^{page 101 note 1} CQ N.S. 28 (1978), 267 ff.CrossRefGoogle Scholar

^{page 101 note 2} De Lucretiani libri primi conducione ac retractione (1912), pp.136, ff.Google Scholar

^{page 101 note 3} Lucrezstudien, in Sitz. Preuss. Akad. (1918), pp.912 ff.Google Scholar

^{page 101 note 4} Titi Lucreti Cart De Rerum Natura libri sex (1947), i.32–3.Google Scholar

^{page 101 note 5} See in particular the excellent bibliography by Dalzell, A., in CW 66 (1973), 389–427, (especially 425–7) and 67 (1973), 65–112.Google Scholar

^{page 101 note 6} Il problema del testo e delta composnione del De rerum natura di Lucrezio, especially pp.150–74.Google Scholar

^{page 102 note 1} Pizzani, p.150, cites with approval Reitzenstein's, R.Das erste Prooemium des Lucrez (Nachricht von der königlichen Gesellscbaft zu Göttingen, 1920), pp.83–96Google Scholar (which I have not been able to consult), to the effect that the order in which the books are described in these lines is not significant but is intended to give proper emphasis to the subject-matter of book 4. Pizzani himself virtually implies that ‘cum primis’’ in line 130 indicates actual chrono– logical priority as well as greater importance.

^{page 103 note 1} See Giussani, ad loc.

^{page 103 note 2} For another indication that the discussion of appearances of the gods in 5.1169–82 was written before the full investigation of perception in 4, see Nichols, J. R., Epicurean Political Philosophy (1976), p.152, on the inconsistency of the unquestioned acceptance of appearances of the gods in 5 and the recognition in 4.455–68 that inferences about the validity of visions is often fallacious. The whole doctrine of false inferences on the basis of simulacra, including optical illusions, is worked out in 4, and would surely have influenced Lucretius' treatment of the gods in this passage of 5 if it were written later.Google Scholar

^{page 103 note 3} Hermes 43 (1908), 286–95.Google Scholar

^{page 103 note 4} So Pizzani, pp.158–9.

^{page 104 note 1} Pizzani (pp.161–7), while accepting that the passage is extremely awkward (p.165), believes that the two summaries have different but complementary functions and were intended to stand beside each other more or less as they do now. Although this theory explains the word sed in line 45 more convincingly than other explanations do, I find Pizzani's argument impossible to follow.

^{page 104 note 2} p. 101 n.l. It is there suggested that the whole opening of 2, especially lines 29–33 and 54–62, may not have been written until 5 and 6 were more or less complete.

of this line, see Addendum.

^{page 105 note 1} For a summary of recent views on Lucretius' belief in progress, see Dalzell, , in CW 67 (1973), 75–6, revealing strong reasons for adopting an equivocal position.Google Scholar

^{page 105 note 2} See now Kenney, E. J., in Greece & Rome, New Surveys 11, Lucretius, pp.21–3,Google Scholar convincingly answering the attempt of Smith, M. F., in the new Loeb of Lucretius (1975), pp.578–9, to argue that ‘this dark and hellish picture of what life is like without the guidance of Epicurus’’ serves to emphasize the brightness of the Epicurean vision. Its position at the end of the poem would, as Kenney points out, oblige the reader to ‘turn back again to the earlier books of the poem for reassurance’’. If the work was originally planned as I have suggested, this difficulty would not arise.Google Scholar

^{page 106 note 1} Op. cit., p.138.Google Scholar

^{page 106 note 2} Cf. 1.133, ‘morbo adfectis somnoque sepultis'.

^{page 106 note 3} Dalzell, A., in Phoenix 14 (1960), 103–4, points out how the repetitions from this passage in lines 788–801 indicate a lack of finish of the whole section.CrossRefGoogle Scholar

^{page 107 note 1} Op. cit., p.121.Google Scholar

^{page 107 note 2} Storia delta letteratura latina (1945), ii.321–2.Google Scholar

^{page 107 note 3} Pizzani, pp.167–72, produces further arguments for the unfinished state of book 4, suggesting that the second summary (45–50), lines 218–29 (from 6.924–33), 687–705, 777–817, 822–57 are all provisional drafts, inadequately adapted to their contexts.

^{page 108 note 1} So perhaps also the mutilated passage 6.47–8, if we could be confident of the sense to be restored (so Pizzani, p.174).

^{page 108 note 2} Op. cit., pp.290–1.Google Scholar

^{page 108 note 3} Lenaghan, L., in TAPA 98 (1967), 221–51, shows how excellently the lines fit into their context in book 1.Google Scholar

^{page 108 note 4} Bruns, I., Lucrez-Studien (1884), p.7, believed that the passage in 1 marks the introduction of a fresh motive in composition, distinct from the conversion of Memmius, which Lucretius was already abandoning for the pursuit of poetical glory. This is an over-simplification, but may contain some truth.Google Scholar

^{page 108 note 5} The second-person opening also provides an owner for the mysterious anitnum in 1.932, where the omission of tuum is more easily understood if Memmius has been referred to by implication ten lines earlier. In 4.7 no such owner for the animum is to be found; and one can understand why Lactantius (Inst. 16.3) quotes the line, as from book 4, with the more general animos.

^{page 109 note 1} In a similar way, the second linkpassage in 4.45–53 was preserved in that place despite the use of the more important lines as 3.31–4.

^{page 109 note 2} On this whole problem, see Giussani, i.118–20.

^{page 109 note 3} Bailey, , vol. i, pp.71–2,Google ScholarGiussani, , i.184,Google ScholarRist, J. M., Epicurus: an Introduction (1972), p.157,Google ScholarLong, A. A., ‘Hellenistic Philosophy’’, Philosophy (1974), pp.48–9.Google Scholar The passage of Epicurus (ad Menoec. 124) which appears to provide the closest approximation is textually very uncertain, and would hardly have been interpreted to give this sense without the guidance of Lucretius. However, Cicero, (N.D. 1.49) attributes to Epicurus the doctrine that an infinite supply of imagines enters the mind, bringing great pleasure and providing a concept of the blessed and eternal.Google Scholar

^{page 109 note 4} The whole beginning of 6 is a bit of a patchwork, with 56–7 repeated from 1.153–4 (and again in 6.90–1), and 58–66 from 5.82–90, with 67 added as an additional link (cf. the insertion of 2.54 before a similarly repeated section); but it still forms a complete unit, with or without 68–77, the last line of which seems intended to take the place of 67 as a statement of general demoralization to be avoided in the light of the following observations.

^{page 110 note 1} Contrast the careful work of Virgil's editors, Varius and Tucca, after his death.

^{page 110 note 2} So Kenney, E. J., Lucretius, De Rerum Natura, Book III (1971), pp.12–13).Google Scholar

The closeness of the link between the beginning of 3 and what was written to precede (or even the less gloomy picture at the end of book 2 which precedes in our present text) clearly justifies Bailey, and the majority of other editors, in accepting ‘e tantis tenebris’’ in 3.1 from cod. Monacensis (echoing ‘fluctibus e tantis vitae tantisque tenebris’’ in 5.11), rather than o of O and V. Despite the arguments of Timpanaro in Philologus 104 (1960), 147– 9,Google Scholar who fails to account for the variants in the Italian manuscripts (best explained on die assumption that their parent, P, like L, one of its best attested descendants, and also Q, had a lacuna at this point, resulting from a failure of the scribe of the archetype to insert the capital), neither reading has any more authority than a of some of the Italians. Clarke, M. L., in CQ N.S. 27 (1977), 354–5,CrossRefGoogle Scholar argues that in Latin poetry an address beginning with a relative or participial phrase normally starts with o. While the negative of this would be difficult to establish, it is clearly true that such an address never opens with a bare relative. In Aen. 8.511 it is covered by tu; here the relative ‘qui primus potuisti’’ is covered by the whole previous line, linking the argument closely with the end of the preceding book, in a way paralleled by none of Clarke's other examples, nor by any other I have discovered. The sense thus runs: ‘Out of such darkness (as just described) to raise so bright a light, you were the first to be able … and I follow your lead as a result, o glory of the Greek race.’’ Perhaps no less significantly, Clarke asserts that He knows of no example of o separated from the relative by such an interval. In fact, a survey of the poems of Cicero, Catullus, Vergil, Horace, Propertius, Ovid, Manilius, Lucan, Valerius Flaccus, and Statius (which would often have been an easier task if Deferrari had admitted that o could have any stylistic importance) reveals no example where it is separated at all, except Lucan 1.195, ‘o magnae qui’’, and two examples where a pronoun precedes, Prop. 4.9.33, ‘vos precor, o luci sacro quae’’, and Val. F. 1.7, ‘tuque, o pelagi qui’’. The anomaly of separation by a whole line, in addition to the lack of parallel for any sort of o qui followed by a second o plus vocative (also pointed out by Clarke), is enough to guarantee the judgement of the scribe of Monacensis rather than of those of O and V.