IE thematic dual

[Jens Elmegaard Rasmussen]

On Mon, 10 May 1999, Richard M. Alderson III wrote:

[... (JER: Goth. ahtau has diphthongal form of them.dual., like Skt.
as.t.au, ergo the -u was IE already:)]

> The final <au> in Gothic is a spelling of [O], and does not bear on the
> presence or absence of a final *-u in the Indo-European dual, nor is it
> entirely clear that the word for '8' is a dual.

But what can then be the source of the Goth. -au, how ever pronounced?
Both *-o: and *-a: give Goth. <-a>. And can one really reduce the weight
of Kartvelian "otxo" 'four' to nil? I know people are biting each other's
heads off over the exact shape of the protoform, and perhaps also over the
direction of borrowing (supposing any of the two families was the source).
Does this not still leave the unbiased observer with the impression that
the word for 'eight' _was_ an o-stem dual at one point?

> The Skt. -u is an extension from the u-stems via sandhi variants to the other
> stem formations.  Or so I was taught.

I cannot disprove that. Still, is it likely? Are there so many u-stems
that they can have reasonably been used as models for the thematic stems,
including pronouns? - And does the parallelism between *to- + *-e and
perf. *ple-ploH1- + *-e emerging as ta:(v) and papra:(v) count for
nothing?

Jens