20) allows that Cassandra must have referred to the Judgement in the Alexandros, and sees no inconsistency in making Hecuba reject the story in the Troades, since ‘she has still no cause to believe even though Cassandra's foretelling of the Trojan War has proved correct’. But dramatically the War and its arche, the Judgement, go together.

38. This point is succinctly made (and so with Aesch., Ag. 750 ff.Google Scholar, discussed below) by MrsEasterling, P.E. (PCPS n.s.20 [1974] 43Google Scholar), apropos of Simonides fr.37. For the rhetorical device of controverting a gnome she cites Arist., Rhet. 1393^a19–33Google Scholar.

39. Dodds, E.R., The ancient concept of progress (1973) ch. i, esp. 5–12Google Scholar.

40. For the controversy see Fraenkel, , Agamemnon, p.349 fGoogle Scholar.

41. See Lloyd-Jones, H., The Justice of Zeus (1972) 58–69Google Scholar.

42. See CQ 25 (1975) 249–51Google Scholar, and cf. n.55 below.

43. For this systematic difference between choral and tragic lyric, see Dornseiff, F., Pindars Stil (1921) 85 ffGoogle Scholar. (‘das chorlyrische Ich’).

44. Cf. Parry, H., ‘The second stasimon of Euripides' Heracles', AJP 86 (1965) 363ffGoogle Scholar. When, as here, the chorus refer to their own activities as a chorus, singing and dancing, there is some oddity if these activities are not relevant to the drama, since this seems to stretch the chorus-convention so far as to break the dramatic illusion. Eur., Hcld. 892Google Scholar introduces a priamel with the climax ‘it is delightful to see the unexpected good fortune of friends’ (cf. Diggle, J., CQ 22 [1972] 243 f.CrossRefGoogle Scholar), so that is the standard conventional frame for a gnomic utterance, as in Ion l.c. But it is also clear that the first term of the priamel, ‘dancing is pleasant’, is chosen because it is the chorus speaking (cf. ad loc. in ‘Notes on Greek Tragedy, II’, to appear in JHS 97 [1977]Google Scholar). This is not undramatic, since they dance to celebrate a joyful occasion, so the illusion is not broken, There is no such dramatic occasion for dancing at Soph., OT 863 ff.Google Scholar, and Dodds has argued that 896 must have extra-dramatic reference (G & R 13 [1966] 46Google Scholar = o.c. [n.39] 75). But this too can be accommodated within the dramatic convention: dancing is a normal part of worship, so the phrase simply anticipates the other acts of worship mentioned in the following stanza (so D.M.Bain in an unpublished thesis on ‘Asides and related conventions in Greek tragedy’; cf. now his article ‘Audience address in Greek tragedy’, CQ 25 [1975] 16 n.2Google Scholar).

45. Zuntz, G., ‘Theology and irony in Euripides' Helena’, Entretiens Hardt, vi (Euripides, 1958) 217 ffGoogle Scholar.

46. The corruption is easy; cf. Ba. 1113, where the papyrus reading (for codd. ) is probably right (see Dodds' note), (Kirchhoff) does not give the necessary contrast.

47. Euripides, , Helena (1969) ii.24Google Scholar.

48. Poets' stories stand for falsehood because .

49. Cf. Page, D.L., Actors' interpolations in Greek tragedy (1934) 169–71Google Scholar. But Page does not regard the closing lines as in any way suspect.

50. E.g. (Elmsley), better ; not Scaliger's , which introduces a rare form of choriambic dimeter (see Page, , PCPS n.s.6. [1960] 52)Google Scholar.

51. Griechische Verskunst (1921) 560Google Scholar.

52. The substance of this paragraph I owe to Dr M.Schofield, who kindly elaborated in correspondence his valuable criticism of the original paper.

53. I take to refer to Either is possible, but it is one thing to say that truth is irrelevant to the deterrent effect of a story, quite another to suggest that the culprit should have remembered a story which is by implication false, even though it does illustrate a general truth. (Denniston, taking as antecedent, argues that it is the story of Aerope, a faithless wife like herself, which Clytemnestra should have remembered. But Aerope's punishment plays no part in the chorus' account, and Clytemnestra's crime here is not her adultery but the murder of her husband.)

54. Burnett, A.P., Catastrophe survived (1971) 175 ffGoogle Scholar.

55. Knox, B.M.W., CP 67 (1972) 276 fGoogle Scholar. Nor do I believe that we are meant to think of Heracles' apotheosis, as Burnett (o.c. 182) and Lloyd-Jones, (The justice of Zeus, 153–4)Google Scholar suppose.

56. When the paper was originally read at Oxford, this final section evoked various expressions of disbelief. I am not sure that I now believe it myself. I have let it stand, on the principle that the best way to see how far a given line of interpretation can be taken is to take it too far, and in the hope that my own expression of doubt, after the manner of Lucian (p.62 above), may at least encourage credence in what has gone before.

57. The main argument against my interpretation is that picks up Theseus' at 1325, so that Heracles is emphatically denying the truth of the stories, not just disapproving of them. But my whole purpose in this paper has been to show that emphatic denials of truth are not always what they seem, and serve to express not disbelief but some other attitude; and that one of the attitudes they serve to express is disapproval.

58. Hipp. 120; Ba. 1348. Cf. Lloyd-Jones, o.c. (n.53) 153, where these passages are given a rather different emphasis.

59. Xenophanes, DK 21 B 11.