In Primitive Elvish, [s] became [z] before voiced stops. In Quenya, this sound change was reversed, with [s] as well as the following stop being unvoiced. This change was parallel to, and around the same time as, combinations of voiced stops becoming unvoiced. Tolkien described this change in both the Outline of Phonetic Development (OP1) from the 1930s and the Outline of Phonology (OP2) from the 1950s:

[z] had only originally appeared in CE in the groups: zd, zg, and rare zb. These groups were unvoiced in Q. Cf. p. 21 on voiced stops. The resultant st remained; sk, sp were usually metathesized to ks, ps (OP1: PE19/48).

In conjunction with s. After s in CE zb, zd, zg were produced. These were unvoiced in PQ > sp, st, sk, later usually ps, st, ks (OP2: PE19/95).

As indicated above, the resulting pairs [sp], [sk] became [ps], [ks]. This metathesis seems to be later than the other Quenya metatheses; see that entry further discussion. Examples of the unvoicing of z + stops are plentiful in Quenya. A representative sample includes:

• ✶buzbō > puspo > Q. pupso “blow fly” [1930s] or “large fly” [1950s] (OP1: PE19/48; OP2: PE19/101).
• ✶ezdē > Q. Este (Ety/EZDĒ; WJ/403).
• ✶mazgō > Q. makso “mire” (OP1: PE19/48, OP2: PE19/101).

The combination st was also the result of zd produced from d+d by suffixion, as in ✶red+dā > rezdā > Q. resta “sown field” (EtyAC/RED; OP2: PE19/91).

Conceptual Development: Finding clear examples of this development in Early Quenya is quite difficult, but given the prevalence of s + voiceless stop combinations (and the complete absence of z) it was likely a factor in the phonetic development of the earliest conception of the language as well.