26 May 2011
I've been looking back recently at my deviations and notice that my attention has veered away from Proto-Indo-European for quite some time as I obsessed over the separate topic of all things Etruscan, Rhaetic and Minoan. No particular reason for this, really, it just happened that way. Sometimes one's learning journey can be quite meandering like that, much like a spider weaving her web, connecting the dots of a large area, slowly filling in the spaces one thread at a time. Recently though I came across something that gives me cause to come back to that topic.
Among the pile of unlikely roots Julius Pokorny had attempted, *kais- (that is, *qais- in the revised PIE notation) is an ugly one but an interesting one. Just the three following cognates are indicated:
*a, etc. This alleged root then remains more a set of questions than a single, decisive answer. Because of this, I for one don't accept it as part of genuine Proto-Indo-European vocabulary. A better solution to explain these terms seems to involve non-Indo-European substrate.
Simultaneously, Proto-Aegean *zira 'hair' (presumably from Old Egyptian *θīra 'hair, thread', later written sr) explains away Greek kithara through Minoan *ki-zera 'three-stringed', exactly parallel to Persian seh-tar (سه تار), Chinese san-xian (三弦) and Japanese san-gen (三絃) which all mean the same. That Greek term has long been suspected to be a Minoan term anyway so I'm not saying anything terribly deviant. I can only expect that the proper reflex in Etruscan is *zir /tʃir/ which looks nice beside Latin cirrus 'tuft of hair, a lock, a curl' whose etymology otherwise appears unknown.
And so, addressing the unlikely PIE root above, in light of our collective ignorance on the matter, would it hurt to suggest an Aegean alternative, something like **kai-zera 'hair on head', for consideration?
 Pokorny, Indogermanisches etymologisches Wörterbuch (1959), *kais- (see link); Wyatt, Indo-European /a/ (1970), p.31 (see link).
 Upon some reflection, it's of course possible that, under the revised notation, the [a] is a uvularized allophone of *e in underlying *qeis- (ie. traditional "vowel colouring"). Nonetheless this helps naught in explaining its derivation.