222 Austin Morris Hamton, 



Secondly, if it must be conceded that certain pentesyllables had 

 their secondar}' accent on the first syllable whereas others had it on the 

 second, this would mean that there is a slight disagreement between 

 different clausulae of form III, type ;', some presenting the scheme 

 r^ r>^ ^n^j r<j r\j r<j r\j and others the scheme r<^ r\j\r'(j r\j r>sj i<sj r\j. 

 Now I find no inclination evinced by Ammianus to avoid this hypo- 

 thetical difficult}' ; there are no signs that he made any attempt at 

 all to pick and choose between pentesyllables in the construction 

 of this cadence. It follows, therefore, either that all pentesyllables 

 had their secondary accent on the same syllable (which does not 

 seem in the least likely) or that the discrepancy in the location of 

 their secondary accents did not bother Ammianus, — in which case 

 the secondary accent must have been light.* 



SYLLABICATION. 



Under the head of s3dlabication we may first consider the matter 

 of elision and hiatus, which can be summarily disposed of. A glance 

 at the text of Book XXI will show that Ammianus admits hiatus 

 with great freedom. Under certain circumstances, of course, his 

 desire to secure length by position bars it out: for example, he 

 would not admit praecipue artis or Caesarem absens. In such clau- 

 sulae, however, it is not the hiatus that is objectionable, but the 

 failure to make position. Wherever a syllable preceding caesura 

 is not required to make position, it may stand in hiatus, as in dia- 

 demali utebatur, cogitatione intcnta, eoruni ecclesiaui etc.^ Elision and 

 ecthlipsis are not employed at all. 



Aphaeresis appears to occur in the following clausulae.' 



1 V. p. 182-183. 



* V. pp. 192, 195. The statistics there given show tliat iu clausulae 

 with y caesura hiatus occurs in about 10 per cent of the instances, where- 

 as in those clausulae with cT caesura in which the c)' syllable is not re- 

 quired to be long it occurs in about 3 per cent. This discrepancy can 

 hardly be attributed to a desire to avoid hiatus in (f clausulae : in the 

 one case {cogitatione intcnta^ etc.) it is the last syllable of a penultimate 

 word that is affected, in the other {diademati tttcbatur) the last syllable 

 of an antepenultimate word ; hence we could not expect the proportion 

 of hiatus to be the same in both. 



^ In Ammianus there is no reason to assume that when the auxiliary 

 is the last word it stands, as it were, extra metrum and does not form 

 part of the clausula. Wherever it occurs last it gives a regular clausula 



