Starting in the 1930s, the combination wo sometimes reduced o, but the exact conditions under which this change occurred are unclear. Tolkien gave 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:

When stressed the sequence wo was usually changed to wa. A similar change occurs in Noldorin. But when unstressed wo was often reduced (before the unrounding of the o) to o with loss of w. Cf. the development of *gwo- “together” > wa- and o-´. The sequence wō remained unchanged (OP1: PE19/53).

When stressed the sequence wo was usually changed > wa. A similar change also occurred in Sindarin. But when unstressed wo was often reduced to o with loss of w. An example of both is provided by CE gwo “together” > wa- as a stressed prefix or element, but o- as an unstressed pretonic prefix. The sequence wō remained unchanged (OP2: PE19/106).

Here the rule is clear: wŏ was only reduced if short and unstressed. However, in the Quendi and Eldar essay from around 1960 Tolkien said:

WO ... This does not remain in Q as an independent word. It is however a frequent prefix in the form ó- (usually reduced to o- when unstressed) ... Telerin use does not materially differ; but in form the w- (lost in Quenya before ō) is retained: prefix vō, vo- (WJ/367).

This note indicates w could be lost before long ō. Furthermore, in The Shibboleth of Feanor from 1968 Tolkien said:

Connexion with Telerin vola “a roller, long wave”, which was sometimes made by the Teleri themselves, was not a serious etymology but a kind of pun; for the king’s name was not normally Volwë (Common Eldarin *wolwē) but Olwë in Telerin as in Quenya, and w was not lost before o in Telerin as it was in Quenya (PM/357-8 note #20).

These two notes from the 1960s seem to imply that wo > o more generally in Quenya. There are, in fact no obvious examples of wo surviving to become vo in Quenya, with the exception of a (deleted) etymology of Q. miruvórë < ✶miru̯(a)-wōrĭ (PE17/38), elsewhere derived from Val. mirubhōze (WJ/399). The most common example of the reduction of wo > o is in the etymology of the prefix o- “together” given above, but the pronominal suffix -ngo that seems to be the dual of -ngwë in a 1964 pronominal paradigm could be another example (PE17/75; VT49/48); hat tip to Vyacheslav Stepanov for pointing this out to me.

Conceptual Development: Both wo- and qo- appear in Qenya words from Qenya Lexicon of the 1910s, so it is unlikely this reduction of wo applied in Tolkien’s earliest conception of Quenya. There is evidence that quo- remained valid in Tolkien’s later writings, since qo- was given as a possible initial combination in a discussion of reduplicating syllables from the Quenya Verbal System written in 1948 (PE22/112). This qo- postdates Tolkien’s 1930s discussion of wo > o from OP1 discussed above (PE19/53), which is evidence that this rule did not apply to quo as late as 1948.

Neo-Quenya: For Neo-Quenya, I prefer to assume that only initial wo > o, mainly so that I can continue to use early Early Qenya words starting with qo-. However, other Neo-Quenya writers assume kwo > ko in Late Quenya, rendering early words beginning with quo- invalid.