> In Old Nordic there was a similar development, but the literature
> says that it affected /f/ and the interdental /th/ only.
> Vernerian /z/ had become 'palatal r', i.e. /z~R/ at that stage in
> word-final positions and had been assimilated or lost in other
> positions. Now my question: why the hell was there no new /z/
> from /s/ like in English??

I don't know :) But I can think out loud a bit...

First of all we don't know for sure what the phonetic state of the
Old Norse descendant of Vernerian /z/ was at any given time. It may
well have remained close to [z] for quite some time. Perhaps it was
still [z] at the time of this second voicing. This might have discouraged
/s/ to join its brethren fricatives in acquiring a new allophone.

> Or was there, but it was not reflected in
> orthography (voice was normally not marked in runic writing)??

I suppose that's also possible. This would not necessarily be
reflected in the orthography (runic or latin). I suppose the
suprising thing would be why this voicing didn't survive into
the modern languages? Or did it maybe?

> And if
> there were new instances of /z/ (Brondum-Nielsen, Gammeldansk
> Grammatik argues that occasional <z> spellings in later manuscripts
> could reflect voiced /s/), what happened to these?

A devoicing in Danish seems entirely reasonable to me.

Kveðja,
Haukur