¹ Many of these manuscripts are listed, and their symbols explained, at the head of my text of Hecuba (OCT i. 334–5).Google Scholar

² It is reported by Wecklein and Biehl from the lost Thessalonicensis (16th cent.), on which see Pappageorgiou, P. N., Ἀθ⋯ναιον 10 (1881), 286–309Google Scholar, Turyn, 151–2Google Scholar, Mastronarde-Bremer, 172–3Google Scholar. It was also probably the original reading (later corrected) of J (Cambridge, University Library, Nn 3. 13 [15th cent.]), on which see Turyn, 206–8Google Scholar, Matthiessen, 50.Google Scholar

³ This Lexicon, compiled c. 1300 (Mastronarde-Bremer, 72Google Scholar), was edited by Nauck, from Vind. 169Google Scholar. But it exists in at least two other manuscripts, Vat. 12 and Vat. 22 (Benedetti, F., BPENC 14 [1966], 85–92Google Scholar, A. Colonna, ibid. 19 [1971], 13–16). Vind. 169 and Vat. 12 have γε, Vat. 22 has с⋯.

⁴ It would not be sufficient to accept only τι for τε, leaving сε … с⋯ν Øρ⋯να as an example of the καθ᾽ ὅλον κα⋯ μ⋯ροс construction (for which see Kühner-Gerth 1.289–90, Schwyzer 2.81, Hahn, E. A., TAPA 85 [1954], 219–36Google Scholar, Renehan, R., Studies in Greek Texts [1976], p. 60Google Scholar, my note on Phaethon 90ff.Google Scholar, Barrett, on Hi. 571–4Google Scholar, Bond, on Herc. 162Google Scholar, Johansen, Friis and Whittle, on A. Su. 379, 515Google Scholar), since с⋯ν after сε would be otiose (contrast Horn. Il. 1.362 = 18.73, 3.442, 6.355, Od. 18.331 = 391, A. Su. 379, Eum. 88Google Scholar, S. El. 147, Ant. 319Google Scholar, E. Tr. 408, Or. 608).Google Scholar

⁵ The same explanation is offered by Weil: ‘Si la leçon est bonne, chacune des deux phrases qui suivent est divisée par le poëte en deux idées, le sujet et l'attribut: πατ⋯ρ et ⋯ø⋯τενσ⋯ν με, σ⋯ παîς et ἔτικτε.’ Possibly something like this was in the mind of the scholiast (Σ^mve δ⋯ο Πρ⋯γματα δ⋯ο Πρ⋯γμαсιν ⋯ντ⋯εс, τι πατρ⋯ τ⋯ сπ⋯ρμα, τι μητρ⋯ τ⋯ν ἄρουραν).

⁶ Hermes 59 (1924), 257Google Scholar = Kl. Schr. 4 (1962), p. 351.Google Scholar

⁷ Mém. Acad. Impér, des Sciences de St.-Pétersbowg, sér. vii. 1.12 (1859), 44.Google Scholar

⁸ GRBS 14 (1973), 155–6.Google Scholar

⁹ Some credit should go to Facius, I. F. (1778)Google Scholar, who, according to Beck (I have not seen Facius' edition), ‘monet, hunc versum glossematis speciem habere, eiici tamen non vult’. Biehl's defence (Textprobleme in Euripides Orestes [1955], p. 31Google Scholar and commentary [1965] ad loc.) is ineffectual.

¹⁰ See also Page, D. L., Actors' Interpolations in Greek Tragedy (1934), p. 53.Google Scholar

¹¹ See Cropp, M. and Fick, G., BICS Suppl. 43 (1985), 34Google Scholar. There is only one parallel in Aeschylus (ScT 212Google Scholar πεδιον⋯μοιс), if we discount [A.] PV 817Google Scholar ⋯παναδ⋯πλαζε (⋯πανδ- Dindorf). And S. fr. 725.2 Radt ⋯νυμεναιοντεс must be considered dubious.

¹² Cropp, and Fick, 35.Google Scholar

¹³ See Page, , op. cit. 188Google Scholar. Willink cites 597 η ουκ āĭŏχρс as another anomalous linebeginning. There are, in fact, parallels for this (IT 1284Google Scholar, Hel. 493Google Scholar, Archel. fr. 257.2 [fr. 31.2 Austin]; Cropp, and Fick, 42Google Scholar). But we might treat it as āĭŏχρс (for the synizesis see Studies 93, 120).Google Scholar

¹⁴ Blaydes, Hence' παραсχο⋯сηсGoogle Scholar (Adversaria critica in Eur. [1901], p. 398).Google Scholar

¹⁵ Observe how νν δ⋯ с⋯ν ταρβ τρ⋯χα rounds off the prooemium with an appropriate recapitulation of δειμα⋯νω in 544. In 548 I have accepted Paley's δ⋯ for δ⋯: see Willink.

¹⁶ CQ n.s. 37 (1987), 286–7.Google Scholar

¹⁷ See above, p. 100.

¹⁸ Turyn, 123.Google Scholar

¹⁹ Turyn, 148–9Google Scholar, Mastronarde-Bremer, 172.Google Scholar

²⁰ Turyn, 141–3Google Scholar and (for Xf) Mastronarde-Bremer, 171.Google Scholar

²¹ See n. 2 above.

²² Turyn, 124.Google Scholar

²³ Zuntz, Similarly G., An Inquiry into the Transmission of the Plays of Euripides (1965), p. 158Google Scholar, who however regards it as an unhappy accident. For the confusion of ⋯γώ and ἔχω see Willink, , CQ n.s. 39 (1989), 52 n. 32.CrossRefGoogle Scholar

²⁴ Kritische Studien zu den griechischen Dramatikern, ii (1886), pp. 361–2.Google Scholar

²⁵ Greek Particles, p. 169.Google Scholar

²⁶ So, evidently, thought Herwerden, who proposed сο θρ⋯сουс (Exercitationes criticae [1862], p. 131Google Scholar), and H. Stadtmüller (as reported in Bursian, , 1910, 332Google Scholar), who proposed οὐδ᾽ εὐψυχ⋯αс.

²⁷ For the dochmiac + bacchiac clausula see Stinton, T. C. W., BICS 22 (1975), 84–8Google Scholar, Willink, , p. 253Google Scholar, and below, p. 121. Willink prefers to join 1388–9 as a single colon.

²⁸ BICS Suppl. 21.3 (1983), 124.Google Scholar

²⁹ Conomis, N. C., Hermes 92 (1964), 46–8Google Scholar and Dale, A. M., BICS Suppl. 21 (1971–1983)Google Scholar, especially the section on dochmiacs (Suppl. 21.3, 21–152), may be consulted. But I have ignored some of the passages which they cite, based as they are on an unreliable text.

³⁰ See below, p. 110.

³¹ For the divided resolution in the last element of the dochmiac see n. 53 below.

³² An alternative remedy is suggested by Willink, , CQ n.s. 38 (1988), 93.Google Scholar

³³ The transmitted choriamb calls to mind Hi. 1275, where an anomalous choriamb is interposed among dochmiacs. See Barrett ad loc., and my app. crit. The theoretical alternative at Ph. 169 is ba + 2 do (see (iii) above).

³⁴ See Stinton, , BICS 22 (1975), 85.Google Scholar

³⁵ See below, pp. 109 and 120. Ph. 178 is corrupt. So too is the spondee at Tr. 260–1 (2 do + sp).Google Scholar

³⁶ On these two passages in Med. see CQ n.s. 34 (1984), 62.Google Scholar

³⁷ See CQ n.s. 33 (1983), 347 n. 33Google Scholar, and for the resolved bacchiac see Tr. 564Google Scholar (Studies 19–20Google Scholar, Dale, , Lyric Metres, p. 74).Google Scholar

³⁸ On Ph. 1350–1Google Scholar see n. 90 below.

³⁹ ø⋯νου BZu^1s (accepted by Willink), rather than ø⋯νου (cett.). The structure of the dochmiac dimeter is then the same as Med. 1273 ⋯κο⋯ειс βο⋯ν ⋯κο⋯ειс τ⋯κνων, Hec. 1063Google Scholar τ⋯λαιναι κ⋯ραι r⋯λαιναι Φρυγν (Seidler: τ- τ- κ- Φ- codd.).

⁴⁰ And conceivably Or. 1500 πολ⋯πονα δ⋯ π⋯θεα/πολ⋯ποναGoogle Scholar (πολ- δ⋯ πολ- π⋯θεα codd.), for Which see Willink. The conjecture which I have printed at Hi. 826Google Scholar (τ⋯νι λ⋯γωι, τ⋯λαс, τ⋯νι [τ⋯να λ⋯γον … τ⋯να codd.] τ⋯χαν с⋯θεν) also conforms to this pattern. For further examples of the pattern see CQ n.s. 34 (1984), 65.Google Scholar

⁴¹ With Page's 〈ἰώ〉 and Hartung's π⋯θεν for π⋯θεν π⋯θεν. Alternatively, (Wilamowitz) πγ⋯ναι, π⋯θεν/π⋯θεν πἔλαβεс ⋯μ⋯ν (hypodochmiac + dochmiac), which falls within category (e) below.

⁴² But not Hel. 682Google Scholar (see Dionysiaca: Nine Studies … presented to Sir Denys Page [1978], p. 163Google Scholar, Willink, , CQ n.s. 39 [1989], 66).CrossRefGoogle Scholar

⁴³ See above, p. 108, below, p. 120.

⁴⁴ See n. 85 below.

⁴⁵ Only a possibility: see above (under (iv)).

⁴⁶ See above (under (v)).

⁴⁷ Only a possibility: see above (under (i)).

⁴⁸ But probably not dochmiac: see n. 65 below.

⁴⁹ See above (under (v)).

⁵⁰ Note also IT 869–70Google Scholar τ⋯λμαс δε⋯ν᾽ ἔτλαν/ἔτλαν δε⋯ν᾽(δε⋯ν᾽ ἔτλαν L), ⋯μοι (PCPS n.s. 22 [1976], 42–3Google Scholar). Willink suggests to me as an alternative ⋯μοι, δε⋯ν᾽ ἔτλαν.

⁵¹ See above, p. 107.

⁵² See Dionysiaca (n. 42 above), p. 166Google Scholar, Willink, , CQ n.s. 39 (1989), 59.CrossRefGoogle Scholar

⁵³ ⋯π⋯δοс, Ø⋯λοс, ⋯/π⋯δοс gives divided resolution in the last element, for which parallels are El. 1170Google Scholar, Herc. 1070, 1212Google Scholar, Tr. 244, 253Google Scholar, IT 871Google Scholar, Ph. 1295Google Scholar (with Elmsley's ⋯χ⋯сω for ἰαχ⋯сω: see n. 74 below), Or. 1364Google Scholar, Ba. 995 = 1015Google Scholar, Rh. 131Google Scholar, and possibly Hyps. fr. 64.99 (see above, under (ii)). See Parker, L. P. E., CQ n.s. 18 (1968), 267–8CrossRefGoogle Scholar, Willink, , p. 298Google Scholar and CQ n.s. 39 (1989), 59Google Scholar. Normally such division follows after a run of shorts (). The exceptions (to which Andr. 843 would have to be added) are Herc. 1070 (unless we follow Willink, , CQ n.s. 38 [1988], 96Google Scholar), Tr. 253, Rh. 131 .

⁵⁴ Rh. 821Google Scholar μ⋯γαс ⋯μο⋯ μ⋯γαс must be restored to dochmiac shape, but convincing restoration is hard to find (see Ritchie, W., The Authenticity of the Rhesus of Euripides [1964], p. 309Google Scholar). Murray's conjecture at Or. 1483Google Scholar (δ⋯ τ⋯τε διαπρεπεс|τ⋯τ᾽ …) is certainly wrong.

⁵⁵ See above (under (i) and (iv)).

⁵⁶ See Zuntz, , Inquiry, pp. 27–35Google Scholar, and Drei Kapitel zur griechischen Metrik (Sitzb. Ösi. Akad. Wiss., phil.-hist. Kl., 443 [1984]), 50–8Google Scholar; Barrett, , Hippolytos, pp. 84–90Google Scholar; Mastronarde-Bremer, 151–66.Google Scholar

⁵⁷ It is not necessarily the right one. But, if we reject it, we must have good reason for doing so.

⁵⁸ Square brackets enclosing manuscript symbols indicate that this is the place at which these manuscripts divide the text. I have placed a vertical stroke at the end of each line, in order to indicate where I believe the Alexandrian division occurred. The manuscripts which follow in square brackets are witnesses to this division. I have omitted a few aberrant divisions by single manuscripts or by a very few. A plus sign indicates that the division is found in the great ‘majority of manuscripts in addition to those listed. I have printed each dochmiac (or hypodochmiac) as a separate line. Whether they were originally written singly or in pairs is of no consequence.

⁵⁹ There are no aeolo-choriambics hereabouts.

⁶⁰ The colometry of P. Berol. 17051 at 325 and 326 can be restored with certainty. The possibility of mid-word division was overlooked by Lenaerts, J., Pap. litt, grecs (Pap. Bruxell. 13 [1977]), 19–23Google Scholar, Luppe, W., Archiv f. Pap. 27 (1980), 241Google Scholar, and O'Callaghan, J., Stud. Pap. 20 (1981), 19–24.Google Scholar

⁶¹ See below, p. 123.

⁶² We must keep the exclamation separate (so Biehl, Willink, West). Benedetto, Di and Dale, (BICS Suppl. 21. 3 [1983], 134)Google Scholar print Ἐριν⋯ν ⋯ττοτο the former analysing as ba + cr, the latter as'long doch'.

⁶³ The manuscripts give this as a hypodochmiac, but other divisions are at least as plausible.

⁶⁴ See below, pp. 114–115.

⁶⁵ If we divide ἄπερ ἔδρακον ἔδρακον/⋯ν δ⋯μοιс τυρ⋯ννων (do + ith). But since there are no other dochmiacs hereabouts, Willink's analysis () may be preferable.

⁶⁶ We shall have another instance at Or. 838, if we follow West in writing Ἁγαμεμν⋯ιοс(Ἁγ-codd.) παс. But the reizianum is of a common shape (Held. 373, 750 ~ 761, 757 ~ 768, El. 700 ~ 714, IT 396 ~ 411, Ion 458 ~ 478, 460 ~ 480), and metre does not afford us the liberty to change Andr. 1034 Ἁγαμεμν⋯ιοс(Ἁγ κ⋯λωρ.

⁶⁷ CQ n.s. 33 (1983), 352.Google Scholar

⁶⁸ I suppose that γε would give a semblance of sense, but it would not give acceptable metre, unless we adopted a colometry different from that which I postulate in 1387.

⁶⁹ This (for the most part) was probably the ancient colometry (see above, p. 110). P. Oxy. 3718 (5th cent.) divides at ἔсω| (with only AbPrRfSa; the majority divide at μολ⋯ντεс|) and Π⋯ριс| (with the majority) and probably at δακρ⋯οιс| and ταπεινο⋯| (with the majority). Thereafter the papyrus fails us. Only Sa (which divided at πεøυρμ⋯νοι|) divides at δ⋯. The remainder divide at κεθεν|ἄλλοс.

Incidentally, the gloss]‥ τ ‥ οс in the margin of P. Oxy. 3718 next to προνο⋯αс will be øρ]τοс. This word glosses προνο⋯α. in Hesychius, and in MV at Ph. 736 (Schwartz 326.20), and is found in the paraphrase on our line in Gudian. gr. 15 and Barocci 74 (as reported by Dindorf [schol. ii (1863), p. 307]), which I have found also in Zb.

⁷⁰ ‘Pendant close is always a mark of period-end, if it is followed by a short or anceps’ (CQ n.s. 27 [1977], 39).Google Scholar

⁷¹ See ICS 6 (1981), 84–7.Google Scholar

⁷² For an alternative treatment of this passage see Willink, , CQ n.s. 38 (1988), 94–5.Google Scholar

⁷³ For an alternative treatment see Willink, , loc. cit. 96.Google Scholar

⁷⁴ Murray's colometry here is impossible. Mastronarde, D. J. (Teubner, 1988)Google Scholar, retaining (like Murray) the transmitted text, analyses 1294–5 ~ 1306–7 as hexasyllable dochmiac + dochmiac, and notes that ῾γ⋯ους, ὕμνους vel sim. ἠχεν dici potest non potest ν⋯κυν ἠχεν. But ν⋯κυν ἰαχεν is no less unusual in itself, and is supported only by Hel. 1147 (Hermann: see below, p. 116). In fact, a personal object for either ἰαχεν or ἠχεν though unusual, is not more surprising than that which is attested for ὑμνεν (IT 1457Google Scholar) and χορε⋯ειν (Herc. 871Google Scholar), or than expressions like IT 367–8Google Scholar αὐλεται … μ⋯λαθρον, El. 691 ⋯λολ⋯ξεται … δμα, and βοâсθαι passive in Hdt. See also Willink, on 103.Google Scholar

⁷⁵ But we could write λ⋯γοιсι〈ν〉 (Fritzsche); and Murray's deletion of 315 (⋯κεсε κα⋯ τ⋯ δερο) may be right.

⁷⁶ The defence of ἅπαντα by Renehan, R., Greek Textual Criticism: A Reader (1969), pp. 107–12Google Scholar, does not touch on the metrical issue.

⁷⁷ BICS Suppl. 21.3 (1983), 136.Google Scholar

⁷⁸ cf. Kühner-Gerth 1.441.

⁷⁹ See Barrett, on Hi. 760.Google Scholar At Or. 839–40 ὅτ|χρ- (on which see Stinton, , JHS 96 [1976], 126Google Scholar) read ὅτ᾽〈κ〉 (Willink). At Ph. 166 β⋯λοιμī χρ- I propose β⋯λοιμεν (SIFC 1989).Google Scholar

⁸⁰ And possibly Su. 599 ~ 609 (GRBS 14 [1973], 250–1).Google Scholar

⁸¹ See above, p. 111.

⁸² Griechische Verskunst (1921), p. 271.Google Scholar

⁸³ BICS Suppl. 21.3 (1983), 138.Google Scholar

⁸⁴ For this figure see GRBS 14 (1973), 265Google Scholar and Studies 18.Google Scholar

⁸⁵ ἕπεс᾽ ἔπεсε Seidler: ἔπεсεν ἔπαιсεν(ν) fere codd.: ἔπαιс᾽ ἔπαιсε(ν) FT.

⁸⁶ And possibly fr. 453.9 (Cresph. fr. 71.9 Austin) 〈ἴθ᾽〉 ἴθι μοι, π⋯τνια, ππ⋯λιν (Papyrologica Florentina 7 [1980], 59Google Scholar). Harder, A., Euripides' Kresphontes and Archelaos (1985), pp. 104–5Google Scholar, expresses reservations about a lekythion in an aeolic context (in addition to Hi. 67Google Scholar and El. 153Google Scholar which she cites see Hi. 530 ~ 540, 531 ~ 541; also Itsumi, K., CQ n.s. 34 [1984], 72–4CrossRefGoogle Scholar) and about resolution of the penultimate long (which is found at Hel. 180, 199, 203, 373Google Scholar, Ph. 649Google Scholar [s.u.l.], 652 ~ 671 [s.u.l.], 1286, 1288 ~ 1299, [Ph.] 1721).

⁸⁷ The variant сυν⋯πεс(εν) ἔπεсε at Or. 1309 is certainly wrong.

⁸⁸ We must follow Willink (and Biehl) in writing οἰχ⋯μεθ᾽ for -μεθα.

⁸⁹ GRBS 14 (1973), 265Google Scholar, Studies 18–21, 119.Google Scholar

⁹⁰ The conjecture is neat, but it leaves a cretic (-κυτ⋯ν ⋯π⋯) interposed between dochmiacs (see above, pp. 107–109). If these lines are worth emending (I do not believe them to be Euripidean), we might accept the conjecture and follow it with ⋯π⋯ κ⋯ρα 〈τ⋯θε〉τε, giving dochmiac and hypodochmiac twice.

⁹¹ See Willink, ad loc., and his Addenda, p. 362.Google Scholar

⁹² Or 2 tro.

⁹³ See ICS 6 (1981), 95–8Google Scholar, CQ n.s. 33 (1983), 347Google Scholar, Stinton, , JHS 97 (1977), 143.CrossRefGoogle Scholar

⁹⁴ If we accept Willink's сυγγ⋯νων for с⋯γγονον (CQ n.s. 38 [1988], 97).Google Scholar

⁹⁵ See ICS 6 (1981), 91 with n. 25.Google Scholar

⁹⁶ Note also (in a dactylo-epitrite context) Tr. 515 ~ 535 . I should not accept the transmitted text of Ph. 309, where appears as a clausula to (and in synapheia with) dochmiacs, but (like Murray) should accept Fritzsche's transposition.

⁹⁷ See PCPS n.s. 20 (1974), 15Google Scholar, Dale, , Lyric Metres, p. 171.Google Scholar

⁹⁸ And possibly Cycl. 77 (Fritzsche)Google Scholar; but not Or. 1447, which is corrupt or lacunose. At Or. 1401–2, where it is difficult to make metrical sense of the transmitted text, I hazard λ⋯οντεс Ἕλλανεс|δ⋯ο διδ⋯μωι (MOMt: -ω cett.) 〈ῥυθμι〉 ().

⁹⁹ For an alternative dochmiac interpretation (with Dindorf's δ⋯κρυ〈α〉 see Studies 20Google Scholar, Willink, , CQ n.s. 39 (1989), 46.Google Scholar

¹⁰⁰ See Dionysiaca (n. 42 above), p. 166Google Scholar, where I wrongly said that IA 1290 is iambic.

¹⁰¹ See above, p. 108. For the position of the repeated element in the dochmiac see p. 109 (under (b)).

¹⁰² See Dale, , Lyric Metres, p. 181.Google Scholar

¹⁰³ Wilamowitz, (Verskunst, p. 271 n. 1)Google Scholar appears to have contemplated the deletion of μοι, but his analysis would then have been (as 18).

¹⁰⁴ This assumes the division -ραс μεταμειβομ⋯να π⋯λιс ἅδ᾽ ⋯π᾽ ἅκροιс|, which will then be followed in 833 by ἕсτακ᾽ Ἀρη⋯οιс сτεø⋯νοιсιν (), an unwelcome length (but an acceptable colon can be restored by ἕсτακ᾽ Ἀρε⋯οιс сτεø⋯νοιсιν [fere Brunck (Ἀρ⋯ιοιс)] or ἕсτακ᾽ Ἅρεοс сτεø⋯νοιсιν [Person]). Alternatively, divide -ραс μεταμειβομ⋯να π⋯λιс ἅδ᾽ ⋯π᾽ ἅκροιс ἕсτακ᾽|Ἀρη⋯οιс (or Ἀρε⋯οιс) сτεø⋯νοιсιν (for the catalectic hexameter see immediately below).

¹⁰⁵ See n. 104.

¹⁰⁶ See GRBS 14 (1973), 250–2.Google Scholar

¹⁰⁷ I see even less cause for ⋯μοс (JMnRRwST), adopted by Murray, giving an encomiologus (), which belongs among dactylo-epitrites (as Ion 1504, fr. 901.1) and is less well suited as a continuation after the preceding dactylic tetrameter. In dactylic lengths the clausular does not abut directly onto dactyls ending in double short (hence Rh. 530 ~ 550 should bedivided[see Studies 102, 121Google Scholar; Ritchie, W., The Authenticity of the Rhesus of Euripides (1964), pp. 314–15]Google Scholar, and we must reject Murray's text of IA 1485–6), but may follow after catalexis (see above, p. 115). I add that ἔλθ᾽ ⋯π⋯κουροс ⋯μοîс Ø⋯λοιс π⋯ντωс would provide a parallel for candidate 1; but I do not recommend it.

¹⁰⁸ See PCPS n.s. 20 (1974), 15–16Google Scholar; Bond on Herc. 1024Google Scholar; West, , BICS 28 (1981), 62Google Scholar, CQ n.s. 32 (1982), 285–6Google Scholar, Greek Metre (1982), pp. 100, 111.Google ScholarWillink, (CQ n.s. 38 [1988], 94)Google Scholar introduces a second dochmiac in both Herc. 1024Google Scholar and Rh. 832.Google Scholar I shall deal with the corruption in Rh. 466 (~ 832) elsewhere.Google Scholar

¹⁰⁹ See n. 27 above.

¹¹⁰ On fr. 117 see Hose, M., Mus. Crit. 21–2 (1987), 143–4.Google Scholar

¹¹¹ See Dale, , Lyric Metres, p. 140.Google Scholar

¹¹² For the alleged spondee see n. 35 above.

¹¹³ See n. 103 above.

¹¹⁴ See above, p. 107.

¹¹⁵ See n. 35 above.

¹¹⁶ See above, p. 107.

¹¹⁷ See Dale, , Lyric Metres, p. 155.Google Scholar

¹¹⁸ See Dale, , Lyric Metres, p. 140.Google Scholar

¹¹⁹ Neither of the two instances alleged by Conomis, N. C., Hermes 92 (1964), 27, is to be counted as dochmiac.Google Scholar

¹²⁰ See Willink, , p. 362.Google Scholar

¹²¹ Wecklein, Since (III. vi [1902], p. 90)Google Scholar attributes to Elmsley a proposal to delete οἱ δ᾽ ἕκειντ᾽, I had better say that what Elmsley proposed (on Held. 838Google Scholar) was πολλο⋯ for νεκρο⋯. Wecklein was misled by Blaydes, (Adversaria critica in Eur. [1901], p. 418).Google Scholar

¹²² Murray's text of Hel. 637Google Scholar presents a catalectic iambic trimeter elided, at sense-pause, before dochmiacs, an even greater anomaly, and generally recognised as impossible (see, most recently, Willink, , CQ n.s. 39 [1989], 52–3).CrossRefGoogle Scholar Nor should I accept Murray's text of Ph. 294–5Google Scholar τ⋯ν οἴκοθεν ν⋯μον с⋯βονс᾽|ἔβαс ⋯ χρ⋯νωι γν πατρώιαν. There is no parallel for elision of an iambic dimeter before an iambic line beginning with a bacchiac, and such elision at a strong sense-pause is unthinkable. To write с⋯βουсα (so Mastronarde and others) gives an unparalleled length. Write с⋯βουсα ν⋯μον. Parallels for resolution in the last element of the dimeter, at sense-pause, are given by Parker, L. P. E., CQ n.s. 18 (1968), 255.CrossRefGoogle Scholar A comparable phenomenon is resolution before change of metre, as in Tr. 565Google Scholar, Ion 212 ~ 230, Ba. 1170 ~ 1186.

¹²³ See above, p. 108.

¹²⁴ The manuscripts mostly divide at Πριρμ⋯сι| (none at π⋯λαιсι|). Murray's conjecture (accepted by Willink) is a small price to pay for avoidance of the metrical analyses offered by other editors.

¹²⁵ HMB and others divide at ⋯|⋯κλεπτον, and I take this to be the Alexandrian division (see above, p. 111). Most of the others divide at οἵδα|or γ⋯ρ|, none at δρα|π⋯την.