^{page 89 note 1} ‘Enniana I’, C.Q. xxxvii (1944), 79.Google Scholar

^{page 89 note 1} ibid. 80.

^{page 89 note 3} ‘Enniana II’, C.Q. xlii (1948), 99.Google Scholar

^{page 89 note 4} Nonius' first example is Virg. Am. 6. 383Google Scholar, where he is patently wrong. His third and last example is Enn. ann. 63Google Scholar, where either meaning would fit. In our line parumper is characteristic of the modesty with which one addresses oneself to the higher power; so ann. 208 diui hoc audite parumper. Cf. Virg. Aen. 2. 690Google Scholaraspice nos—hoc tantum—; Ciris 404 supprimite o paullum turbati flamina umti; Ovid, am. 3. 6.Google Scholar 2 siste parumper aquas. Cf. also Juv. 10. 250 oro parumper attendas.

^{page 89 note 5} This wholly unreasonable piece of chartolatry calls to mind a passage in Propertius, 2. 26. 7 quam timui ne forte tuum mare nomen haberet/atque tua labens nauitafteret aqua. Here Heinsius wrote teque because he saw that atque was flat; the poet would stress the person and state directly and not merely by implication that the sailor's tears are for Cynthia. But atque was not flagrantly impossible, and so teque is not even mentioned in the apparatus of Butler and Barber. Since then Axelson has shown that the elegists avoid atque as a trochee, and teque has now appeared at least in the apparatus of the Oxford text. For the Teubner edition it still does not exist.

^{page 90 note 1} In the margin of his copy of Vahlen's Ennius, which the courtesy of Mr. C. H. Roberts, then Librarian of St. John's Col lege, Oxford, permitted me to inspect many years ago.

^{page 90 note 2} seen. 177 V. = trag. 141 R³ Lindsay's colometry, involving ted for te (ted already Bothe and L. Mueller), is certainly to be preferred to Vahlen's trochee pentameter, which is impossible per se and further disfigured by hiatus; see Timpanaro, S., S.I.F.C. xxi (1946), 80.Google Scholarmed is probable in Caecil. 9; cf. Lindsay, , E.L.V., p. 160.Google Scholar

^{page 91 note 1} In his notes on Orosius published long after his death in Cholinus's edition of Orosius (Mainz, 1615), and repeated in Haver-kamp's edition of 1738.

^{page 91 note 2} Ennius und Vergilius (Lpz. 1915), p. 85 n.Google Scholar

^{page 91 note 3} uincere in acie, of course, is found (e.g. Caesar, , B.G. 7. 29)Google Scholar; in certamine Livy 10. 6. II; 37. 53. 7; cf. Hor. ep. I. 10. 35.Google Scholar

^{page 91 note 4} Mnemos. xix (1891), 50.Google Scholar

^{page 92 note 1} Norden, , Aen. VI³, pp. 389 f.Google Scholar According to Norden's figures Lucretius has a pause after the fifth trochee once in 177 lines, after the fifth dactyl once in 239 lines. In the Annals I find fifth trochee pause six times (107, 166, 244, 245, 359, 499), fifth dactyl pause three times (16, 192? viz. the present example, 246). There are 550 hexameter endings and the proportion for fifth trochee pause thus is one in 92 lines, for fifth dactyl pause one in 185 lines (or one in 275 if the present example be omitted). A pause as heavy as that before qui antehac is not found either in Ennius or in Lucretius; compare, however, Virgil, , Aen. 4. 593Google Scholardiripientque rates alii naualibus? ite, / ferte citi flammas; 12. 425 ‘arma citi properate uiri, quid statis?’ Iapyx / conclamat; 526 Aeneas Turnusque ruunt per proelia. nunc nunc / fluctuat ira. In these three examples a special effect is clearly intended. It must, however, be allowed that the epigrammatic form creates special conditions for our line which may well account for the unusual pause.

^{page 93 note 1} Virg. Aen. II. 440uobis animam henc … / Tumus ego … / deuoui; ecl. 5. 41 Daphnis ego in siluis (Theocr. 1. 20 .Google Scholar

^{page 93 note 2} For a fictitious dedication transmitted by an historian compare Ephorus in Diodorus 11. 33, who cites an epigram, unquestionably a later composition, as belong ing to the snake-column commemorating Plataea, (Wilamowitz, Hellen. Dicht. i. 27).Google Scholar

^{page 93 note 3} quia in at the beginning of the second line is to be read as qui ui or perhaps qui illinc. The proverbial use of Osculana pugna to denote uictos uincere may have profited from a popular etymology connecting Osculum with (osculum)oscillum, on which Festus 194. 20 M. remarks: per oscillationem, uelut imaginem quondam uitae humanae, in qua altissima ad infimum interdum, infima ad summum efferuntur.

^{page 93 note 4} Harvard Studies lxiv (1959), 167 f., n. 22.Google Scholar

^{page 94 note 1} euomo similarly but with uirus as the object: Cic. Lael. 87Google Scholar; in an aggressive sense with in: Ter. ad. 312, 510; hec. 515; Cic. Mil. 78Google Scholar, Phil. 5. 20Google Scholar (Thes. v. 2. 1071.Google Scholar 66, 1072. 63. 1073. 1 ff.). exspuo in Lucr. 2. 1041Google Scholar, wrongly grouped with Ter. eun. 406Google Scholar by the Thesaurus (v. 2. 1909. 51 ff.)Google Scholar, corresponds to ‘to spurn’.

^{page 94 note 2} Of secrets: Callim. fr. 75. 7 Pf.; of grief: ibid. 714. 4. Similarly Aelian, , not. an. 4. 44Google Scholar; Greg. Naz. carm. II I. 34. 50.Google Scholar

^{page 95 note 1} I am not convinced by the observations of Leeman, A. D. (Mnemos. xi ⁴ [1958], 319 f.)Google Scholar, who sees in doctus fidelis a thematic proposition of two aspects, the personal and intel lectual, and the social and moral, of the man's personality, to be particularized in chiastic order, in groups of two attributes.

^{page 95 note 2} Körte, , Arch. f. Papyrusforsch. vii (1924) 257Google Scholar, and others also consider him an officer of a royal court. I accept this view, but A. Momigliano in conversation with me questions it: he feels that the attributes speak against it.

^{page 95 note 3} 247 multa tenens antiqua, sepulta uetustas quae facit. Thus the sentence is to be stopped; cf. Hom. Od. 2. 188 al.Google Scholar (Skutsch, F., R.-E., s.v. Ennius, 2611).Google Scholar

^{page 95 note 4} Stilo can only have been thinking in terms of an historical disguise, not of an open self-description. He knew the chrono logy and circumstances of Ennius well enough not to make him actually assume such a role in the early years of the Second Punic War. Leeman finds a further argument for identifying the friend with Ennius in the fact that he is given no name; but he may well have been addressed by name in the subsequent speech of Servilius. This would hardly have deterred Stilo from making his guess. The only thing reasonably certain is that the name was not Quintus, since Stilo, Varro, and Gellius would not have let that point slip.

^{page 96 note 1} There are difficulties in the other two as well, but I do not propose to deal with them here.

^{page 96 note 2} I omit instances where a proper partitive construction is taken up by a partitive object: Plaut. most. 1017 quod … negoti gessit:: … gesserit… negoti?'; Pom. 641 si quid boni…:: boni de nostro tibi nec ferimus nec damus; etc. impertite sermonis curiosum is conjectured by Helm, at Apul. met. I. 2Google Scholar (sermones MSS.).; but see Wiman, G., Text-krit. Stud, till Apuleius (Göteborg, 1927), p. 17.Google Scholar

^{page 97 note 1} The verb itself is expected here. Vahlen and others have compared Virg, . Aen. II. 822:Google Scholar

Accam ex aequalibus unam adloquitur, fida ante alias quae sola Caraillae, quacum partiri curas.

We may add Stat, . Theb. 8. 279:Google Scholar

quicum ipse arcana deorum partiri et uisas uni sociare solebat Amphiaraus aues.

This passage is especially noteworthy because it shows a certain resemblance to the construction rejected above: here sociare in the sense of impertire follows upon quicum. But it does so only in a zeugma: both quicum and sociare have their regular constructions with partiri and uni. Statius' critical attitude towards the Musa rudis ferocis Enni (silv. 2. 7. 78) and the fact that carmen tuba sola peregit (Theb. II. 56Google Scholar = ann. 519) is his only borrowing from Ennius make it very unlikely that in writing Theb. 8. 279 ff. he had Ennius in mind.Google Scholar

^{page 97 note 2} Three examples in Plautus, two in Lucilius, one each in Ennius (seen. 327 V.), Afranius, Accius, Lucretius, Sallust, and Tacitus (Neue-Wagener, iii. 74).

^{page 97 note 3} For consilium compare credere omnem exercitum / consilia in the passage from the Eunuchus discussed above, p. 94, and, e.g., Val. Flacc. 5. 280 intima secum / consilia et uarias sociabant pectore curas. For rerum suarum compare Cic. Lael. 103 in hac (amicitia Scipionis) mihi de re publico consensus, in hoc rerum priuatarum consilium.

^{page 97 note 4} I overlooked that consilia inpertit was con sidered but not adopted by Bergk, , Kl. Schr. i. 276.Google Scholar

^{page 98 note 1} Timpanaro's defence of agebant (S.l.F.C. xxi [1946], 53Google Scholar) is another example of predilection for what is odd and irregular. Ennius has one or two examples of hiatus after a dactyl (above, p. 91), none after a trochee. And why should he choose hiatus here when the frequentative, which he has at the end of the line in 271, lay to hand? There is no bacchiac synonym of octo, or we may be sure he would have used it and avoided hiatus in 332. Compare also Virg. georg. 4. 154agitant sub legibus aeuom.Google Scholar

^{page 99 note 1} Zillinger, W., Cicero u. d. altröm. Dichter (Diss. Erl., 1911), p. 104Google Scholar, n. 2, though he follows the vulgate, rightly considers the omission strange. He felt compelled to assume that Cicero quoted from memory and in exactly, overlooking that Gellius' text derives from Cicero's.

^{page 99 note 2} The copula is often omitted by Ennius, and not only where a metrical advantage is gained: 30–31 Assaraco natus Capys optimus, isque pium ex se / Anchisen generat.

^{page 99 note 3} olim would seem to be an obvious replacement for an unfamiliar ollis.

^{page 99 note 4} I want at least to mention that, if Ennius had used both ollis and olim, he would have written olim popidaribus ollis / qui turn uiuebant, thus avoiding the clash of olim and turn; turn illi homines in Cicero may indicate that ollis and tum were not far separated.

^{page 99 note 5} Vahlen: ‘Tuditano collega m. (h.e. marci) primus Schuetzius restituisse dicitur.’ The rumour is false: not only Merula (1605) has the restored reading but Lambinus (1566), and even he does not claim it as his own.

^{page 100 note 1} I take this opportunity to correct an inaccuracy in ‘Enniana IV’ (C.Q.N.s. xi [1961]).Google Scholar Basing myself on earlier maps, and slanting my computations too strongly against my own thesis, I gave (p. 256) the distances of Mt. Murcus and Mt. Aventine from the sacellum Murciae as 225m. and 185m. According to the map of Lugli and Gismondi (1949) the plateaus of the two hills seem to be equidistant at c. 320m., whilst the foot of Mt. Murcus is actually a trifle nearer (c. 215m.) than that of Mt. Aventine (c 230m.).