Pavel A. da Mek wrote:
> Another thing which seems to me suspicious is -ddh- > -d-;
> is not there a soundlaw which produces rather a sibilant in such clusters?
Nothing unusual here. In Indo-Iranian, the PIE sequences *-d[z]d(H)- and
*-t[s]t- (the affrication of the first member of the geminate was
allophonic) were inherited unchanged. The fricative portion of the first
segment was then lost in Indic, where the result is -dd(H)-, -tt-, while
in Iranian the plosive portion was deleted instead, yielding -zd-, -st-,
cf. Gathic Av. v&r&zda- 'big' vs. Skt. vr.ddHa-. This is why we have Av.
zrazda: but Skt. s'raddHa:.
Piotr