^{page 324 note 1} C.R. lxxxvii (1973).Google Scholar

^{page 324 note 2} Having looked at some of Heinsius's collations (Bodl. Auct. S V, 7 and to) and ten manuscripts in the Bodleian, I can now, amplify my charge of inaccuracy in Dörrie's apparatus criticus: 2. 135 portent cited by Dörrie from only five minor manuscripts but it is in nine of the manuscripts that inspected in the Bodleian, including Ob and Heinsius gives no indication that he found anything else in his manuscripts; 6. 162 exspes in Vb as well as G; 7. 100 not quern but quas P (checked in the original); 8. 41 egerat also in Gi, Vb, Ob, Of; 11. 129 to Ob; 12. 8 vitam also in Mi; 13. 133 omen revocantis Vb; 16. 323 nostri K and Pb¹ (bot1 by implication); 19. 171 clam also in Mi and Pb; 20. 6 dolere also in Gi, K, Pb, Vb (all by implication); 232 vigilem v.1. in Gi. These are only a selection of the discrepancies I have noticed. Heinsius's collation of Sp, the lost Argentinensis, is either more full or more accurate than Loers's, which I have not seen; the manuscript broke off in Ep. 20 not after 239 but after 233 or 234, and it read e.g. 4. 93 herbam, 5. 128 ante, 6. 89 sparsis, 93 male, 111 vir non, 7. 175 v.l. certius, 8. 104 v.I. munus et hoc, 10. 96 rabidis, 11. 113 rabidarum, 13. 13 mandantis, 16. 323 nostri, 17. 173 relicta, 188 fuit, 212 Asiae, 20. 191 cum, 203 ignorent. Other mistakes of Dörrie's, mostly connected with manuscripts used by Heinsius, are pointed out in my forthcoming article ‘Heinsius's Manuscripts of Ovid’.

^{page 325 note 1} I owe this information to Mr. Kenney. Dörrie gives ‘es a. nocent F (?)’.

^{page 325 note 2} Sedlmayer in his apparatus maintains. that Ovid is translating Il. 9.682–3 but he is manifestly not; and why should he have thought that in Greek justified fama est dare in Latin ? Sedlmayer's other parallel, Met. 7. 739, is vitiated by an uncertain text.

^{page 326 note 1} For est at successive stresses cf. 20. 111 suspected by Dilthey but unemendable.

^{page 326 note 2} Palmer's note on his text quid non censeri inique? tells the reader everything but wha it means. In his earlier edition of Epp. (London, 1874), he renders it ‘what do you not rate unfairly?’, which in the context is obscure to say the least; and his parallels are far from proving that censeris can mean aestirnas.

^{page 328 note 1} Mnem. xxxiii (1905), 34–5.Google Scholar Though Dörrie includes a great quantity of rubbish in his apparatus, he mentions not a single conjecture of Damsté's. No doubt he was scared off by Housman's, assertion in C.R. xliii (1929), 196Google Scholar that ‘among all that has been written on the heroides Mr. Damste's, paper in Mnem. 1905 pp. 1–56 is conspicuous for shallowness and futility’; but someone who so often sets Housman's scholarship aside has no reason to heed his strictures on other people's. At the risk of being condemned for Damsté's vices, I should like to record the opinion that amongst much triviality and error he says something worth saying on at least 1. 1 hanc, 44 at, 2. 126 illa, 5. 74 has, 8. 81 nam coniunx aberat, 14. 42 vina, 121 et, 15. 178 et (his conjecture habe is ingenious), 18. 141 et.Google Scholar

^{page 328 note 2} When I say that caeco is not far removed from tecto, I mean not that c and t were frequently confused, which I have no reason to believe they were at the time when this corruption must have taken place, but that the cs and ts in et caeco tutus could easily have become muddled in a scribe's mind.

^{page 328 note 3} A parallel for caecus hostis would be welcome, since it might be taken in the obvious sense ‘blind enemy’; but the context and the commonness of expressions like caecum vulnus greatly reduce the risk.

^{page 328 note 4} ‘Phoebus has a chariot, the moon has not’ Goold, H.S.C.P. lxix (1965), 46; but cursus cannot be right at Met. 15. 790 sparsi lunares sanguine currus, and she must do something with her horses in the present passage and at Am. 2. 5. 38, Rem. 258, Med. 42, Met. 2. 208–9, Fasti 3. 110, 4. 374, 5. 16, Trist. 1. 3. 28.Google Scholar

^{page 328 note 5} See my review of Dörrie's edition (cf. above, p. 324 n. 1). Housman mentions this conjecture on Manilius 4.451 (I owe the reference to Professor R. Kassel) together with a palaeographical experiment of his own, et nova, which fails to state the essential fact that nine months were complete, and thereby leaves itself ambiguous: would nova mean the ninth, just mentioned, or the tenth ?

^{page 329 note 1} Owen is so often lacking in judgement that it is a pleasure to find him arguing sensibly in favour of two other variants in Heroides not printed by Dörrie, 14. 11 ense (ibid. p. 166) and 17. 261 faciam (C.Q. xxxi [1937], 14–15).Google Scholar

^{page 329 note 2} On Am. 2. 19. 20 see Goold, op. cit. (above, p. 328 n. 4) P. 44.

^{page 330 note 1} Alternatively, Epp. 16–21 may be assigned to some other poet than Ovid. Two of Lachmann's reasons for taking this step, the polysyllabic endings and the metre of 19. 29, have been repeated by Courtney, , B.I.C.S. xii (1965), 63–4,Google Scholar who points out that parallels can be found only in the poems from exile; but why should Ovid not have composed Epp. 16–21 in exile? One of the other things that Lachmann objected to is worth recalling, 17. 215 qui. It is offered by the oldest manuscript at Trist. 3. 4. 21, where Courtney, Gnomon xliv (1972), 80, will not allow Luck to accept it; but if the genuineness of Ep. 17 is granted, the two instances support each other, the more so because they both occur in late works (another archaism, haud, makes its one appearance outside Met. at Trist. 1. 3. 73). On the whole, Lachmann's observations seem to me much too weak to establish the existence of a second poet as talented as Ovid, or more talented, it might be thought, than the Ovid of Epp. 1–15.Google Scholar

^{page 330 note 2} Goold, op. cit. (above, p. 328 n. 4) p. 44.

^{page 330 note 3} It has a long history: in Bodl. Auct. F 2 17 (15th cent.), oris is glossed verecundie.

^{page 331 note 1} Most of the spurious couplets in Heroides have already been suspected by someone other than Lehrs, and where I cannot bring forward new reasons I have no wish to repeat old ones until they are ignored in a better edition than Dörrie's.

^{page 332 note 1} When Daedalus at A.A. 2. 37 considers escaping by air, he is disastrously sane. Hypsipyle's injunction aera temptet at 6. 161 is a deft touch of Ovidian irony: in the context it is triumphantly sarcastic, but anyone who knows Euripides has seen it fulfilled.

^{page 332 note 2} Exceptions in Ovid: 15. 211, 217 (problematical: see Burman), Trist. 3. 89 egrediar, sine illud erat sine funere fern. Fasti 4. 749 sine…ve…si…ve…si… hardly counts as an exception.

^{page 332 note 3} The probability that another couplet beginning with sine preceded 97–8 makes it tempting to suppose that a scribe omitted it by jumping from one sive to the other; 85–96 would then be an attempt to fill the gap and at the same time to supply enough illustrations of pereundi mine figurae in 81. I have three reasons for resisting this temptation: (1) the gap before sine in 97 would not have been much more evident if it followed 84 than in its present position; (2) as the hypothetical interpolator felt, one illustration of pereundi mille figurae is not enough; (3) it is extremely hard to think up a couplet beginning with sive that would provide a smooth transition from 83–4 to 97–8. Anyone who accepts these reasons but finds the temptation irresistible will be driven to deleting 83–4 as well.

^{page 333 note 1} Bearing in mind how Professor Douglas Young has recently opened our eyes to certain refinements of thought and expression in Longus (Proc. Camb. Phil. Soc. cxciv [1968], 65–74, cxcvii [1971], 99–107), I feel it would be rash to exclude this interpretation, supported as it is by the parallel of Xen. Eph. 5. I. 9–11. Indeed, one might go further with Professor Young and regard funere and vulnera as author's variants; Ovid first wrote vulnera, but then he realized that it would be difficult for Macareus to shed a tear on the wounds of a mummified corpse, and so he altered vulnera to funere (funus does not seem to occur elsewhere in the sense ‘anniversary of death’, but Romans knew from childhood the word natalis, ‘anniversary of birth’). Pedants will object that in 126 Canace has asked for her remains to be put in an urn; but Ovid well understood the psychology of his heroines, and he would have expected Canace to be somewhat confused on the brink of suicide.Google Scholar

^{page 334 note 1} On the other ill-attested couplets, e.g 7. 98–9, I have nothing to add to Housman C.R. xi (1897), 200–2,Google Scholar and Sicherl, , Herme. xci (1963), 190–212.Google Scholar

^{page 334 note 2} Kraus, , Wien. Stud. lxv (1950–1951), 66;Google ScholarLatta, B., Die Stellung der Doppelbriefe in Gesamtwerk Ovids (Marburg 1963), 120–38;Google Scholar Goold, op. cit. (above, p. 328 n. 4) p. 3; Kenney, , Philologus cxi (1967), 212–32Google Scholar anc H.S.C.P. lxxiv (1970), 179–85Google Scholar; Luck, , Unter suchungen zur Textgeschichte Ovids (Heidelberf 1969), 12.Google Scholar

^{page 334 note 3} See especially Fischer, Uta, Ignotum ho, aliis ille novavit opus (Berlin diss., Augsburg 1969), 132–52, 196–222.Google Scholar

^{page 334 note 4} If this use of quidem…sed is ‘ganz un. passend’ (Fischer p. 138), so is the one just before (131–4).

^{page 334 note 5} The parenthetic uses of credo to be found in Ovid can all be reduced to four syntactical forms: crede mihi, si credis, quis credere possit?, and vix equidem credo. Cf. also Fischer p. 79: ‘Versteht man es [credis et hoc] als Frage, so kennzeichnet er seine Komplimente als incredibilia, von denen er nicht sicher ist, ob Helena sie ihm glaubt. Wie muss aber ein Kompliment wirken, das von seinem Urheber selbst in Frage gestellt wird! Liest man dagegen das leicht herzustellende “crede sed hoc”, so hat man eine Beteuerungsformel, wie sie sich ähnlich bei Ovid immer wieder findet. Paris fordert Helena damit auf, etwas zu glauben, was sicher wahr ist, was sie aber vielleicht aus Bescheidenheit nicht ohne weiteres annimmt.’

^{page 335 note 1} ‘Lesen wir einmal mit s crede sed hoc in v. 145. “Glaube mir aber dieses. Dein Ruhm (sc. von deiner Schönheit) ist kleiner als die Wirklichkeit.” Der erste Gedanke verlangt einen Gegensatz als vorausgehend, der zweite lasst darauf schliessen, dass dessen Inhalt war “der Ruhm deiner Schonheit ist gross”. Das steht aber nicht in v. 38 …; beide Gedanken stellen auch in keiner Weise einen Gegensatz dar’ Latta p. 121. The piece of literalism italicized is satisfactorily answered by Fischer p. 80: ‘Wenn Paris von der “fama vultus Helenae” spricht, so kann man nur an die Schönheit ihrer Züge denken, in der ihr Ruhm allein bestehen kann’.

If anything is wrong, it is the juxtaposition of plural and singular vultus in 37–8, which Housman did away with by modifying a conjecture of Palmer's and suggesting prima tulit vulnus nuntia fama tui. The legitimacy of this slight alteration is not obvious, since vulnera ferre commonly means vulnera accipere (the exceptions in Ovid are Rem. 44 and Trist. 2. 20 vulnus opemque feret, which could be a mild zeugma); but perhaps nuntia helps. If it is legitimate, it is a great improvement (cf. 12. 33–4 for a similar progression from the hexameter to the pentameter). Legitimate or not, however, it makes no difference to the train of thought.

^{page 335 note 2} Cf. Riese, , Bursian x (1877), 22. In spite of this, it is one of the three props on which Luck, op. cit. (above, p. 334 n. 2), pp. 11–12. supports his theory that the archetype of Heroides had 13 lines to the page; the others are the loss of 21. 15–146 and the appearance of 14. 114 in some manuscripts as 14. 62. I cannot take any of them seriously. Both s6. 39–144 and 21. 15–146 must be reduced by a couplet before they will fit into pages of 13 lines, and I see no reason for supposing that any couplet in either, albeit spurious. was absent from the archetype (or that Luck has more right to delete 16. 97–8 than other scholars have to postulate a lacuna after 50); as for the disturbance in Ep. 14 something can perhaps be made of it, but the knots that Luck ties himself into p. 34; might rather suggest that he had an awkwarc fact to explain away.Google Scholar

When Luck begins to elaborate his theory, the loss of 16. 39–144 is matched by another coincidence: Ep. 15 fell out in its entirety, and nothing of Ep. 14 or Ep. 16 went with it. Unfortunately 220 (or 221 if you count the title) is not divisible by 26; but presumably the front of the first leaf that fell out was blank (though the scribe of P, it seems, was so inattentive to such gaps that he actually ran some epistles together).

Finally, when he has to decide what lines were in the archetype, Luck employs three assumptions: (1) where P has fewer lines than the archetype requires, some lines were undecipherable and the scribe simply ignored them; (2) where P has more lines than the archetype requires, there are interpolations; (3) where the archetype requires lines that Ovid did not write, the wording is corrupt. (1) is applied in the most shameless manner: as the line with which P ends, for instance, would not have been the last of a leaf in the archetype, the remaining lines were no longer legible (p. 16); and P omitted 5. 25–6 because 25 stood at the damaged foot of one leaf and 26 at the damaged head of the next (pp. 19–20). (2), as I have already pointed out, arbitrarily identifies accretions to the original with accretions to the archetype. (3) is superfluous and betrays an irrational dread of ascribing interpolations to the archetype; it is also in some cases highly improbable, because the wording is free from objection (e.g. I. 37–8, 13. 63–4, 17. 248). In short, I can only agree with Luck himself: ‘dies alles hat zuweilen eine spielerische Note’ (p. 7). Dr. Winterbottom, M. in C.R. lxxxv (1971), 208–9 takes the same view and shows that weaknesses in Luck's method are not confined to Heroides.Google Scholar

^{page 335 note 3} Cf. Fischer p. 146; the whole section pp. 140–7, ‘Die Verse als notwendige Voraussetzung für Helenas Brief ?’, is worth reading.

Fischer borrows from Asteroth an argument of the same kind that would be even more elegant if it were valid: that quondam in 16. 165 is impossible after the detailed narrative in 53–88. 20. 216 seems to show that it is not, unless Asteroth's formulation of the argument, which differs rather from Fischer's, can be sustained: ‘so spricht mar im Hinblick auf eM dem Gesprächspartne: bekanntes Moment, welches, bereits lange vertraut, in die Erinnerung zurückgerufen werden soil’ (p. 33).

^{page 336 note 1} Cf. Maia xxii (1970), 6. The couple cannot be deleted without a further change (99 at for sed Itali). Another difficulty in the vicinity is subeunt in 99, which at first seems: to be a genuine present but turns out to be a historic present.Google Scholar

^{page 336 note 2} The present iacent is defensible (cf. 15 46), but te vigilans oculis videbam is not Bentley's conjecture oculis animi improve: the sense but leaves the second half of the antithesis without an instrumental ablative to match oculis animi (or perhaps it cancel: the antithesis altogether: ‘waking and sleep ing, I saw you in my mind's eye’).

^{page 336 note 3} Only rewriting will achieve results, a Kenney is forced to admit (H.S.C.P. lxxiv [1970], 579–81). His idea of separating nec tibi par usquam Phrygiae from nomen habet is ar unhappy one: the sense required by the antithesis of which it forms part is not ‘yot have no equal in Phrygia’ but ‘your fame ha no equal in Phrygia’. The couplet could admit tedly be deleted without damage to the senseGoogle Scholar

^{page 336 note 4} Kraus, op. cit. (above, p. 334 n. 2) 66.

^{page 336 note 5} The unlikelihood is increased by Fischer's observation, pp. 99–100, 140, that in 1–38+ 145-end, though Paris is writing at a particular time and is not telling a story, the illusion of a narrative progression is created The main story in 39–144 puts the clocl a long way back, the prophecies in 49 and 123–4 a long way forward.

^{page 336 note 6} The first problem in these lines is tht meaning of 161–2. The second is how ter in 159 and saepe in 163 are to be reconciled The third is the point of consurgere in 169.

^{page 336 note 7} Most of the problems here have beer adequately ventilated: see e.g. Palmer anc Fischer pp. 193–5. Fisher, Elizabeth, H.S.C.P lxxiv (1970), 198–205, tries to solve them, bui without success. Her arguments rest largely on false parallels: for her conjecture Legeres in 230, Trist. 3. 8. 36 legenda (corrupt); for quaeritur in 234, Met. 15. I (unambiguously a historic present); for her conjecture nescioquem in 235, 13. 91 (cf. Fischer); for deus et vates in 237, Fasti 5. 97 et matri et vati (‘Apollo says this both as a god and as a prophet’ is absurd); for mea carmina in 237, A.A. 2. 3 and 3. 792 (Cydippe is not a poet). She says nothing about the propriety of sponsae in 230 (cf. Fischer), the meaning of 238, or the point of munc ut uaga fama susurrat in 235 (did vaga fama make the inquiry?). Her conjecture nescioquem in 235 has the further drawback of being incompatible with her desire to bring the passage into line with Callimachus by supplying a suitable couplet before 237 (cf. Call. fr. 75. 22–7; as Pfeiffer says, ‘del responsum minime obscurum est’). My inclination is to agree with the view expressed by Ehwald in another connection and applauded by Luck, op. cit. (above, p. 334 n. 2) p. 21: ‘Es hat wirklich wenig Sinn, einen interpolierten Text durch Konjekturen heilen zu wollen’. It is a premature answer, however, to anyone who thinks that emendation can save a passage from condemnation. In 229–38, one reasonable emendation is to delete 229–30 and alter petas in 232 to petes; but I have yet to see any others, and the only remedy I can devise for 235–8 is to scrap all four lines and start again. After 238 there are three more difficulties: que in 241 and again in 243, and the content of 243–4 (cf. 19–20). The threefold que in 241–3, which Fischer finds ‘unschön’, has parallels at 14.32–4, 90–2, 16. 334–6, 19.47–9.Google Scholar

^{page 336 note 8} It seems to be an overambitious adaptation of insidias legi in 112.

^{page 336 note 9} If it is a fancy way of saying tuo or illo, where are the parallels ? Perhaps the author thought he had found one in 109: mittitur ante pedes malum cum carmine tali ei mihi! iuravi nunc quoque paene tibi. Here, of course, tali means ‘like the following’, but as the carmen does not actually follow, it could have been misunderstood.

^{page 337 note 1} There are sections in 16. 39–144 that seem to me linguistically unexceptionable and in manner worthy of Ovid, notably 53–88, the account of the judgement; and I dare say a zealous executor may have found this fragment among Ovid's papers and with additions of his own incorporated it into the only possible context. Such speculation I do not find congenial, and it would never persuade me to leave the whole of 39–144 in the text for fear of sacrificing a fragment that Ovid might have written. Not interfering with finished poems is surely more important than not sacrificing fragments, and I do not believe that Ep. 16, when Ovid finished it, included any part of 39–144.

Even if I were relegating the two passages to an appendix and ascribing them to a forger, I should make some emendations that are not made by most editors who retain them and ascribe them to Ovid. My text of 16. 39–144 would differ from Dörrie's in at least the following places: 60 veri (Heinsius), 91 laeta domus nato post tempora longa recepto est (so Palmer; post Bentley), 103 faceres (Heinsius; this ‘beseitigt den Anstoss nicht’ according to Fischer p. 77, but I have no idea why not), 98 a te (Heinsius), 140 in dubio (Micyllus; Heinsius's defence of in dubium finds no support in T.L.L. D 2120. 5–2120. 80).

^{page 337 note 2} Cf. Kenney 196.

^{page 337 note 3} Cf. Kenney p. 196 n. 1. In assailing Haemonis Kirfel may be following the example of his supervisor Luck, who suspects the wording of 8. 19 because repetitor is a (op. cit. [p. 11 n. 2], p. 25).