178 Austin Morris Harmon, 



perfect as to be nearly the worst possible.^ Consequently, in case 

 Ammianus consistently restricted himself to the use of the regular 

 forms, we could not expect to tind a consistent reproduction of these 

 forms in our text of his work. We should look for the introduction 

 of a moderate number of irregular cadences of all descriptions ; 

 naturally some few of these cadences would be in close enough 

 proximity to suggest responsion, and in some few of them the cor- 

 ruption would be latent. And this is precisely the state of affairs 

 that we find in our text. The probabilities, then, are strong against 

 the validity of any of the irregular cadences. 



We may therefore bring to a close our long discussion of the 

 forms employed by Ammianus with the statement that except for 

 two forms V and VI, used for the handling of long words and so 

 infrequent as to play but an infinitesimal part in his rhythm, he 

 seems to have used only the four forms I-IV, which, even in our 

 corrupt text, are well-nigh universal. 



TYPES. 



To name the types in which the four regular forms occur I have 

 adopted the system devised by Zielinski for the quantitative clau- 

 sula. This consists simply in taking the longest form (IV) and as- 

 signing a Greek letter to each place in it where a caesura may 

 occur. Thus : 



^ y (V t ^ 1] & 



Consequently a clausula having but a single caesura will be denoted 

 by a single Greek letter, which will be /3 if the caesura follow the 

 first syllable, ;' if it follow the second syllable, and so on: a clau- 

 sula having two or more caesuras will be denoted by two or more 

 Greek letters, each determined after the method just described. 

 A clausula having no caesura is signified by the letter u. 



Under this nomenclature the types which occur in Book XXI 

 (Coll. I) and among my sentence-endings (Coll. 11) are classified in 

 the following list, which, for the second collection, also gives com- 

 plete statistics as to the length of the first word in each type. 



» See p. 122. 



