230 Austijt Morris Harmon, 



we are governed by the fact that the ordinary pronunciation of 

 quod would create an abnormally large number of irregular cadences 

 containing this word— a fact that hardly calls for proof, as it should 

 be evident merely from the number of instances cited in the list 

 above. But to remove any possible doubt we may point out that 

 in the collection of sentence-endings the relative occurs in only 8 

 instances, of which 4 are irregular if we give the word its usual 

 pronunciation, whereas the monosyllable et occurs in 97 clausulae, 

 not one of which is irregular. While this comparison is not as im- 

 pressive as it would be if we took the entire text of Ammianus 

 into consideration, it nevertheless shows clearly that the relative is 

 not handled like other monosyllables. Hence we must infer that it 

 is not on all occasions monosyllabic. 



Again, in the three words quies, quibus, qiiidein we have asserted 

 not only that the // counts as a syllable but that it bears the accent. 

 This remarkable phenomenon is far harder to admit than any other 

 application of the princijile of syllabic w. Yet it must be admitted, 

 for in every single clausula in which the word qnics appears it is 

 treated as if it were an antepenultimate trisyllable like requies instead 

 of a dissyllable. 1 Since this fact cannot possibly be due to the blind 

 working of chance, the onl}' admissible explanation is the obvious 

 one that to the ear of Ammianus the word had the effect of an 

 antepenultimate trisyllable— an effect which can have been secured 

 only by pronouncing it qi'iies. Thus the evidence for quies is just 

 as strong as the evidence for, let us say, subinde. And from quies 

 we get the key to the explanation of the apparently irregular ca- 

 dences in which quibus and quidem appear.^ It is very clear, then, 

 that in Ammianus the syllabication of u after q must be admitted. 

 In trying to account for a feature of his pronunciation so contrary 

 to good usage as exemplified in the poets, one is naturally tempted 

 to see in it an idiosyncracy due to his Greek birth, inasmuch as 

 transliterations like Koivtoz. indicate that to the Greeks the Latin u 

 seemed syllabic.^ This consideration should not be ignored, yet we 

 must not overrate its importance. In the first place it should be 



^ Except in patitur quies (II. 67. 32), where either trisj'-llabic or dis- 

 s^'llabic pronviiiciation of qides would give a regular form. 



^ That quibus should occur in 3 appai-eiitly irregular clausulae is in 

 itself suspicious, in view of the fac^t that it is rareh' to lie found in a 

 regular cadence. 



^ This is shown not only by the fact that the u is represented by a 

 vowel, but even more conclusively by the accent which falls upon the 

 vowel. 



