In Ancient Quenya, any aspirates that followed a voiceless stop lost (most of) its aspiration, so that the result was a pair of voiceless stops. This must have happened before aspirates became voiceless spirants. Already in Primitive Elvish, voiced stops were unvoiced both before aspirates and after aspirates, and also any aspiration moved to end of group of stops. The ultimate result is that any stop-aspirate or aspirate-aspirate pairs became a pair of voiceless stops in Quenya: g-th, kh-d, kh-t, k-th all > kt > (later) ht. The same was true for combinations of voiced stops (gg > kk).

Tolkien provided nearly identical descriptions of this sound change in both the Outline of Phonetic Development (OP1) from the 1940s and the Outline of Phonology (OP2) from the 1950s:

The aspirates and stops fell together and developed in Q. identically (when not interfered with by grammatical or other analogy) in the following cases: ... (b) when dynamically lengthened: as pp, pph etc. (c) in combination with other stops: as pht, pth etc. ... In other cases the two groups coalesced in stops with a weak aspiration (insufficient to cause opening and spirantalization): so pp, pph > ppʰ; pht, pth, pt > ptʰ whence later developments (OP1: PE19/40).

Aspirates and stops fell together and developed identically in Quenya, when not interfered with by grammatical or other analogy, in the following cases: ... (b) when dynamically lengthened; (c) in conjunction with other stops: as in pht, pt, &c. ... the two varieties coalesced in stops with a weak aspiration (insufficient to cause opening and spirantalization). Thus: ... pp and pph > pp (ppʰ); ph-t, p-th > pth > ꝑt; kht, kth > kth > hth > ht (se. htʰ) (OP2: PE19/84).

The ellipsis (...) in the above quotes omit the descriptions of the phonetic developments of aspirates after s, which occurred much earlier (in Primitive Elvish) and is discussed in another entry. As indicated above, there was a slight aspiration in Quenya pronunciation at the end of the cluster (from stop-pairs of all origins), but this was not reflected in spelling. Any geminate pair of voiceless stops was preserved, but later on [pt], [kt] became [ɸt], [xt], spelled pt and ht; see that entry for further details.

Examples of this sound change include:

• √NAKH [> *(a)nakkha] > Q. (a)nakka “narrows” (PE17/166).
• ✶netthi > Q. nette “(little) girl, sister” (VT47/16).
• ✶okh’tā [> *oktā] > Q. ohta “war” (PE18/87).

This phonetic development can also been seen in the nasal-infixed past tenses of basic verbs whose roots end in an aspirate, since earlier nasals became voiceless stops before aspirates. See, for example, past tenses rappe, lakke, pitte for verbs ᴹQ. raf-, lah-, pis- [from piþ-] (PE22/102-3), from (unattested) ramphe, lañkhe, pinthe > rapphe, lakkhe, pitthe.

Conceptual Development: Tolkien’s earliest conception of Primitive Elvish from the 1910s and 1920s had voiceless spirants rather than aspirates. These voiceless spirants tended to remain spirantal with various assimilations and transformations in combination with voiceless stops, as described in the Qenya Phonology from the 1910s (PE12/17-21), so the phonetic developments discussed above only became relevant after the introduction of aspirates into Primitive Elvish in the 1930s.