² On André's edition see Kenney, E. J., CR N.S. 20 (1970), 340–2,Google ScholarCourtney, E., Gnomon 44 (1972), 80–2.Google Scholar

³ Luck had published a text of Tristia and Epistulae ex Ponto, with a translation by W. Willige, four years earlier (Zürich, 1963). There is a valuable review of it by Albrecht, M. von, Gnomon 37 (1965), 491–6.Google Scholar

⁴ Rather, sparsum…uenenum (Tanaquil Faber); cf. Kenney, E. J., CR N.S. 28 (1978), 252.Google Scholar

⁵ OLD s.u. passus cites Enn. Ann. 349 as an instance of comis passis, but this phrase depends upon a speculative reconstruction of the fragment by Vahlen.

⁶ It is not clear how well attested illi is here. Luck ascribes it to ‘G¹ Val.’, but Owen ascribes it only to ‘GVξ’ (ξ is Luck's O_r).

⁷ Ribbeck printed illi from one manuscript of Arusianus Messius (vii.485 Keil) at Verg. Georg. 1.54 (illic the MSS of Vergil) and proposed illi for illic at Georg. 1.251.

⁸ It is not even recorded by Owen in the repertory of conjectures appended to his 1889 edition.

⁹ Housman's treatment of this passage (CR 20 (1906), 37 = Classical Papers, 637–8) is supplemented by Hikanson, L., Statius Siluae: Critical and Exegetical Remarks (1969), pp. 44–5.Google Scholar

¹⁰ Cf. also Cat. 63.19, Tib. 1.3.16.

¹¹ Nor, as far as I can see, does any other author.

¹² See Henry, J., Aeneidea, iii, 832–5 and R. D. Williams (ed. Aen. 7–12 (1973)) ad loc.Google Scholar

¹³ For the wish to die on dry land uttered by a person in danger of shipwreck see Horn. Od. 5.306–12, Verg. Aen. 1.94–6, Met. 11. 539–40, Fast. 3.597–8, Sen. Ag. 514–19; and cf. F. Bömer in Ovid (edd. M. von Albrecht and E. Zinn (1968)), 181–2, and Nisbet and Hubbard on Hor. carrn. 1.28.23.

¹⁴ Luck's text (1967) has the misprint sepulcra for sepulcrum.

¹⁵ M. von Albrecht in his revision of this edition (1966) has expressed reservations about some of these instances. For the use by other authors of the indicative in indirect questions see Kühner-Stegmann, 2.494.

¹⁶ See Kühner-Stegmann, 2.489.

¹⁷ pararis is accepted by Anderson, W. S. (Teubner, 1977).Google Scholar

¹⁸ Commentarius exegeticus ad Ovidii Tristia (1913).

¹⁹ discedat is wrongly rejected by Dörrie, H., P. Ovidius Naso: Der Brief an Phelan (1975), p. 136.Google Scholar

²⁰ In the following pentameter (1.7.8) editors continue to print quae instead of qua (M_o, Heinsius), oblivious of Housman's discussion in CQ 10 (1916), 130 = Classical Papers, 917–18.

²¹ neue quibus scribam possis dubitare, libellus / quattuor hos uersus e tribus unus habet. If one wanted a simple test of an editor's feeling for Ovidian style, one could hardly do better than note the choice which he makes here between libellus and libellos. The test is passed by Heinsius and Luck, failed by nearly everyone else.

²² Compare his own bowdlerization of Ars 1.33 in his quotation of the line at Tr. 2.249. One might, alternatively, argue that the expression uetus libellus does not have to be taken as referring to all three books of the Ars.

²³ It was accepted by Ehwald-Levy (see Ehwald, in Bursian, 21 (1882), 160) and by Owen (in deference to Ehwald) in 1889 and 1894. Owen had printed prima in 1885 and he returned to this in 1915 in deference to the futile objections of K. P. Schulze (Woch. f kl. Phil. 7 (1890), 577).Google Scholar

²⁴ For details see Bömer on Fast. 2.156.

²⁵ See, in addition to the commentaries ad loc., Bömer on Fast. 3. 659–60 and the Lexicon der frübgriecbiscben Epos s.u.

²⁶ For Ovid's fondness for recherché proper names in -is see Linse, E., De P. Ovidio Nasone vocabulorum inventore (1891), pp. 19–22,Google ScholarKenney, E. J. in Ovid (ed. Binns, J. W., 1973), p. 126.Google Scholar

²⁷ Such is their devotion to M that Owen is able to translate pauca…uerba…pati as ‘bear with (uttering) a few words’, Wheeler ‘suffer a few words…to escape you’.

²⁸ At first sight Burman's te <in> prole tua is appealing. For this use of in see OLD s.u. § 43 (‘in the person or shape of’). But here is a parallel for the bare ablative: Met. 13.503–4 cinis ipse sepulti/in genus hoc saeuit: tumulo quoque sensimus hostem, where tumulo means (so I should assume) not ‘in his tomb’ but, as Loers paraphrases it, ‘per tumulum’.

²⁹ Goldbacher, , WS 26 (1904), 271; for Nérnethy see above, n. 18.Google Scholar

³⁰ In the edition of O. Güthling (1884).

³¹ Wheeler actually translates the text as if it had sic: ‘and thus art half present…’ ita would be closer than sic to what is transmitted, but such an ita would not conform with Ovidian usage.

³² Ovid has Martia Roma at 3.7.52, Ex P. 1.8.24, 4.9.64; cf. Bömer on Fast. 3.79.

³³ ‘tuta: undemolished’ (Owen); ‘et tamen nemo ullum istius generis simulacrum loco mouendum censet’ (Pontanus).

³⁴ Luck's transposition of 303–4 to follow 305–6 is very attractive. The objection of Albrecht, M. von, Gnomon 37 (1965), 493, that ‘Die Argumentation aus den “Sakralrecht” gewinnt erst ihre Pointe, wenn der Hinweis auf die Warnung in der Ars vorausgegangen ist’ is not convincing. If my interpretation of 302 is accepted, it loses all validity. But in 305 read quaecumque irrupit (Ld, Luck: irrumpit c: erupit uel erumpit cett.) quo (ζ: qua ω, Luck) non sinit ire sacerdos. For quo see 5.1.78 quae tamen irrumpunt quoque uetantur eunt, Ex P. 1.7.23 nec tamen irrumpo quo non licet ire, Ars 3.636 quoque sui comites ire uetantur eunt. The two last passages tell against Kenney's hesitant quoque which is proposed in Luck's apparatus criticus at 5.1.78.Google Scholar

³⁵ The early editors print no parenthesis, nor does André. The result is hopeless.

³⁶ A ‘gross violation of Ovidian usage’, says Kenney, E. J. (CR N.S. 20 (1970), 342)Google Scholar of ferens in André's text. The same must be said of habens at Her. 12.96, which is commended by Goold, G. P., Gnomon 46 (1974), 477.Google Scholar

³⁷ Palmer, A., Hermathena 7 (1890), 268, proposed the atrocious conjecture carmina (for plurima)… fere. But I should like to commend his conjecture in electu (for euentu) poenae at 2.125 (cf. Her. 2.144 in necis electu). I commend it the more keenly because editors are unaware of it and because it was suggested to me independently by Mr W. A. Camps. Owen's parallel for a defining genitive with euentu is not good enough: at Ex P. 2.2.32 nam timor euentu deterioris abest, read euentus with the majority of manuscripts.Google Scholar

³⁸ For the corruption of the parenthesis see my note on 2.331–2 above.

³⁹ The conjecture was published in the apparatus criticus of Owen's text in the Corpus poetarum Latinorum and was reported again by Owen in his Oxford Text. So far as I am aware it has not been reported elsewhere and has never received a word of comment. It is interesting that Housman himself nowhere refers to it, although it offers such a good example of the kind of trajection of letters which he was fond of illustrating. Perhaps he later saw the weakness which I have tried to expose.