Glen:
> m. i-dem + f. ea-dem + n. i-dem = Jens analysis is wrong
Jens' analysis is right, Glen. There's a lot of rearrangement going on.
id-em (neuter)
gives as part of its original paradigm, the ablative eod-em, and the
feminine ead-em.
With the loss of the final d on ablatives, the pairs
eo / eodem and ea / eadem arose
so eodem was re-analysed as eo-dem, and eadem as ea-dem.
The suffix -dem was then used in place of the original -em.
This is why the Latin masculine is i:dem < is-dem (also attested), and not a
formal equivalent of the Sanskrit ayam. The neuter remained idem, not
iddem. The late formation is suggested by eosdem & easdem which retain
the -sd- which is lost everywhere else.
So I think Jens must be right when he says:
>since it is the same pronoun that occurs in Latin id and Sanskrit
idam, it is worth trying to look for a match for the Indo-Iranian -
am which sits on practically every pronoun of that branch. And in
the one pair idem : idam we can really have a nucleus from where it
can have started.
Peter