> The source of <s.> is IE *s in the "ruki" context, not a velar
stop, so it's quite clear that <kH> is secondary. Quite likely, the
spelling was employed to represent a velar fricative pronunciation
[x] in the dialects. The development of "ruki" *s^ to *x is attested
in Slavic. A similar development has taken place in Spanish (the old
pronunciation of <j/x> was [s^]).
So then what we're seeing is probably s. > x > kh? In other words,
it's an uncommon but not unheard of "un-satemization"?