¹ Lines 1369–71 of this messenger's report appear interpolated (Fraenkel, E., Zu den Phoenissen des Euripides, SBBA i [Munich, 1963] 64–5)Google Scholar, but, interpolated or not, do not affect my discussion.

² Cf. Eteocles' earlier argument (604–8) that by invading his fatherland Polyneices no longer has any claims upon its deities.

³ The Athena whom Eteocles invokes may have been Athena Onka, on whom see Schachter, A., Cults of Boeotia, BICS Suppl. xxxviii (1981) 130–1Google Scholar. Euripides may well have used the epithet χρυσάσπιδος (1372) to give his Athena a Theban colouring; its only other occurrence is in the opening lines of Pindar's Isthmian i, Μᾶτερ ἐμά, χρύσαπι Θήβα.

⁴ For recent studies and bibliography on Greek prayers, see Corlu, A., Recherches sur les mots relatif à I'idée de prière d'Homere aux tragiques (Paris 1966)Google Scholar; von Severus, E., ‘Gebet I’, in Reallexikon für Antike und Christentum viii 1134–52Google Scholar; H. S. Versnel, ‘Religious Mentality in Ancient Prayer’, 1–64 in Faith, hope, and worship, ed by Versnel, H. S. (Leiden 1981)CrossRefGoogle Scholar; and Burkert, W., Greek religion (Cambridge, Mass. 1985) 73–5Google Scholar.

⁵ For the reaction, often violent, to unanswered prayers in later antiquity, see Versnel (n. 4) 37-42.

⁶ On which see Corlu (n. 4) passim.

⁷ Such unanswered wishes involving, but not addressed to deities I exclude from consideration. Those interested may find examples in S. Aj. 185–6, Ph. 133–4, 314–16, and OC 1689–92. Such wishes in A. Suppl. (e.g. 1030–2, 1052–3, and 1062–73) are varied forms of unanswered direct prayers made throughout the play. On these see below, pp. 93–4.

⁸ The σε of 221 is surely Apollo (with Denniston, ad loc.), not Orestes (Langholf, p. 47).

⁹ In this play, though not elsewhere, Sophocles associates the epithet λύκειος with Lycia (203–8). Apollo Lykeios is a major cult figure in Argos, Athens, and much of mainland Greece (Winkler, , RE ii [1896]Google Scholar cols. 58–60; Kruse, , RE xiii (1927)Google Scholar cols 78–80; and Nilsson, , GGR i ³536–8)Google Scholar. The tragedians alone (A. Th. 145–6, S. OT 203–8 and 919) place him in Thebes. Sophocles' assimilation of him to Apollo Agyieus here and in El. 634–56, 1376–83 may reflect cult realities or may result from a literary adaptation. It is not improbable that a Theban cult of Apollo Lykeios/Agyieus is merely a Sophoclean fabrication. For this and other cults of Apollo in Thebes, see Schachter (n. 3) 77–88.

¹⁰ Apollo is ‘nearest’ in spatial terms because here he is Agyieus, of whom an image or altar regularly stood outside the door of a house. See Fraenkel on Ag. 1081 and Gomme and Sandbach on Men. Dysc. 659. Apollo may also be, in metaphoric terms, the ‘nearest of the gods’ in that through his oracles he is the deity most intimately involved in the fate of the Labdacidae. Cf. Dawe on OT 919.

¹¹ In Sophocles' OC Oedipus is not exiled from Thebes immediately (427–44), but there is no indication in the OT that there would be a significant delay. Given Apollo's oracle that the murderer of Laius must be banished (95–8), Oedipus' desire for it, and Creon's decision to leave it up to Apollo (1518), we must conclude Oedipus is to be exiled forthwith. Otherwise Apollo's oracle must contradict itself, which it never does in tragedy.

¹² Cf. 149–50. Compare the similar dilemma in Oedipus' threefold request to Teiresias (312–13): (1) ‘Save yourself and the city’; (2) ‘Save me’; and (3) ‘Ward off all the pollution of the one who has died.’ Unbeknownst to Oedipus, fulfilment of the first and last excludes the possibility of the fulfilment of the second.

¹³ Akestor, ‘Healer’ or ‘Mender’, is a typical literary epithet. It is more formal than a common adjective, but is not used in cult. Unlike a true divine epithet, it can serve as a personal name (e.g. Ar. V. 1221, on which see MacDowell ad loc.).

¹⁴ In much the same terms is answered Pylades' prayer (1242–5) to Zeus and Justice, for success for Orestes, Electra, and himself.

¹⁵ Euripides introduces Dionysus into this prayerlike request, but he is not the recipient of it nor is he made responsible for its fulfilment.

¹⁶ Einodia, ‘She in the Streets’, is polyvalent, suggesting as a proper name Enodia, the Hecate-like goddess of Pherai in Thessaly (Kraus, T., Hekate [Heidelberg 1960] 77–83)Google Scholar and, as an adjective, one important sphere of activity of this ‘daughter of Demeter’. Note ὅδωσον, 1051. Similar resonances may be found in S. Ant. 1199–1202 and frag. 492 (N) and E. Hel. 569–70. For the identification of Hecate with Kore in literature see Diggle on E. Phaëth. 268.

¹⁷ Cf. S. Ph. 1464–8.

¹⁸ On Hecate in fifth-century Athens, see Kraus, Hekate 84–94 and Nilsson, , GGR i ³722–5Google Scholar.

¹⁹ On the chthonic side of the Thessalian Enodia, see Kraus 77–83. In tragedy this negative side of Hecate prevails: S. Ant. 1199–1202; E. Med. 395–8, Hel. 569–70, Hipp. 141–2, Ion 1048–60, and, perhaps, Tr. 323–6. Her identifications with Artemis in tragedy each hang on the word ἑκάτη (adjective or noun?) and are at best allusive, at worst doubtful: A. Suppl. 674–77 and frag. 388(N); E. Ph. 109–11. For a recent discussion see H. Friis Johansen and E. W. Whittle (Aeschylus, The Suppliants [Copenhagen 1980]) on A. Suppl. 676.

²⁰ E.g. Wünsch, Defixionum tabellae, 101 and 107. From the fact that these tablets are addressed to either Hecate or Persephone, never to both, one has evidence that Hecate's identification with Persephone was extensive in practised religion.

²¹ One must concede to Knox (YCS xxvii [1977], 204Google Scholar n. 37= Word and action 219 n. 37) that nothing in Medea's words explicitly indicates the chthonic over the beneficent Hecate. But Medea's unusually strong devotion and personal domestic cult of her, the context of the statement, and what later results all point to Hecate's negative side.

²² Alternately in such a situation one may pray to Apollo (Agyieus) in his usual role of ‘averter of evils’. See, e.g. S. El. 634–59 (discussed below, pp. 89–90) and A. Pers. 176–214 and Broadhead on 203. Cf. Xen. Smp. 4-33.

²³ E.g. A. Pers. 176–214, Ch. 523–51 and 928–9, and S. El. 417–23.

²⁴ See Schmid, W., Geschichte der griechischen Literatur i 3 (Munich 1940) 720Google Scholar nn. 4 and 5; Schadewaldt, W., Monolog und Selbstgespräch (Berlin 1926) 101–4;Google Scholar and Langholf 69–76.

²⁵ Even the κατίδετ', ‘look down upon’, of 1252 indicates the prominence of Helios over Ge in this passage.

²⁶ On Helios in Corinth see Jessen, , RE vii (1912) cols. 58–93,Google Scholar esp. 64; Ed. Will, , Korinthiaka (Paris 1955) 209 and 233–5Google Scholar. It may be more than chance that Aphrodite, another major cult figure in Corinth, receives a hymnic ode in this play (627–41).

²⁷ On Helios as a god, see Appendix.

²⁸ Ar. Pax 409–13.

²⁹ Page's one-sided emphasis on the foreign and barbaric character of Medea (Medea xviii-xxi) is properly corrected by Knox, in YCS xxv (1977) 193–25Google Scholar = Word and action (Baltimore 1979) 295–322Google Scholar. Despite several conventional appeals to the gods Medea remains, however, a self-proclaimed devotee of Hecate and of Helios—both, if not totally unGreek, at least un-Athenian.

³⁰ Corlu, 110 and 114, understands ηὐξάμην simply as ‘hope’ here. Though not impossible, this interpretation weakens or ignores the force of 1020. For much the same prayer, but directed to the future and to θεοὶ ἐπόψιο, see 1040–2. Cf. 314–16.

³¹ Cf. S. Aj. 507–9 where Tecmessa imagines Ajax' mother back on Salamis praying ‘to the gods’ for her son's safe return. It would be impossible as well as inappropriate for the foreigner Tecmessa to specify who these Salaminian gods would be.

³² For οἱ θεοί in prayers, see A. Ag. 1, Ch. 212–15, 462; S. OT 879–82, Tr. 46–8, Ph. 1075–7; E. El. 415–16. 563–6, 590–5, Hel. 855–6, Ph. 586-7, Ion 422–4. In wishes, A. Th. 417–19, 422, Ch. 1063–4; S. Ph. 314–16, 528–9, 627, Aj. 949; E. Andr. 750–1, Hel. 1405–6.

³³ Cf. Hesiod fr. 148a (Merkelbach and West). See Weinreich, O., ‘Helios, Augen heilend’, in Ausgewählte Schriften, i (Amsterdam 1969) 7–12CrossRefGoogle Scholar and Jessen, cols. 60 and 85.

³⁴ Mikalson, , Athenian popular religion (Chapel Hill 1983) 99Google Scholar.

³⁵ Mikalson, 94–5.

³⁶ E.g. Pentheus in E. Ba. 255–62, 345–51; Creon in S. Ant. 1055, 1061–3, 1077–8; Tydeus in A. Sept. 377–83; Eurystheus in E. Heracl. 1027–40; Polymestor in E. Hec. 1280; Menelaus and Agamemnon in E. IA 518–20; and Eteocles in E. Ph. 766–73.

³⁷ For a general account of the Delphic oracle and its relations to Athens in this period, see Parke, H. W. and Wormell, D.E. W., The Delphic Oracle i (Oxford 1956) 188–200Google Scholar. Note also IG i² 76, 78, and 80. On the Athenians' unsuccessful use of oracles during the plague years of 430–426, see Mikalson, , 217–25 in Studies presented to Sterling Dow (Durham 1984)Google Scholar.

³⁸ Criticisms of soothsayers and oracles are not themselves termed impious in tragedy, but impious characters regularly indulge in them.

³⁹ For differing views on Jocasta's piety or impiety in regard to oracles, see Whitman, Sophocles 133–8; Knox, , Oedipus at Thebes (New Haven 1957) 42–7, 171–6;Google ScholarWinnington-Ingram, , Sophocles (Cambridge 1980) 179–84;CrossRefGoogle ScholarReinhardt, , Sophocles (Oxford 1979) 120;Google ScholarNock, , ‘Religious Attitudes of the Greeks’, Proc. Amer. Philos. Soc. lxxxv (1942) 474–5Google Scholar = Nock, Essays 538.

⁴⁰ Had Euripides constructed the scene to allow Orestes the choice of recipients of his prayer, they would have been very different. See, e.g., A. Ch. 1–19, 246–63, and 315–480, and S. El. 67–72, 1372–5.

⁴¹ On the Nymphs as cult figures see Nilsson, GGR i³ 244–53. Euripides suggests (El. 626) that his Nymphs are to be classed with those promoting successful human births. In Athens Nymphs were worshipped often together with Pan or Hermes, in caves on Mt. Pentelicon and at Phyle and Vari. The best known and excavated of these is the cave of Pan and the Nymphs at Vari (Weller, C. H., AJA vii [1903] 263–88)CrossRefGoogle Scholar. For the cave of Pan and the Nymphs at Phyle see Gomme and Sandbach on Men. Dysc. 1–49. Several dedications survive from these cults (IC i² 778–800 and ii² 4650–6 and 4826–40), many of them made by slaves and foreigners. These dedications reflect personal and individual, versus state and familial, concerns of the worshippers.

⁴² Neither Aeschylus (Ch. 554–84, 838–77) nor Sophocles (EL 1442–1507) has Orestes kill Aegisthus in such a religiously charged environment. The closest parallel in the complete tragedies to Euripides' version of the killing is the ambush and murder, organized and led by the same Orestes, of Neoptolemus in the temple and even on the altar of Delphic Apollo (E. Andr. 1069–1166). For other murders at sacrifices see Lloyd, M., Phoenix xl (1986) 16Google Scholar.

⁴³ E.g., Adams, S. M., CR xlix (1935) 121;Google ScholarConacher, D. J., Euripidean drama (Toronto 1967) 206–7;Google ScholarStoessl, F., Rh. M. xcix (1956) 64, 89;Google ScholarGrube, G. M. A., The drama of Euripides (London 1941) 308;Google ScholarZeitlin, F., TAPA ci (1970) 652, 660Google Scholar. But cf. Lloyd, M., Phoenix xl (1986) 15–16Google Scholar.

⁴⁴ As Apollo did Neoptolemus, if we take the mysterious voice in E. Andr. 1147–9 to be Apollo's.

⁴⁵ The thought that Aegisthus was in turn Orestes' victim and was sacrificed successfully, and that therefore Orestes' prayer was answered, will no doubt occur to some. But had Euripides intended this, he would have written lines 838–43 quite differently.

⁴⁶ Electra and Orestes are likewise ‘sailing very near the wind’ in religious terms when she entices Clytemnestra into her house to assist in ‘birth-sacrifices’ (653–63, 1141–4, 1122–38, on which see Zeitlin, F., TAPA ci [1970] 652)Google Scholar and he proposes maltreating Aegisthus' corpse and leaving it exposed to beasts and birds (894–9)

⁴⁷ Cf. Grube, 308.

⁴⁸ See Lloyd, M., Phoenix xl (1986) 15–16Google Scholar.

⁴⁹ The need for studying and understanding all prayers in tragedy is well exemplified by Kells', J. H. (Sophocles Electra [Cambridge 1973])Google Scholar comments on this prayer (634–59):

Since the prayer was thus a kind of bargain with the deity and nothing more, it could contain (according to popular belief) wicked and immoral, as well as innocent, proposals. The deity was supposed to tolerate such propositions: cf. Persius (Sat. 2.3 ff.). When a person prayed for something improper, he prayed silently or murmured sotto voce, so that the public might not hear him and become aware of his evil intent. For this reason, other people's ‘silent prayers’ were regarded with suspicion by the ancients: cf. Lucan v 105 f.

Nothing said here can be shown to be true for classical Greeks. Persius and Lucan are hardly adequate sources for understanding Clytemnestra's prayer.

⁵⁰ Such prayers of άποπομπή and ἐπιπομπή, though common and explicit in the Hellenistic period, are found rarely in tragedy. See Weinreich, O., ‘Primitiver Gebetsegoismus’, in Genethliakon Wilheltn Schmid (Stuttgart 1929) 169–99Google Scholar and Fraenkel on Ag. 1573.

⁵¹ On whom see Kruse, , RE xiii (1927) cols. 78–80Google Scholar.

⁵² See above, n. 22.

⁵³ Though differently motivated the substance of Clytemnestra's prayer is virtually identical to that of Aegisthus in E. El. 805–7.

⁵⁴ Clytemnestra,

ταῦτ̕, ὤ Λύκει̕ ̂Απολλον, ίλεως κλύων δός πᾶσιν ήμῖν ὥσπερ ἐξαιτούμεθα (655–56)

Electra,

ᾶναξ ̎Απολλον, ἵλεως αύτοῑν κλύε (1376) and νῦν δ; ῶ Λύκει̕ ̎Απολλον, έξ οἵων ἔχω αἰτῶ (1379–80)

⁵⁵ Cf. Orestes' prayer, 67–72. Note also the objects of Ischomachus' prayer in Xen Oec. 11.8 and Mikalson, Athenian popular religion 22–5.

⁵⁶ Cf. 1181.

⁵⁷ See 124 and 275–81. For the persistent concern with piety in this play, see 245–50, 464, 967–9, 1093–7.

⁵⁸ Even here the interpretation is uncertain. See Kannicht on Hel. 1447–50.

⁵⁹ S. Phil. 1019–21 and E. Tr. 1280.

⁶⁰ His sanctuary in Athens was the stoa of Zeus in the Agora, and he was credited with ‘saving’ the Athenians from slavery in the Persian wars. On this Athenian Soter see Farnell, , Cults i 60–1Google Scholar and Mikalson, , TAPA cxvi (1986) 90Google Scholar note 2.

⁶¹ TAPA cxvi (1986) 89–98Google Scholar.

⁶² On the ‘Iliadic’ character of this prayer, see below, p. 96.

⁶³ See, e.g. Barrett ad loc.

⁶⁴ For extended discussion of the problems, see Ritchie, W., The authenticity of the Rhesus of Euripides (Cambridge 1964)Google Scholar.

⁶⁵ See Mikalson, Athenian popular religion 65–6 and Burkert, Greek religion 179–81.

⁶⁶ For examples from the Hipp, see 1092,1401, 1406.

⁶⁷ Mikalson, 65–6. Of the dozens of examples in tragedy I give only those from the Hipp.: 241, 771, 832–3.

⁶⁸ Thymbraios, like Delios, is a localizing epithet, specifying Apollo of Thymbra, a city near Troy. See Kruse, , RE vi (1936) col. 697Google Scholar.

⁶⁹ Apollo's identification, in tragedy, with daylight and Helios may be as early as Aeschylus' Bassarai ([Eratosthenes] Catasterismoi 24, on which see West, M. L., BICS xxx [1983] 63–71)Google Scholar and is assured in E. Phaëthon 224–6 (Diggle). On the topic see Diggle ad loc.

⁷⁰ Apollo's role as protector and exegete for Orestes (A. Eum. passim) offers no basis for comparison.

⁷¹ For a similar situation involving both Apollo and Hermes, but properly worked out, see A. Eum. 89–93.

⁷² For implicit or explicit expressions of Zeus' superiority over Apollo, see A. Eum. passim, S. OT 898–910 and OC 1085–95.

⁷³ Cf. βαστάοῃ 827.

⁷⁴ In A. Ch. 622 one may assume that Nisos is already dead when Hermes ‘reaches’ or ‘overtakes’ (κιγχάνει) him.

⁷⁵ Zeus, A. Pers. 915–17 (cf. S. Aj. 387–91); Hades, S. Tr. 1040–3 and OC 1689–92; Thanatos, A. fr. 255 (N) and S. Ph. 797–8. For Thanatos see also Garland, R., The Greek way of death (Ithaca 1985) 56–9 and 155Google Scholar.

⁷⁶ On which see Stanford on Aj. 841–2.

⁷⁷ Elsewhere Helios usually reports to one god the activities of other gods, e.g. Od. viii 270–1 and Hymn to Demeter 62–87.

⁷⁸ Rosenmeyer, T. (The masks of tragedy [Austin 1963] 186)Google Scholar, without noting the anomalous character of the prayer, claims it is appropriate to Ajax' major concerns in the play. See also Winnington-Ingram, Sophocles p. 45.

⁷⁹ See above, pp. 85–6.

⁸⁰ If as E. Fraenkel, C. W. Macleod, and M. L. West argue (see West, , BICS xxv [1978] 113–14)Google Scholar, lines 854–8 are interpolated, we must remove this final prayer to Thanatos (854) from consideration.

⁸¹ Olbios is here a literary and not a cult epithet, despite scattered attestations in later cult. See Friis Johansen and Whittle on 524–6 and 526.

⁸² Whether the ‘polis, land, and shining water’ are recipients of the prayer or merely possessions of the gods invoked depends on which text one prefers. See Friis Johansen and Whittle on 23–9.

⁸³ On Zeus Soter see above, p. 90. He had, at least in later times, a sanctuary in Argos (Pausanias ii 20.6). On the triad of Olympian deities, heroes, and Zeus Soter, see Friis Johansen and Whittle on 26.

⁸⁴ For the likely presence of a chorus of Aegyptiads, and not just their herald, see Friis Johansen and Whittle on 825–902.

⁸⁵ Cf. 1045–6.

⁸⁶ On the impossibility of reconstructing the lost plays, and on errors of those who have attempted to do so, see Garvie, A. F., Aeschylus' Supplices (Cambridge 1969) 141–233Google Scholar and Friis Johansen and Whittle i 21–55.

⁸⁷ E.g. scholia to E. Or. 872 and Hec. 886, [Apollo-dorus] ii 1.5

⁸⁸ In 1059–61 one chorus claims that the Danaids' prayers are not μέτριον. See Friis Johansen and Whittle ad loc.

⁸⁹ ἔχων παλιντροπον ὄψιν έν λιταῖσιν

⁹⁰ E.g. 1–2, 104–10, 144–50, 359–60, 625–9, 811–16, 1030–2, See Fraenkel on Ag. 461 and 1270. In Homer, by contrast, the gods ‘hear’ successful prayers: Il. i 43 (cf. 380–1 and 453), 457; v 121; x 295; xv 378; xvi 527 (cf. 515–16); xxiii 771; xxiv 314; Od. iii 385; iv 767; vi 324–8; ix 536; xx 102.

⁹¹ On pollution in regard to Ant. 1016–20, see Parker, Miasma 33 and 44.

⁹² See Parker, Miasma 104–43.

⁹³ On which see Seaford, R., Euripides Cyclops (Oxford 1984) 1–59Google Scholar.

⁹⁴ For a quite different interpretation of line 355, see Seaford, ad loc.

⁹⁵ 606–7,

ἢ τὴν τύχην μέν δαίμον̕ ἡγεἴσθαι χρεών, τά δαιμόνων δέ τῆς τύχης έλάσσονα

⁹⁶ See above, p. 91.

⁹⁷ On occasion, however, the perceived lack of wisdom or justice of the gods causes a character to question their existence. See, e.g., E. El. 583–4, Suppl. 731–2, IA 1034–5, Tr. 356–8, and Frr. 286, 292, 577, and 900 (N).

⁹⁸ The views, though not many of their supporting arguments, of religion in Euripides which Wilhelm Nestle presented in Euripides, Der Dichter der griechischen Aufklärung (Stuttgart 1901) 51–152,Google Scholar esp. 115–16 remain pervasive. About prayer he concludes, in Vom Mythos zu Logos ² (Stuttgart 1940) 500,Google Scholar ‘Das Gebet [bei Euripides] ist meist fruchtlos’. A similar, equally mistaken view of prayer in Euripides may be found in Zuntz, G. (The political plays of Euripides [Manchester 1955]) 20:Google Scholar ‘There are gods: powerful impersonations of the uncontrollable forces that make mankind their sport. No prayers reach them.’

⁹⁹ With the same words Zeus gives delayed fulfilment to the prayer of Achaean and Trojan soldiers who ask Zeus that whichever side first violates the oath concerning the duel of Alexander and Menelaus die along with their children and that their wives be enslaved (iii 295–302). Although the specific terms of the oath as stated by Agamemnon (iii 276–91) were not violated (but cf. iii 456–60 and iv 13–16), clearly Homer treats the oath as broken by Pandarus' bowshot at Menelaus (iv 66–7, 71–2, 155–68, 269–71, vii 69). The eventual fulfilment of the soldiers' prayer may be seen in the death of Pandarus (v 290–6) and the sack of Troy (cf. iv 160–8).

¹⁰⁰ See Kirk, G. S., The Iliad: a commentary i (Cambridge 1985), 160CrossRefGoogle Scholar. In Od. ix 550–5 Zeus did not ‘care for’ (οὐχ ἐμπάʒετο) Odysseus' sacrifice of the ram after the Polyphemus' episode and devised destruction for Odysseus' ships and comrades.

¹⁰¹ For Vows in tragedy, see A. Th. 271–8, Ag. 933–4, 963–5; S. Tr. 237–45, 610–13; E. IT 15–24,fr. 82.22–6 (Austin). Of these only in A. Th. 271–8 is a request possibly not fulfilled. Eteocles promises, ‘if things turn out well (εὗ ξυντυχόντων) and the city is saved’, sacrifices, trophy-monuments, and spoils for the city-protecting gods, Dirce, and Ismenus. The city is saved, and that is the major import of the prayer, but, from Eteocles' point of view, things hardly could have turned out worse. On Eteocles' ἀσεβὴς διανοία which contributed to the unhappy result, see 829–31.

¹⁰² Helenus had recommended and Hector ordered that the women vow to Athena twelve cattle ‘if she pities the town and the wives and children of the Trojans and if she keep Diomedes from Ilium’ (Il. vi 86–101, 269–78)Google Scholar. Had the women not reformulated the vow to include the breaking of Diomedes' spear and his death, Athena could have accepted the bargain, in the short term at least, since Diomedes was soon driven back from Troy. That would, of course, have spoiled the pathos of Book vi which looks to the long term fortunes of Troy.

¹⁰³ ii 66–7, 211; vi 206–8; viii 575–6; ix 175–6, 267–71, 477–9; xiii 201–2; xiv 56–8, 283–4, 388–9; xvi 422–3; xvii 475, 483–7; xxi 25–9; xxii 39, 413–16; xxiii 64–7.

¹⁰⁴ Cf. the bad omens accompanying Aegisthus' sacrifice to the Nymphs in E. El. 803–43 (PP. 88–9 above).

¹⁰⁵ Artemis is invoked as the deity who, in epic but not in tragedy, gives non-violent death to women. See Penelope's same desire for death, but expressed as a wish, in xviii 205–8. Cf. Od. xi 172–3, 198–9, 324–5; Il. xix 59, xxi 483–4.

¹⁰⁶ In addition to references in note 103 above, see Od. i 29–47; iii 132–6; iv 806–7; v 108–9; xiii 213–14; xiv 83–4; xxiv 351–2.

¹⁰⁷ I should like to thank Hugh Lloyd-Jones and Richard Hunter who both offered valuable comments on an early version of this paper. I thank also Glen Bowersock and Christian Habicht for the opportunity to pursue this study in the pleasant environment of the Institute for Advanced Study.