^{page 94 note 1} For semina concipere compare 6. 271–3, resuming this passage.

^{page 95 note 1} Kranz, W., ‘Lukrez and Empedokles’, Philologus xcvi (1944), 74–9.Google Scholar

^{page 95 note 2} Kenney, E. J., ‘Doctus Lucretius’, Mnemosyne xxiii (1970), 366–92, finds in this line an allusion to Alexandrian ars.CrossRefGoogle Scholar

^{page 96 note 1} The phenomenon has often been ob served, for example by Ernout on sequ gregari I. 452, ‘une tmise dans le mot même qui exprime l'idée d'une désagrégation’.

^{page 99 note 1} For such play between anima and animus, see the commentators on Virgil Aen. s. 57; H. Trankle, ‘Die Sprachkuns des Properz’ (Wiesbaden, 1970), 38 Virgil, Aen. to. 357 ‘uenti proelia cei tollunt animis’; and Plato, Phaedo 77 C.

^{page 99 note 2} To concede the point. But I believe tha the corpora multa uaporis et aeris (5. 490) ar particles of steam and mist. The fiery aethe is already formed, 458. The sun and moo are in position between the aether and tit, earth, 472. The heat of the sun forces th, earth to sweat and increase the volume o the sea, 487–8. Now, in 456 and 459 the atoms of fire are rare; so if in 490 atoms of uapor and aer are making the glowing quarters of the sky dense, then uapor is not heat, but vapour or steam. Lucretius is explaining the genesis of cloud.

^{page 99 note 3} Kenney obelizes incohibescit, unable to accept it because the sense ‘strives to hold it in’ is more than the inceptive verb can provide. The subjunctive incohibessit is the smallest change that produces a similar sense ‘could be holding it in’ (see Neue, 3. 510, citing habessit in Cic. de Leg. 2.8. 19 and half a dozen examples from prohibere).

^{page 101 note 1} P. Friedländer would have realized this more clearly than some of those who quote him (see ‘The Pattern of Sound and Atomistic Theory of Lucretius’, A.J.P. lxii( ), 24 and 30). See also Norden on Virgil, Aen. 6. 204 fr.

^{page 102 note 1} D. West in (ed.) C. D. N. Costa, Horace (London and Boston, 1973), 29–57.

^{page 102 note 2} ‘“No day will remove the eternal grief from our hearts.” We must ask this person the question, “If everything reverts to sleep and stillness, what is there so bitter that it with eternal grief?”’ This argument has always caused dissatisfaction because possit has not been translated. The point is scientific. ‘Given atomic physics, how could anybody waste away for ever?’ This interpretation is supported by the same argument at 3. g86.

^{page 103 note 1} P. M. Brown refers me to a similar oxymoron stragemque propagant 1. 280.

^{page 104 note 1} Kenney takes causa as ‘disease’, corno paring morbi … causa 502. But at 502 cause clearly means ‘cause’ and in 486 there is no need to restrict the word to disease. The durior causa might well be a knock on the head.

^{page 106 note 1} Notably in Proof 2 above. For suck fraudulency with metaphors see West, The Imagery and Poetry of Lucretius (Edinburgh. 1969), 81–92: and with similes ibid. 74–8, and Philologus cxiv (1970), 273–4.

^{page 107 note 1} have not found flecti used as a medica term. Virgil, Aen. 12. 46 (haudquaquam dicti uiolentia Turni fiectitur; exsuperat magic ae grescitque medendo) betrays its Lucretiar origin by the close collocation of flecti aegrescit (only here in Virgil) and mederi. Compare also uiuescit et inueterascit alendo (4. 1°68). The odds are therefore that Aen. 4. 35 is to be traced to the same source, aegram rudli quondam Pure mariti.

^{page 108 note 1} Compare scinditur et discedit, 3. 640. If the suggested reading were to be accepted, this would be the only occurrence in Lucretius of quoniam before its apodosis without a conjunction or relative pronoun to link it to what has preceded. Büchner reads animo and supports it by the suggestion that animo haec in the manuscripts may arise from an attempt to correct animo by writing ae over the o. This might be so, but we would suppose against Büchner that the corrector was right to read animae.

^{page 109 note 1} In the Phaedo 118 a 3–4 and 153 d 3–4 ‘Socrates makes some play with this amobiguity in ’: Gill, C., ‘The death of Socrates’, C.Q. xxiii (1973), 27.Google Scholar

^{page 112 note 1} W. M. Lindsay, The Syntax of Plautus (Oxford, 1907), 53 cites half a dozen possible examples of the government of an accusative case by an impersonal third person singular, and they are all transcribed by H. D. Jocelyn, The Tragedies of Eimius (Cambridge, 1967), p. 338 on fragment 202. Only two of the six survive any scrutiny: Ennius frg. 202 (Jocelyn = 241 Vahlen) praeter propter uitatn uiuitur, where J. B. Hofmann Ph.W. lxiii (1943), 20 accepts the conjecture uita (see A.L.L. v. 331), Plautus, Casina 185, where the text is very uncertain and Leo's emendation probably correct. But neither of these impersonals is parallel to mouentur motus. Merrill cites Horace, Epistles 2. 2. 125 Cyclopa mouetur and Lucretius 4. 1274 id moueri. Professor G. B. A. Fletcher has provided three other examples, governing tanta, Manilius 4. 84; quodcumque Lucan 9. 850; and Satyrum, Persius 5. 123. Lambinus comes nearest with (Aristotle, De Anima 1. 3. 406^a).

^{page 113 note 1} With peste maloque 347, a responsion lost by Heinze and by Bailey (‘wrack and ruin’).

^{page 113 note 2} P. M. Bròwn suggests that the link may be the proposition that a compound which passes through another must be divided in the process (3. 695–6, 700).

^{page 113 note 3} For other dialectical exploitations of building images see 3. 773–5 and West, The Imagery, 64–72.

^{page 114 note 1} cadere omnia membra OQV cadere onmit corpore membra F trunco cadge omnia membn Lachmann. Heinze finds no point it Lachmann's distinction between trunk anc limbs, ‘all the limbs fall slack on the blood less trunk.’ For surely the limbs are blood less too. He adopts the reading of F, whicl provides a more meaningful distinction ‘All the limbs fall slack and the body goo bloodless.’ Giussani had objected to F on the grounds that the false reading in 594, de corpore omnia membra, is clearly derived from a version of 596 which ended omnia membra. Kenney correctly replies that the corruption of 594 may have occurred after the corruption of 596 to omnia membra, but seems to make a slip in his apparatus in suggesting that 594 may have derived its error from 596 before 596 was corrupted.

^{page 116 note 1} I am very grateful to P. M. Brown of Glasgow University and J. Y. Nadeau of Edinburgh University for many helpful criticisms.