Re: Convergence in the formation of IE subgroups

There is a considerable

> difference between on the one hand picking one of several
> *related* variant forms simply because that specific variant
> fits one's theories, and on the other hand choosing one of
> two *unrelated* etyma. In the former case, it seems to me,
> one ought if possible to use a reconstructed ancestor of the
> dialect variants; any etymology or connection that you're
> proposing for one variant must (with suitable modifications
> or additions) apply to the related variants as well.
> Inferences drawn from <hound>, however, say nothing at all
> about <dog> and vice versa.
>

I agree.

Torsten