J. McKEEN LEWIS,
of all Attic inscriptions of the fourth century, we might well find that
stonecutters had, in the absence of any orthographic standard, here
and there written o-KeX.r> and o-KeA.ee for o-Ke'Ae, just as they wrote
XoAA^St?? for XoXXySrji, and 'AAaiees for 'AArnJs. As it is, however, in
view of the unparalleled elaborateness of the Greek vowel-system, the
comparative infrequency of such errors bears witness to the mar-
vellous accuracy of the Attic ear.
It is thus evident from the interchangeableness of e, et, with e, ee, 77,
in the fourth century, that e was not changed, in pronunciation as in
writing, to et, but that, conversely, the diphthong must have become
simplified until its symbol could represent both sounds. This degra-
dation began earlier, and was more quickly consummated, where et
was followed by a vowel — a well-known instance is the word 8u>pad,
later Attic Sojpeu.1 This change is only a manifestation of one of the
most important laws of Attic speech, — that law by which a semi-
vocal t is avoided through the dropping of t between almost any pair
of vowels. As et before vowels becomes e at a very early date,2 so
words like cAata, kA<xiu>, Iletpatevs, crrota, 7rotet, vto's, yeyovuto, lose their
l by the best Attic usage. The same principle is observed in all
erases where final t occurs, as in Ka/xoi, ov-ixwpwi, ^JSt™?, rrjKKX-qirLi^
twttlovti? But the degradation of diphthongal et, independently of
this law, is illustrated by the equal corruption of -qi to et or e during
the fourth century.
Confusion of rji with e and et. Not long after the year of Eucleides,
and simultaneously with the falling together of e and et, the diphthong
7)i becomes interchangeable with these. After 375, such forms as
7rdA?7 for 71-oA.et, f3ov\d for f3ov\fj, grow frequent; for rjpeOr/ is found
with e. Probably the dual of ti6Xis in Attic was Tr6\e (7r<iAei); if ir6hr) arose by
contraction, its parent form was not Tr6\ee, but 7roA?je.
1 Older and newer forms are sometimes found side by side, as Supelav, Swptav,
II. I b (circa Eucl. arm.), Upias, tepeias, 573 b (circa 350?) ; the difference in these
•cases is, of course, merely orthographic, et and e standing alike for a simple closed
^-sound. Cf. Meisterh., p. 19 and notes.
2 Cf. Alvea, I. 478 (sixth century); TleAeaTat, 230 (450), AiVecurat, 234 (446);
Te'Aeoj, IV. 3 (before 444); Neaj/Spea, I. 240 (440); BpvAAeauot, 247 (432);
'AAanre/tee?, 184 (412); 'AvSpea, 324 (408); Trpvrayeov twice, II. I b (c. Eucl. ann.),
iepea, IV. 553 a, 3 (fourth century).
3 So there is every reason to write in Attic rovTuii, ixetvoi, etc., in place of
the unpronounceable tovtou, Ixeivoit