On rare occasions an initial s before p, t, k became syllabic, producing a preceding vowel i in most cases. Tolkien mentioned this sound change in both the Outline of Phonetic Development (OP1) from the 1940s and the Outline of Phonology (OP2) from the 1950s:

A third form, parallel to that seen in the nasal groups but rarer, was the development of s before p, t, k to syllabic ṣ > Q is. As in ... [Tolkien never filled in the example] Hence for sp, sk also ips, iks (OP1: PE19/36).

A third development is also found, parallel to the development of syllabic nasals, whereby s became syllabic ṣ > PQ is (before all consonants). This is only seen in individual derivatives or isolated words, never in a stem or base with all its derivatives. The PQ results were: ups, iks with metathesis, and ist (OP2: PE19/79).

In the 1930s, Tolkien said the result was ṣ > is in all cases. In the 1950s Tolkien said ṣp > ups. It isn’t clear whether ṣp > usp first or whether the u was introduced after the reversal of sp to ps, but Tolkien’s statement that “syllabic ṣ > PQ is (before all consonants)” strongly indicates the latter.

Tolkien said in OP2 that the PQ [Parmaquesta] result was is, but ṣ > is probably occurred before initial [s] plus voiceless stops became voiceless spirants, a sound change that itself was probably in AQ [Ancient Quenya] (@@@ revisit). His reference to the “PQ results: ups, iks” probably refers to later metathesis of sp, sk, so that ṣp, ṣk > isp, isk > ups, iks.

The only Late Quenya (1950s or later) example of this sound change I’ve been able to find is: ✶skalā > (archaic) Q. †ixal “a cast shadow” (PE17/184).

Conceptual Development: Similar developments were mentioned in the Qenya Phonology of the 1910s:

If ṣ̄̆ or ẓ̄̆ is to be assumed: ṣ̄̆ gave -is-, ẓ̄̆ gave ir. ṣ before explosives or voiceless consonants, ẓ before spirants, nasals, or voiced explosives, cp. siste, “ulcer, boil” (sṣtē), mirde, “mist” (mẓđē) (PE12/14).

In addition to the above, Early Qenya examples include:

• ᴱ√SṢTYṢ > ᴱQ. sist “ulcer, sore” and ᴱQ. sastya “sore, galled” (QL/86).
• ᴱ√pṣt- > ᴱQ. piste “spit”, past tense pastie or pastye (PE14/58).
• ᴱ√ṢQṢ > ᴱQ. usqe “fog” (QL/98).

The first two examples imply that long ṣ̄ had a distinct development: to as. There were similar differences in the developments of long syllabic ṇ̄, ḹ, ṝ > an, al, ar in Early Quenya (PE12/10). The example ᴱQ. usqe implies the vowel u was produced before labials and labialized consonants, as was the case in OP2 in the 1950s.