I'd been wondering why some forms with final vd.stops show lenition (e.g. 
*-ib > iw) while others do not (e.g. *-ib via **i:b or **-iwb then **i:p, 
**-iwp, ult. modern /-i? ~i:? ~˙?/. Surely not just the usual culprit, 
dialect borrowing.... Now I know: it's due to ancient fossilized 
morphology-- some C (voiced; a nasal?), that perhaps marked some case or 
other. Thus *-ib# lenits, but *ib+C does not, in fact the 
lengthening/breaking is compensatory. (And if the final is vl, the +C will 
voice it, non-leniting)

There are already fossilized derivational prefixes, m-, s-, i- or infix -y- 
and others yet to be, discovered, that produce variant forms.

Another discovery just last night: the ancient God of Healing is called /ah 
tiN/, < *as pirín  ;-))  The modern medical guild/association will be _ah 
ting GUILD_.

A couple bright spots in the really boring process of summarizing ALL the 
damn sound changes.

The "Proto (Baw Da) Gwr" I'm working with is probably comparable to 
Proto-Italic; we have yet to find the PIE equivalent.