196 Austin Morris Ilarnion, 



Our investigations thus far have been based on statistics from 

 the collection of sentence-endings. If we extend them to the inte- 

 rior clausulae we shall find that there also the same tendencies to- 

 ward the avoidance of position in the first unaccented syllable and 

 the effort to secure it in the second assert themselves, but to a less 

 extent. It will be of advantage to make this evident in such a way 

 as at the same time to summarize our results for the final clausulae. 

 The estimates for interior clausulae are founded upon Book XXI. 



In form I, type ;' {nosse confingit) when the clausula is final the 

 syllable preceding the caesura fails to make position in 97 per cent 

 of the cases, and the syllable following the caesura makes position 

 in 79 per cent; when the clausula is interior the respective per- 

 centages are 87 and 68. 



In I y6 (vilis et parva) final clausulae show the syllable preceding 

 the ;' caesura positionally short in 97 per cent of the cases and 

 the monosyllable long by position in 98 per cent ; interior clausulae 

 in 95 per cent and 91 per cent respectively. 



In I (3 (passihus citis) the first unaccented syllable must always 

 be short in virtue of the penultimate law: the second displays posit- 

 ional lengthening in approximateh^ 90 per cent of the clausulae 

 both final and interior. 



In II ;' [semper innocui) the syllable preceding the caesura fails 

 to make position in 94 per cent of the final clausulae and in 82 per 

 cent of the interior; the syllable following the ceesura makes position 

 in 76 per cent of the final clausulae, and in 72 per cent of the 

 interior. 



In II y(^ (asper et vehemens) the syllable preceding the / caesura 

 fails to make position in 100 per cent of the final clausulae, 94 per 

 cent of the interior ; the monosyllable makes position in 98 per 

 cent of the final clausulae and 85 of the interior. 



In III ;' and ytS {magna decernebatur) the syllable before the cae- 

 sura is not lengthened by position in 94 jjer cent of the final clau- 

 sulae and in 86 of the interior. 



There are onl}' 5 cases of IV with ;' caesura among the sentence- 

 endings, and 9 among the clausulae in Book XXI ; one example in 

 each case has the syllable before the caesura positionally lengthened. 



In I ^, in II /:; and II 6 and in all types of III and IV without 

 ;' caesura quantity by position plays no part. 



We find, then, that Ammianus' inclination to observe quantity by 

 position is nowhere carried out with strict regularit}', and that it 

 is considerabl}' less prominent in interior clausulae than in final. 

 The explanation of this fact is that observance of quantity in the 



