446 H. B. Dewing, 
8. ree Eis tas “Adners.} 138, 4. 
9. wet dvvauEews coxovons. 147, 18-14. 
10. xareoxerdace ToLovde. 149, 17. 
11. xa haxxivagioe xei Bixtwoes. 156, 1. 
12. avaody mohooxias. 156, 16she 
3. énecta dé mAéious. 156; tap 
14. of LHépow mheovexteiv edogay - 165, 14. 
15. 1 oroat@ naoédwzer. 162, LO—195 
16. ov xahotor “Poucioe udycotoor * ? 165, 6. 
1%. tov mhoiwy éntBuivery. 159, 19—20. 
18. xaréotnouy eis yeious. 165, 4. 
19. dia méors FEwpovuevov Tis iuéous. 140, 15. 
20. cerijoews a5 elyor. 150, 15. 
21. Mehevxias ovoualouérngs. 158, 8—9. 
22. nohéue yeyvuraouévor. 125, 20. ° 
23, ano dequatwy dé mEvtaxootur. 142, 28, 148, 1. 
24. t@ Tovisavod cawate. 172, 8. 
Among these 30 cases the irregular forms occur as follows: 
Form 0 6 cases. 
? 1 9 ” 
fy MeL a5 > 
3? o 3 ee 
It is noticeable that almost one half of the irregular cases are of the 
3 form. There seems reason to suppose, therefore, that this form of 
clausula was the least objectionable form, at least less objectionable 
than forms 0 and 1. This seems the more reasonable because in 
prose where no accentual cursus is present both form 0 and form 14 
occur more frequently than form 3. 
The situation in Zosimus may be summarized thus: 80+ °/9 are 
absolutely regular, reading by the last two written accents; 10+ °/o 
are regular by counting out the accents of weak words ; 9 + °/) remain 
which seem distinctly to violate the cursus rhythm.? 

1 yee ’o te ‘Adnecs would be a regular 2 form, (considering the form of 
the preposition é¢, with apheeresis). 
* ucycotoor, if pronounced according to the Latin accent of “ magistrum ” 
makes a regular form 2, but there seems to be no justification in as- 
suming such a license; in the following case the use of the Latin accent 
would destroy a regular form: 
os GZOLTMOLO’S TODGKYOQEVOLOLY 165,10 
8 This method of counting regular cases first as marked by written 
accents, brings into the regular forms cases which do not really show 
strict conformity to the form which the last two written accents alone 
ew 
