Matt Pearson (I think?) wrote:
> > > /kp/ --> /kw/, on the other hand, I find hard to picture.
> > Epenthetic
> > > vowel /k@p/ or simplification (/k/ or /p/), or even a click /p!/ I
> > find
> > > reasonable descendents of /kp/, but not /kw/.
> > >

To which, On 25 Jan 00, at 21:32, Steg Belsky added:

> Maybe a transition stage where the /p/ becomes an approximant, which is
> then voiced?

I agree.  I was actually fiddling with putative sound-change rules from
Proto-Indo-Slavic into Proto-Wene-Thagojian (and beyond), based on my
current ad-hoc lexicon and stumbled across a believable kw -> p sequence
(in x-sampa):

k_w -> x_w -> W -> p\ -> p

As to why the initial velar weakened in the first place, I leave that as an
excersize for the reader, though in doing so I point generally in the
direction of centum/satem for no apparent reason.