21.1 Introduction

The style of De rerum natura is perhaps more varied than that of any other classical Latin poem. Taking the poem on its own terms, we may interpret this variety as part of Lucretius’ attempt to bring the full range of poetic resources to bear on his didactic enterprise. Here I am concerned with just one aspect of that varied style, namely Lucretius’ use of the language of early Latin, as well as with its range of possible effects on his readers. By ‘early Latin’ I refer not to a real stage in the history of the language but rather to a certain linguistic and cultural construct: a set of features (phonological, morphological, syntactic, lexical) associated by Lucretius and his readers with the styles and languages of an earlier period, and familiar to them primarily from said earlier period’s surviving texts (literary, dramatic, legal, monumental). Most if not all of such features were optional alternatives to the more regular classical forms of the contemporary literary language. The extent to which a poet selected markedly ‘early’ forms over more regular classical alternatives is the extent to which that poet’s language may be called archaising.

In Lucretius, metrical considerations often play a role in the selection of early forms at a given point in the line (see section 21.2). The metrical usefulness of these forms, however, does not in itself explain Lucretius’ decision to adopt them as a central feature of his Kunstsprache. Lucretius was contemporary with the neoteric circle that included such figures as Catullus and Gaius Helvius Cinna, with whom he shared a mutual acquaintance in the form of his patron, Gaius Memmius. When it came to establishing a poetic style, however, Lucretius reached not to his contemporaries, but back into the past, taking Ennius – qui primus amoeno | detulit ex Helicone perenni fronde coronam, | per gentis Italas hominum quae clara clueret (1.117–19) – as his primary Latin poetic model. An Ennian epic style (see the material gathered by Pullig Reference Pullig1888 and Nethercut Reference Nethercut2021) was a wise choice for a philosophical radical hastening the Epicurean invasion of Italy (Cic. Tusc. 4.7 Italiam totam occupauerunt, with Harrison Reference Harrison2002: 9). Noster Ennius was the national poet, his Annales the epic of the Roman Republic. Adopting an Ennian style lent authority and familiarity to a prima facie unpalatable doctrine (1.935–50), as well as reanimating the aesthetics of a national classic just as the Republic was entering a state of crisis (1.29–30, 41–3; on linguistic archaism and national identity see Munro Reference Munro2013: 22–6). We would be wrong, however, to understand all early Latin features in De rerum natura as Ennianisms. Lucretian archaism is not simplex, either genetically or stylistically: sacral language, the language of Roman law, and of the tragic and the comic stage all have a role to play. My intention in what follows is not to provide a full overview of early Latin elements in the poem (for which see Bailey Reference Bailey1947: 1.72–108), but rather to focus in detail on a select few archaic features in morphology, syntax and phrasing, asking what contribution they may make to the overall style of the poem and, in some cases, to the particular contexts in which they are found.

21.2 Optional Early Latin Forms and the Hexameter

That Lucretius’ selection of early Latin forms and constructions was not motivated solely by metrical convenience is demonstrated by his use of archaisms that have the same metrical shape as their regular classical equivalents. Take, for example, the use of the indicative mood in causal or adversative cum clauses (K–St 2.349; Bailey Reference Bailey1947: 1.98; Bennett 1.133–5, 141–2): examples such as 1.565–6 solidissima materiai | corpora cum constant (with which contrast 4.460–1: seuera silentia noctis | undique cum constent) and 2.859 quae cum ita sunt (passages in which there is no indication of textual variance) prove an interest in archaism regardless of metre; likewise the use of the early form olle for ille, found ten times in the poem (on which see in this context Wald Reference Wald1968: 161–2). Nevertheless, the option to choose between forms with the same meaning but different shapes is of clear value to a poet composing in metre, and that many optional early forms were selected by Lucretius metri causa can be demonstrated by considering the average distribution of optional forms across the hexameter line. When the distribution of optional forms of a given type is not evenly spread across the line, but is rather clustered at a point where the poet has less freedom to manipulate the metre (e.g. in the fifth foot of the hexameter, usually a dactyl), we may surmise that, overall, metrical concerns did play a role in the poet’s decisions to select that form.

The early Latin passive infinitive in -ier was already on the way to becoming an archaism in the time of Plautus (de Melo in this volume: pp. 103–5). By Lucretius’ time it had become very rare in prose, being used by Cicero only a few times in the set legal phrase bene agier, and otherwise only once, in a letter of Vatinius to Cicero of 45 bc, where the legalism causam dicier is parodied in mock-serious tones, with Vatinius referring to himself loftily in the third person (Fam. 5.9). We find 48 instances of this form in Lucretius, as opposed (Bailey Reference Bailey1947: 1.84) to 468 regular passive infinitives in -i, a proportion of just under 10%; compare de Melo’s figures of 18% and 13% for Plautus and Terence respectively. The form is found once among the extant hexameters of Ennius (Ann. 574 Sk., in the fifth foot), and six times in the surviving lines of Cicero’s translation of Aratus’ Phaenomena (another important influence on the archaic style of Lucretius: see the material collected in Wreschniok Reference Wreschniok1907, Buescu Reference Buescu1966: 331–54; also Gee Reference Gee2013a: 81–109), where in fact it outnumbers regular forms of the passive infinitive by a ratio of two to one. In both Lucretius and Cicero, the form is not used freely across the line, but is clustered towards the end, with the two short syllables of -ier usually completing a dactyl in the fifth foot (Table 21.1).

Whereas the use of dicier in Vatinius’ letter to Cicero is so rare as to suggest a particular archaising purpose (in that case mock-serious lawyerly badinage), these forms are widespread enough in both Cicero’s Aratea (once every 92.5 lines on average) and Lucretius (once every 167 lines on average) to suggest that, besides their metrical value, one of their functions is to contribute to the overall archaic effect of each poet’s Kunstsprache. That does not rule out, however, the possibility of individual examples bearing their own particular expressive function, as determined by the contexts in which they are employed, and indeed there is some limited evidence to suggest that Lucretius selected these forms particularly when concluding arguments or when emphasising argumentative features of particular importance (Castillo Herrera Reference Del Castillo Herrera2011: 55–7; Wald Reference Wald1968: 165–6).

The early Latin form a reader most regularly encounters in Lucretius is the disyllabic genitive singular of the first declension (with the suffix spelled -ai in modern editions), a form already archaic by the time of Plautus (cf. de Melo: pp. 109–11), and by Lucretius’ time entirely confined to verse. Lucretius uses the monosyllable 153 times, and the disyllable 166 times (Bailey Reference Bailey1947: 1.72; see further Wald Reference Wald1968: 167–70 and Nethercut Reference Nethercut2021: Appendix 2). As with infinitives in -ier, we find that these disyllabic genitives cluster towards the end of the line, in this case in the sixth foot. Table 21.2 lists surviving disyllabic forms which can be securely placed in the line (Ennius’ terrai frugiferai (Ann. 510 Sk.), perhaps to be placed at line-end on the basis of its imitation at Lucr. 1.3 terras frugiferentis, is not included).

Notwithstanding the low number of surviving examples in Ennius and Cicero – numbers too low to allow secure statistical conclusions to be drawn – the similarity is noticeable: the evidence suggests that all three of these authors clustered disyllabic genitives in the sixth foot. Lucretius’ reasons for doing so will have differed slightly from those of Ennius. Ennius inaugurated the Latin hexameter and so, with regard to the distribution of given Latin forms within the line, had no direct model from which to work. We may assume two reasons for his placement of disyllabic syllables at the end of the line: first, the provision of a spondee to fill the (relatively inflexible) sixth foot and, second, the creation of rhyming spondaic sound effects of the sort siluai frondosai (Ann. 179 Sk.) and Albai Longai (Ann. 31 Sk.), which may have been deemed particularly effective at line-end. Lucretius (and, no doubt, also Cicero) had an additional motive, namely stylistic allusion. In his emulation of the verse style of his greatest Latin predecessor Lucretius mimics Ennius’ tendency to use disyllabic genitives at line-end. Further connections between Ennian imitation and the use of early Latin forms will be discussed in this chapter, pp. 439–43.

At the same time, and despite this clustering of disyllabic forms in the sixth foot, Lucretius’ overall use of these forms shows a surprising degree of freedom when compared to the use-patterns of his two predecessors. First, Cicero’s use of the disyllabic form is, with one exception (terrai at fr. 34.58), limited in the surviving fragments to proper names (Aquilai once, Nepai five times); Lucretius shows no such limitation, using it with proper names, ordinary nouns and adjectives, and Greek words (Iphianassai, Geryonai, harmoniai). Second, surviving examples of the form in Ennius’ Annales are ‘restricted to the end of the line … and to words in attributive rhyme with the end’ (Skutsch Reference Skutsch1985: 61; note that Skutsch (750–1) follows Housman in attributing Ann. 16 Vahlen Lunai portum, est operae, cognoscite ciues to Ennius’ Satires, in which he is followed by Goldberg and Manuwald). So, the two surviving Ennian examples listed earlier, of disyllabic forms placed elsewhere than the sixth foot, both occur in lines where the sixth foot also contains a disyllabic genitive form: Albai Longai and siluai frondosai (terrai frugiferai, if correctly placed at the end of the line, would count as a further example of this effect). Again, Lucretius shows no such apparent restriction: on the contrary, while most of his uses are clustered at line-end, others are found throughout the line, demonstrating a freedom of placement not evidenced in what remains of Ennius or Cicero. This relative freedom of placement in the latest of the three poets, we may note, is the opposite of what might have been expected. Lucretius appears not only to be emulating the style of his models, but also to be reanimating the early forms which, even by the time of Ennius, had become fossilised.

While early Latin optional forms are often treated as semantically equivalent to their regular counterparts, they may nevertheless bear very different connotations of tone and style. Here is just one example:

Escit is an inceptive verb with existential meaning first found in the Twelve Tables and also used in Cicero’s De Legibus. Here it is used as a metrical variant of erit (compare the use of superescit by Ennius as an equivalent for supererit at Ann. 514 Sk.; see Pezzini Reference Pezzini2015: 243 n.44). That metre played a role in Lucretius’ decision to select escit at 619 is plausible: the penultimate syllable of the hexameter must be long, and accordingly the regular form erit is not an option. The presence of a metrical motive, however, does not undo the archaising effects of this language. Escit, a form which ‘was probably already archaic at the end of the third century bc ’ (Pezzini Reference Pezzini2015: 243, with further bibliography), is a startlingly archaic form to encounter in a poem written during the 50s bc.

The two other surviving examples of this verb make clear what significance it would have held for a mid first century bc reader: si morbus aeuitasue escit, iumentum dato (‘If there is illness or age, he is to provide a yoked beast of burden’) (Twelve Tables 1.3 Crawford; transl. Crawford Reference Crawford1996); ast quando duellum grauius, discordiae ciuium escunt, oenus ne amplius sex menses, si senatus creuerit, idem iuris quod duo consules teneto (‘But when grave war or civil discord arises, let one man hold, for no longer than six months, and if the senate so decrees, the same power as that wielded by two consuls’, Cic. Leg. 3.9). The verb, then, carries with it the authority and gravitas of the law code. Lucretius, as an early composer of philosophical literature in Latin verse, needed a language in which to describe a radical and unfamiliar atomistic conception of reality. One domain of the existing language he drew from in order to describe this world was that of politics, law and diplomacy (hence such well-known expressions as foedera naturai). The early Latin form escit forms part of this pattern: transferring the authority and gravitas of the Roman legal code into Lucretius’ account of the nature and laws of the universe.

21.3 Archaism and Ennianism

Direct imitation of Ennius is one route by which early Latin forms may find their way into Lucretius’ text: see, for an example, 3.1025 lumina sis oculis etiam bonus Ancus reliquit, which lifts the old monosyllabic form sis (instead of regular suis: see Skutsch Reference Skutsch1985: 293–4 and Festus p. 387.14 L.: ‘sos’ pro ‘eos’ antiqui dicebant … interdum pro ‘suos’ ponebant) directly from Annales 137 Sk.: postquam lumina sis oculis bonus Ancus reliquit. Here, however, I am interested in a different question: the effects of direct allusion to Ennius on the selection and interpretation of early Latin forms in the context surrounding that allusion. In such passages, I argue, the presence of Ennian imitation activates the archaic resonances of early Latin forms and constructions which are not always so activated when used elsewhere in the poem.

One example is found in an important Ennianising passage in Book 1. We sometimes find in early Latin texts variation between third and fourth conjugation forms in cases where, in classical Latin, we would expect the third conjugation only, for example fodīrī in Cato, parīrī in Ennius, morīrī in Ennius and Plautus, aggredīrī in Plautus (Leumann Reference Leumann1977: 568). Skutsch (Reference Skutsch1985: 420) identifies ‘fourth conjugation’ forms of cupio at Plautus, Cur. 364 (laudato quando illud quod cupīs effecero) and Ennius Ann. 238 Sk. (alter nare cupīt, alter pugnare paratust). We find another in Lucretius:

The lines are taken from the proem to the first book, where Epicurus is cast in Ennian terms (Graius homo at 1.66 is borrowed from Annales 66 Sk. – see Harrison Reference Harrison2002: 9) as a conquering general, exerting control over the natural world and bringing back knowledge of its limits. Harrison points out that lines 70–1 adapt the description of the opening of the Gates of War from Annales book 7: postquam Discordia taetra | Belli ferratos postes portasque refregit (225–6 Sk.; Harrison Reference Harrison2002, 10). Cupīret is Lucretius’ own addition to the Ennian idea: in Ennius’ narrative, the bursting open of the gates has already happened (postquam … refregit); in Lucretius, it is an idea in the mind of Epicurus (effringere… cupīret). The early Latin morphology of cupīret, then, is not here borrowed directly from Ennius, but added by Lucretius as part of his adaptation of the Ennian original (for the form in Lucretius see also cupītum at 3.770 and 5.847, and contrast cupe˘rent at 5.169; Bailey Reference Bailey1947: 1.86). We have, then, the presence of an early Latin form alongside an Ennian allusion. While such a form in and of itself would not require explanation other than as a general feature of Lucretius’ archaic style, when juxtaposed with a direct imitation of a famous Ennian passage its salience as an archaism of a kind used by Ennius (see earlier on cupīre, parīrī, morīrī) is enhanced. The presence of Ennian allusions activates the archaic resonance of an early Latin form, which in turn adds to the overall didactic effect of the passage – in this case the representation of Lucretius’ hero, Epicurus, in terms borrowed from the discourse of Roman national epic.

For a more complex example we may look to Book 3:

Lines 654–6 have long been recognised as a re-working of two lines of Ennius preserved by Servius (ad A. 10.396):

The remainder of the passage is replete with morphological and syntactic archaisms. While de subito (643, elsewhere in DRN only at 2.265) has a long history in Latin (see Adams Reference Adams2007: 499; Reference Adams2013: 583–4), from a first-century bc perspective it may be termed an archaism, being common in early Latin drama (including tragedy) but rare in classical Latin (see Diels Reference Diels and Diels1922: 47 = 1969: 366; in classical texts outside Lucretius it is found only twice: Cicero Rep. 6.2 and Celsus 5.28.11). Concessive cum with an indicative verb (quit 646) is a feature of early Latin syntax (K–St 2.349; Bennett 1.141–2) found elsewhere in Lucretius at 1.565–6, 1.825 = 2.690, 3.107, 6.140 (see Bailey Reference Bailey1947: 1.98). Although the verb survives only in classical Latin, that the desiderative form petessit (648) would be considered archaic relative to the time of Lucretius is suggested by Festus: petissere antiqui pro petere dicebant, ea quidem forma uerbi, qua sunt lacessere, et incessere; sed, ut mihi uidetur, cum significabant saepius petere; (p. 226.19–23 L.); the verb is found once elsewhere in Lucretius (5.810, also at line-end), twice in Cicero’s poetry (fr. 6 Büchner (Div. 1.17), from De Consulatu Suo 2, also at line-end) and once in prose at Cic. Tusc. 2.62; on the development of forms in -esso see Thomas Reference Thomas1935 and Leumann Reference Leumann1977: 555 (‘nur die seit Plt. oder Enn. belegten’). At 650 we have the unusual syncopated form of the s-perfect abstraxe (Leumann Reference Leumann1977: 598). As demonstrated by Norden (Reference Norden1957: 140–1 ad A. 6.57), while these forms seem to have been considered colloquial by Catullus (who avoids them in Poem 64 but employs them elsewhere), and by Horace (who employs them only in the Satires), they were also capable of carrying an archaic tone, and indeed there appears to have been just such a usage in a high-register passage of Ennian tragedy (note the legal archaism quis, on which see Jocelyn Reference Jocelyn1967 ad loc.):

Finally, disyllabic animai (656) is a familiar form in DRN, occurring twenty-nine times (twenty-four of which are at line-end). While the majority of instances of animai do not function as archaisms in their specific contexts, the salience of the archaism here (given that it is a familiar feature of the Ennian style) is enhanced for those readers who recognise the Ennian allusion.

The Lucretian passage 3.642–56 describes the grisly realities of battle in vivid epic style, and in so doing alludes directly to a battle-scene of Ennius’ Annales (see Skutsch Reference Skutsch1985: 644–6 and in particular Nethercut Reference Nethercut, Ker and Pieper2014: 447–50; Reference Nethercut2018: 79–82). It also shows the influence (perhaps via Ennius) of the Hellenistic Alexander tradition (see Heinze Reference Heinze1897: 141 on falciferos). It is possible, of course, that more extensive imitation of Ennian lines now unknown to us could account for the striking cluster of early Latin forms in this passage. We should note, however, how closely the passage as a whole, including the early forms it contains, is fitted to Lucretius’ explanatory purpose, namely the provision of an example of how a sudden (3.636 subito) division of the body will also result in a division of the soul, thus proving the soul’s mortality. More likely than sustained imitation of specific passages of Ennius here, I suspect, is the close imitation of a general Ennian/archaic epic style throughout a passage otherwise original to Lucretius, culminating in a direct imitation of lines from the Annales, for which the archaisms, being as they are located in a description of a battle, have prepared the reader. The subject-matter (battle: a topic peculiarly suited to Ennianism), as well as the direct allusion at 644–5, serves to enhance for the reader the latent archaic resonances of the early Latin forms. As such, the early forms and constructions found here, which would not have a particular archaising function when used individually passim throughout the poem, take on a more specific allusive purpose, namely the appropriation of the style and subject-matter of traditional martial epic in the service of Epicurean psychology.

21.4 Two Difficult Cases

I discuss here two cases in which the analysis of a given form or construction is particularly difficult, due to the scarce or inconclusive nature of evidence for its use outside the text of Lucretius.

We find in DRN six sigmatic subjunctive forms of audeo: ausim twice (2.178 = 5.196) and ausis four times (2.982, 4.508, 5.730, 6.412, the latter being the only one of the six not positioned at line-end). The sigmatic subjunctive (‘already on its way out in Archaic Latin’ – de Melo Reference de Melo2007: 191) had by Lucretius’ time long since given way in most contexts to its regular equivalent the present subjunctive; productivity of the sigmatic forms decreased between Plautus and Terence (de Melo Reference de Melo2007: 194), and ‘subsequently [sigmatic forms] survived only in specialised legal/religious written contexts and in a few expressions such as haud ausim’ (Clackson and Horrocks Reference Clackson and Horrocks2007: 165; see also Leumann Reference Leumann1977: 622). Ausim, as opposed to audeam, is in fact the standard first-person form in late republican poetry: in Virgil, Horace, Propertius, Tibullus and Ovid ausim is encountered sixteen times (ten of which are negated or virtually negated), audeam never; conversely, we find audeat eleven times, ausit only twice (both negated). The distribution within genres is notable: Virgil uses ausim in the Eclogues and Georgics but not in the Aeneid; Horace uses it only in the Satires; with two exceptions (Met. 6.561, a parenthetical aside to the reader, and Met. 8.77, from a passage drawing on elegiac topoi – cf. Am. 2.16.21), Ovid’s uses of it are limited to elegy. Virgil’s use at Eclogue 3.31 is in a lightly humorous passage – Menalcas worrying about his mother-in-law – reminiscent of comedy in language and tone (Clausen Reference Clausen1994 ad loc.; Karakasis Reference Karakasis2011: 109).

Ausis is much rarer, surviving once in (tragic) Accius (quid est, cur componere ausis mihi te aut me tibi? – trag. 147, from the Armorum Iudicium), twice in Lucretius and once in Avianus (39.15). The line of Accius may, on the basis of an imitation at Ovid, Met. 13.6 mecum confertur Ulixes, be attributed to the character Ajax, to whose speech the elevated tone of an archaic form may be considered appropriate (thus Dangel Reference Dangel1995: 302). Less rare, but still unusual, is the sigmatic third person singular ausit, used three times by Lucretius’ contemporary Catullus in the highly traditional and formulaic encomium of Hymenaeus in the marriage song 61 – quis huic deo | compararier ausit? (61.64, 69, 74) – where it is combined with the early form of the passive infinitive compararier (for archaism here see Fedeli Reference Fedeli1972: 60; Jocelyn Reference Jocelyn, Adams and Mayer1999: 352; Fordyce Reference Fordyce1961 ad 61.42); Catullus uses ausit once more, at 66.28, where the subject is alis, an early alternative form of alius. These Catullian examples, combined with the gravity of Ajax’ ausis at Accius com. 147, appear to point, in contrast with the use and distribution of ausim outlined earlier, to an elevated register for ausit and ausis.

A consideration of exactly how ausim and ausis are used by Lucretius may help elucidate the matter. Both forms are used in contexts where Lucretius seeks to engage his reader directly, with ausim describing his own boldness and commitment to the didactic enterprise: a) hoc tamen ex ipsis caeli rationibus ausim | confirmare, 2.178–9 = 5.196–7; and ausis his reader’s anticipated engagement with or response to that enterprise: b) inde alia ex aliis, nusquam consistere ut ausis: | quippe sequar (2.982–3); c) nisi credere sensibus ausis (4.508); d) aut minus hoc illo sit cur amplectier ausis (5.730); e) an hoc ausis numquam contendere factum? (6.412). We may note that the one instance of ausim in Virgil’s Georgics is in much the same context: a response by the poet-teacher to a putative question from the reader: forsitan et scrobibus quae sint fastigia quaeras. | ausim uel tenui uitem committere sulco (2.288–9). Each one of these passages of Lucretius is a forceful and direct attempt to engage the reader: a deals with those who insist that the world was created with human beings in mind. The preceding line 177 contains a disarming suggestion that Lucretius himself may be ignorant rerum primordia quae sint – an ironical, perhaps even sarcastic (thus Fowler Reference Fowler2002 ad loc.) rhetorical feature to which an informal tone would be appropriate – and the surrounding lines contain colloquial features, namely anacoluthon at 172–3, and de + ablative replacing a partitive genitive at 183 nunc id quod superest de motibus expediemus (on which see Reinhardt Reference Reinhardt, Dickey and Chahoud2010: 215 with n.16); b is from a reductio ad absurdum which caps a series of more serious arguments demonstrating the insensate nature of atoms, a deflationary passage in which an elevated tone would be quite inappropriate; c and e are forceful passages (e borders on the incredulous: qui credere possis? ‘how could you believe it?’, 6.411), each of which insists that Lucretius’ reader accept the evidence of his senses, the urgency of didaxis in c being enhanced through Lucretius’ use of an ethic dative tibi (for whose regular use in informal contexts see Adams Reference Adams2013: 348; for ethic datives in Lucretius see Volk Reference Volk2002: 78 with n.29); d urges the reader to adopt multiple explanations of the moon’s light as all true, in the face of fierce yet unenlightened debate between Chaldaeans and the astrologi. These are all contexts of didactic urgency and intimacy between poet and reader, contexts in which a colloquial rather than an elevated tone would be appropriate. This is consistent with the findings of Reinhardt (Reference Reinhardt, Dickey and Chahoud2010), who demonstrates the roles played by multiple syntactic colloquialisms in contexts of close readerly engagement in DRN. Lucretius, writing a poem at whose heart is a relationship between first-person teacher and second-person student, has extended the colloquial use of ausim, as described above, to ausis.

We find in DRN the following nine instances of the impersonal use of the gerundive governing a direct object (not including 3.626, on which see Risch Reference Risch1984: 186–7):

The construction (for which see Risch Reference Risch1984: 186–8; Penney Reference Penney, Adams and Mayer1999: 259–60; Pinkster Reference Pinkster2015: 290) is common in another didactic text, almost contemporary with Lucretius, namely Varro’s De re rustica, where it is used to give instructions to the addressee, for example: faciendum quoque saepta (2.2.8); faciendum haras quadratas (3.10.8); canes potius cum dignitate et acres paucos habendum quam multos (1.21.1), but very rare everywhere else, until it crops up again in the jurists (H–S 1965: 372). In earlier texts it is encountered just once in Plautus, and once in Afranius:

Risch (Reference Risch1984: 111) suggests, in the light of the content of Trin. 869, that this may be an example of the language of soldiers, an idea to which I will return. There is one further example in each of Catullus (39.9), Cicero De senectute 2.6, spoken by Laelius and identified as a characterising archaism by Powell (Reference Powell1988: 22 and 114) and Penney (Reference Penney, Adams and Mayer1999: 260), and Virgil (A. 11.230), described as an archaism by Horsfall Reference Horsfall2003 ad loc. Evidence for the early use of this construction, however, is very thin on the ground.

This lack of early evidence for the construction should give us pause before we label it as a feature of Lucretius’ early Latin style. An alternative is suggested by Ernout (Reference Ernout1908–1909: 297; see also Ernout and Robin Reference Ernout and Robin1962 ad 1.111) and Risch (Reference Risch1984: 188), who argue for Grecism, with impersonal Latin gerundives rendering Greek verbal adjectives in -teos (cf. H–S 1965: 372 on Apul. Apol. 69 ualetudinem… medicandum). Verbal adjectives of this sort are indeed very common in Epicurus and, as with this construction in Lucretius, are often used to offer advice or direction to the reader. For a similarity in expression, albeit in an evidently different context, we may compare 1.111 aeternas quoniam poenas in morte timendum with Epicurus On Nature Book 11 fr. 26.22.2 Arrighetti τὸ δὲ ὅτ[ι μ]ὴ φοβητέον τὸ βαρ[ὺ τῆ]ς γῆς πρὸς τὸν μ[ε]τε[ω]ρισμόν. We should note however that the translations of Cicero (Aratea and Timaeus), never render a Greek verbal adjective in -teos with an impersonal Latin gerundive. Nor does the suggestion of Grecism help explain the construction’s regularity in Varro.

Its usage across the poem suggests that it may in fact be at home in a variety of registers. 1.138 levies an obligation on the poet himself (cf. 4.777–8), coming in a passage in which Lucretius expresses desire for Memmius’ patronage. It therefore calls for a formal tone appropriate for addressing social superiors. 2.1128–9, on the contrary, is a strikingly bold statement of domination over the reader, instructing him to submit to an acceptance of the facts (see 2.1043 for the same metaphor of surrender in the face of the truth); likewise 6.917–18, where the no-nonsense firmandumst multa is followed up at 920 by something approaching a direct command: attentas aures animumque reposco. High formality here would be out of place; what is called for is directness, and here we may compare Risch’s identification of the language of soldiery at Trin. 869 agitandumst uigilias. An association with direct and forthright instruction may go some way to account for the fact that the construction occurs primarily in two didactic texts (Lucretius and Varro). Similar, if perhaps less stark, are 1.381 and 2.491–2, in each of which the reader is instructed by the poet on an appropriate attitude or course of action to take in order to achieve a putative desired goal (cf. si uoles, 2.491); in these cases we may imagine a need for a tone expressing a degree of informality and social intimacy between poet and reader, and here we may again recall the arguments of Reinhardt on the role of colloquialism in Lucretius’ attempts to engage the reader. I suggest, then, that to label the impersonal gerundive in Lucretius simply as an archaism gives an insufficient account of the fairly complex evidence for its usage and register.

21.5 terra mare caelum

I end with a question not of language in the strict sense, but of what might be called poetic style. Given the role of Greek models in shaping the style of Latin poetry, this can be a difficult category in which to isolate the influence of early Latin. We regularly find in the text of DRN versions of the merism terra marique and of the triad terra mare caelum. While terra marique certainly existed in the language of early Latin (it first appears in C. Laelius orat. 17 ORF; cf. Pl. Poen. 105 mari terraque with Bennett 2.373), it later became a very common phrase in classical Latin descriptions of war, geopolitics, and imperial administration (see e.g. Cic. Leg. Man. 9; BAfr. 24; Momigliano Reference Momigliano1942), and indeed is used by Lucretius himself as a piece of semi-technical military language (3.837 on the Punic War terraque marique (the additional -que being necessary to avoid a cretic), and 1.30 on the fera moenera militiai | per maris ac terras). More promising for analysis under the rubric of early Latin is the triad terra mare caelum, called by Cicero a poeticism (Fin. 5.9 nulla pars caelo, mari, terra, ut poëtice loquar), which is found in a range of Lucretian contexts including accounts of natural processes, cosmogony, cosmophthory, and philosophical contemplation (for a full list and exhaustive analysis of Lucretian instances see Schmidt Reference Schmidt, Urushadze and Gordeziani1975). Triads, syndetic or otherwise, are a feature of Latin literature of all genres and periods and cannot be particularly associated with sacral or religious language (see Adams and Nikitina sections 14.3–4 in this volume, and Adams Reference Adams2021: 14–16 with cross references). As will be seen, however, the particular triad terra mare caelum is indeed prevalent in early Latin literary contexts representing sacral language and thus will have carried an archaic religious significance for the readers of DRN, a significance exploited by Lucretius to suit his didactic purposes.

First, however, one must acknowledge the influence of Greek (on which see Schmidt Reference Schmidt, Urushadze and Gordeziani1975: 185–6; Reference Schmidt1981). The Urtext for the terra mare caelum triad is the opening line of Homer’s description of the shield of Achilles:

lines which exert a direct influence on Lucretius:

5.417–18, a passage of cosmogony in which matter’s fundarit matches Hephaistos’ ἔτευξ᾽, recapitulates the first five items of Homer’s list (18.483–4) in exact order, this close intertextual echo being implicitly marked by the following phrase ex ordine. These lines are themselves a close repetition of 5.67–9: et quibus ille modis congressus materiai | fundarit terram caelum mare sidera solem | lunaique globum, also on cosmogony, which effect a slight re-ordering in order also to include the stars of Iliad 18.485. Lucretius’ use of the triad at 5.68 and 417, then, is influenced by the Greek model.

The shield of Achilles marks the beginning of the Greek cosmological tradition as it has come down to us, and Homer was recognised by Lucretius (citing Ennius) as a philosophical poet: unde sibi exortam semper florentis Homeri | commemorat speciem lacrimas effundere salsas | coepisse et rerum naturam expandere dictis (1.124–6). Commenting on the sixth Eclogue – itself an important chapter in the history of Latin literary cosmology – Pseudo-Probus preserves two important fragments of Ennius:

The pair of fragments attest to the significance of the triad for Roman religious language; the first (from Medea Exul) clearly belongs to a prayer (with Goldberg and Manuwald Reference Goldberg and Manuwald2018 I take the antecedent of quique to be Sol: for an alternative, that is, aether, see Jocelyn Reference Jocelyn1967: 372–3, adducing Empedocles B38 DK), and the subject of Ann. 555–6 Sk. is clearly Jupiter (thus Jocelyn Reference Jocelyn1967 ad 275; see further Flores et al. Reference Flores2002: 161). To these we may add a paratragic (and possibly quoted: Sharrock Reference Sharrock, Augoustakis and Traill2013: 56–7; Maltby Reference Maltby and Papaioannou2014: 211–12) line of Terence o caelum, o terra, o maria Neptuni (Ad. 790), Stasimus’ (comically elevated?) oath from Pl. Trin. 1070 mare terra caelum, di uestram fidem, and Cicero’s archaising account of the power of Sulla at Rosc. Am. 131: etenim si Iuppiter Optimus Maximus cuius nutu et arbitrio caelum terra mariaque reguntur…; see further Jocelyn Reference Jocelyn1967: 374. The significance of the terra mare caelum triad for Roman sacral and religious language, insofar as it was represented in early Latin literature, is clear.

At 5.564–91 Lucretius defends the difficult and widely-contested Epicurean claim that the sun is the size it appears to be (on the theory, and its critics in antiquity, see Barnes Reference Barnes1989 = 2015). At 5.592–5 he introduces arguments to demonstrate how, despite the sun’s small size, it could nonetheless fill the world with light and heat:

In Ennius’ Medea Exul the sun is prayed to as the divine being who surrounds sea, land and sky with light; in Lucretius’ defence of the Epicurean theory of the sun (a theory which, among other things, denied it the divinity it is granted in Ennius: see Lucr. 5.114–16, where the six entities of Iliad 18.483–5 are fitted into a single line, ne forte rearis | terras et solem et caelum, mare sidera lunam | corpore diuino debere aeterna manere), the traditional triad of Ennius is adopted, keeping the same order of its elements, and repurposed to the rational ends of Lucretian didactic, an example of what Philip Hardie terms ‘demythologisation’ (1986: 177–80, 185–7; see also Gale Reference Gale1994: 164–8, 172–3, 185–9, Taylor Reference Taylor2016). Nor is this passage a one-off: we encounter a similar use of the triad in an atomistic description of the behaviour of sunlight at 4.203: perque uolare mare ac terras caelumque rigare.

In Annales 555 Sk. (qui fulmine claro | omnia per sonitus arcet, terram mare caelum) the triad may be considered a universalising expression, as is made clear by omnia. This is a familiar aspect and purpose of such triads – taken together, their separate elements cover everything (see, on this aspect of such expressions, the in-depth treatment of Hardie Reference Hardie1986: chapter 7), and indeed they are occasionally used in this way by Lucretius, for example at 4.783 si mare, si terram cordist, si denique caelum, where the triad seeks to encompass all things of which we have perceptual experience, or at such passages as 1.340 or 5.92 (principio maria ac terras caelumque tuere), where the contemplation of all three elements is within the power of the student of philosophy (a godlike perspective: compare inspicis in the Medea fragment). At 6.639–79 Lucretius gives an account of the fires of Etna, during which (650–1) he asks the reader, who may be so over-awed by such a sight as to deny the possibility of rational explanation, to bear in mind how tiny a fraction of the whole universe is one sky (et uideas caelum summai totius unum | quam sit paruula pars et quam multesima constet), thus providing a multiple-worlds version of Aristotle’s admonition (Meteor. 352a27, adduced by Ernout and Robin Reference Ernout and Robin1962 ad loc.) ὁ δὲ τῆς γῆς ὄγκος καὶ τὸ μέγεθος οὐθέν ἐστι δή που πρὸς τὸν ὅλον οὐρανόν (‘the mass and size of the earth are surely nothing compared to the whole universe’). Lucretius then proceeds to turn the (normally) universalising triad of terra mare caelum on its head:

The universal triad that, in Ennius, stresses the breadth of Jupiter’s control, is revealed, in the multiple-worlds context of Epicurean cosmology, not to be universal at all: what looks to us (cf. 4.783) to be the sum of things is in fact no more than one sky, one earth, one sea, among an infinite number of skies, earths and seas. The power of the traditional terra mare caelum triad, familiar to readers from early literary texts which stress the universal power and reach of the god, is thus deflated. Lucretius’ adoption and deployment of the triad, then, may be read as an attempt to neutralise the sacral connotations it bore in early Roman literature.

21.6 Conclusions

An aspiration to consider linguistic and literary features fairly closely has meant that only a small range of the early Latin elements of DRN have been treated in this paper. Its conclusions may be expressed succinctly: Lucretius is a brilliantly inventive manipulator of the Latin language, a fact which extends to his adoption and creative adaptation of early Latin forms and constructions. We find in his archaising style a paradoxical sense of reanimation and renewal: he deploys fossilised linguistic forms and poetic expressions with a degree of ingenuity and freedom that is surprising for a poet working in the 50s bc. Many early Latin features of DRN resist interpretation at the individual level (other than as evidence for metrical motives in the selection of optional forms), but must instead be read as part of a broader archaising texture that is central to the poem’s identity as a philosophically radical text firmly rooted in a national epic tradition. There are, however, as I have sought to demonstrate, a number of instances where the archaic status of a given element in the poem serves a specific expressive function relative to the didactic aims of its context.