¹ I am most grateful to Mrs P. E. Easterling and Dr A. F. Garvie for generous help.

² The Stagecraft of Aeschylus (Oxford 1977) 369–74Google Scholar, with references to earlier discussions. (References to Taplin are to this book unless otherwise stated.) Taplin's view has now been accepted by Melchinger, S., Die Welt als Tragödie i (Munich 1979) 349Google Scholar.

³ The treatment of the chorus as a major party to the action, which we find in Supp. and Eum., is likely to be a bold experiment rather than a survival from archaic tragedy; see Garvie, A. F., Aeschylus' Supplices: Play and Trilogy (Cambridge 1969) 106–20Google Scholar. And certainly the non-human and unsympathetic character of the chorus of Eum. rules out most of the normal functions of a tragic chorus. There may always have been precedents in lost plays, but few known titles offer any scope for them; that the Phorkides (three in number by all accounts) could have formed the chorus of the play named after them (Garvie 114 n. 8) seems most improbable.

⁴ Il. ii 20, xxiii 68 (Patroclus)Google Scholar; Od. iv 803Google Scholar, vi 21; cf. Il. x 496Google Scholar. Similarly ἐπιστῆναι is regularly used of dream figures in Herodotus and elsewhere; see Dodds, E. R., The Greeks and the Irrational (Berkeley/Los Angeles 1951) 105Google Scholar.

⁵ See Kessels, A. H. M., Studies on the Dream in Greek Literature (Utrecht 1978) 162Google Scholar.

⁶ The difficulty comes not in 104 but in 105, a line which is hardly translatable as it stands (even with ἀπρόσκοπος) and seems to imply that the chorus are βροτοί. But to delete 105 alone, with Prien, leaves 104 looking uncomfortably isolated. I am inclined to retain both lines, reading Hermann's φρενῶν (ἡ τῆς φρενὸς μοῖρα schol. M) for βροτῶν, despite the repetition, and regarding 105 as ‘foil antithesis’ to the main point in 104.

⁷ If 104 f. are retained, then καρδίᾳ σέθεν must be retained also, to motivate these lines. Only if both 104 and 105 are deleted should Hermann's ὁρᾶτε . . . καρδίας ὅθεν be read (because καρδίαι σέθεν is barely intelligible without the explanation in 104, and because of the plural verb in 106).

⁸ So apparently Hermann, G., Opuscula vi (Leipzig 1835) 2.163Google Scholar; Hickman, R. M., Ghostly Etiquette on the Classical Stage (Cedar Rapids 1938) 32Google Scholar.

⁹ Hermann loc. cit. (n. 8) perhaps did better in claiming that ‘You see, they're asleep’ can naturally be said to someone who has just seen them. (The word νῦν cannot be pressed hard, since it may simply serve to contrast the particular instance in 67 with the general statement in 64–6.) But one would still expect the audience to be able to see what Apollo is pointing out.

¹⁰ I agree with Taplin, on the other hand, that no weight can be given to the schol. on 64 or to other secondary evidence; also that μαντικῶν μυχῶν at 180 must refer to the whole precinct and not to the interior of the temple. Nor can the Italian vase paintings of Orestes and the sleeping Furies provide even slight evidence that the scene was shown on stage, seeing that some of these vases also depict Orestes' purification.

¹¹ For a curtain see Pickard-Cambridge, A. W., The Theatre of Dionysus in Athens (Oxford 1946) 128–30Google Scholar. He reasonably concludes that it is not required in any of the scenes he discusses, but not that it is impossible.

¹² I am assuming that Sophocles and Euripides did use the ekkyklema to reveal interior scenes, though even this is by no means undisputed; to Taplin's references (442 f.) add (in favour) Melchinger, S., Das Theater der Tragödie (Munich 1974) 192–4Google Scholar; Blume, H.-D., Einführung in das antike Theaterwesen (Darmstadt 1978) 67–9Google Scholar.

¹³ He accepts the device in Sophocles and Euripides but questions it in Aeschylus (325–7, 357, 442 f.). The only solid reason that he gives for these doubts is that the revelations in Ag. and Cho. are not signalled by any words about the doors being opened, such as tend to be found when the ekkyklema is used in later tragedy. But it would not be uncharacteristic of Aeschylus to be less explicit than Sophocles and Euripides in such a matter; cf. Taplin 280: ‘Perhaps he [Aeschylus] thought it unnecessary to clarify entrances and exits from the new skene door …; though in that case he was not followed in this by Sophocles and Euripides.’ The words about doors being opened are surely not required to explain the ekkyklema (it would have been understood well enough if it had been used even once before) but are a matter of convention, which might have taken time to become established. Anyway, why should the use of stagehands to carry out the bodies (which is what Taplin favours) require any less explanation?

¹⁴ Greek Theatre Production ² (London 1970) 9Google Scholar.

¹⁵ The chairs (line 47) are a curious detail, and one which the vase paintings of this scene understandably omit. It is tempting to explain them in terms of the staging of the tableau, as ensuring that the masks of the sleeping Furies are clearly visible, or even as saving space on the ekkyklema, since a body on a chair takes up less room than one lying on the ground.

¹⁶ If a seated man takes up, say, 15ʺ × 24ʺ, then a 10ʹ × 6ʹ platform allows room for 3 rows of 8. If there were in fact 15 choreutae, this leaves 22½ sq. ft. to spare for Orestes and the Omphalos and for a less regimented seating plan.

¹⁷ I am leaving PV out of account here, since, even if the play is Aeschylean (and at the present stage of the debate we must assume that it is not), its staging is far too problematic to be used as a parallel for anything.

¹⁸ Another reason is perhaps their desire to provide an opportunity for Apollo's purification of Orestes; see below. There is also Apollo's action in putting the Furies to sleep, about which Aeschylus is oddly reticent (though καὶ νῦν work), but I suppose the god could achieve this from on high, just as madness is caused from on high by Athena in Sophocles' Aj. (see n. 24 below) and Lyssa in Euripides' HF.

¹⁹ But Blass's commentary still makes Apollo and Hermes appear within the temple. I cannot discover that anyone has made use of Burges's transposition in the way that I propose.

²⁰ The transposition leaves 88 looking obviously out of place (not that it is really any more appropriate as a reply to 87). The line doubtless belongs before 84 (Dawe), i.e. one line earlier; surely not between 87 and 64 (Maas).

²¹ Or more than slightly, if, as is possible, Orestes' prayer originally consisted of more than the three lines we now have. Against this it can be argued that Apollo's entrance should follow soon after the Pythia's virtual invocation of him at 60–3. In any case I do not believe Taplin's ‘wild speculation’ (383 n. 3) that some lines of Apollo's have dropped out here, since 64 makes such a good response to Orestes' prayer (note also the echo of Cho. 269, which begins a speech).

²² Cf. Smyth's, H. W. stage direction (Loeb, Aeschylus ii 277Google Scholar): ‘The interior of the temple is disclosed. Enter, from the inner sanctuary, Apollo.…’ But, if Smyth means that there was actually a separate entrance within the skene, the audience could hardly have seen this properly.

²³ I see that this is also the view of Melchinger (n. 2) 115 (though there is much else in Melchinger's account of this scene that I cannot accept). Schol. vet. 64b, having talked of the epiphany of Apollo, refers to the revelation of the inside of the temple as a δευτέρα φαντασία; if this means that the scholiast placed Apollo on the mechane (Taplin 370) or the theologeion, he came closer to my view than most modern scholars.

²⁴ See Calder, W. M., CPh lx (1965) 114–16Google Scholar. Taplin, , indeed, thinks otherwise (Greek Tragedy in Action [London 1978] 185 n. 12CrossRefGoogle Scholar; cf. Stagecraft 116 n. 1, 366 n. 1), but lines 14–17 must, I think, imply that Athena is invisible to Odysseus (Buxton, R. G. A., JHS c [1980] 22 n. 2CrossRefGoogle Scholar), and, if this is so, then the parallel of Eur., Hipp. 1391 fGoogle Scholar. (where Artemis, presumably on the roof, is invisible to Hippolytus) suggests that both scenes were staged in the same way.

²⁵ 405 is doubtless spurious (Paley). Then, although 403–4 are probably meant to be more impressive than precise, it hardly seems necessary to specify that Athena has no wings unless some notion of flying is present; cf. ἀπτέροις ποτήμασιν at 250. I am thus unconvinced Taplin's vision (390) of the goddess skimming across land and sea while whirring her aegis, like a sort of hovercraft. For deities imagined as flying before they appear on the rooftop cf. Eur., Andr. 1226 ff.Google Scholar, El. 1233 ff.

²⁶ If the gods are on the roof, Hermes cannot exit at Orestes' side at 93, but this is hardly a disadvantage, seeing that he will not re-enter with him at Athens. 89 ff. need only mean that he is to direct Orestes' steps from on high.

²⁷ ‘The Evidence for Apolline Purification Rituals at Delphi and Athens’, JHS lxxxix (1969) 38–56Google Scholar.

²⁸ Verrall's notion that the blood at Eum. 41 f. is that of sacrificed animals is actually accepted in the commentaries of Blass, Thomson, Groeneboom and even Lloyd-Jones. But, quite apart from the fact that any audience will assume the blood to be that which they have seen on Orestes' hands throughout the last scene of Cho., any attempt (short of extensive and arbitrary textual alteration) to argue that Orestes is not polluted when the Pythia sees him is rendered futile by θεομυσῆ and προστρόπαιον at 40 f. Equally out of the question, though neglected by Dyer, is Thomson's belief that by the purification is completed off stage during the Ghost Scene; Orestes cannot hang about at Delphi after 74–93.

²⁹ I have some sympathy with the suggestion of Dirksen, H. J., Die aischyleische Gestalt des Orest (Nuremberg 1965) 9 n. 15Google Scholar, that Apollo's presence and the power of his words form a substitute for an actual ceremony. But this will hardly account for the quite specific χοιροκτόνοις at 283 (cf. Dyer, , Gnomon xxxix [1967] 189Google Scholar).

³⁰ Most edd. emend καθαρμούς in 277 and delete or transpose 286 (for reasons that have to do with the immediate contexts rather than the whole purification issue), but the other passages are in no way suspect.

³¹ I suppose Aeschylus could have contrived, if only at the cost of some distortion of ritual practice, to show the essence of the ceremony without showing the pig-sacrifice. Anyway, the assumption that used (justifiably) to be made, that sacrifices were never shown on stage (Arnott, P., Greek Scenic Conventions [Oxford 1962] 53–6Google Scholar), is not borne out by the new anapaestic fragment plausibly assigned to Aesch. Psychagogoi (Kölner Pap. iii [1980] 11–23Google Scholar), unless we can believe that Odysseus goes off stage for the sacrifice of τοῦδε σφαγίου and then returns for his meeting with Tiresias (fr. 478 M). Note also the libations seen at Cho. 129–63 and the victims at Eum. 1007.

³² Even when we reckon with the possibility of lacunae in the exchange between Orestes and Apollo, it is not easy to insert a report of the purification (still less an enactment of it). E.g. after 87 (cf. n. 21 above), ‘It is true that you have purified me, but this does not seem to have worked very well, so I still need your help’; surely not. Or after 73, ‘And this sleep of the Furies is the effect of your recent purification’; but why should it have had such a temporary effect? And a reference to previous purification would not be in place any later than this.

³³ The subtle discussion of Jones, John, On Aristotle and Greek Tragedy (London 1962) 105–8Google Scholar, is vitiated by neglect of the simple fact that whenever Orestes is on stage the audience can see whether his hands are bloody or clean, and no ambiguity would be possible in production. The point has, indeed, been generally ignored (see, however, Dingel, J., Das Requisit in der griechischen Tragödie [Tübingen 1967] 43 n. 1Google Scholar), but the only way of evading it that I can see would be to claim that blood on the hands, like tears on the face, might be left to the audience's imagination. And this would not be convincing, for blood can be seen at a far greater distance than tears, and the proper comparison is rather with the bloody eyes of Oedipus and Polymestor, which, so everyone assumes, were physically represented.

³⁴ Pointed out by Hammond, N. G. L., GRBS xiii (1972) 442 fGoogle Scholar.

³⁵ Erfurdt's εἶς σοι καθαρμός is plausible. M's Λοξίου, making Orestes ‘touch’ Apollo and not vice versa, is in some ways more attractive than Auratus's Λοξίας, but I doubt whether the anacoluthon can be defended. With Λοξίας I take Apollo's ‘touch’ to be metaphorical, since it is not mentioned in Eum. and since the chorus here cannot know that Apollo will be physically present at Delphi.

³⁶ I believe that the solution is to take καθαρμός less literally, as referring to any process that can free Orestes from the Furies, and not necessarily to an actual ceremony. This usage is natural in the context of Cho., where the Furies are closely associated with pollution, even if they will cease to be so in Eum. Taken in this sense the prophecy will be fulfilled, since Apollo, when once supplicated at Delphi, will indeed rescue Orestes—by sending him to Athens and defending him there.

³⁷ So Lesky, A., Hermes lxvi (1931) 209 fGoogle Scholar. ( = Ges. Schr. 106 f.).

³⁸ It could possibly be these lines that have ousted the misplaced 286 from its rightful position.

³⁹ The line creates an undue separation between μαρτυτήσων and ξυνδικήσων; and, if it were absent, we would be free to refer the words ἔστι γὰρ νόμῳ ἱκέτης ὃδ᾿ ἁνὴρ καὶ δόμων ἐφέστιος to Orestes' present supplication of Athena, thus taking account of the present tense and of the parallel at 669. We then have to ask whether this interpretation provides Apollo with an adequate reason for his intervention; perhaps so, if enough stress is placed on νόμῳ.