From: Piotr Gasiorowski
> The intermediate form must have been -kWkW- which could be changedI agree. The 'horse' word may be related to Gk. o:kús, but the
> by analogy to -kYkY- if obviously related to a word with -kYu-.
> This only works with kYw not kw; *LakYu+s has e>a by L.But Gmc. laxsa-, Lith. la~s^is, Slavic *lososI all mean 'salmon', and
> *LakYko+s 'little pond dweller, fish'