In Ancient Quenya, any voiced velar spirants ʒ [ɣ] between liquids r, l and low vowels e, a developed into the semi-vowel y [j]. This was mentioned as a specialized development in both the Outline of Phonetic Development (OP1) from the 1930s and the Outline of Phonology (OP2) from the 1950s, but only for the vowel e:

while lg, rg > lʒ, rʒ > ll, rr; but lgē̆, rgē̆ > lye, rye (OP1: PE19/46)

The products of rg, lg; rgw, lgw were in (N.) TQ rr, ll (but rgē̆, lgē̆ > rye, lye); rw, lw (OP2: PE19/93).

This sound change was similar to how [ɣ] became [j] between [i], [e] and a vowel. The quotes above indicate this sound change only occurred before the vowel e, but all the attested examples show the same development before the vowel a:

• ✶phelgā > Q./ᴹQ. felya “mine; cave” (PE17/118; Ety/PHÉLEG).
• ᴹ✶targā > ᴹQ. tarya “tough, stiff” (Ety/TÁRAG).

There are examples before other vowels where lg followed its more usual development to ll:

• Val. delgūmā > Q. telluma “dome” (WJ/399).

Note that in the 1930s, this may have been a Lindarin [pre-Vanyarin] only development, see the entry on how: [lɣ], [rɣ] became [ll], [rr] for discussion.

Conceptual Development: In the 1910s and 1920s, it seems that g between l and a became ll, and this was the usual development of lg in this conceptual period:

• ᴱ✶balga > ᴱQ. palla “paunch” (PE13/138).
• ᴱ✶mḷgo > ᴱQ. millo “oil” (PE13/139).