^{page 27 note 1} The division between Parts I and II of this article has been made according to one main principle: the avoidance, so far as possible, of cross-references from part to part. In Part I are included those passages in which lacunae and interpolations (rightly or wrongly) have been postulated by editors; for these the introductory remarks about the transmission of the text are especially relevant.

^{page 27 note 2} It is a pity that Zuntz was unable to take account of Barrett's study of the Euripidean tradition.

^{page 28 note 1} Reluctantly, since it necessitated some fundamental reappraisal when this article was already in proof, I am driven to accept Zuntz' arguments for this. Arguments based on handwriting can never be conclusive, since a twin of L could have been in a closely similar hand; more significant are cases like IT 98. But the strongest argument of all is a negative one: probably nowhere in the Alphabetic Plays (excluding orthographica and the conjectures of p) has P preserved a reading genuinely superior to a legible reading in L; whereas P follows L in innumerable nonsensical (and therefore probably recent) errors, some of which at least would have been avoided by a twin of L.

^{page 28 note 2} Dodds, pp. lii, etc.

^{page 28 note 3} Zuntz, pp. 135 ff.

^{page 28 note 4} Here I differ from Zuntz: (a) it seems to me incredible that L's exemplar can have had the whole of Ba. and Tr. all the time; (b) note further that there is nothing ‘Triclinian’ about the divergences of P: if, as Zuntz thinks, π was simply the archetype of L as thoroughly revised by Triclinius, we should certainly expect to find some signs of his handiwork; at the same time there are innumerable ‘l’ corrections in L that P knows nothing of.

^{page 28 note 5} Zuntz speculatively, but not implausibly, conjectures that the copyist of P transcribed cahiers (usually of four plays) alternately from L and π. A possible objection to this is that L and π can never have been brought together in order that L might be supplemented; further it is clear that L's thorough revision by Triclinius (1) was subsequent to the use of L as P's exemplar, and done at a time when neither P nor π was available (or at least used) for reference: otherwise Tr. and Ba. 756 ff. would surely have been added, and other corrections made.

^{page 28 note 6} Δ is Barrett's symbol for HCDE in Hp. (pp. 68 ff.). 7 H: Palimpsest Codex no. 36 in the Patriarch's Library at Jerusalem.

^{page 28 note 8} Al. is included (with Hp., Med., An.) in the early 14th-century D (Laurentianus 31. 15), a manuscript of the same affiliation as H.

^{page 29 note 1} The transmission of Ba. is linked with that of Tr., for which we have a widely diver gent text in V (Vaticanus graecus 909, of the late 13th century). V is a manuscript of hybrid affiliation (Barrett, pp. 73 ff.), but adheres more often to Ω. Now Ω presumably had the complete decad (the other eight, apart from Ba. and Tr., are trans mitted by 0—Laurentianus 31. 10), and it is not unlikely that V obtained Tr. from this source; since V's ending is damaged, it may originally have contained Ba, as well (so Dodds). Since V is so clearly distinct from P in Tr., it must be unlikely that P can have had Ba. 756 ff. from Si, a tradition with which P has elsewhere nothing in common. We thus find tenuous confirmation of what is anyway probable—that both Λ and Ω originally contained all ten select plays.

^{page 29 note 2} I choose in deference to Dodds, who uses the symbol [Q] for P's lost source.

^{page 29 note 3} See further on 1300, 1329 f. below. In general I prefer to absolve the copyist of P of the sin of incompetent omission: he omits nothing in Ba. 1–755 (unlike L).

^{page 29 note 4} At one time L, at another π, and perhaps further subdivided into separate cahiers of four plays. But at any one moment he would be following only a single exemplar: there is something incongruous in picturing this copyist—a mere hack, we may suppose—laboriously transcribing one text but with an eye on another in case he had omitted something.5 For Ba. 756—a vitally significant crux—see below, pp. 42–44.

^{page 29 note 6} E.g. 227 L, P.

^{page 29 note 7} As perhaps in Christus Patiens, which is something of an enigma. The author had a manuscript containing at least Hec, Or., Hp., Med., Tr., Rh., Ba., which vacillated between Λ and Ω (so far as his peculiar habits allow one to judge). This text may have been either an early hybrid, or a lone survivor of an otherwise unknown tradition (Barrett, p. 77).

^{page 30 note 1} Barrett's account of ‘The Tradition in Antiquity’ (op. cit., pp. 45–57) provides an invaluable theoretical basis for conjectures that postulate pre-Alexandrian corruption. All commentaries on Euripides are compelled to do this at not infrequent intervals; the only difference in Ba. is that such conjectures are necessarily much more hypothetical.

^{page 30 note 2} The structure Introduction—Past—Present—Future—Conclusion (announcing the approach of someone) is standard (e.g. Hp., Ion, Med., etc.) and I know no other case of a reversion to past time between the fourth and final sections.

^{page 30 note 3} See now Barrett, on Hp. 41–50, where there is a clever double entendre. The commoner technique is to mislead (but never to tell a lie) by putting the false prediction in a subordinate clause: this is a final clause in Ion 71–73, a conditional clause in Ba. 48–52, and a fearing–clause in Med. 37 ff.; the last of these is a corrupt passage, but it suffices, I think, to delete the obvious doublet 40–41 and read (Giesing) (Hermann) in 42; the opening of 42 will naturally have been adjusted when 40–41 was inserted, in order to introduce an explicit forecast of the death of Creon, and to avoid tautology with 40–41 (which was clearly intended to refer to the assassination of Glauce).Google Scholar

^{page 30 note 4} Though a literal–minded actor or editor may well have thought otherwise, it is in fact mistaken to consider 53–54 dramatically valuable as ‘making it clear to the audience that the speaker, whom they accept as a god, will be accepted as a man by the people on the stage’ (Dodds): the fact that Dionysus is next seen (434) being dragged before Pentheus in bonds is obviously sufficient, and fully explained in advance by the rest of the Prologue, 233–47 and 352–7; till these events occur, and especially during the Parodos, we do well to refrain from overnicety in distinguishing the human from the divine Dionysus: we are not meant to ask at this stage whether the Chorus accept their Leader as a man (see on 135–43 below).

^{page 31 note 1} See Barrett, (Index: ‘interpolation’, op. cit., p. 444)Google Scholar for a more elaborate subdivision. I differ from him, however, in his readiness to postulate comparatively motiveless ad hoc composition, and hope in a later article to show that a number of the lines which he de letes can be defended, or saved by emendation.

^{page 31 note 2} The sentiment is likely enough to have been available in other plays—an essential criterion in postulating this sort of intrusion. See 201–2 (discussed below) for another example. I agree with Dodds (against Barrett) in defending 316.

^{page 31 note 3} A thorough survey by Barrett, p. 358. There is no certain example in Ba., but I conjecturally assign 229–30 (q.v.) to this category. 716 probably qualifies also.

^{page 31 note 4} Norwood, Gilbert, Essays on Euripidean Drama (Cambridge, 1954) pp. 112–81.Google Scholar

^{page 31 note 5} 182 may be assigned to this category, an emasculated reminiscence of 860, possibly combined with some statement about Dionysus' ‘manifestation’ () in the lost lines after 1329. So also 1091–2 (v. Dodds).

^{page 31 note 6} The three reduplicated echoes are particularly striking, and seem calculated to draw the listener's attention to a parallelism which might otherwise escape notice. In addition to the direct verbal echoes, it seems not too fanciful to find other features of symmetry, e.g. , and Note also the similarity between 134 and 155.

^{page 32 note 1} It is unfortunate that 115 should also be corrupt; but it seems tolerably certain that ‘when’ is in fact the right sense. Murray's gives a very harsh parenthesis, and Dodds is surely right to reject it. We must read either Elmsley, or perhaps —a skip, rather than a gloss.

^{page 32 note 2} 547 ( 551–2 I take to refer to the whole ).

^{page 32 note 3} For the interpretation of 576–656, see Gilbert Norwood's brilliant in Essays on Euripidean Drama (1954), pp. 52–73Google Scholar, which are no more than mentioned in Dodds' 2nd ed. The extraordinary excitement and unreality of this scene should certainly prevent the spectator raising such an issue at the time, and I should add this as another motive to explain Euripides' peculiar handling of what was doubtless an obligatory part of the plot.

^{page 33 note 1} Not to be pressed, perhaps, in view of

^{page 33 note 2} Some other examples of ‘epicism’—perhaps rather traditional ‘dithyrambic’ style— will be discussed in Part II (on 421 ff.).

^{page 33 note 3} Although is adjectival in Il. 24. 271 (), it seems illegitimate to interpret the existing words as equivalent to ; but this is the sense one requires.

^{page 34 note 1} Easier still, of course, if one reads : for such a corruption cf. Hp. 659. But on the whole seems likelier, in view of the parallels mentioned.

^{page 34 note 2} Resolution of dactyls () is a great rarity (cf. Ar. Av. 1752, Ibycus fr. 285. 3), but it need not surprise us here in a context where so much is strange to us. Long sequences of pure dactyls are also rare, and Euripides is in general fond of total resolution to short syllables (e.g. 905).

^{page 34 note 3} are of equal tragic authority, but gained ground in later Greek, as acquired specialized meanings (e.g. is the Scholiast's comment on , E. Or. 1416); Euripides used them both frequently, especially , with apparent indifference to gender.

^{page 34 note 4} Cf. , E. HF119 (lyr.); alternatively might represent the reason for the bacchanal's fall. If a lacuna is once admitted, it could of course be of greater length, though it is unlikely, I think, to be very long, if the argument from broad symmetry with 157 ff. is worth anything.

^{page 34 note 5} See Introd. and on 753–8 below.

^{page 35 note 1} For such a separate, non-dramatic use of a lyric passage, cf. the use of Ar. Ra 454 ff. (Dover, K.J., Lustrum 1957/2, 98).Google Scholar

^{page 35 note 2} is the smallest correction but seems hard to support by parallels.

^{page 35 note 3} Also on 427, which will be considered in Part II.

^{page 36 note 1} ' is not metrically indispenable with , and the is more directly explicable if was replaced by . But on balance ' seems worth including; if such a compound corruption could occur, an original ' could well have dropped out.

^{page 36 note 2} The evidence is discussed by Page, D. L. on Med. 149 (Oxford, 1938).Google Scholar Although he takes the stricter view that is everywhere doubtful, the number of ‘doubtful’ instances seems to favour the view diat either scansion was possible, and seems here more rhythmical (if not ).

^{page 37 note 1} is of interest—surely a marginal comment, rather than a variant reading. One would expect a scholiast to write ; looks like a survival of an ancient comment to the effect that ifoef or has to be supplied from .

^{page 37 note 2} I should guess that the interpolation of 151 is associable with the separate, non-dramatic use of this ode, suggested above on 135–43 (last para). ‘The more joy-cries the better’ in such circumstances, and it will have been thought important to make it clear that it was not the assembled congregation (of mixed sexes) that is being addressed in tS ire …

^{page 38 note 1} Cf. Ar. Ra. 1309, attributed by the Scholiast to the contemporary I A. 1 Wilamowitz achieved the same effect with little transcriptional justification, this is open to the further objection that causes serious confusion between the two ‘personal’ objects.

^{page 40 note 2} Cf. (Cornutus, De Nat. Deorum, 30), coined as an explanation of Dionysus. Whether Euripides intended the word-play is less clear: the point is not laboured like (if right) is hardly likely to be a mere coincidence in an etymological passage, and may have been unmistakable to the poet's contemporaries. Such an intention would account for the silence about Nysa in Tiresias' version of the myth.

^{page 41 note 1} This seems enough to dispose of the view that 652 is equivalent to which also disregards the emphatic position of . It is surely obligatory to understand (internal ace.) with , not some thing like . Campbell made the best of a bad theory by altering : but the dative is needed with , not with .

^{page 41 note 2} When follows an initial verb, it tends to have something of the force of — e.g. , S. Ant. 823, and , Anacr. 24. That makes sense here (’you've said it’!). But there is also ironical emphasis (= ‘in fact’, ‘of course’): in this usage there is commonly a pronoun, quite often following, as here, when belongs with a phrase which includes a demonstrative, e.g. , PI. Grg. 514 e.

^{page 41 note 3} More commonly the ironical begins the sentence (so above), but it can end a line also—e.g. A. Eu. 209 and S. OC. 1003. The effect here seems to correspond with the dash in Kitto's rendering, and the heavily loaded takes the position normally occupied by the sarcastic word. is thus more violent than the more routine tragic line , which one may suppose that Euripides considered and rejected.

Closely similar, but unhelpful, is IA 304–5: M∈.

Two renderings are possible and appropriate: (a) ‘I glory in the reproach you cast on me’, (b) ‘That's a fine reproach!’ (ironical). I do not think it is only the exigencies of my argument that make me prefer the latter: it seems better to take (unemphatic) with the verb than closely with is obviously well suited to the more effective scornful repartee. In view of Med. 514, it is certainly false to say (with England) that was proverbial for ‘a reproach in which one glories’. The proverbial phrase was (Ph. 821 and Diogenianus 4. 85, quoted by the Suda s.v. ): this is different in that (a) there is no ; (b) the superlative is less apt to be ironical; (c) in IA 305 and Ba. 652 the is used predicatively, replaceable indeed by .

^{page 43 note 1} See Dodds: the best example is Bri Mus. E 775 (Beazley, A.R.V., p. 833).

^{page 43 note 2} So also Jackson, John (Marginah Scaenica, p. 17)Google Scholar, who wished to delete , and to reinsert after 761; but such a complication strains credulity too far.

^{page 44 note 1} Robert, C. (Hermes, xiii [Y], 137)Google Scholar, supported by Dodds.

^{page 45 note 1} Presumably the idée fixe (one can call it no less) about Agave's compositio mem-brorum is based on the absurd conclusion of Sen. Hipp. (1256 ff.). If this proves anything, it is that no Greek dramatist would have sunk so low. There is not the slightest reason to suppose that this compositio is adapted from any scene in Greek drama, let alone the Bacchae; it is sufficient to suppose that it was suggested by some narrative speech (perhaps, but not necessarily, Ba. 1261–21).

^{page 45 note 2} Also I4 118.

^{page 45 note 5} For example:

(1449).

The last line was identified by Dodds, but wrongly, I think, assigned to Agave. Such a wry sentiment would seem intrinsically better suited to Cadmus, with used metaphorically, like ‘consolation’, rather than referring literally to Agave's ‘address’ (which is rather a consolation for herself).

^{page 46 note 1} Of Agave's lost lines, perhaps I o can be guessed at from the adaptations preserved in Chr.Pat. (1311–13, 1315, 1256–7, 1469–70, 1122–3). These lines can hardly be fitted together jigsaw-fashion, and the greater part is evidently missing. In addition, the citations (Antiatticista in Bekk. Anecd. pp. 87 and 105) may belong here; and the Oxyrhynchus fragment of line-openings, ‘madness’ … ‘mountains’ … …, is suggestive but of doubtful assignation. Any attempt to use all this hypothetical material for a reconstruction is likely to require 25–30 lines at least.

^{page 46 note 2} This is on the assumption that the other lacuna after 1329 is about 50 lines (v, p. 49).

^{page 46 note 3} Perhaps 6–7 lines: Dodds plausibly locates Chr. Pat. 1011 immediately after Ba. 1329; it could be that Chr. Pat. 1123 derives from this request (e.g. with ); and here I should locate , Schol. Ar. Plut. 907. She will have ended with something like .

^{page 48 note 1} A minimum link, but the original may well have been longer here, including perhaps some statement of Dionysus' assumption of human form: conceivably this was the source from which 53–54 was interpolated, with some degree of adaptation (q.v. and cf. 229–30).

^{page 50 note 1} Not known elsewhere, but of the nightingale in S. El. 107 (anap.), and Suid. means ‘child-slaying’ (not ‘child-losing’) at Med. 849.