Greek z and IE *j

From: Andrew Jarrette
Message: 58227
Date: 2008-04-30

Diverging from recent topics and going back to a subject more typical
of earlier threads on cybalist, here is my question:

Why does IE initial *j appear sometimes as Greek <z-> and sometimes as
Greek <h->? I delved into the archives but could not find any
satisfactory discussion of this problem, mostly by-discussion that produced
theories for which there was contradictory evidence. Can anybody offer a
solution now?

Andrew