The Clausula in Amiiiia/iits Marcellinus. 169 



Of V I can cite five examples in Ammianus : II. 24. 25 pondere cir- 

 cumveniretur ; 212. 27 venerint allegationum; 242. 18 ve'ritus ut adhiic 

 durdiHem ; 268. 20 eminuit at Sebastidiii ; 269. 19 e'rat cum moderatione. 

 To these should perhaps be added three ambiguous clausulae in 

 Book XXI.^ Of VI I have only two instances : II. 79. 19 postea 

 Co}istantinop6leos \ 214.6 Ae'sopi cavillationibus. It is possible that 

 some instances of one form or the other may have escaped me, 

 but they cannot be numerous. Consequently both the forms are 

 extremely rare. But we must not let this move us to doubt their 

 legitimacy, for all of the texts in which they occur seem to be 

 sound and are very difficult to assail. Moreover, their rarity finds 

 adequate explanation in the motive that prompted their use : they 

 were brought into employment simply as a means of handling long 

 words which' could not easily be disposed of otherwise.^ It is 

 noteworthy that, if we except the three ambiguous cases of V in 

 Book XXI, all the examples come from the last 9 books — in fact, 

 5 of them froni the last two books. This appears to indicate a late 

 recognition of these forms, and a slightly increasing tendenc}' to 

 use them, once they are recognized. The fact that their occur- 

 rences are not in responsion but isolated is somewhat surprising. 

 I can account for it only on the hypothesis that Ammianus regarded 

 them as extensions of III and IV, and did not feel that they broke 

 the rh3thm. It is obvious that forms V and VI stand in the same 

 relation to III and IV in which the latter stand to I and II. Using 

 metrical terms to define accentual rhythm (as we do in dealing 

 with English poetry) we may say that each of these clausulae has 



1 P. 142, 1. 14 trahendumque si reniterehir^ listed as doubtful (see below) ; 

 p. 154, Q>S fal/acissintae geittis obse7-7'dntiir axidi tirgentia velut pugnatiiriis — two 

 clausulae in apparent responsion, listed as examples of A (see below). 



* I note tlu-ee cases in wliicli a final word of 6 syllables witli penul- 

 timate accent is not preceded by an antepenultimate Avord, but they are 

 all corrupt. In I. 83. 5 correxerit et libera I /tate the Ms. reads correxit et 

 libertaie (III), whicli should be retained in spite of tlie change in tense 

 eget-it-correxit. (v. Lofstedt, Beitrdge ziir Kcnnt)iis d. spdter. Lett. p. 70 ff.) 

 In II. 12. 14 amittit intempestuosos the ms. lias inteinpesiiosos., corrected into 

 intempesiuosos by a later hand : the reading of Accursius intempestivos is 

 in better accord with the usage of Ammianus. In IT. 64. 20 somniorum 

 adsiduitate we have a case of hidden corruption : a verb has fallen out 

 (v. Giinther, Phil. 50, p. 69) and it is most natural to supply it at the 

 end of the clause, after adsiduitate. As for words like cavillatioiiibus.^ I 

 have not found any instance of their employment before a pause except 

 in the two cases cited above. 



