Dennis wrote:

> > But looking now at the list of
> > exceptions given there I note that 'assae' gives 'assu' and that DIL
> > itself gives the comparative of gasta as gastu. Does 'gasda'
> > resist/lose the palatalisation for similar reasons to 'assae'? Can
> > anyone walk me through that?
> I don't know why we have "ard > ardu" instead of "airdiu" in OI.
> The odd thing is that the general rule is observed in the modern
> language: "ard > níos airde".  OI also has "tromm > trummu", while
> today it's "trom > níos troime".  I bet David can explain this.

The pre-OIr. comparative suffix was *-iyu:s the "i" of which 
palatalised all it could palatalise at the time of the great 
palatalisations. But as a rule, consonant cluster such as in "ard" 
and some other consonants after certain vowels (e.g. in "tromm") 
resisted palatalisation. Early OIr. retains these differences to a 
high degree. Therefore the apparent irregularities in OIr. The later 
language tended to introduce palatalisation as a morphological marker 
analogically wherever it was possible.

As to "assae": The reason why there's no palatalisation there is, I 
think, that the word ended in *-owiyo- in pre-OIr. The intervening 
"w" prevented the palatalisation of "ss".