The ancient change of labialized velars to labials in Telerin is extremely well documented (PE18/104; VT41/8; WJ/374-375), and it is characteristic of the Telerin branch of the Eldarin languages. For example, in a note associated with The Shibboleth of Fëanor essay from 1968:

In Sindarin the back sounds k, g, kh, ng (not n alone) [+ w] had all at a very early period (as in Telerin) become labials p, b, ph, mb (VT41/8).

The “[+ w]” is an editorial addition by Carl Hostetter, but it is clear from context (Tolkien is discussing the development of final -w in Sindarin). The exclusion of “n alone” in this note is peculiar, since elsewhere Tolkien makes it clear that [ŋw] > [m] (e.g. PE18/104-105), but perhaps he is referring to dental n rather than velar ñ.

As a result of this rule, Telerin and Sindarin frequently had p, b, m where Quenya had qu [kw], v/w [< gw], and ñ(g)w [ŋ(g)w], for example: T. alpa and S. alph “swan” vs. Q. alqua < ✶alkwā (UT/265); S. balch “fierce” vs. Q. walda “excited” < √GWAL (PE17/154); N. nem “nose” vs. ᴹQ. nengwe < ✶neñwi (Ety/NEÑ-WI). There are no attested words where [kʰw] > [pʰ], but this change appears occasionally in Tolkien’s description of this phonetic rule (VT41/8, PE19/18). Similar though less comprehensive changes occurred in the Celtic branch of Indo-European languages (WGHC/§89i, §92ii, §92v).

The change also occurred in the “strengthened” labialized clusters [ŋkw] and [ŋgw], such as: S. nimp “pale, white” vs. Q. ninquë [< niŋkwi] (PE17/168); S. (m)baw “howling” vs. Q. nwámë [< ŋgwāmē] (PE19/106) with initial [mb-] becoming [b-] in this second example, as was usual in Sindarin. In fact, the change of [ŋw] > simple [m] could only occur initially, because already in Primitive Elvish medial [ŋw] became [ŋgw].

This phonetic change occurred with both “natural” labialized velars that were part of roots, as well as combinations that were the result of w-suffixion, such as: ᴹ√TAK > ᴹ✶atak+wē [> atape] > N. adab “building” (Ety/TAK). As suggested by David Salo (GS/§4.1, 4.23), such suffixal combinations likely first became labialized, and then changed into velars: [kw] > [kʷ] > [p]. Since the development of [kʷ] and [k+w] seem to be identical in both the Sindarin and Quenya branches, this lexicon doesn’t distinguish them in phonetic discussions, writing them both as [kw].

Conceptual Development: This phonetic rule dates all the way back to Solosimpi, the earliest form of Telerin (PE12/16-17, 21; QL/38). In one note from the 1910s:

But it is to be noticed Solosimpi had already split. The variants in brackets [from the table preceding this note: p, b, m, f vs. Qenya ku̯, gu̯, nu̯, hu̯, the last derived from primitive x͡w] denote regular Solosimpi forms ... (1) [q-series] and (5) [p-series] fell with (5) in Solosimpi except before ū̆ (PE12/17).

However, in the early Gnomish of the 1910s, the only similar change was that non-initial [kʷ] become [p] (HPG/§2.7), such as G. alfa “swan” < ✶alchwa [< alkʷā given ᴱQ. alqa] (GL/18). At this early stage, initial [kʷ] was preserved in Gnomish, along with other labialized velars, for example: G. cwed- “to say” (GL/28) vs. later S. ped- (LotR/305). This was perfectly reasonable, since at this early conceptual stage Gnomish and Solosimpi (Early Telerin) were unrelated languages.

Tolkien introduced this sound change into the Early Noldorin of the 1920s, such as G. cweth “word” (GL/28) becoming ᴱN. peth (PE13/152); perhaps this was out of a desire to make Noldorin more “Welsh-like”. However, throughout the 1920s to 1940s, there was no direct relationship between the Telerin and Noldorin languages. At this stage the phonetic rule must have been a parallel change, though perhaps Tolkien imagined the two languages influencing each other early in their history.

In the 1950s when Tolkien reconceived of Noldorin as Sindarin, he also made Sindarin a member of the Telerin language branch, which allowed him to postulate this development as occurring in the ancient common ancestor of the two languages. Simplified formulations like this probably helped motivate Tolkien to change Noldorin into Sindarin. In that light, it is somewhat surprising it took thirty years for Tolkien to decide to create such an ancestral relationship between these languages.

GS/§4.23 WGHC/§89i [PIE kʷ, kʷʰ > Kelt kʷ > Brit p] §92ii [PIE gʷ > Kelt b] §92v [PIE ngʷ > (Kelt mb?) > mm > W m] @@@

HPG/§2.7