I wish to express my thanks to Professors A. D. Fitton-Brown and F. H. Sandbach for their help in the preparation of this article.

¹ The readings of this passage and the one following are questionable, but most of the alternatives are equally acceptable for my purposes here.

² I translate ποταίνιος ‘novel, strange, unexpected, unprecedented’ in all its early occurrences, against LSJ and Italie, who attribute to it here and at Eum. 282 a purely temporal sense, ‘fresh, recent’. Typical instances are Soph. fr. 153 3–7, where the novel pleasures of a new, exciting sport, playing in the snow, are contrasted with the tedium when that sport has grown dull (as a comparison for the νόσημα of sexual, perhaps homosexual, pursuits); Ant. 849, where the tomb of Antigone is not newly built but a novel sort of tomb, one for a living person; Pi. O. x 60, where the poet, writing of the first victors at the Olympic games, asks who won this novel crown. The lexicographers and scholiasts give the word two senses, ‘unexpected; new’, but νέος is itself always ambiguous between a temporal sense and the sense ‘novel’ (in which I believe it to be used here by the lexicographers). Thus we have: (EM 685.8); (Hsch.); (sch. Aesch. Pr. 102); cf. sch. Aesch. Th. 239, sch. Pi. O. x 72 (=60 Snell), Phot. 445.13, Eust. Comm. 1106.23. The analogy to πρόσφατος has misled the commentators, as the scholiast to Soph. Ant. 859 (=848) realised:

I thus translate Ch. 1055–6, ‘Yes, for there is still strange (unprecedented, because a mother's) blood on your hands; from them the disturbance drips into your mind’; and Eum. 282–3, ‘For, although it had no precedent, the stain was driven away at a hearth of the god Phoebus by purifications with a slain pig’. The senses in which the passages are normally taken: ‘still fresh blood’ and ‘while it was fresh’, seem to me weak and unparalleled. This point is central to my interpretation of the Eumenides (§7 below).

³ The interpretation of the scholiast and other commentators, that Apollo here claims to have purified Orestes, is a natural one if and only if lines 282–3 have already drawn attention to such a purification. See also §2 and footnote 20 on the meaning of καθάρσιος.

⁴ Amandry, P., ‘Eschyle et la purification d'Oreste’ in RA xi (1938) 19 ff.Google Scholar; cf. ‘Eschyle et Eleusis’ in Mélanges Grégoire i (1949) 27–41.

⁵ This dilemma may be reduced to the following question. Does Aeschylus leave his audience in suspense for 280 lines after the tense conclusion of the Choephoroi before he tells them quite casually that Orestes was purified at Delphi?

Yes: Aeschylus, for artistic reasons, wishes to play down an unsuccessful purification by Apollo, and neither stages a purification nor wishes the audience to think of it again until indirectly in the scenes at Athens (a solution first suggested to me by T. B. L. Webster).

No: (I) The fact that Orestes had been purified (or was being purified) was obvious to the first audience through the staging of the play, although no mention of this happens to stand in our text.

1. Orestes has been purified before the Pythia enters the sanctuary, and appears visibly purified at line 64. In this case, the Pythia's description of Orestes at lines 40–5 is of the purified Orestes, which she either (i) realises or (ii) does not realise (Blass, Verrall and Groeneboom advance versions of this).

2. Orestes is in the process of being purified when the Pythia enters the sanctuary, but either (i) she does not see Apollo, or (ii) Apollo is assumed always to be present in the sanctuary (a possible compromise between (1) and (3)).

3. Apollo appears to Orestes after the Pythia leaves the sanctuary and, when the interior of the sanctuary appears to the audience, either (i) has just completed purifying him, or (ii) is still engaged in purifying him (Thomson, , Mazon, , and Séchan, L., Etudes sur la tragédie grecque dans ses rapports avec la céramique [Paris 1926] 97–99Google Scholar).

4. Apollo appears first to Orestes at line 64 and proceeds to purify him, an action not in the text but required by the staging of the play (Smyth, Weir, and Hauser, F. in Furtwangler, A. and Reichhold, K., Griechische Vasenmalerei ii [Munich 1908] 330–3Google Scholar).

(II) Our text must be corrupt.

1. The scene with the Pythia is an intrusion into the prologue, which was originally the scene between Orestes and Apollo containing the purification. The innovation in the prologue was possible only after the introduction of the ekkyklema.

2. The scene which the Pythia describes at lines 40–5 is in fact the purification, as she realises. The description of Apollo has been deleted by an editor who could not understand why the Pythia should find the purification so frightening. This explains the clumsy sentence.

3. The description of the purification has fallen out elsewhere, perhaps between lines 63 and 64.

The Wrong Question. The argument of this paper will suggest that the audience's expectation was satisfied by the instructions at lines 74–83, and that they did not expect a purification ritual at Delphi. In that case, either (1) lines 282–3 are an interpolation, or (2) they require special explanation (cf. §7 below).

⁶ Ginouvès, R., in the opening remarks of his chapter, ‘Le bain et la purification du meurtre’, in Balaneutike: Recherches sur le bain dans l'antiquité grecque (Paris 1962) 319–25.Google Scholar Some confusion results from a casual use of the words from which ‘cathartic’ comes. The Greek words for wash come from the root λου-, but καθαρ- is often used synonymously, particularly for washing off any substance that implies defilement (μίασμα), such as excrement, traces of the sexual act, menstruation, blood from a dead man. To what extent καθαρ- in any instance implies purification, rather than mere washing, depends on how deeply the μίασμα is regarded as having religious significance. Whether the words were ever used of an entirely spiritual purification before Plato's εἶδος καθάρσεως (Sophist 226d–31e) is doubtful. See Moulinier, L., Le pur et l'impur dans la pensée des Grecs (Paris 1952) 149–68Google Scholar; ‘καθαρμός’ in Pauly-Wissowa; Eitrem, S., Beiträge zur griechischen Religionsgeschichte ii: Kathartisches und Rituelles (Kristiania 1917).Google Scholar

⁷ Gernet, L. et Boulanger, A., La genie grec dans la religion (Paris 1932) 176.Google Scholar Apolline religion was much concerned with questions of purity, but it is not clear that its solution was always purification (cf. §3 below).

⁸ Defradas, J., Les thèmes de la propagande delphique (Paris 1954)Google Scholar; Delcourt, Marie, L'oracle de Delphes (Paris 1955).Google Scholar For Amandry, 's views see La mantique apollinienne à Delphes (Paris 1950).Google Scholar

⁹ Farnell, L. R., Cults of the Greek States (5 vols., Oxford 1896–1909).Google ScholarNilsson, M. P., Geschichte der griechischen Religion (=GGR) i (MüllersHandbuch v 2, i, 2nd edn., 1955).Google Scholar

¹⁰ Cults iv 411–12 n. 222; cf. the discussion 295–306.

¹¹ Farnell also lists as Apolline purifications those of Heracles at Amyclae and the daughters of Proteus (under instruction of Melampus) at Sicyon (cf. the Melampus krater). Delcourt adds the tale of Alcmaeon to this list, and Ginouvès mentions others not related to Apollo (op. cit. 320).

¹² See Allen's O.C.T. of Homer v 105. 28 ff.; cf. EGF i 33 Kinkel.

¹³ Cults iv 300–5, cf. 428–9 n. 273. Amandry (129–30) has attempted to interpret the shaking of the laurel as an actual sign of purification, but this seems to have indicated no more than that the god was present in his sanctuary. Cf. Ginouvès passim. However, no treatment of purification rituals has yet superseded Rohde, E., Psyche 294–7, 588–90.Google Scholar The latter denies that such rituals were a fresh step in the history of Greek ethics and stresses their essentially primitive character. ‘The purification is effected by religious processes directed to the external removal of the evil thing; it may be washed off (as by water from a running spring or from the sea), it may be violently effaced and obliterated (as by fire or even smoke alone), it may be absorbed (by wool, fleece of animals, eggs), etc. It must be something hostile and dangerous to men that is thus removed' (295). He has illuminating comments on the practices of the καθαρταί themselves.

¹⁴ Ginouvès also associates purification with pig's blood with Delphi itself, arguing (321–2) against the conclusions of Amandry (cf. n. 4) and Defradas (188–90) that such purification was unknown there.

¹⁵ Nilsson claims that the best attempt is Persson's, a book rightly castigated by JaCoby throughout his Atthis (236 n. 1, 265 nn. 172, 174, etc.). As Jacoby understood (e.g. Atthis 256 n. 102, 266 n. 174), there is no subject in Delphic ‘religion’.

¹⁶ Parke, H. W. and Wormell, D. E. W., The Delphic Oracle, 2 vols. (Oxford 1956).Google Scholar References to Nos. in Parke and Wormell are to the numbered responses in vol. ii.

¹⁷ Plut. 388f, from Ruiperez, M. S. (Emerita xxi [1953] 14 ff.)Google Scholar has used Hesychius to postulate a root ^*bheig^u ‘to purify’, but this seems untenable, as Nilsson sees (559). It is equally difficult to postulate a root for Λοξίας, probably with the name-formantsias, hence ^*lok, ^*lokh, ^*lok^u, etc. It is conceivable that this is the o-grade of ^*lekh, in a more primitive sense of ‘laying low’, appropriate to the god whose arrows bring sudden death.

¹⁸ Plut. 418b (de defectu orac. 15); cf. Aelian, VH iii 1.Google Scholar See the discussion in Nilsson, , Gr. Feste 151 ff.Google Scholar

¹⁹ For a full discussion on these preliminary rites, see Ginouvès 327–44, but on the unimportance of the preliminary καθαρμοί see AP xiv 71 and 74 quoted below on p. 44.

²⁰ It is to be observed that this word is found in Greek before κάθαρσις (first Herodotus unless in Arctinus) from which it is apparently derived. I have no doubt that Aeschylus associated the word closely with καθαρμός, καθαρός and καθαίρω in the sense of ‘having to do with purification’, but it is possible that he also saw in it the opposite of ἀνάρσιος, used by Homer in the sense of ‘hostile, implacable’. Aeschylus uses ἀνάρσιος in this sense at Ag. 511 of Apollo, and it may be that he wishes to underline the contrast of Apollo ἀνάρσιος towards Agamemnon and the Greeks at Troy and καθάρσιος towards Orestes (Eum. 578). Herodotus uses ἀνάρσιος regularly in the sense of ‘strange and terrible’ (i 114, iii 10, v 89–90, ix 37, 110), and LSJ may be correct in giving the derivation as α- privative + ἄρσιος. However, the root present here, and in καθαρός and cognates, may also be the simple one from which αἱρέω is a secondary formation. Neither of these possibilities argues against a popular ἀνα-/κατα- opposition, for the root of ‘fitting’ is also found aspirated in such words as ἅρμα, ἁρμονία (Aeschylus may well feel a connection between ἁρμονία and καθαρμός). Clearly, if Aeschylus conceives καθάρσιος as an opposite to ἀνάρσιος, its connection with ritual purification at his time becomes even less.

²¹ This analogy requires more documentation than is relevant here. The case of Oedipus, viewed in the light of conventional Athenian religion, is an interesting one. Had he the religious authority to appoint himself καθαρτής of Thebes, or should he have persevered with his search for a divinely-inspired λύσις κακῶν? Would Teiresias have found a less drastic solution to Oedipus's involuntary sins?

²² If Schoell, (Hermes vi [1872] 14 fr.)Google Scholar, Jacoby, (Atthis 8, 237 n. 3)Google Scholar and others are right in reconstructing on IG ² i 77, an inscription of the 430's, it would be evidence that the exegetes appointed by Apollo, presumably the Pythochrestoi, were active in Athens at least soon after the Oresteia.

²³ MacDowell, D. M., Athenian Homicide Law in the Age of the Orators (Manchester 1963) 70–81.Google Scholar See also Hewitt, J. W., ‘The Necessity of Ritual Purification after Justifiable Homicide’ in TAPA xli (1910) 99–113Google Scholar; and the handbooks: Bonner, R. J. and Smith, Gertrude, The Administration of Justice from Homer to Aristotle (Chicago 1930–1938)Google Scholar; Jones, J. W., The Law and Legal Theory of the Greeks (Oxford 1956).Google Scholar

²⁴ MacDowell has drawn up a long list of the types of homicide where a plea could be entered that the deed was κατὰ τοὺς νόμονς. The first question to be determined for such a plea was naturally whether the killer's actions and intentions fell within the terms of the applicable laws. These laws came from Draco, who, although he prevented all killers from taking part in religious observances (unless purified, I presume), ‘yet did not remove the prescription of justice, but laid down on what conditions it was permitted to kill and defined that if one killed in this way he was pure’ (Dem. xx 158). This is the category of homicide where law states there is no crime and religion (e.g. the Delphic oracle) agrees there is no miasma. From this must be distinguished justifiable homicide (φόνος δίκαιος), as MacDowell has shown from the Athenian law which told citizens ‘to kill neither justly nor unjustly’ ( quoted by Antiphon, 3b. 9, 3c. 7, 4b. 3, 4d. 8; cf. Maidment, K. J., Minor Attic Orators i 46Google Scholar). Where men have no legal sanction for killing they should not take the law into their own hands, even with justification. Yet they do and are sometimes acquitted through justification. This is Orestes' case. He acts in an unprecedented way, and despite the Delphic command receives miasma from his act. This must be washed out before he can be tried at the Areopagus on a plea of justifiable murder. Where the Delphinion judged a man with a plea of murder κατὰ τοὺς νόμους in such a case, with visible signs of miasma, it is plausible to suggest that they would prescribe exclusion from Athenian religious practices until the stain was washed out and then a trial before the Areopagus on the question of φόνος δίκαιος.

²⁵ Dem. xxiii 67, trans. MacDowell, op. cit. 42.

²⁶ On ἐφέσις in Attic law see Sealey, R., ‘Ephialtes’ in CP lix (1964) 11–22 and nn. 12–13Google Scholar; cf. Ruschenbusch, E., ‘Φόνος: zum Recht Drakons und seiner Bedeutung für das Werden des athenischen Staates’ in Historia ix (1960) 129–54Google Scholar (cf. Zeitschrift der Savigny-Stiftung für Rechtsgeschichte lxxviii [1961] Röm. Abt. 386–90). Sealey writes (21), ‘Ephesis as voluntary reference of a case by a first judicial authority to a second, when the first has not given a verdict, is attested by Dem. xxxiv 21’.

²⁷ Thus Porphyrius assumes that ‘even homicides condoned by law receive the customary purifications. … Whence the first who became aware of it not only set up punishments but also hung other irrational fear over people, ordaining that those who killed a man in any way at all were not pure unless they had recourse to purifications’ (de Abst. i 9). He is wrong insofar as some homicide did not need purification, but correct in pointing to irrational feelings of guilt or horror as the reasons why men needed purification for homicide. These feelings may need to be blotted out either when there is no legal guilt or before a fair ‘judgment as to justice’ (δίκη) can be made in the courts.

²⁸ The principal discussions are in Jacoby, F., Atthis (Oxford 1949)Google Scholar and Oliver, J. H., The Athenian Expounders of the Sacred and Ancestral Law (Baltimore 1950).Google Scholar See also Philippi, A., Der Areopag und die Epheten (Berlin 1874)Google Scholar, still valuable; Ehrmann, P., De juris sacri interpretibus atticis (Giessen 1908)Google Scholar; Boethius, A., Die Pythaϊs: Studien zur Geschichte der Verbindungen zwischen Athen und Delphi (Diss. Uppsala 1918).Google Scholar

²⁹ Jacoby discusses and justifies this punctuation on p. 250 n. 59, against that suggested by Toepffer and Ehrmann. He suggests that Timaeus owed the reference to a Hellenistic grammarian.

³⁰ And they would think, in my opinion, of the Alcmaeonids, whose position must be at least at the back of Aeschylus's mind. How thorny such a question remains may be judged from the reviews of Podlecki, A. J., The Political Background of Aeschylean Tragedy (Ann Arbor 1966).Google Scholar Most reviewers feel sceptical that the political and social questions of the day influenced the tragic stage, for ‘great poetry is not written that way’ (CR xviii [1968] 30). Yet the evidence of this century is that myth is most often used as a medium precisely for this purpose, to seek for a comprehensible structure in contemporary intellectual, social or political problems through the clarifying analogy of a chosen myth.

³¹ Reverdin, O., La religion dans la cité platonicienne (Paris 1945) 96–9Google Scholar; cf. Platon, , Lois ix, ed. Gernet, L. (Paris 1917)Google Scholar; Boyancé, P., ‘Platon et les cathartes orphiques’ in REG lv (1942) 217–35.CrossRefGoogle Scholar

³² The standard typologies of South Italian vases are abbreviated as follows: APS = Cambitoglou, A. and Trendall, A. D., Apulian Red-figured Vase-painters of the Plain Style, Monograph X of the American Institute of Archaeology (New York 1962)Google Scholar; LCS = Trendall, A. D., The Red-figured Vases of Lucania, Campania and Sicily (Oxford 1967)Google Scholar; PP = id., Paestan Pottery (London 1936; with addenda in BSR xx [1952] 1–53, xxvii [1959] 1–37); VIE = id., Vasi antichi dipinti del Vaticano: Vasi italioti ed etruschi a figure rosse [Vatican City 1953–5]. I owe much to Trendall for the opportunity, while he was Master of University House, Canberra, to use his library in this field and his files of Apulian vases of the ornate style, and for his advice and caution in discussing this topic.

I am grateful to the appropriate authorities for permission to illustrate this section with vases from the collections of the Staatliche Museen zu Berlin; the Fogg Art Museum (David M. Robinson Bequest), Harvard University, Cambridge, Mass.; the Ermitage, Leningrad; the British Museum, London; the Finarte Institute, Milan; the Paestum Museum; and the Louvre, Paris.

³³ For published lists see Brommer, F., Vasenlisten Zur griechischen Heldensagen (2nd ed., Marburg 1960) 324Google Scholar; Webster, T. B. L., Monuments Illustrating Tragedy and Satyr-play (2nd ed., BICS Suppl. xx, London 1967 = MITS ²), Appendix 138–141Google Scholar; cf. Vermeule, Emily, ‘The Boston Oresteia Krater’, AJA lxx (1966) 1–22CrossRefGoogle Scholar (with comments by Elizabeth G. Pemberton, ibid. 377–8; Clairmont, C. W., Antike Kunst ix [1966] 125–7Google Scholar; Dyer, R. R., AJA lxxi [1967] 175–6CrossRefGoogle Scholar; J. G. Griffiths, ibid. 176–7). By my count, a total of 100 vases show Orestes in one myth or other.

³⁴ Besides, MITS (1st edn., 1962; 2nd edn., 1967)Google Scholar, see ‘South Italian Vases and Attic Drama’ in CQ xlii (1948) 15–27 (cf. Pickard-Cambridge, A. W., CQ xliii [1949] 57CrossRefGoogle Scholar); ‘Fourth Century Tragedy and the Poetics’ in Hermes lxxxii [1954] 294–308; Greek Theatre Production (=GTP; London 1956); cf. his revision of Pickard-Cambridge, , Dithyramb, Tragedy and Comedy (Oxford 1962).Google Scholar Among earlier studies of this relationship should be recorded Séchan, L., Etudes sur la tragédie grecque dans ses rapports avec la céramique (Paris 1926)Google Scholar; Bulle, H., ‘Von griechischen Schauspielern und Vasenmalerei’, in Festschrift für James Loeb (Munich 1930)Google Scholar; and the work of Pickard-Cambridge.

³⁵ A sensitive account of the painters' freedom will be found in Page, D. L.'s introduction to his edition of Euripides' Medea (Oxford 1938; 1952²) lvii–lxviii.Google Scholar

³⁶ See Pickard-Cambridge, , The Dramatic Festivals of Athens (Oxford 1953).Google Scholar

³⁷ See Philippart, , ‘L'iconographie de l'Iphigénie en Tauride d'Euripide’ in RBPh iv (1925) 5 ff.CrossRefGoogle Scholar; Séchan, op. cit. 379–88; Löwy, E., ‘Iphigenie in Taurien’ in JDAI xliv (1929) 86 ff.Google Scholar

³⁸ ‘The Oresteia of Aeschylus as illustrated by Greek vase-painting’ in Harvard Studies xxi (1910) 111 ff.

³⁹ Op. cit. 86–101.

⁴⁰ ‘The Choephoroi Painter’ in Studies in Honor of David Moore Robinson (St. Louis 1953) ii 114–26; cf. LCS 118–26.

⁴¹ ‘Oreste’ in Enciclopedia dell' Arte Antica Classica ed Orientale v 741–3.

⁴² Pickard-Cambridge, , The Theatre of Dionysus (Oxford 1946) 75 ff.Google Scholar; Webster, , GTP 14, 102 ff.Google Scholar

⁴³ By using the iconographic type of the Eumenides, the artist throws our minds forward to Athena's role in the later stages of the play. On the distinction between the two iconographie types, Erinyes and Eumenides, see Roscher, 's Lexicon i 1332 ff.Google Scholar

⁴⁴ Jebb suggests in the introduction to his edition of Sophocles' Electra (p. li) that the vase suits Eur. Or. 1345 ff., but he may overestimate the need to find a staged scene to equate to the painting.

⁴⁵ Attic vases: BM 1923. 10–16.1, column-krater, by the Orestes Ptr.; Syracuse 41621, bell-krater, by the Hephaestus Ptr.; Louvre K 343, column-krater, by the Duomo Ptr. Apulian: Naples 3249, volute-krater, by the Black Fury Ptr.; Berlin F 3256, volute-krater, from the Iliupersis circle; Leningrad inv. 1714 (St. 523), volute-krater, by the Lycurgus Ptr. Campanian: Milan (private coll.), bell-krater, by the Parrish Ptr. (LCS 251, no. 154).

⁴⁶ Attic vases: Italian mkt. (Beazley, , ARV ²1097Google Scholar, 21 bis), column-krater, by the Naples Ptr.; Berlin 2380, hydria (shoulder), by a late mannerist ptr. Apulian: Copenhagen 332 (217), calyx-krater, from the Lycurgus group. Lucanian: Harvard, Fogg 1960.367 (ex de Cambacères coll.), Nestoris II, by the Choephoroi Ptr. (= plate III 4, n. 47); Vatican W1, volute-krater, by the Primato Ptr. (the hand is stretched towards Orestes, while Athena occupies the chief protective position—clearly in an attempt to portray their joint role). Campanian: Berlin inv. 3164, hydria, by the Ixion Ptr. (LCS 340, no. 803); lost vase (Tischbein ii 16), from the Ixion group (LCS 347, no. 879). In other scenes Apollo either is not present or sits quietly by, his mighty presence in the sanctuary serving, seen or unseen, to balance the Erinyes. For the sanctuary is a holy place, whether the god appears in miraculous epiphany or iconographie convention before his suppliant or not, whether he is seen to extend a gesture of purification or not.

⁴⁷ This vase is fully discussed by Bock, M., AA 1935, 493–511.Google Scholar figs. 1–2; cf. LCS 125, no. 644; pl. 62.4.