PIE *Pek-u in Celtic languages

This discussion of Hamp's Dog got me thinking about a problem that I have
been working on for quite some time - I don't know quite how relevant it is,
but let me go ahead anyway.

In the Gaulish language Calendar of Coligny (inscription dates to +/- late
2nd century), we see a month name EQUOS, which most scholars prefer to see
as a religious Q-Celtic archaism for (P-Celtic) Gaulish Epos "horse." This
follows the rule that PIE -Kw- gives Gaulish -P-. The problem is that the
calendar otherwise contains technical terms containing -P-, which
demonstrates, to my mind, that this theory might be false - why would some
terms retain archaic features and not others? Why then doesn't every Gaulish
religious term retain archaisms (ie, Epona "horse goddess" should be
*Equona).

In any case, it must be noted that the Gaulish grapheme -QU- may have
sometimes been employed to represent a Gaulish -C- (from PIE *K-)+-U- (PIE
*U), perhaps to be seen in Sequana from *Sek-u-an-a (?), or in compounds
using PIE *Kom- when followed by a -U- - thus, we get the Gaulish personal
name Quigon from *Kom-Ueg- "companion" (note Welsh Cywain "convey" from
*Kom-Ueg-n-) and possibly the tribal name Quariates from *Kom-uari-at-es. I
should add that the Sequani and Quariates are often noted as proof of the
survival of Q-Celtic on Gaul, but all the inscriptional material from their
tribal area points only to P-Celtic usage - therefore the unlikelyhood of
their name preserving an archaic form of Gaulish.

Following the -QU- from -K-+-U- theory, we might see EQUOS not as "horse,"
but rather as *Ek-u-os (the masculine singular -os ending is found attached
to the majority of months in the calendar -seeming to match Latin [-us] and
Greek calendric practice and may not be normal grammatical usage). This
*Ek-u-os may come from a PIE *Pek-u- (with Celtic loss of PIE -P-) with an
added masculine stem for what might have otherwise been a neuter stem in
Gaulish (perhaps the root of Welsh Gwartheg "cattle" from Common Celtic
*Uerto-ek-(u) "value-flock").

For other alleged proof of Q-Celtic survivals in late Gaulish, we also see
in Gaul some placenames which have been reconstructed as coming from an
original *Equo-randa (I am not sure at the moment if this form is attested
anciently - it may be). This is normally interpreted as an archaic
"horse-territory," once again where we would otherwise expect an *Epo-randa.
Might it rather refer to some development of PIE *Pek-u- and have the
meaning of "flock-territory," or some such derivation?

On the "dog" front, we have the following forms:
PIE *Ku-on = Welsh Ci, Irish Cu (should make Gaulish *Cuon)
PIK *Kun-o = Gaulish Cun-o-, Old Welsh Cyn-, Irish Con-
Along with my proposed "flock" forms:
PIE *Pek-u- = Gaulish Equ-os (?) and *Equo-randa (?), Welsh (Gwarth-)eg (?)

Sorry if my post was a ramble - anyone have any comments?
Chris Gwinn