¹ Tarrant, R.J., Texts, Editors, and Readers: Methods and Problems in Latin Textual Criticism (Cambridge, 2016), 85–104CrossRefGoogle Scholar. He notes that the diagnosis of interpolations was once an integral part of textual critics’ toolkit, but that in the last century the subject has become contentious. His discussion draws on a typology of interpolation proposed in his ‘Toward a typology of interpolation in Latin poetry’, TAPhA 117 (1987), 281–98Google Scholar, and on his edition of Ovid's Metamorphoses (Oxford, 2004)Google Scholar, in which he brackets approximately 100 lines.

² I reproduce the text and apparatus criticus of Hall, J.B. (ed.), Claudii Claudiani carmina (Leipzig, 1985)Google Scholar, supplemented at the bottom with two variant readings for line 345 from the apparatus criticus of Charlet, J.-L. (ed.), Claudien: Œuvres. 1. Le rapt de Proserpine (Paris, 1991)Google Scholar. In the face of an extremely complex tradition (over a hundred manuscripts from the twelfth to the fifteenth centuries, multiple routes of transmission, pervasive contamination), both editors adopt an eclectic approach (Hall [this note], x–xi; Charlet [this note], liv–lv). The readings each of them cites are largely drawn from a list of representative manuscripts (fifteen for Hall, twelve for Charlet), chosen, significantly, ‘non ut optimos … sed ut notitia dignos’ (Hall [this note], xi), but exceptions are frequent. For instance the variant ac (for line 347 et), which Hall's asterisk marks as a lectio fortasse recta, is found in none of his fifteen chosen manuscripts.

³ Gruzelier, C. (ed.), Claudian: De Raptu Proserpinae. Edited with Introduction, Translation and Commentary (Oxford, 1993)Google Scholar, ad loc. (following Clarke, A.K., ‘Claudian's De Raptu Proserpinae’, PCA 27 [1930], 38–41, at 40Google Scholar) notes the double meaning of flexisque minis, which describes both the general figurative turning aside of the Furies’ usual menace (flectere minas describes the averting of Hercules’ threats against Troy [Sen. Tro. 719–20] and Neptune's against the Argo [Val. Fl. 1.219]) and the specific literal bending down of their snakes to drink wine (cf. Verg. G. 3.241 tollentemque minas and Stat. Theb. 11.313 cecidere minae, of snakes assuming and relaxing menacing postures; the concrete sense for minae of the pinnacles of a wall [e.g. Verg. Aen. 4.88; cf. TLL 8.994.3–15] helps to suggest the idea here of the Furies’ snakes as threatening projections from their heads, temporarily bent down).

⁴ Charlet (n. 2) sees the repetition in lines 335–7 as typical of Latin poets from Virgil onwards, and keeps all three lines, as do Birt, T. (ed.), Claudii Claudiani carmina (Berlin, 1892)Google Scholar and Koch, J. (ed.), Claudii Claudiani carmina (Leipzig, 1893)Google Scholar. Hall, J.B. (ed.), Claudian: De Raptu Proserpinae. Edited with Introduction and Commentary (Cambridge, 1969)Google Scholar leaves open the possibility that lines 335–6 are spurious, and line 337 genuine. Gruzelier (n. 3), Jeep, L. (ed.), Claudii Claudiani carmina. I: Carm. I–XXIV (Leipzig, 1867)Google Scholar and Onorato, M. (ed.), Claudio Claudiano: De Raptu Proserpinae (Naples, 2008)Google Scholar reject the line.

⁵ Cited by Hall (n. 4).

⁶ These passages are the only instances of nominative Ixion and the clausula Tantalus und- in extant Latin epic (nominative Ixion also appears twice in Horace and Senecan tragedy, and once in Phaedrus).

⁷ Both lines might therefore have originated from the same source: a lost hexameter poem on the wedding of Pluto and Proserpina, or a composition exercise to recast the theme of Claudian's passage, or simply the pen of a learned glossographer.

⁸ Tardioli, M., Claudien: De Raptu Proserpinae. Introduction et commentaire (Nancy, 1971)Google Scholar, cited by Charlet (n. 2), ad loc. For the correct interpretation, see Charlet (n. 2) and Gruzelier (n. 3), ad loc.

⁹ Cited by Hall (n. 4).

¹⁰ Gruzelier (n. 3), ad loc. notes the first two passages.

¹¹ Gruzelier (n. 3), ad loc.

¹² Statius’ Furies have cerastae for hair at Theb. 1.103, 4.55, 9.174 and 11.65. Cerastae are found in Megaera's hair at Claud. Ruf. 1.96, but the only other description of a Fury's hair in DRP names hydri rather than cerastae: 1.39–40 crinitaque sontibus hydris | Tisiphone (cf. Verg. Aen. 7.447 tot Erinys sibilat hydris).

¹³ A line from earlier in Claudian's scene (2.319 uertunt emeritos ad pascua nota iugales) may have provided another model for the structure of the interpolated line. The composition of line 346 would thus follow, more subtly, the combinatorial method used in composing line 337: uoluitur Ixion + Tantalus undam produced soluitur Ixion inuenit Tantalus undas, and uertunt emeritos ad pascua nota iugales + poto Phlegethonte cerastas produced extendunt socios ad pocula plena cerastas.

¹⁴ Stylistic assessments of Latin poetry are perilously subjective, although the judgement of Gruzelier (n. 3), xxiii–xxiv is worth citing: ‘no word is ever wasted and all, especially his verbs and adjectives, work extremely hard. This results in lines of felicitous brevity …, in the verbal paradox to which he is addicted …, and also in clever sententia …’. On the other hand, repetition and variation are of course pervasive in Latin poetry, and are especially characteristic of what Roberts, M., The Jeweled Style: Poetry and Poetics in Late Antiquity (Ithaca, 1989)Google Scholar describes as the ‘jewelled style’ of Late Antiquity. The challenge thus lies in determining when a repetition departs from an author's usual style. (Gnilka, C., Prudentiana I, Critica [Munich, 2000]Google Scholar negotiates this problem throughout his study of interpolations in the text of Prudentius, whose repetitions eclipse those of Claudian; pages 16–67 address interpolations in Claudian.) Repetition that serves only to simplify and explain a compressed and clever expression is not typical of Claudian's style. Compare, for example, Roberts's discussion ([this note], 25–7) of the repetitions and variations in DRP 1.231–6 and 3.344–52; DRP 1.89–91, where the repetition is a succession of three equally epigrammatic phrases; DRP 1.133–7, where an elaborate antithesis is developed through repetition and parallelism; DRP 2.262–3 and 2.313–15, where an initial longer expression (sed mihi uirginitas pariter caelumque negatur and mox ipse serenus | ingreditur facili passus mollescere risu) is followed by a repetition that is actually more compressed and epigrammatic (eripitur cum luce pudor and dissimilisque sui). Judgements in these matters inevitably differ: for instance Charlet (n. 2), ad loc. does not feel that even the repetition in lines 335–7 is uncharacteristic of Claudian's style.

¹⁵ The initial syllable of line 347 is transmitted variously as et, at, ac and nunc, while faustas is a variant for festas in one fifteenth-century manuscript (and was once endorsed by Heinsius). See also below for alio corrupted to aliae, which suggests confusion over Claudian's meaning.

¹⁶ Val. Fl. 8.240, Stat. Silv. 3.5.70, Claud. IV Cons. Hon. 644; note also festae faces at Claud. Nupt. 202–3 and DRP 3.409.

¹⁷ Cf. also Claud. Ruf. 1.83.

¹⁸ infesto is a minor variant here.

¹⁹ von Albrecht, M., ‘Proserpina's tapestry in Claudian's “De raptu”: tradition and design’, ICS 14 (1989), 383–90, at 386Google Scholar.

²⁰ Cf. also Sen. Med. 15 atram cruentis manibus amplexae facem, of the Furies at Medea's wedding.

²¹ This and ferunt for bibunt (and Reinhardt's ore for crine) may be added to leta for lene and to minantes for canentes to suggest persistent confusion with this image of the Furies at play.

²² Interpolation via intrusive gloss or parallel is common enough, but a likely instance at Claud. Stil. 1.302–5 illustrates the process that could have occurred in our passage: line 304 quis Briareus aliis numero crescente lacertis entered the text as an elaboration on line 303 bracchia centum, leading to subsequent changes in lines 303 (non to quae) and 305 (possent to posset) to repair the syntax (see Gnilka [n. 14], 8–27 with further discussion by A. Coşkun in his review of Gnilka [BMCR 2001.11.17]). For interpolation to fill a perceived lacuna, see, for example, Tarrant (n. 1 [1987]), 288–9. The other generally accepted interpolations in DRP were motivated by real or perceived lacunae: see Hall (n. 4), on 1.139–41 (lines 139 and 140 were each attempts to complete the syntax of lines 137ff. after the loss of lines 141–214 from part of the tradition) and Gruzelier (n. 3), on 2.118 (an attempt to bridge an abrupt scene transition, which reflects either a small lacuna or typical late antique style).