MORGAN. — ON THE LANGUAGE OF VITRUVIUS. 487 



schou bei Plaut., bei Verg., Hor., Ov. u. a.' It is obvious that we have 

 here what may be called a distinct division on a point of style. Though 

 the Ciceronian must be taken to be the better, yet we see that late 

 authorship cannot be proved from the other usage. On the second 

 point, the setting of negatives foremost in the sentence, no evidence is 

 presented that this was a habit of late authors. In phrases like non 

 putavi praetermittendum Praun, who cites (p. 27) eleven occurrences of 

 it in Vitruvius, holds that the attaching of the negative to puto is the 

 Greek idiomatic use as in ov (fjrjfxi He might have compared ovk oto^ou, 

 ov vofjLL^w, etc. ; see Ki'ihner-Gerth, Gr. Gramm. II, p. 180. And W. 

 Schmidt in Jahresbericht Alter tumsw., CVIII, 1901, p. 119 draws atten- 

 tion to Caesar B. G. 2, 31, 2 : qui ad hunc modum locuti : non se existi- 

 mare Romanos sine ope divina helium gerere. But I think it probable 

 that this position of non was, in the less polished speech, commoner than 

 is usually supposed, for it appears not only in the Bellum Africum 59, 

 1 : Non arbitror esse praetermittendum quemadmodum, etc., and 84, 1 : 

 Non videtur esse praetermittendum de, etc., but also there is a similar use 

 in the eighth book of the Gallic War, by Hirtius, 48, 10 : quod ego non 

 existimavi mihi esse faciendum, propterea quod, etc. Finally, in Ussing's 

 last example we have in non enim quae . . . non eae possunt nothing but 

 the rhetorical figure of anadiplosis, found (to compare great things with 

 small) in Demosthenes 9, 31 : dAA' oi)(virep Qikiinrov kcu. o>v e/ccivos 7rpdi-- 

 T« vvv, ov-% oitcos e^ovcrLv. And the recurrence of non once again in non 

 ruinosae may be compared with Cic. Fam. 13, 18, 2 : non potest mihi non 

 summe esse iucundum (see also Drager I, p. 135). Neither of these 

 usages is any proof of late authorship. 



Taking up a new topic, Ussing says : ' It is a well-known fact that in 

 the Silver age the conjunction num is gradually replaced by an, and 

 later on disappears entirely from the language. In Vitruvius num does 

 not exist at all, neither do we find (the single) an, ne, nor nonne. The 

 only particle by which he introduces a dependent interrogative clause is 

 si, e. g. 53, 14: si est firma probatur ; cf. 32, 4: quaesiit si essent agri ; 

 133, 20: quaerebant si koneste essent educati ; 156, 20: quaesiit si quern 

 novisse?it ; 183, 10: de aqua . . . quibusque rebus si erit salubris et 

 idonea probetur explicabo. Only in double clauses we find utrum . . . 

 an, as 18, 26: dubitantes utrum morbo an pabuli vitio laesa essent. But 

 si occurs equally, cf. 53, 12 : de ipsa autern testa, si sit optima seu vitiosa 

 ad structuram, statim nemo potest iudicare; 173, 17: neque animadver- 

 tiint si quid eorion fieri potest necne. Si in this sense already occurs in 

 Plautus ; so we do not wonder that it is found in Vitruvius, but we won- 



