In both Sindarin and Noldorin, the cluster [sk] usually became [sg], at least in writing:

• ✶skalā > S. esgal “cast shadow” (PE17/184).
• S. lisg “reed” in Lisgardh (UT/34) vs. ᴹQ. liske (PE19/51), though the primitive form of the Quenya word is given as ᴹ✶lisge.
• S. rasg “wain” vs. Q. raxa [raksa] (PE19/51), probably primitive *raskā.
• ᴹ√MISK > mesc > N. mesg “wet” vs. ᴹQ. miksa (Ety/MISK).
• ᴹ✶askarā > ascar > N. asgar “violent, rushing” (Ety/SKAR).
• N. osgar “cut round, amputate” (Ety/OS), probably os- + car-.

There are a few cases where this does not occur: N. rhosc “russet”, S. rusc “fox” and the river name S. Ascar (compare N. asgar above). The last of these might be an archaic name. For the other cases, perhaps final -sc only became -sg after front vowels a, e, i, and was preserved after back vowels o, u. This limitation does not seem to be true medially, for example: N. osgar (though this could simply be ordinary lenition as well).

There is an example indicating a similar sound change for [sp]:

• ✶spini- > S. esbin “thin thread, tress” (PE17/17, 119).

As a counter-example, we see N. osp “smoke” (Ety/USUK). Perhaps [sp] followed the same pattern as [sk], preserved (finally?) after o, u, but there are not enough examples to formulate a clear theory. Note that the cluster [st] is preserved medially and finally, for example: S. esten(t) “short” (PE17/185; WJ/311), N. ast “dust” (Ety/ÁS-AT).

In notes on The Feanorian Alphabet from the 1930s, Tolkien said:

The letters sp, st, sc are in later orthography often replaced by full writing [special symbols]. For sp, sc there also appear sb, sg. The phonetic intention is voiceless s + voiceless media b, g [p, c] - since in these combinations p, c have not their slight[ly] aspirated normal pronunciation (PE22/36).

Thus in this document it seems the sound change was not to full voicing, but rather merely the loss of aspiration. It is hard to tell if this pronunciation applied to Tolkien’s later examples. @@@ make pronunciation notes in relevant entries.

Conceptual Developments: There are quite a few Gnomish examples from the 1910s that show sc after every possible vowel: G. fasc “clean” (GL/34), G. hesc “withered” (GL/49), G. tisca- “tickle” (GL/70), G. nosc “damp, wet” (GL/61), G. musc “grey” (GL/58). Other examples show sc/sg variants: G. asg/asc “bone; stone of fruit” (GL/20), G. losc/losg “rye” (GL/54), G. usc/usg “fog, mist” (GL/75), G. lisg/lisc “reed, sedge” (GL/54); compare the last example to S. lisg “reed” above. The variety of forms makes it hard to tell what is going on.

The few Early Noldorin examples from the 1920s mostly show sg: ᴱN. asg “bone” (PE13/137, 160), ᴱN. lhesg “sedge” (PE13/148), ᴱN. esg “sharp rock” (PE13/143), the last with variant esk. Of these, ᴱN. asg “bone” has a cognate ᴱT. axas which hints at a primitive form with [sk] (PE13/160), but the phonetic developments of the other examples are unclear.

@@@ exception: rhosc