In Ancient Quenya words beginning with d that had an n (or in theory, other nasals) at the end of the first syllable, the initial d- often became n- by assimilation. This sound change was mentioned briefly in both the Outline of Phonetic Development (OP1) from the 1940s and the Outline of Phonology (OP2) from the 1950s, and was described in more detail in several other places:

n [was derived from] n, nd, sn. rarely d as in nerno [< *derno?] (OP1, 1940s, PE19/38).

n was derived from n, nd, sn (which possibly included prehistoric sñ). It also appears occasionally as product of d, instead of l, by assimilation to succeeding nasals, as in dorno > norno (OP2, early 1950s, PE19/80).

Quenya initial d- normally > l-. Occasionally n-, usually by assimilation to an n elsewhere in stem or word, though base-variation nd- from d- is always possible (linguistic note in the late 1950s or early 1960s, PE17/17).

The Q forms norna, Norno [“tough” and “Dwarf”] may also contain nd-, though S dorn does not; but this is probably one of the cases in which Q initial d became n-, not l-, by assimilation to an n occurring later in the word (Quendi and Eldar, circa 1960, WJ/413-4).

This was far older and distinct from the normal Quenya change of initial d > l; though the existence of the variation [between primitive d/l] already in the language may have assisted in making l the normal change, instead of the medial intervocalic change d > ð > r, or the occasional change > n- when the second consonant of a stem was a nasal: as in Q. (archaic and chiefly surviving in topographical names) nuine “river (of large volume, and liable to flooding)”, S. duin, T. duine, beside Q. luy-, luita- “to flood, inundate, drench”, T. duita- (Variation D/L in Common Eldarin, 1968, VT48/30).

Tolkien rejected the last of these notes, instead deciding that Q. **luine would have been the normal result of primitive ✶duinē, and that this word was abandoned due to conflict with Q. luinë “blue” (VT48/23).

Despite the apparent late (1968) rejection of this sound change, there is ample evidence of it from the 1930s up through the late 1960s. The last example seems to be Q. nonda “fist” < ✶dond(a) appearing in one of the versions of Eldarin Hands, Fingers and Numerals from the late 1960s (VT47/23 note #25). In addition to the examples already given above, other early examples from the 1930s include:

• ᴹ√DEN > ᴹQ. nenda “sloping”, ᴹQ. nende “slope” (Ety/DEN, a rejected root).
• ᴹ√DÓRON > ᴹQ. norno “oak” (Ety/DÓRON).

As for counterexamples, along with √DUY > ✶duinē > Q. **luine “(large) river” given above, another early counterexample is:

• ᴹ√DAT > ᴹ✶danta- > ᴹQ. lanta- “to fall” (Ety/DAT; PE21/58, 63).

One interesting feature of both of these counterexamples is that the nasal n was not part of the root, but was an added element: duy-nē, da-n-ta-. Both these roots had other derivatives that followed the more common development of d- > l-: Q. luita- “flood”, ᴹQ. latta “hole”. Perhaps these related words pressured the variants with d- > n- to reform back to l-.

Conceptual Development: The normal phonetic development of initial d- in Early Quenya was to voiceless t-: ᴱ✶dagniı̯- > ᴱQ. táne (PE13/161). Thus neither of these later sound changes to l- or n- were relevant in the earliest conceptual stages of the 1910s and 1920s.