² Lachmann's micanti was adopted by Munro (1864), Brieger, Giussani, Heinze, Bailey (1900) and Duff; it later fell out of favour with editors and is not even mentioned in the apparatus of Flores. A reconsideration of its merits is in order. It is corroborated by descriptions of the tail's movement in Nic. Ther. 229 (σκωλύπεται) and 476 (μαστίζων) and in Ap. Rhod. Argon. 4.1402 (σκαίρεσκεν). In Latin we find references to the tail's lashing (Ov. Met. 3.94 flagellari; and Stat. Theb. 5.538 uerbere). Even if these verbs, with the exception of σκαίρεσκεν, denote movement more violent than micare, they still represent movement of the same type. Cf. also Cic. Marius fr. 3.4, [anguem] micantem (Soubiran p. 249 = Courtney fr. 17.4, p. 175 = Blänsdorf fr. 20.4, p. 166). At German. Arat. 592–93, anguis | ultima cauda micat, it seems unlikely that anything more is meant than ‘gleams’, of starlight, although the combination is suggestive. Above all, micanti denotes movement, indicating the presence of the anima; see the comments of Brown on 657–660; it also suggests the movement of the tail after, as well as before, being cut; cf. Enn. Ann. 483–4 Sk., a severed head, and Verg. Aen. 10.396, the fingers of a severed hand; minanti, in this context, is imprecise by comparison. Emphasis on the snake's length and movement in the description, as I will explain later on, contributes much more to the proof than the detail of a threatening posture. The verb minari would properly be used of the head and forepart of the snake when upraised and poised to strike (Verg. G. 3.421), for that is where the threat exists; compare Sil. Pun. 6.278. The attempt to interpret minanti as meaning ‘raised up and menacing’, with a reference to Verg. Aen. 1.161–2, uastae rupes geminique minantur | in caelum scopuli (Ernout–Robin, Bailey [1947])Google ScholarPubMed overlooks the fact that in the Virgil passage the perspective is that of an observer who sees something looming far overhead. Our snake's tail, presumably, is not that long. In addition to the explanations found in the commentaries, minanti is defended by Barrett, D.S., ‘Lucretius’ snake, DRN 3.757–9’, LCM 15 (1990)Google Scholar, 144 and Wellesley, K., ‘Lucretius’ snake dissected’, LCM 16 (1991), 41–2Google Scholar, who adopts Pius's conjecture minantis modifying serpentis.

³ My translation is based on those of Bailey (1947) and Smith. In 662 ipsam is generally taken as modifying se but this interpretation runs counter to standard usage. In the intensifying combination of a form of ipse and the reflexive pronoun, the form of ipse normally modifies the subject of the verb, not the object, to emphasize that the subject is acting upon itself in an unusual or unexpected way. Here ipsam modifies partem the subject of petere and represents ipsa of the direct statement ipsa … pars … prior. Compare 4.639 [serpens] sese mandendo conficit ipsa. In 663 Kenney rightly points out that morsu premat means ‘bite’, with the reflexive understood as direct object, not ‘assuage’ its pain, as it is sometimes translated; see also Heinze on 663. The poet's language seems to express a grim satisfaction at the snake's pain; uolneris ardenti … icta dolore might describe a human victim of a snake's bite.

⁴ utrumque is printed in Merrill (1907) and (1917), Diels, Ernout, Martin, Bailey (1947), Valentí, Orth, Büchner and Smith; and now it has a foothold in the twenty-first century in Flores. With the notable exception of Heinze, who printed utrumque, most nineteenth-century editors after Lachmann followed him in replacing it with conjectures (Bernays, Brieger, Giussani and Bailey [1900]). Heinze cautiously conjectured utramque in the apparatus, meaning utramque partem, on the hypothesis that lines 642–56 on the wounds inflicted by scythe-bearing chariots were a later addition by the poet and that, in the passage's original form, the meaning of the conjectured utramque, coming six lines after partem … utramque in 637, would have been clear. The hypothesis of an original structure in which line 641 was immediately followed by line 657 and the assumption that utramque can mean utramque partem still leave unexplained what parts are identified by that word and why these parts are specified before the division.

⁵ The translation ‘each of the two’, rather than ‘both parts’ is intended to preserve the pronominal function of the word and to bring out the abruptness of the shift, in both sense and grammar, from three nouns in the ablative specified as attributes of the snake to a pronoun in the accusative denoting two of those attributes and functioning as the object of discidere. There is no evidence that utrumque can mean utramque partem, a meaning which misleadingly clarifies the reference of the pronoun by substituting the general term ‘parts’ for the two things denoted by ‘both’, whatever they may be.

⁶ When there is a genitive dependent on a form of the pronoun uterque, it is a normally a demonstrative pronoun, or relative pronoun, which may be accompanied by its antecedent in the same case, or, much less often, a noun; the genitive expresses the two things referred to by utrumque; e.g. utrumque eorum, ‘each of these two’, i.e. animus and anima (3.421); cf. quorum utrumque (1.758, 975, 2.565, 5.1101, 6.365). The phrase serpentis utrumque is highly unusual because the genitive expresses not the two things referred to by utrumque but to their source, ‘each of the two [i.e. tail and body] of the snake.’ The oddity of serpentis utrumque lies in the combination of a pronoun with a noun in the genitive that is properly part of the pronoun's own antecedent; if utrumque refers to the tail and body of the snake, then serpentis properly belongs with cauda and corpore, not with utrumque. The normal form of expression would be serpentis micantem caudam et procerum corpus utrumque discidere; here utrumque is in apposition to caudam and corpus. Lucretius uses a similar construction at 3.472 and 6.499. Apart from utrumque in 3.658, the only other problematic occurrence of the word is utraque in 6.517–18: sed uemens imber fit, ubi uementer utraque | nubila ui cumulata premuntur et impete uenti; for different views see Lachmann, Munro (1886⁴), and Bailey (1947). My own view is that utraque refers to ui and impete uenti, with utraque attracted into the gender of the nearer noun; ui itself means the pressure exerted by the clouds on themselves when massed together, a phenomenon described at 6.510–12 where that pressure and the wind's onslaught are identified as the two causes (dupliciter 510) of rain (cf. 6.734). I translate ‘when the clouds, after massing together, are violently pressed by each one, their own force and the wind's onslaught’.

⁷ It is sometimes suggested that utrumque at 3.668 and partem utramque at 3.637 are parallels that support the use of utrumque in 3.658; see Ernout and Robin and Smith. Neither one is relevant to the problem. The meaning of utrumque at 3.668 is unmistakably clear. In 667–8 the poet twice mentions anima and corpus; when he concludes in 668 that ‘each one of the two’ (utrumque) must be thought to be mortal, the reference of utrumque is clear and precise, i.e. anima and corpus. In 3.633–9 Lucretius uses the adjective, not the pronoun. He refers to the two parts of a body that has been cut in half (medium, 3.636) as partem utramque (637); one body, divided at the middle, yields two parts. The poet's clarity of expression in these instances only highlights the obscurity of serpentis utrumque.

⁸ Munro (1886⁴), alone among supporters of utrumque, recognized the inconcinnity of the genitive serpentis dependent on utrumque. In his translation he treats the ablative phrases as an ablative absolute construction with serpentis dependent on lingua, and, by extension, on cauda and corpore as well, to which utrumque refers. Although this solution creates problems of its own, it does show a regard for the proper use of the pronoun utrumque. In his text Munro adopted a different solution, positing a lacuna after 658 and offering the following supplement in the apparatus, et caudam et molem totius corporis omnem, which neatly provides a clear reference for utrumque. One may find this a radical solution but that does not diminish its value for showing that utrumque stands in need of clarification; it also shows how utrumque creates tautology of expression. Munro's successors have failed to appreciate the diagnostic importance of his proposal.

⁹ Since the point to be proved is that the anima, diffused throughout the whole body, undergoes division with it, the proof begins with strong emphasis on the wholeness of what is to be divided (3.634 toto corpore, 3.635 totum esse animale).

¹⁰ Candidus' text reads: quin etiam tibi, si lingua uibrante minacem, | serpentem, et caudae procero corpore, utrinque [sic]. Lachmann adopted in his text serpentem and utrimque, adding micanti (see n. 2 above) and cauda e, for caude, the reading of O before correction and Q. His version of 657–8 reads, lingua vibrante, micanti | serpentem cauda, e procero corpore utrimque, which he interpreted as follows: utrimque e procero corpore, id est ex utraque parte, lingua vibrat, micat cauda, i.e. a snake with a darting tongue and a quivering tail at each end of its long body. The weakness of utrimque, construed with the prepositional phrase e procero corpore, lies in the superfluous detail of specifying the location of the tongue at one end of the body and the tail at the other end; the poet's use of lingua and cauda makes the disposition of parts perfectly clear. In Lachmann's version the poet fussily specifies that the flickering tongue is found at one end of the snake and the darting tail at the other. Needless to say, too high a price is paid to retain a form of uterque. Vtrimque merely repeats, in adverbial form, the problem of utrumque. Those who follow Lachmann in reading e procero corpore but retain the transmitted utrumque interpret the prepositional phrase differently. Munro (1864) takes it with micanti: ‘as its tail is darting out from its long body’; Merrill, W.A., ‘Notes on Lucretius’, UCPCP 3 (1918), 265–316Google Scholar, at 292 regards it as an ablative of separation: ‘Both ends of the snake are chopped from the long body’; Richter, likewise W., Textstudien zu Lukrez (Munich, 1974), 49–50Google Scholar, who makes it dependent on his conjecture, caudam … truncam, i.e. the tail cut away from the body.

¹¹ The attribution of conjectures on the De rerum nat. to Marullus, primarily on the evidence of Candidus, has been called into question because his conjectures have been found in the marginalia or interlinear notes of certain manuscripts that are not directly linked to Marullus by any written testimonia in the books themselves, such as an owner's inscription testifying to Marullus as the source of the marginalia, and, further, these marginalia and interlinear notes are said to antedate the period of Marullus's work on the text of Lucretius and therefore cannot be his property; see Flores, vol. 1, ‘Introduzione’, 10–14 and vol. 3, ‘Introduzione’, 13–16. Flores, who has eliminated the name of Marullus from the apparatus, assigns both serpentem and utrimque to the marginalia of two manuscripts, Laurent. 35.25, written before 1450 (D), and Monac. Lat. 816, written before 1475 (I), and to the text of a third, Ambros. P 19 sup., written after 1475 (Ea). See also Reeve, M.D., ‘The Italian tradition of Lucretius’, IMU 23 (1980), 27–48Google Scholar, at 44–8, and ‘Lucretius from the 1460s to the 17^th century: seven questions of attribution’, Aevum 80 (2006), 165–84Google Scholar, at 169–71. More recently, however, Brown, A., The Return of Lucretius to Renaissance Florence (Cambridge, MA and London, 2010), 113–22Google Scholar has attributed the corrections in Laurent. 35.25 and Monac. Lat. 816 to Marullus. Important questions concerning the identification of the hand in which the marginalia are written and the date when they were entered into these manuscripts remain to be settled before there can be a definite determination that they are the intellectual property of someone other than Marullus. In my view, for the purpose of understanding the impact of the conjectures serpentem and utrimque on the editorial tradition of the poem, the most useful historical statement that can be made is to attribute both to the primary agent of their diffusion among readers of the poem, i.e. the edition of Candidus 1512, where they make their first appearance in print.

¹² Editions published after Lachmann with serpentem and utrimque, Bernays and Brown: with serpentem, Kenney, who obelizes utrumque, and Müller, who prints in the text his conjecture toruum for utrumque and records in the apparatus a second suggestion, taetrum; both conjectures strike me as purely ornamental epithets in this context since neither the snake's savage disposition nor repellent aspect is in point here but rather the length and movements of the snake's body. Müller's conjectures raise the question of the gender of serpens. At 4.60 and 4.638 serpens is unambiguously feminine; at 5.33, asper, acerba tuens, immani corpore serpens, it is unambiguously masculine. In the case of Müller's toruum, it could just as well be toruam, since there is no clear sign of the gender in the passage, unless icta in 663 is taken to refer to the snake rather than specifically to priorem partem in 662.

¹³ Modern editors attribute caudam to Lambinus; Lambinus himself says that he found it in certain manuscripts and printed editions.

¹⁴ Caudam as object of discidere was revived by Kannengiesser, A. (review of Nencini [n. 19 below], BPhW 15 [1895], 1132–4, at 1134)Google Scholar: minanti | serpenti caudam e procero corpori’ trunco; similar is the proposal of MacKay, L.A., ‘Notes on Lucretius’, UCPCP 13 (1950), 433–45Google Scholar, at 440–1: minanti | serpenti caudam e procero corpore truncam; W. Richter's proposal (n. 10 above, 48–50) essentially repeats MacKay's: minantis | serpentis caudam e procero corpore truncam. On the shortcomings of Richter's solution see the review of Kenney, E.J., CR 26 (1976), 180–1Google Scholar, at 181. The appearance of the adjective truncus in all three of these proposed solutions represents a modification of a suggestion first made by A. Brieger, on which see below.

¹⁵ Quaestiones Lucretianae (Salzburg, 1857)Google Scholar, 25.

¹⁶ Susemihl, F. and Brieger, A., ‘Bemerkungen zum dritten Buche des Lucretius’, Philologus 27 (1868), 28–57CrossRefGoogle Scholar, at 50–1. Bockemüller appears to be the only editor who reads serpentes as the object of discidere.

¹⁷ Cf. 3.654 calido uiuenteque trunco; see also 3.404.

¹⁸ Susemihl and Brieger (n. 16), 57 n. 21.

¹⁹ More ambitious versions of the line have been proposed. Wellesley (n. 2), 41–2, working on the assumption that the poet was describing a snake with its forepart upraised, rewrote the line as follows: minantis | serpentis scandens procero corpore dorsum. Working with a similar image of the snake's upraised posture, Nencini, F., ‘Emendationum Lucretianarum Spicilegium’, SIFC 3 (1895), 205–24Google Scholar, at 216–17 combined conjecture with a convoluted construction of the Latin: minenti serpenti cauda e procero corpore utrumque. Rebac, A. Savic, ‘Lucretiana’, Ziva Antika 1 (1951), 102–9Google Scholar, at 106–7 proposed the unmetrical minanti | serpens it cauda procero corpore unumque. The complexity of these proposals underscores the syntactic clarity and simplicity achieved by reading serpentem.

²⁰ It may not be too far afield to suggest that the pair of famous snakes at Verg. Aen. 2.214, serpens … uterque, exercised a surreptitiously corrupting influence on the wording of the line as well.

²¹ Additional examples of the coiling snake in poetry, celestial or terrestrial, expressed with the verb torqueo or the noun tortus are found in Cicero, [anguem] intorquentem (Marius 3.5, Soubiran p. 249; Courtney fr. 17, p. 175; Blänsdorf⁴ fr. 20, p. 166); Quintus Cicero, on the zodiacal constellations 17, squama … torta Draconis (Courtney p. 179, Blänsdorf⁴ p. 185); Catullus 64.258 tortis serpentibus; Varro Atacinus, torta … angue (Courtney fr. 23, p. 253 = Hollis fr. 130. p. 176 = Blänsdorf⁴ fr. 9, p. 235); Verg. G. 3.38 tortos Ixionis anguis, Aen. 5.276 longos … tortus, 12.481; Hor. Carm. 2.13.35–6 intorti … angues; Ov. Her. 12.102 torto pectore, Met. 2.138 tortum … ad Anguem, 3.42 orbes torquet, 4.483 tortoque angue; German. Arat. 593 tortus … timendos; Aetna 46–7 per orbes … intortos, Sen. Med. 961–2 anguis … tortus.

²² The text is quoted from Goold, G.P., M. Manilii Astronomica (Stuttgart and Leipzig, 1998 ²)Google Scholar. In 332 torto is Scaliger's certain emendation for the transmitted toto; see Housman's, A.E. note in M. Manilii Astronomicon Liber I (Cambridge, 1937 ²).Google Scholar

²³ See n. 24.

²⁴ Since the perfect passive participle torto is here used as an attributive adjective it can be taken as the equivalent in meaning of a present intransitive participle, i.e. ‘coiling’ or ‘winding’ rather than ‘having been coiled’. Lucretius has the same usage at 1.293 uertice torto, ‘whirling eddy’, which is followed by a synonymous phrase in 294 with a present participle, rotanti turbine; cf. Verg. Aen. 7.567, torto uertice. In four of the passages cited above in n. 21 the perfect participle tortus clearly functions as a present: the fragment from Varro Atacinus, Ov. Her. 12.102, Met. 4.483, and Sen. Med. 961–2; similarly intortus at Hor. Carm. 2.13.35–6 and Aetna 46–7. In the passages quoted above from Cicero's Aratea, fr. xv.1–2 and Manilius, 1.331–2, torto is a true perfect passive participle.

²⁵ See the collection of examples in Munro (1886⁴), on 1.258 and 5.13.

²⁶ In 661–3 the sound pattern of the ‘-or-’ syllable continues to underscore the snake's sinuous movement after it has been cut up: sor-, tor-, or-, -or-, mor- and -or-. The material presented in this paper will suggest other possibilities for conjecture. One such possibility is the participle tortam/-um in agreement with serpentem; the result is similar in sense to torto in agreement with corpore. In my view, however, tortam/-um interrupts the syntactic momentum of the three ablatives of description, in each of which, if torto is read, a participle describes a type of movement appropriate to each part of the snake as it is mentioned, lingua uibrante, micanti cauda and procero corpore torto. Another possibility is to retain the transmitted serpentis and read the noun tortus, either in the accusative singular tortum or the accusative plural tortus. The shortcoming of this approach is that the specification of the coils as the object of discidere again results in an expression that substitutes a part of the snake for the whole. I want to thank the editor, Bruce Gibson, and the anonymous reader for comments, criticisms and corrections that led to the improvement of this paper.