^{page 113 note 1} Cf. J. Phil, xvi, p. 245.

^{page 113 note 2} Prot. 343 e, with Adam's note.

^{page 113 note 3} In the exceptionally complicated (but not really awkward or difficult) passage of Ovid, , fast. 1. 263Google Scholar, cited by Housman, inquit is neither A nor B, but C.

^{page 114 note 1} But is or is not significando res saepius in usu (for saepius usurpatas) good enough Latin for Vitruvius? In any case, the implied comparison with articles in less frequent use, and therefore not as yet named, introduces (although true) an oddly negative detail. Far more cogency will be given to this account of the origin of language by reading ‘significando res saepe’ and then ‘ut in usu’; for by this reading saepe is forced to go with significando as the thought requires, and ut in usu appears as the justification of ‘saepe’. (Granger removes the comma after usu and gives us two mistranslations thus: ‘indicating things more frequently and by habit, they came by chance to speak according to the event.’ What nonsense! ex euentu means ‘in the result, eventually’. ‘Through indicating certain objects frequently according as these were in use, they ultimately developed by accident the institution of language.’)

^{page 114 note 2} As will presently appear, I do not myself accept the word ‘alternative’; but that makes no essential difference here.

^{page 115 note 1} The passage quoted by Gow from Livy, 28. 22.12, to show that after a negation of fearing ut could have the same sense as ne, had already been dismissed by Palmer on the ground that ut auderent is converted by the preceding nihil into the equivalent of illud, ut auderent. Housman accepted that explanation, and the doctrine is even more precisely stated by Gildersleeve and Lodge, §550, note 3. I completely disagree. From another passage of Livy, virtually on all fours with this, where the tradition gives us ne, Gow would deduce that in such circumstances ut and ne could have the same force. I deduce that ne is the proper conjunction. What could be easier for a scribe than to ‘correct’ quamn into quam and eegredi into egredi? In 28. 22. 12 I suggest that ne thus fell out and that ut in our archetype was somebody's too amateurish supplement.—For my argument's sake I leave this note as I wrote it; Prof. Fletcher, again greatly strengthening my hand, points out to me that ‘the proposal … and the explanation of the error are to be seen in Madvig, , Emendaliones Liuianae ², p. 408’Google Scholar.

^{page 115 note 2} And if you come to bare facts, he has mentioned Persius twice, but Rex three times. Even if he were now to give us a thumb-nail sketch of Rex, that would be ad Regem iam adeo, not redeo. When Horace says redeo he means it; see serm. 1. 1. 108 (after 107 lines), 6. 45 (after 22 lines).

^{page 115 note 3} Equis, that is ἴπποις in the Homeric sense, chariot; not ‘mount’ which would require the singular; cf. quadrigis albis, Plaut, . asin. 279Google Scholar.

^{page 116 note 1} J. Phil. xviii. 25. So far as it goes, H. demonstrably meant to imply; since he himself objected to Romae.

^{page 116 note 2} suasor (Hamacher, , sched. crit., Trier, 1858)Google Scholar, auctor (Palmer), and praemostrator (Housman, loc. cit.); all ‘one-sided’, and none easily to be reconciled with furor.

^{page 116 note 3} ‘… rhetor, ut(erque | alterius laudum sic admirator, ut) alter | alterius sermone … laudum admirator is a pleonasm for which I can no support in Latin; how un-Horatian it is may be seen from Zangemeister's index, s.v. laus. Actual ‘admiration’ misses the point; it should be ‘partiality’ or ‘laudation’. And two brothers may well ‘admire’ one another's ‘reputation’.

^{page 117 note 1} Postgate's attempt to compromise by retaining the actual word with previous insertion of ita, in the sense ‘so brotherly’, fails through ambiguity; this brotherly orator might also have been an actual brother, so far as expression goes; in such matters, as in all, Horace says neither less nor more than he means; see epist. 1.10. 3–4 (not brothers), serm. 2. 3. 244–5 (brothers).

^{page 117 note 2} serm. 2. 1. 59; 6. 23 (cf. 1–19); 7. 13, 28; epist. 1. 2. 2; 8.12; 11.21; 20.10 (cf. 13); 2.1.103; 2. 41 (cf. 43), 65; eleven instances.

^{page 117 note 3} Wilkins, introd. to ed. Cic, . de orat., p. 22Google Scholar.

^{page 117 note 4} Bentley's note by inadvertently confusing the two makes it refer to both.

^{page 118 note 1} At serm. 2. 1. 86 editors compare de orat. 2.58. 236; see also Fiske, G. C. and Grant, Mary A., Cicero's De Oratore and Horace's Ars Poetica (1929), p. 134Google Scholar.