For both Sindarin and Noldorin, in compounds where a voiced dental spirant [ð] came in contact with a nasal after various vowel losses in compounds, the [ð] vanished, as noted by David Salo (GS/§4.119). There are several examples in both Sindarin and Noldorin:

• Kh. kheled + S. morn > S. Hele(ð)vorn “Black Glass” (RS/466; PE17/37).
• ✶Gala(da)ndil [> *geleðndil] > S. Gelennil “Lover of Trees”, with variant Gleðennil (PE21/83).
• N. Golodh + N. mîr > N. Golo(ð)vir “Silmaril, *(lit.) Noldo-jewel” (Ety/MIR, ÑGOLOD).
• ᴹ✶Eledandore > Eleðndor > N. Elennor “Land of the Elves” (Ety/ELED).

The last example is the clearest demonstration of this change. The examples Hele(ð)vorn and Golo(ð)vir seem to hint that the loss of [ð] occurred after the change of m > v, but I think this was simply notational convenience on Tolkien’s part, indicating there used to be a [ð] that was lost earlier.

In notes related to The Rivers and Beacon-hills of Gondor from the end of the 1960s, Tolkien indicated that [ð] survived before [m]: “tm, dm > ðm, later > ðv, ðw” (VT42/26). There is a couple of example of this development in notes from 1969: ✶khadmā > chaðw > haðw “seat”, along with the similarly derived verb haðwa “to seat” (PE22/148). Since these examples occur in the late 1960s, this could be a conceptual change. However, these examples of dm > ðw occurs in the interior of a word, while the examples of ð-loss above occur in compounds, so it is possible that ð-loss is conditioned by environment (e.g. at morpheme boundaries). Compare this to the vanishing of nasals at morpheme boundaries.

Conceptual Developments: The examples G. bridh(n)ir “queen” (GL/24) and G. goredhnir “cousin (f.)” (GL/41) indicates this change did not occur in Gnomish. There aren’t enough examples in the Early Noldorin of 1920s to figure out whether or not this sound change occurred at that conceptual stage.

GS/§4.119 @@@