In cases where velar nasal ñ [ŋ] survived before a dental nasal n, it likewise became dental so that eñn, añn, oñn > enn, ann, onn. Tolkien mentioned this in both Outline of Phonetic Development (OP1) from the 1930s and the Outline of Phonology (OP2) from the 1950s:

In early PQ ñn remained as in tekna “a letter” > teñna. But in TQ pronunciation this is represented by nn, a development that may be due to actual natural pronunciation in later classical PQ, since the spelling nn is also there frequently found ... But iñn > ín; uñn > un (OP1: PE19/43).

In Quenya dissimilation, by loss of nasality, occurred ... probably in iñn, uñn > ín, ún (via iʒn, uʒn?). The only assimilation was the comparatively late assimilation of eñn, añn, oñn > enn, ann, onn ... (OP2: PE19/98).

As indicated in these quotes, the combinations iñn, uñn were an exception to this rule, with the ñ vanishing with compensatory lengthening to produce ín, ún. The OP1 quote indicates this change took place in late Parmaquesta [PQ]. Aside from the example ᴹQ. tenna “letter” given above, this sound change can also be seen in:

• √TEÑ > Q. tenna “thought, notion, idea” (PE19/97).
• ✶laŋna > Q. lanna “athwart” (PE17/65).

This sound change was mentioned in other contexts, each indicating it was a relatively recent phonetic development:

In TQ ñn > nn, hence 5{ often appears = g5. (Feanorian Alphabet, 1940s, PE22/149).

ŋn = ŋn remaining, but later > nn (phonetic notes from the 1960s, PE22/149).