Torsten:
> An interesting little idea suggested by Prof. Witzel concerns an
> alleged alternation k/zero,
Actually, you're mixing up a lot of things together as usual.
Greeks sometimes used "k", sometimes "h", sometimes nothing to represent
foreign 'h'-like sounds that may not completely have corresponded to the
Greek 'h' as far as the ears of its native speakers were concerned.
Another thing is that IE itself seems to have, for example, an alternation
of its own between *q- and *x- in the 'bone' root as well. However, this
isn't particularly mysterious if you accept that they're both uvular. A
tendency for a uvular stop *q- to erode into a fricative *x- is a common
and natural one in many languages. As you've pointed out with similar
dichotomous sets, IE does have a few roots showing this pattern but
it doesn't show us anything other than an irregular sound change during
the period of a fracturing IE.
= gLeN