¹ D. Kiss, Catullus Online (http://www.catullusonline.org/CatullusOnline/index.php), last checked on 24 April 2015.

² C. Fea, Quincti Horatii Flacci opera … tomus primus (Rome, 1811), XI–XIV n. Q, at XIV.

³ J.A. Vulpius, C. Valerius Catullus Veronensis et in eum … novus commentarius locupletissimus (Padua, 1737), 313. Fea's familiarity with Vulpius’ commentary can be inferred from his use of Vulpius’ readings as a point of comparison for his own.

⁴ M.P.L. Ginguené, Les noces de Thétis et de Pélée, poème de Catulle (Paris, 1812), 194.

⁵ A. Guarinus, In C.V. Catullum Veronensem per Baptistam patrem emendatum expositiones (Venice, 1521), LXXVII(r).

⁶ I.L. de la Cerda, P. Virgilii Maronis Bucolica et Georgica (Lyon, 1619²), 184.

⁷ N. Pinder, Selections from the Less Known Latin Poets (Oxford, 1869), 23. Cf. C.J. Fordyce, Catullus (Oxford, 1961), 313: ‘the lengthening of the first syllable in this word is found only here: cf. 151 sūpremo’, and similarly W. Kroll, C. Valerius Catullus (Leipzig, 1923), 182.

⁸ Cf. recently Sen, R., ‘Vowel-weakening before muta cum liquidā sequences in Latin: a problem of syllabification?’, Oxford Working Papers in Linguistics, Philology & Phonetics 11 (2006), 143–61Google Scholar, at 147.

⁹ As has been observed by R.I. Sbĭera, Die prosodischen Functionen inlautender muta cum liquida bei Vergil (Chernowitz, 1898), 38, in Virgil words with prosodic scheme –̆ – × are invariably interpreted as ̆– × at hexameter end (and this holds true for other authors, with the only exception of Atlante at Ov. Fast. 5.83). Sbĭera assumed that Virgil's practice was solely motivated by the avoidance of spondaic endings, but there are sufficient examples of a fifth-foot spondee produced by trisyllabic Latin words both in Virgil (Ecl. 7.53 hirsutae; G. 3.276 conuallis; Aen. 7.634 argento, 12.863 desertis) and in other poets (most relevantly, Catullus: 64.44 argento, 96 frondosum, 297 praeruptis, 65.23 decursu). Of course, since the overall number of spondaic hexameter-endings in Latin poetry, especially those produced by trisyllabic words, is not high, the lack of parallels for cupressu producing a fifth-foot spondee (i.e. –̆ – × interpreted as – – × at hexameter end) may be due to chance.

¹⁰ A fifth-foot spondee can also be produced by quadrisyllabic Greek words with prosodic lengthening before mute and liquid, e.g. Catull. 64.11 Amphitriten (ultimately modelled on Od. 3.91 Ἀμφιτρίτης, etc.), Ciris 239 Adrastea (cf. Callim. Hymn 2 47 Ἀδρήστεια), and of course Ov. Fast. 3.105 Atlanteas (no Greek precedent survives). But, again, there are no comparable examples with Latin words.

¹¹ In particular, it can be noted that in Homer (Od. 1.52, 7.245) and Hesiod (Th. 509, 517) forms of Ἄτλας invariably have a long first syllable. And so in Virgil (note especially Aen. 6.796 caelifer Atlas), whereas in Ovid the first syllable can also be short.

¹² The pronunciation of Greek words, especially proper names, in Latin is a complex topic: cf. recently Barnett, F.J., ‘The second Appendix to Probus’, CQ 56 (2006), 257–78CrossRefGoogle Scholar, at 268–71. The important point is that Greek words could sometimes retain not only morphological (declension) but also phonological features (such as accentuation and presumably syllabification) alien to Latin. Of particular interest is an anecdote told by Quintilian concerning two different pronunciations of a Greek name, aptly discussed by J.N. Adams, Bilingualism and the Latin Language (Cambridge, 2003), 108: ‘It may be deduced from Quint. 12.10.57 that names of Greek origin might sometimes be pronounced with Greek prosody even if Latinised morphologically: the name Amphionem pronounced in the Greek manner (but apparently with a Latin inflection) had to be repeated in Latinised form (with an unaspirated stop and a short i) by the speaker because a “rustic” witness had at first not understood.’ Ovid's Atlante is presumably a parallel case: it is in the ablative (a case not available in Greek), but retains its Greek syllabification and possibly accent.

¹³ Note the fifth-foot spondee, though of a more regular type, in the parallel passage 64.108 radicitus exturbata (both contexts speak of trees being uprooted; radicitus is repeated at 64.288).

¹⁴ Despite its apparent Greek origin that would be obvious to any educated speaker of Latin, the word cupressus was firmly rooted in the Latin phonological and morphological systems (it can regularly take forms of the fourth declension), and had long been established in Latin poetry (note Enn. Ann. 223 longique cupressi, 511 rectosque cupressos, both at hexameter end). Moreover, its Greek counterpart κυπάρισσος has no prosodic lengthening effected by mute and liquid which could serve as a model.

¹⁵ See especially Oksala, T., ‘Zum Gebrauch der griechischen Lehnwörter bei Catull’, Arctos 16 (1982), 99–119 Google Scholar. Cf. also G.A. Sheets, ‘Elements of style in Catullus’, in M.B. Skinner (ed.), A Companion to Catullus (Malden, MA, 2007), 190–211, at 197–8.

¹⁶ Oksala (n. 15), 100–4 has a list of Greek words used by Catullus with an indication of their first occurrence in Latin.

¹⁷ R. Ellis, A Commentary on Catullus (Oxford, 1889²), 420.

¹⁸ See e.g. P. Kyriakou, Homeric Hapax Legomena in the Argonautica of Apollonius Rhodius: A Literary Study (Stuttgart, 1995).

¹⁹ For the concept in general, see J. Wills, Repetition in Latin Poetry: Figures of Allusion (Oxford, 1996), and specifically 22, on the use of Homeric hapax legomena in Latin poetry.

²⁰ The parallel is noted e.g. in Kroll (n. 7), 185. The only other occurrence of calathiscus in Latin is at Petr. 41.6. In Greek καλαθίσκος is more frequent, but has only one other context in dactylic poetry, at Theoc. 21.9.

²¹ Ellis (n. 17), 328, citing Dousa the younger. Strictly speaking, the Heliades were transformed into black poplars (αἴγειρος: see e.g. Ap. Rhod. Argon. 4.604) rather than white poplars (λεύκη), which moreover have a different aition of their own. The two species are consistently distinguished in Greek (though a Suda entry identifies αἴγειρος as λεύκη), but in Latin populus is used for both, which may partly account for Catullus’ substitution of one for the other. A further reason may be that thus Catullus comes closer to Theocritus’ Homeric model, Od. 5.64 κλήθρη τ’ αἴγειρός τε καὶ εὐώδης κυπάρισσος, on which see A. Sens, Theocritus: Dioscuri (Idyll 22) (Göttingen, 1997), 109. Note that the first narrative part of Theocritus 22 (27–134) treats an episode from the Argonautic expedition, a topic closely related to Catullus 64.

²² It may be interesting to note that the standard epic form since Homer is πλατάνιστος, while πλάτανος is largely avoided in narrative hexameters. The present context is the only one in Theocritus to use the latter form, whereas the former is attested four times (18.44, 18.46, 22.76, 25.20).

²³ Note also that, especially after the periphrastic reference to the poplar-tree as sorore flammati Phaethontis, the Greek form would more readily recall the aition of the cypress-tree.

²⁴ The parallel is noted e.g. in R.A.B. Mynors, Virgil Georgics (Oxford, 1990), 112.

²⁵ By contrast, there are three contexts of cupressus in the Aeneid (2.714, 3.64, 6.216); likewise in the Georgics, in addition to one occurrence of cyparissus (2.84), there are two of cupressus (1.20, 2.443). Valerius Flaccus, the only other Latin poet to use the Greek form, has one context of each (1.774, 7.405).

²⁶ J.M. Trappes-Lomax, Catullus: A Textual Reappraisal (Swansea, 2007), 181 seems right to argue for coniferam, which is also accepted by G. Trimble, ‘A commentary on Catullus 64, lines 1–201’ (Diss., Oxford, 2010), 184. Kiss (n. 1) names B. Realinus, In nuptias Pelei et Thetidis Catullianas commentarius (Bologna, 1551), 21(r) as the first to propose this reading. But it is already found in I. Parrhasius, Cl. Claudiani Proserpinae raptus (Milan, 1505), on 1.205 (no pagination), whom Realinus cites, on another topic, at the same page. Also prior to Realinus's commentary, Basilio Zanchi wrote coniferam pinum aut pulsantem sydera quercum in a close imitation of the Catullan context (Panis et Aegis Solis filiae epithalamion, in L.P. Zanchus, Poemata varia [s.l., ante 1540], no pagination).

²⁷ On the unenviable fate of Greek words in the manuscript tradition of Catullus, cf. Trappes-Lomax (n. 26), 13–14, who even postulates ‘a deliberate campaign of Latinization by Greekless scribes who thought Greek forms were simply errors’. Trappes-Lomax's most relevant example is the split of the manuscript tradition between Serapim and Sarapim at 10.26, which implies that in all likelihood Catullus wrote Sarapin (cf. the similar split between Serapin and Sarapin at Mart. 9.29.6). More generally, Housman, A.E., ‘Greek nouns in Latin poetry from Lucretius to Juvenal’, Journal of Philology 31 (1910), 236–66Google Scholar provides ample evidence of Greek forms being ousted by Latin ones in the course of transmission. While the majority of Housman's examples concern case endings, the pair Hecuba – Hecabe is similar to cupressus – cyparissus: as he points out (261), in Ovid's Metamorphoses, Seneca's tragedies and the Ilias Latina, the Latin form has systematically replaced the Greek one even when the latter is required by the metre. Housman ([this note], 266) also notes an exception that confirms the rule: at Juv. 7.11 the much rarer Latin word tripedes has in most manuscripts given place to the more familiar Greek word tripodes. Although in our case the available editions report no variants for forms of cyparissus at G. 2.84, Aen. 3.680 or Val. Fl. 7.405 (the only contexts in classical Latin poetry), such a variation is attested in Vida's Christiad at 1.621 coniferis cyparissis, where a manuscript predating the first edition, possibly the dedication copy, reads cupressis (which, given Aen. 3.680 coniferae cyparissi, can hardly be the original reading): see J. Hankins, ‘Variants of the first redaction’, in J. Gardner (ed.), Marco Girolamo Vida: Christiad (Cambridge, MA, 2009), 385–90, at 385. Finally, Bentley, R., ‘Emendationes Bentleii in Ovidium’, CJ 19 (1819), 168–77Google Scholar, at 175 may be right in conjecturing cyparissifero for cupressifero at Ov. Her. 9.87 on the basis of cuperisifero read by manuscript 598 at Trinity College, Cambridge; if he is right, then also at Ov. Fast. 5.87 cyparissiferae should be read for cupressiferae.