Rob wrote:
> Couldn't that be from analogy with _as-_ 'be'?
I don't think so, because <a:sat-> is the antonym of <s-at->, not of
*<as-at-> 'existing' (nonexistent :-)). The analogy of <sat-> produced
the "regularised" alternative negated form <รก-sat-> (with a short /a/,
also occurring in the RV). As there is no analogical explanation for the
form with <a:->, it's rather clearly of phonological origin.
> Or do you think there
> was still a phonetic schwa adjacent to the laryngeal?
Why a schwa? The initial sequence *Hs- regularly gives RV s-, which
means that the laryngeal (still preserved in Proto-Indo-Iranian) was
consonantal in this cluster. It caused the lengthening of the reflex of
*n. as it also did in word-internal contexts.
Piotr