The Clausula in Aunmanus Marcellinus. 191 



mate words together. Let us now classify them according to length 

 and see what ratio we find in Ammianus in trisyllables, in tetra- 

 syllables and in pentesyllables. The facts are expressed in per- 

 centages in this table. ^ 



The results are illuminating. Each sort of word shows a different 

 ratio, which is constant no matter in what clausula or in what place 

 in the clausula the word occurs.^ Such a condition cannot be due 

 to anything but the working of natural laws of frequency, and we 

 ma}^ spare ourselves the pains of seeking confirmation of each ratio 

 from outside evidence. 



In all words, then, in which the Latin language admits short ac- 

 cented syllables, we find short syllables occurring in the clausulae 

 of Ammianus in a proportion which shows no trace of intentional 

 tampering with quantities, and which can only be ascribed to the 

 influence of natural laws of frequenc}'. The conclusion therefore is 

 inevitable that Ammianus was indifferent to quantity under accent. 



QUANTITY IN UNACCENTED SYLLABLES. 



It remains to be demonstrated that in unaccented syllables Kn\- 

 mianus observed quantity by position and neglected vowel-quantity. 

 His regard for position evinces itself in two ways : 



(1) in an effort to avoid positional lengthening in the case of 

 the first unaccented syllable, and 



(2) in an effort to secure positional lengthening in the case of 

 the second unaccented syllable.^ 



The effort to avoid positional lengthening asserts itself in all forms, 

 but onl}^ in types with ;' caesura [nosse confingit). For in all other 



* Words of 6 syllables are too few to give a ratio of any value. 



* In the two percentages bracketed the count is a very small one 

 (19 tetrasyllables as the first word in IV, 24 pentesyllables as the first 

 word in I). The percentages given under III are based on the largest 

 counts. Statistics as to the length of the first word as well as that of 

 the last are given for all forms in the tables of types, p. 179 ff). 



* The first of these two observations is Meyer's {Ges. Abh. II. 264 — 265), 

 the second is my own. 



