te 
_ ia 
Accentual Cursus in Byzantine Greek Prose. 455 
above; they are perfectly regular if the hiatus be allowed to stand. 
Here, then, are 229 cases which will not show conformity to the 
regular forms if elision be consistently introduced. The number of 
syllables in arsi being lessened by one, 2 forms become 1 forms, 
and 4 forms become 3 forms. In the same way 35 forms fall to 
2 forms, and 5 forms to 4 forms; since the 3 forms as printed, 
where elision is possible, number 13, and the 5 forms 7, we should 
have only 20 regular cases left from the 421 before us. This strange 
result makes it seem unlikely that elision is to be consistently ad- 
mitted; taking this body of cases alone, and avoiding elision, we 
find the six forms of clausulz standing in about the same proportions 
as those found above to be normal for the Persian War; whereas 
consistent elision makes the proportions of the various possible 
forms absolutely different from the normal. The percentage of 
regularity is divided by ten; not only this, but form 1 would oc- 
cupy almost half the field (191 out of 421). But since this body 
of 421 clausule is not taken at random from the whole, it is not 
safe to insist that the different forms can be expected to stand in 
the proportions found normal for the whole Persian War. But full 
and satisfactory evidence is at hand in the usage of other writers 
who show an accentual cursus rhythm and in the Political Verses, 
as set forth above. This evidence makes it clear that the following 
principle should be adopted for Procopius: elision need not he 
consistently introduced in reading clausulae, nor yet consistently 
avoided; the manuscripts are not a safe guide in such a matter, and 
elision may be introduced when the cursus rhythm requires it, and 
elided vowels may be restored against the authority of the manu- 
script tradition. On the other hand, the frequent hiatus which is 
produced by this procedure makes it plain that Procopius is not 
a careful writer according to the best standards of his time. 
There are, as stated above, 229 cases which may be read as 
regular clausule by permitting the objectionable hiatus, without 
departing from the written accents. 
Form 2. 191 (2, V) 
a ee * - 
EVEVETO WIE. 122, 7-8. 
colote lve. 88, 24. 
, 3 ’ Lay | lof 
toge Evyoyour. 212, 1%. 
Form 4. 38 (4, IV) 
mony yEhhouevee ExTELELY. 514, 7 
év to) maodvt avdounyadilecdee. 468, 13. 
ibumuoe wvoucdleto. 461, 5: 
