> That opens a possibility to solve the old
> puzzle of why the 2sg pres. ind. of *h1es- "be" has only one -s- (Skt
> ási, Av. ahi, Gk. ei~)...
It is a puzzle? The morphological cluster *-s-s- seems to have been
simpilified in PIE and IIr., not only in *h1es-si > *h1esi but also
elsewhere, cf. e.g. the loc.pl. of -es stems, Skt. -as- + -su --> -asu.
The analogical restoration of the *s in the 2sg. form would have been
understandable (and certainly happened here and there), but why should
*h1e-si have been resegmented as *h1es-i?
Piotr