428 H. B. Dewing, 
75 + /) of the preferred forms, while the remaining 24 -+ °/, fall 
into the avoided forms (five in all), while in Demosthenes, Lysias 
and Pausanias the percentage of these preferred forms (of Zosimus 
and Zonaras) falls considerably below 50 °/o. 
These comparisons show a striking contrast between the two kinds 
of prose, and the characteristics of the accentual cursus are per- 
fectly evident. It is not merely true that writers who use the cursus 
‘avoid forms 0 and 1, but also forms 8 and 5. As to what the law in 
general is there can be no doubt; the figures of themselves are 
convincing. Assuming nothing more than what is plainly demon- 
strated, the law may be stated in negative form thus: The cursus 
law requires that forms 0, 1, 3 and 5 be avoided. The original 
statement by Meyer must be revised; form 3 is as certainly avoided 
as form 1. Exactly the same forms are avoided as in the Latin 
cursus. This is no mere trivial correction, as may be readily seen 
by returning to Litzica’s work. 
In the first place, his mathematical computation of the possibilities 
of the language of Leontius gives an entirely different result when 
forms 3 and 5 are counted among irregular clausule as they must 
be. Counting thus the following result is obtained: out of a pos- 
sible total of 148, 157,564, only 61, 161,874 regular combinations are 
possible—that is, less than half. The proposed test of 90/9 re- 
eularity for writers who have a cursus falls absolutely to the ground. 
The total of regular cases which may be expected where no law 
is present is really 50°/) or less, not 80°/, as Litzica holds. 
The other point in which Litzica arrived at a wrong result through 
his misunderstanding of the law is in the comparisons made between 
prose which conforms to Meyer’s law and that which does not. 
He finds 69°/) regular in Demosthenes, as against 95.35 °/) in Zosi- 
mus; this presupposes only forms 0 and 1 irregular. Now if the 
percentage of forms 3 and 5 are subtracted from this estimate, 
Demosthenes falls to 47 °/, regular, with a loss of 22°/), while Zosi- 
mus loses no more than 4.5°/) and still stands above 90°/) in reg- 
ularity.' The mere fact that the loss by excluding these forms is 
so different in the two cases is of itself convincing. The same 
striking contrast can be pointed out between other writers: for ex- 
ample, of the 3 and 5 form together there are found 24.3°/9 in 
Lysias, as against 4°/) in Zonaras. The contrast in the 3 and 5 
1 These figures show results when written accents are treated as 
Litzica does—counting out the article etc.—in order to have a safe basis 
of comparison with his results. 
