In Quenya the combination ꝑu [ɸu] sometimes became hu, though Tolkien seems to have vacillated on whether this was a general change or restricted to the Vanyarin dialect. In the Outline of Phonetic Development (OP1) written in the 1940s, Tolkien described as a Vanyarin-only change:

ph > ꝑ became the labio-dental f (as in English “fin”), but in Vanyarin before ū̆, ui h was produced, as in huine (PQ phuine) “deep shadow; nightshade” (OP1: PE19/31).

As originally written this note said “Lindarin” rather than “Vanyarin” (OP1: PE19/31 note #11); the switch to Vanyarin was probably sometime in the late 1940s or early 1950s. In any case, it seems that in the 1930s the phonetic development phu > ꝑu > hu was only fully accomplished in the Lindarin dialect. These fu/hu variants can be seen in the contemporaneous examples from The Etymologies of the 1930s: both ᴹQ. Fui/Hui and ᴹQ. fuine/huine appear in the entry for the root ᴹ√PHUY (Ety/PHUY), though in other writings of this time period typically only huine appeared (LR/47, 56; SD/246, 310).

By the 1950s, Tolkien seems to have decided this change occurred before the shift of ꝑ > f and was universal rather than dialectical, and thus probably occurred in Parmaquesta [PQ]. As described in the Outline of Phonology (OP2) from the 1950s:

ph became a bilabial spirant [ꝑ]. Later where preserved this became labio-dental [f]. While still at the stage [ꝑ] the sound became [h] (voiceless breath) before ū̆, ui: as in huine “deep shadow, night shade” < AQ phuinē (OP2: PE19/71).

The universal change of [ɸu] to [hu] was reiterated in notes from the 1968:

Similarly the labial spirant f was bilabial, and so remained in Vanyarin. The shift from dental and labial þ and f to interdental þ and labio-dental f occurred first in Telerin. The labio-dental f soon spread to Noldorin. Probably because it helped to make f and the voiceless w (transcribed hw) of Quenya more clearly distinguished ... The Noldor, before they made the change, accused the Vanyar of confusing the sounds [ɸ and ƕ = hw]. In fact if left to unheeded change they would probably have merged in Quenya hw. Their near approach (by slackening the spirantal friction of f) before the separation of Vanyar and Noldor is seen in the development of phu- > *hwu- > hu-, as in Quenya huine “gloom”, unrelieved darkness (as a night without stars or moon), Telerin fuine of same sense, Sindarin fuin “night” (VT41/7-8).

However, in other notes on darkness and light from the late 1960s, Tolkien indicated that [ɸu] to [hu] was a Vanyarin-only change: “√PHUY: Noldorin Q. fuinë, Vanyarin Q. huinë, S. fuin” (NM/279), so it is clear this vacillation continued late in his life.

Conceptual Development: The fu/hu variation appears all the way back as the Qenya Phonology of the 1910s, but the change of [ɸu] to [hu] was sporadic:

f from ꝑ maintained itself in Qenya before ū̆ (distinguish f < hw), but h- forms occur (PE12/20).

The fu/hu variation was more characteristic of the phonetic development of Primitive Elvish x͡w [xʷ] in the 1910s:

x͡w ... gave generally f in Qenya (all Telellin) ({Solosimpi hw,} Noldorin f) except before ū̆ where h is common to all dialects (OP2: PE12/17).

The presence of fu- in Gnomish vs. hu- in Qenya was a strong sign of ancient ƕ [xʷ] in Early Elvish. In the Qenya Lexicon Tolkien said:

ǶUẎU: hui “fog, dark, murk, night”. huiva “murky”. (ƕ because of Noldo fui) (QL/41).

There did seem to be some dialectical variations, however. Tolkien said “Fui (= hui) wife of Vê = Nûri Nyenna (QL/38)”. There are other fu/hu variations in Early Qenya, not all with clear origins: fuiva as a variant of ᴱQ. huiva above (PME/104) and ᴱQ. furin, hurin “hidden” (QL/39).