The example of the snake cut into pieces, quoted above, comes roughly in the middle of a series of some thirty proofs (3.417–829) presented by Lucretius to demonstrate that the soul is mortal. Within this series the snake example belongs to a proof that is intended to show that because the soul is diffused throughout the whole body and, as a result, is structurally interconnected with the whole body, it is divided when the body is divided (3.634–6), and is therefore mortal (3.667–9). Within this proof, which also includes as supporting evidence the behaviour of limbs hacked off in battle (3.642–56), the serpens discisa serves as the climactic illustration, signalled by the introductory phrase quin etiam, which will persuade the reader that since the soul, along with the body, can be divided into pieces, the soul is therefore mortal, according to the principle that what can be divided must be mortal (3.640–1). Even more effectively than the immediately preceding passage (3.642–56), in which the poet graphically depicts how human limbs and the head show signs of uitalis sensus after they have been lopped off, the cutting up of the snake brings home the point that the soul is coextensive with the body and an integral part of it because the one snake is divided into many writhing pieces, each with a portion of its anima, which continue to exhibit behaviour characteristic of the intact snake. Thus it provides multiple proofs, so to speak, that the soul is divided together with the body, since the many writhing parts of the snake's body (in multas partis, 3.659) reveal that its soul has been cut into many parts (in multas … partis, 3.669), whereas in the human examples there is a limit of just two pieces, the severed appendage and the body. Moreover, the human appendages show modest signs of the continued presence of the anima as they lie on the ground; a limb quivers (3.642–3), toes twitch (3.652–3), and a severed head keeps its eyes open (3.654–6). But the snake's instinct to attack and bite, even after it has been cut into pieces, now directed against itself, gives a startling illustration of the vital presence of anima.

The mutilation of the snake provides compelling evidence that the soul and the body form an interconnected structural complex. The verbal complex, however, in which this serpens is articulated, has long been a problematic one. At the heart of the problem is the meaning of serpentis utrumque, a phrase which has been treated with considerable indulgence and is printed in the majority of twentieth-century editions, though it does not yield a satisfactory sense.Footnote ⁴ It is usually interpreted to mean ‘both parts of a snake’, as if utrumque serpentis were equivalent in meaning to utramque partem serpentis. The word ‘parts,’ however, is an evasion of the semantic value of utrumque because it eliminates the ambiguity, in this context, of the pronoun ‘each of two,’ the reference of which should be made clear by the context, and supplies instead the very thing that is in question here, a clearly defined object, ‘both parts,’ for discidere. This may seem a small point but ‘both parts’ greatly obscures the nature of the problem. If we take a more literal approach to utrumque, we will get a better sense of the frustrated linguistic expectation caused by the pronoun: ‘of a snake with a darting tongue, quivering tail, long body, to cut up each of the two’. The question immediately arises: to what does ‘each of the two’ refer?Footnote ⁵ According to the normal usage of uterque the answer should be apparent. In 3.658 it is not. It has long been assumed that utrumque refers to cauda and corpore but such a reference is not at all clear from the syntax. In the description of the snake we do not find, and this is the essential point, two clearly defined components of the snake to which utrumque (‘each of the two’) can refer in accordance with its meaning and the syntax of the sentence.Footnote ⁶ Instead, we find three components, expressed in three parallel ablative phrases, uibrante lingua, micanti cauda and procero corpore, all of equal importance in delineating the snake.Footnote ⁷ And since the whole construction is dependent on one verb, discidere, the normal expectation would be that, whatever words are the antecedent of utrumque, those words would be in the accusative as well; the shift from cauda and corpore in the ablative to utrumque in the accusative, in what is essentially an appositional relationship, is syntactically jarring.

Furthermore, even if we assume for the sake of argument that cauda and corpore are the two components of the snake referred to by utrumque, we are confronted with an extremely awkward construction in which utrumque in the accusative is referring back to two nouns in the ablative and all three are governed by the same verb. In this syntactic environment utrumque is essentially a bizarre tautology, i.e. the specification of parts already specified, for caudam and corpus, the accusatives one would certainly expect if the poet did in fact want to single out these two anatomical features for cutting. One might as well say: canis rostro procero, uibranti cauda, uilloso corpore utrumque permulcere.Footnote ⁸

In addition to the objections already presented, utrumque is inappropriate to the poet's argument. The series of descriptive ablatives is not an enumeration of parts that divides the snake into two segments, whether they be thought of as cauda and corpus, or cauda and caput (if lingua is standing in for that part of the snake's anatomy), or corpus and caput; rather, the ablatives depict a snake that exhibits through its movement the presence of its anima (uibrante, micanti) and underscore, in their steady accumulation culminating in the six syllables of procero corpore, the impressive length of the snake extending over two lines.Footnote ⁹ The picture presented by the poet is that of a long snake that has movement, not one divided into two parts, which will then be cut into many parts. As Kenney rightly observes in his commentary, there is little sense in having a two-stage division, indicated first by utrumque and then by in multas partis, since, in order to make his point most effectively, the poet should emphasize the division of a whole into many pieces in a single stage.

A new twist was added to the interpretation of serpentis utrumque by Diels in his translation of the poem. In his view utrumque refers to the snake's body and soul (‘Beides, den Lieb wie die Seele’). Diels's interpretation was subsequently defended by Nicoll who emphasized that the passage itself, with its examples of the soul being cut away with parts of the body, naturally points in that direction. This view, however, results, as Nicoll himself was aware, in a conspicuous petitio principii; for the conclusion that portions of the anima are cut away along with portions of the body is incorporated into the premise of the example, namely cutting into pieces the body and soul of a snake. With Diels' interpretation the poet is saying that if one cuts up the body and soul of a snake, then its soul is divided along with the body. The conclusion, however, that the soul is also being cut into pieces along with the body is what the poet seeks to demonstrate and cannot be assumed at the outset. In order to reach that conclusion the demonstration requires the description of the behaviour of the individual pieces because, after the snake has been cut up, the movements of the individual pieces indicate the presence of the anima in those pieces. To say at the outset that both body and soul are being cut up is to undermine seriously the logic of the proof. Moreover, the verbal parallelism, invoked by Nicoll, of utrumque and in multas partis in 668–9 does not help to elucidate the meaning of utrumque in 658 since, as was mentioned earlier, utrumque in 668 clearly refers to anima and corpus, which are both mentioned in 666–8, while utrumque in 658 has no such reference in its vicinity.

The evidence of meaning, usage and the logic of the argument shows that utrumque is indefensible. Out of respect for the Latin language and Lucretius's poetry editors who will stubbornly persist in printing it must do so by obelizing it.

Numerous attempts have been made to supply a direct object for discidere other than the transmitted utrumque, all of which, with one exception, are open to the objection that they specify a part of the snake rather than the indispensably whole serpens required by the logic of the proof. The one exception is serpentem, printed by CandidusFootnote ¹⁰ and attributed to Marullus.Footnote ¹¹ It is astonishing that the convincing result of so slight a change should have had such a negligible impact on the editorial tradition.Footnote ¹²Serpentem supplies the much needed and undivided direct object of discidere, restores proper emphasis to the description of the snake by directing attention to the animate creature itself, and opens the way for dealing with the corrupt utrumque. I will have more to say about serpentem later on. But first, since improbable conjectures have diverted the attention of critics for so long from the simplicity of this partial solution, it will be necessary to review the weaknesses of these various proposals. The results of that review will show just how indispensable serpentem is.

Perhaps the earliest attempt to replace utrumque as the object of discidere is caudam, a conjecture attributed by Lachmann to Avancius, editor of the first Aldine edition (Venice, 1500).Footnote ¹³Caudam was adopted by Lambinus, along with the additional changes of minanti to Pius's minantis and of utrumque to utrinque (Candidus): minantis | serpentis caudam, procero corpore, utrinque. As Lambinus explains in his commentary, procero corpore is an ablative of description dependent on caudam, i.e., caudam procero corpore = caudam proceram. The ‘long-bodied tail’ of a snake is a bit of an anatomical puzzle and does not provide a persuasive substitute for the whole creature that is to be divided into many parts.Footnote ¹⁴

Another possibility for a direct object in 658 was suggested by E. Goebel who changed procero corpore into procerum corpus: minanti | serpentis cauda procerum corpus utrimque.Footnote ¹⁵ Goebel's conjecture provides a better direct object than caudam but it spoils the parallelism of the asyndetic series of ablative phrases; by making procerum corpus the salient feature, the details of darting tongue and quivering tail appear to be an afterthought to the length of the body. Moreover, utrimque, meaning ‘at each end’ is a rather odd specification in connection with procerum corpus. It suggests that the snake is not in fact to be cut along its entire length but rather in the area of the tail and the head.

Two other attempts were made to specify what part of the snake is to be cut up. Both attempts are modifications of a solution first proposed by A. Brieger, who introduced the noun truncus, destined to be popular as noun or adjective among Lucretian critics, into the repertoire of conjectures for utrumque. Brieger, although he followed Lachmann in reading serpentem, reacted against his changes to the remainder of the line, regarding them as a needless complication of an already difficult passage, intelligible only with the help of his commentary. Brieger rewrote the line as follows: micanti | serpentem cauda, procero corpori’ trunco; at the same time he suggested that serpentis could be interpreted as an accusative plural, indicating that the experiment could be repeated.Footnote ¹⁶ The noun truncus is a strange word to use of a limbless creature and does not belong to the vocabulary of poetic descriptions of snakes; it will at best serve to differentiate the body of the snake from the head but beyond that it is not a useful term to specify a segment of a long cylindrical body. Stranger still is the phrase procero corpori' trunco as a periphrasis for procero corpore. Why the specification truncus? And what is gained by describing the snake in this way? Brieger explains that truncus means the body in contrast to the head and the tail. But that meaning is already clearly conveyed by corpus. This pedantic articulation of the serpens spoils the effect of the poet's description by turning it into an anatomical chart: caput, truncus, cauda. Lucretius presents the reader with a snake that is long and full of movement. In addition, trunco, in the company of procero corpori’, almost seems an oxymoron; it suggests a particular segment when the image is one of full length. In the bloody arena of serial dismemberment and mutilation in this group of proofs, truncus will immediately connote that the snake has already in some way been truncated.Footnote ¹⁷

F. Susemihl, Brieger's co-author, offered, with considerable hesitancy, a modified version of Brieger's conjecture in a note appended to their paper. He turned procero corpori’ trunco into the direct object of discidere, procerum corpori' truncum.Footnote ¹⁸ The same objections to Brieger's proposal, i.e. a periphrasis for a meaning that is already clearly expressed and a premature truncation of the body, apply here as well, with the added difficulty of the contorted double genitive, serpentis and corporis, one of which is accompanied by two ablatives of description, ‘the long trunk of the body of a snake with a darting tongue and quivering tail’.

And finally, yet another twist on the snake's truncus was devised by C. Giussani; he introduced one change into the line, truncum for utrumque, reading micanti | serpentis cauda, procero corpore truncum. The only improvement here, if one can call it an improvement, over the proposals of Brieger and Susemihl, is that Giussani's truncum involves fewer changes to the transmitted text. In terms of sense, however, the reader is once again confronted with a conjecture that specifies a part of the snake, whatever part the ambiguous truncus may denote, rather than the whole it.Footnote ¹⁹

The most serious shortcoming of all the proposals discussed above is that they are far surpassed by serpentem printed by Candidus; it produces excellent sense, it is a very economical correction, and it focusses critical attention where it belongs, on the corrupt utrumque. Adopting serpentem as direct object of discidere, I base the following discussion on three hypotheses: (1) As Munro (1860) observed, utrumque is an error that crept into the text from 3.637 or 668; in both places the word occurs in proximity to corpore (639 and 665). Therefore the ductus litterarum is of no help in finding a solution.Footnote ²⁰ (2) When utrumque displaced the genuine word at the end of the line, the original reading serpentem was altered to serpentis to restore sense since utrumque was now taken to be the object of discidere. (3) The word displaced by utrumque can be recovered from descriptions of snakes in hexameter poetry, a source of evidence that has been under-utilized in the analysis of the Lucretian snake.

The first example comes from Virgil's Georgics where we find an articulation of the snake according to three anatomical features, a head (caput), a coiling body (nexus), and a tail (cauda). Here is the poet's description of the poisonous coluber as it retreats from the shepherd's attack into a hiding place (G. 3.422–4):

The passage, with its emphasis on sinuous movement (nexus, soluuntur, sinus, orbis), highlights an essential characteristic that is missing from the transmitted text of 3.657–8, the coiling of the snake's body. After all, what is the hexameter snake without reference to its coils? To take a poet who was Lucretius's older contemporary, Marcus Cicero, in his translation of Aratus's Phaenomena, a poem with which Lucretius was familiar, describes how the constellation Ophiuchus grips a massive snake while he himself is entwined in the coils of its body (corpore torto):

In two other translations, taken from Greek tragedies, Cicero introduces the detail of the coils, not found in the source texts, to enhance the fearsome aspect of a snake or serpent-monster. In his version of Il. 2.299–330 Cicero expands Odysseus’ description of the snake that devours nine nestlings and their mother, a portent that signifies the length of the war at Troy (Il. 2.308–320), by adding that the creature was ‘terrifying with its coil’ (tortuque draconem | terribilem, 11–12, Soubiran p. 266). Likewise in his version of Sophocles, Trachiniae 1046–102, where Hercules lists the creatures he vanquished, the serpent, described in the Greek simply as ‘the guardian of the golden apples’, becomes in Latin ‘a serpent with multiple coils’ (42–3 tortu multiplicabili | draconem, Soubiran p. 275).Footnote ²¹

In this group of parallels the phrase corpore torto stands out for its sound pattern, which seems to have commended itself to the poets for its threefold repetition of the syllable ‘-or-’, perhaps regarded as imitative of a snake's coiling, sinuous movement. Manilius does Cicero one better in his description of the snake held by Ophiuchus, juxtaposing references to the bodies of the snake and Ophiuchus to emphasize how the two figures are entwined:

Here we have a fourfold repetition of the ‘-or-’syllable. In Aen. 5.276 Virgil uses the noun tortus, rather than the participle of torqueo, together with corpore, to achieve the same sound effect: serpens longos fugiens dat corpore tortus. And Ovid, in a variation on the phrase, speaks of the ‘coiling breast’Footnote ²³ (Her. 12.102 torto pectore) of the serpent that guarded the golden fleece.

It follows from these examples of ‘coiling’ as a formulaic element in the description of snakes that the serpens in 3.658 has ‘a long, coiling body’, procero corpore torto:Footnote ²⁴

The image is that of a snake with a darting tongue, quivering tail and a long body that moves with a winding motion; a live snake would not be a stationary target for its human assailant. The double epithet, procero and torto, consisting of an adjective and participle, is very much in Lucretius’ manner.Footnote ²⁵ Here then are the features one would expect in the delineation of the serpens before it is chopped up into many pieces. The poet directs attention first to the darting tongue, then to the quivering tail and then to the long, coiling body. And for each element he specifies the type of movement characteristic of it (uibrante, micanti and torto), a detail which will be of special importance after the snake has been cut into pieces because the focus on the movements of its body before division highlights the movements of its parts after division. Thus the emendation torto provides a characteristic of the snake's movement that is regularly found in poetry and results in a sound pattern that has been identified as one favoured by the poets in such descriptions; in addition, the verb tortari in 661 will now not only be an echo of torto in 658, it will also be evidence that the pieces of the snake continue its natural form of movement (corpore torto) and must therefore contain a piece of its anima since the anima is the source of motion.Footnote ²⁶