Richard Wordingham wrote:
> Non-Latinness!

Eh? :-)

> The same applies mutatis mutandis to Hebrew geresh
> and Devanagari nukta.

Yes, that's what I meant.

> [..]
> English <wh> is a bit of a problem, as <w> isn't Latin,

<W> is not uncommon in medieval Latin, as e.g. in the proper name
"Willhelmus". But in older documents it is still spelled <uu>, which
explains the English name of the letter.

> but then it's also curious in that it is, at least in places, [hw].
> I think <wh> is best regarded as a degeneration of <hw>,

I think in fact that <hw> (or, well, <huu>) in place of modern <wh> is
widely attested in ancient documents.

> As to English <sh> and <th>, I could pedanticaly suggest that 'h' in
> <sh> and <th> represents half-way Irish-style lenition for 'sh'
> and 'th' - there are lenition sequences [s] > [S] > [h] and [t] >
> [T] > [h]. However, it doesn't work for soft 'ch'.

I though that h-digraphs were a relatively recent innovation in Irish
spelling: in the uncial script (which was ubiquitous in Ireland up to at
least the 1970's) aspiration is indicated by a dot over the letter. I guess
that the switch to <h> in roman type might have been caused by the fact that
tall lowercase letters, such as "f" or "t", look ugly with a dot on them:
lowercase uncials have no ascenders.

IMHO, the most likely origin for all H-digraphs are the <CH>, <PH> and <TH>
spelling used in Latin to spell Greek loanwords.

These Greek sounds where originally aspirated stops, so the <stop+H> was
quite motivated. When these sounds evolved to fricatives in Greek, Latin
followed suit, as is demonstrated by the fact that <PH> is now read [f] in
all European languages. The Latin spelling of these Greek fricatives must
have been a became precedents for using <H> together with any letter to form
spelling for new sounds.

E.g., the spelling <dh> for [ð] is already present in the Sarments de
Strasbourg, which are AFAIK the oldest sample of written French and German.