68 JOURNAL AND PROCEEDINGS. 
where we see that the forms with the short vowel are those before an 
accented suffix -ma and -na; whereas the diphthongs (Sanskrit é is 
generally for ez or oz) are eae the accent. 
This is the secret of the difference between daz, bibheda, bad, 
pcpoitha on the one hand, and dzt-on, bebhedema, bidon, e-pe-pith-mén 
on the other; in other words, here gradation is a result of accent 
difference. This conclusion is absolutely certain, being vouched 
for by Verner’s success in explaining by it such differences as : 
A.S. weorthan, pret. wearth, _ pl. wordon, pp. -en. 
O.H.G. werdan, cre awards “ wurtum, ‘“¢ worten. 
O.Fr. wertha, Ce awartine ‘¢ worden, “¢  worten. 
But can we explain the difference between Je/d-e and bad, peith-o 
and fé-poith-a? Would it not be a natural supposition that this 
also is due to difference of accent? ‘This, some writer, Mceller, I 
think, asserted some fourteen years ago in Kuhn’s Zeitschrift, and 
later on Mr. Sweet has adopted without hesitation. He says, 
‘There are three accents in Sanskrit, raised=acute, unraised = 
grave, and ‘swarita=circumflex. The acute is the emphatic accent, 
and was either a rising or a high levei tone. ‘The syllable imme- 
diately following an acute is always circumflex—that is probably a 
falling glide tone—unless an acute follow. Every syllable before as 
acute or after a circumflex is grave. And again, in parent Aryan 
every vowel had a different form under the various accents. The 
most important of these is the ¢/o series, which is evidently a 
modification of original a. Under the acute accent a became e, 
under the circumflex, 0, and under the grave it was dropped alto- 
gether. This would account for méno, memona, mémamen for memn- 
men, and with like treatment of the vowel-element of the diphthong 
peitho, pipoitha, epe-pith-men, e-pe-pith-més, the vanishing of the e/a 
leaving 2, just as the loss of e in ez (for, as Gustav Meyer says, it is 
unmethodic to deny the term diphthongs to such terms as ex, em, 
el, er, parallel as they are in pathology (so to speak) with e7 and ew) 
leaves 2, which, between two consonants, must become the sonant 
nasal represented in Greek by a. So in dinde (for *bende), pret. 
band, pl. bundam, pp. bunden, the ux of the pl. and pret. is the Teu- 
tonic representative of the same sonant nasal. 
This explanation presupposes that the vedic accent, dzbhéda tor 
instance, is not original, but that originally the reduplication as 
