In Quenya words with distinct pairs of voiceless stops, the first of the pair became spirantal. In particular, pt, kt > ꝑt, ht [ɸt, xt], and possibly even tk > þk [θk]. To fully understand these sound changes, you need to examine the contexts in which Tolkien conceived of them and how his ideas evolved over his life.

Conceptual Development: There is no mention of sound changes involving pairs of voiceless stops in the Qenya Phonology of the 1910s, and the combinations pt, kt are common among Early Qenya words from Qenya Lexicon. Where ht appears in the Qenya Lexicon, it seems to be derived from primitive χt. However, examples of kt > ht start to appear in the 1920s:

• ᴱ✶ekta- > ᴱQ. ehta “sword-point” (PE13/158).
• ᴱQ. ehtil “spring, fount, source” (PE13/136, 158); compare with earlier ᴱQ. ektele “fountain” (QL/35).
• ᴱ✶stakta- > ᴱQ. sahta- “to split” (PE14/66).
• ᴱ✶siktā > ᴱQ. sihta “moist, wet” (PE13/163).

There are also indications that pt > ps in the late 1920s:

• ᴱQ. sapta, sapsa “pit, grave” (PE16/75); compare with earlier ᴱQ. sat (sapt-) “hole” (QL/82).

In the Outline of Phonetic Development (OP1) from the 1930s, the normal development of pt, kt is to ps, ht in the original layer of composition for this document:

Quenya did not tolerate combinations of two different stops. When these anciently occurred, as in pt, kt, one of the two, or both, became open and spirantal (OP1: PE19/39).

pt, kt. The development here is obscured by the continuous use of the special single letters Q Z, no doubt originally devised to represent AQ pt, kt in spite of the later changes in their spoken forms.

The normal TQ (and L.) development is ps, ht. The latter being phonetically [χt] with a fairly strong back-spirant [χ], except in hty where the h is fully fronted (OP1: PE19/41 note #77).

Tolkien then elaborated on the development of pt > ps:

In Lindarin AQ pt yields both ht and ps. The divergence of pt, kt, and the development of pt in Lindarin are probably to be explained as follows: The earliest development in PQ, obscured by the maintainence of notation Q Z, was actually to open both elements > [ꝑþ, χþ]. In this combination, however, [ꝑ] did not develop > labiodental [f]; but in spoken Lindarin developed > ƕ > h [χ]: pt and kt thus falling together. The recognizable grammatical relationship of [-p + t-] was thus destroyed, and in most cases, where the relationship was still clear, p was restored. Thus for pt there existed [pþ] in the majority of the cases, but hþ only in isolated cases where the relationship to stems in -p was forgotten. For kt only ht existed. There then set in a reaction against a sequence of two spirants, and one was closed: hþ of either origin > ht; but pþ naturally remained becoming Lindarin and general TQ ps (pþ being maintained only by those Noldor who used their native þ for PQ, TQ th, þ). In TQ only the ps, pþ was used, because of the permanence of the PQ spelling with sign Q related to [p] q and quite distinct from Z kt, ht (OP1: PE19/41 note #78, #79).

These notes replaced several drafts that described more or less the same phonetic developments (OP1: PE19/42): pt, kt > ꝑþ, χþ > ps, χt, with occasional (Lindarin-only?) pt > χt (ht). The pt > ps development is well represented in The Etymologies of the 1930s:

• ᴹ√LEPET [> *lepte] > ᴹQ. lepse “finger” (Ety/LEP).
• ᴹ√KYELEP/TELEP > ᴹQ. telepsa “of silver” (Ety/KYELEP) vs. ᴱQ. telepta (QL/91).
• ᴹ✶libda [> *lipta] > ᴹQ. lipsa (Ety/LIB²).

There is only one clear example of the (Lindarin-only) sound change pt > ht: the archaic variant †lehta of lepsa “finger” from an early draft of the above notes (OP1: PE19/42). Thus pt > ps seems to be the normal phonetic development in the 1930s.

However, at some point Tolkien made revisions in red ink to his OP1 description of pt-developments:

In spoken Lindarin AQ pt Q is pronounced ht or Z. In PQ the pronunciation was probably ꝑt in which ꝑ remained bilabial (and in natural spoken language developed to χ). But in TQ the difference of the letter Q (related to q = p) and etymolog[ically] retained ꝑt which is pronounced ft or pt. In our transcription pt (or pht) is used. In T. pt, kt > pþ, kþ > ps, x, and some forms of this sort invaded TQ (OP1: PE19/41 and note #79).

In this revised system, pt > ht remains the (Lindarin-only) pronunciation, but pt > ꝑt in normal Quenya development written “pt”, whereas the pt > ps sound change appeared only in Telerin loan words. Some examples of this pt survival can be seen in words from the 1940s:

• ᴹQ. opto “back” as a name for the tengwa Q replacing uphto (PE22/50 note #180).
• ᴹQ. rakkalepta “*claw-fingered” in Lord of the Rings drafts (SD/68).

When Tolkien started writing the updated Outline of Phonology (OP2) in the 1950s, he seems to have restored the original OP1 scenario in the first layer of composition (PE19/84 note #75). But this was ultimately switched back to the pt > ꝑt development:

pt, kt. The history of the phonetic developments here is obscured by the continued use of special single letters, such as Feanorian Q A, probably originally intended to represent sounds still little or not at all altered from pt, kt, after considerable phonetic change had occurred in the spoken language ...

The development of kt is the more certain. This had become ht, in which h was a fairly strong spirant, in the spoken language of both Vanyar and Noldor before the Exile, and the opening of the k goes back to PQ. Even the loremasters represented PQ A by ht in recitation ...

Alter this and make change of pt (Q) > ꝑt normal. Transcription pt to show bilabial pronunciation. The combination ft was avoided in Q. and Sindarin (PE19/84).

The “alter this” refers to the replacement of text which originally had pt > ps (PE19/84 note #75). The clearest description of the exact conceptual changes in the phonetic developments appear in the section on the development of aspirates from OP2:

In Quenya there was no distinction between the products of CE pth, kth and pt, kt, probably because pt, kt > pth, kth. Both varieties then proceeded to {[ꝑþ, χþ] >>} [ꝑtʰ, χtʰ]. The ultimate results were {ꝑþ (TQ ps) >>} ꝑt (transcribed pt) and ht (OP2: PE19/87 and notes #84, #85).

Thus the revised system was pt, kt > ꝑtʰ, χtʰ > ꝑt, χt (transcribed pt, ht). As described in OP2, the sound change kt > ht [xt] is reflected in spelling and there are numerous examples of it in the 1950s and 60s. Tolkien even mentioned this pronunciation in The Lord of the Rings appendices:

The Quenya combination ht has the sound of cht, as in German “echt, acht”: e.g. in the name Telumehtar “Orion” (LotR/1113).

However, as noted above the pronunciation ꝑt [ɸt] was not reflected in spelling. There are fair number of examples of written pt appearing in Quenya words in 1950s and 60s, some of which had ps in the 1930s:

• Q. leptafinya “clever-fingered” (PE17/17) vs. 1930s ᴹQ. lepse (Ety/LEP).
• Q. telepta “silver” (PE17/71) vs. 1930s ᴹQ. telepsa (Ety/KYELEP).
• Q. tapta “impeded” (VT39/17).
• Q. lepta- “pick up” (VT47/10).
• Q. apta- “refuse” (PE19/90) vs. 1930s ᴹQ. apsa- (PE19/45).

The conceptual transition wasn’t entirely smooth, however, since Q. lepse “finger” appears in etymological notes associated from the late 1950s or early 1960s (PE17/89), and a form that might be lepþa appears in the late 1960s (VT47/23).

Tolkien’s notes in OP1 and OP2 focus on the combinations pt, kt, but there is evidence of similar developments for tk. This would not be a common combination, since usually in Primitive Elvish tk became kt. The best example of this sound changes is an alternate word for the number “seven” appearing in notes from the late 1960s:

• ✶otok > otko > osko > Q. oxo (VT47/42).

This Quenya word “seven” is derived from the same root that produced S. odog. One possible explanation of otko > osko is that tk > þk > sk, with þ to s as usual and then sk > x [ks] via metathesis. Near this word Tolkien wrote “tk > sk in Q., T., S.” which seems to indicate this was a universal (and thus perhaps Primitive Elvish) sound change. However, AQ. otko is almost certainly the result of the Quenya syncope, strongly implying these were parallel changes in the child languages (and I have found no further evidence of Telerin or Sindarin sound changes tk > sk).

Another example from Quenya Notes from 1957 (QN: PE17/145) seems to show a similar set of phonetic developments:

• ✶(U)RUÞ > Q. rusca, ruxa (PE17/188).

This example might also have tk > þk > sk > ks, except that the Primitive Elvish sound changes were rutʰ-kā > rutkʰā > [AQ] rutkā. The adjectival suffix -kā̆ is unusual but not unknown, so this scenario is reasonably plausible (PE19/87).

To summarize:

• In the 1910s both kt and pt remained unchanged in Quenya.
• In the 1920s Tolkien decided that kt > ht [xt].
• By the 1930s (and possibly somewhat earlier) Tolkien had pt > ps.
• Sometimes in the 1940s he changed his mind and had pt > ꝑt (spelled pt), with some vacillations between ps/ꝑt going forward, but mostly favoring ꝑt.
• Possibly also tk > þk > sk > ks (via metathesis), but here the evidence is more ambiguous.

compare ektele >> ehtil