216 Austin Morris Harmon. 



{de acccntilms liber: K. III. 525. 27) : quae autem habent c inter / et u 

 in paenultimo loco, corripiuntur, ut portions Gallicus Italicus Ale- 

 mannicus Romanicus : excipiuntur ea quae habent a, ut Taurinacus. 

 The phenomenon is perhap.s attril^utable to a tendency to accent 

 the broader vowel.^ 



Against Teredona, in which the accent is probably on the ultima 

 in the nominative, we may set Amdzones i\. 278. 9 : II. 235. 22) and 

 especially Chalccdona (I. 267. 12: 272. 10: II. 89. 27), with recessive 

 accent. There is nothing to show how Ammianus accented the 

 nominative in these two words. In Paphlago (I. 239. 18) it appa- 

 rently falls on a short penult : if this assumption is correct, it may 

 be derived from the oblique cases iPaphldgonis)?- But it is ill 

 arguing from isolated instances. 



From our limited material it wovild seem that only unfamiliar 

 proper names retain their Greek accent when that would fall upon 

 the ultima — that familiar words, and especially common nouns, throw 

 their accent back to a place usually but not invariabl}' determined 

 by the penultimate law. 



Ammianus almost always retains the Greek accent when it falls 

 upon the penult or the antepenult. A notable group of exceptions 

 however, is formed by place-names ending in -la, all of which he 

 accents on the antepenult, whether they come from the Greek or 

 not. Thus 



Bithynia I. 37. 34 : 276. 5 : II. 85. 1 



Sicilia I. 43. 22 



Dacia I. 48. 17: II. 101. 20 



Syria I. 76. 7 



Aegyptia I. 119. 4 



Mesopotamia I. 123. 3: 154. 18: 208. 18 



Pannonia I. 135. 1 : 189. 19 



Armenia I. 222. 2 



Hiberia I. 241. 1 



Mygdonia I. 277. 6 



Paphlagonia I. 277. 27 



Cilicia I. 309. 32 



iMedia I. 311. 22 



Assyria I. 324. 9 



Phrygia I. 324. 25 



Arachosia I. 335. 25 



Cajipadocia II. 59. 19 



> Cf. hidicioli above, p. 212. * See p. 171. 



