In both Sindarin and Noldorin, where it was not otherwise lost, a final [w] usually became [u], most notably after a consonant as described by David Salo (GS/§4.195). In Tolkien’s writing, he sometimes represented this final [u] as -w, but this seems to be an orthographic convention, not the actual pronunciation. He mentioned this sound change a number of times:

• But after other consonants (though not s) when w- became final owing to the loss of a Common Eldarin vowel, w became vocalic, as in [Q] Elwe = [S] Elu; Q. kurwe = [S] curu (VT41/08).
• In Sindarin adoption of Quenya names (as Voronwe > Bronweg) -we was sometimes used to represent -we, which historically had become w̯ or u (as in Elu = Elwe). But this S. -we is of distinct origin, √WEG- “live, be active” (PE17/189).
• N.B. final -w (left after loss of vowels) in Sindarin was dropped after labials (-mw > mm anyway): after other consonants [it] became ŭ or was intruded like y but without alt[ering] of the preceding vowel. So matwā [>] madw̯ > maud or madu. teswā “[?chip]” > teχwā > teχw̯ > tewch (PE17/148).

This last note is interesting because it hints at two possible alternate phonetic developments for final [w] after a consonant: (a) w-intrusion (analogous to i-intrusion) and (b) w-loss after labials. There is no evidence for w-intrusion beyond this one note, so I think this was probably a transient idea. As for w-loss after labials, this would have been quite rare, because w-suffixion after labials was rare in Primitive Elvish:

This is in accordance with the general avoidance of consonantal [w] after labials and special favoring of [w] after k-series, that is seen in Quendian (PE18/82).

Despite this quote, [w] after labials does occasionally appear in Primitive Elvish, and there is one Noldorin example that seems to indicate the loss of [w] finally after a labial consonant:

• ᴹ✶hamwa > N. ham or hanw vs. ᴹQ. hamma (EtyAC/KHAM) [gloss unclear but perhaps “chair”].

Of the two Noldorin forms, hanw [hanu] is the expected development if one assumes that medial [m] became [n] before [w] as it did before [j], as David Salo did (GS/§4.31). But it seems Tolkien wasn’t entirely committed to this sound change, and so the other form ham may represent an alternate development where [m] remained before [w] which was later lost after the [m]. On the other hand, the last Sindarin quote above says that -mw > -mm, so the phonetic development could be hamwa > hamm(a) > ham with the usually shortening of [mm]. The rarity of these sound combinations make it difficult to determine the exact set of phonetic developments.

In addition to becoming [u] after consonants, final [w] also became [u] after single vowels to form diphthongs, though the distinction between the pronunciation of semi-vowel [w] and diphthongal element [u̯] is minuscule. Nevertheless, Tolkien explicitly described this sound change for both Sindarin and Noldorin:

• In S. -w- did not vanish or bec[ome] absorbed while foll[owing] vowel remained, but became u̯ and formed diphthongs when the foll[owing] vowel was lost (PE22/152).
• But the later [Noldorin] diphthongs due to loss of final vowels or other causes ew, iw were written ln `Bn. Similarly final āu̯ — from áw + lost vowel or from au̯ + lost m, b, w, or from stressed final au̯ in monosyllables — was always written ]n: transcribed aw (Feanorian Alphabet, 1930s, PE22/37).

As the second note indicates, this sound change was obscured because Tolkien transcribed the resulting final diphthongs as -aw, -ew, -iw. The final diphthong -ow would not appear, because [ou] became [au]. However, it seems that in cases where there was already a diphthong, the [w] did not reduce to [u]:

• S./N. aew “bird” (SA/lin¹, Ety/AIWĒ).
• S./N. gwaew “storm; wind” (NM/237, Ety/WĀ).
• S. oew “evil deed” (PE17/170).

The Noldorin note shown above also indicates that in monosyllables the result could be a long diphthong, which might be represented in Tolkien’s transcription of words like S. têw “letter” (WJ/396) and S. rhîw “winter” (LotR/1107). Tolkien never wrote -âw in monosyllables, however, so perhaps this always reduced to a short diphthong. It’s hard to say for sure; this could simply be varying orthographic conventions.

Conceptual Developments: Final [w] becoming [u] after consonants also seems to be the common pattern in Gnomish and Early Noldorin in the 1910s and 1920s, as noted by Roman Rausch in his Historical Phonology of Goldogrin (HGP/§2.7) and Historical Phonologies of Ilkorin, Telerin and Noldorin around 1923 (HPITN/§4.1.4). As noted above this was obscured by orthographic conventions, since Tolkien also often wrote -w for final [-u] after consonants in the Early Period, for example: G. colw [kolu] “black” (GL/26), ᴱN. fadhw [faðu] “parchment” (PE13/146).

The same seems to be true for final diphthongs as well, though Tolkien consistently wrote -au instead of -aw in Gnomish: G. mau “soon, early” (GL/57), G. pau “beard” (GL/63), G. rau “lion” (GL/65); compare the last to N. rhaw “lion” (Ety/RAW). The diphthong -ew does not appear in Gnomish or Early Noldorin, but -iw [iu] does: G. gwiw “young” (GL/42), ᴱN. híw “rich” (PE13/147), ᴱN. lhiw “worm” (PE13/149).

GS/§4.195 @@@

HPG/§2.7 @@@