ᴱ√MṆGṆ “*butter”
An unglossed root in the Qenya Lexicon of the 1910s with derivatives ᴱQ. mange “grease” and ᴱQ. manya “butter” (QL/62). G. mang “butter” from the contemporaneous Gnomish Lexicon is also almost certainly related (GL/56). In Qenya Word-lists from the 1920s Tolkien gave ᴱQ. mingwe “butter” (PE16/141). For purposes of Neo-Eldarin, I think it is worth positing a root ᴺ√MANGA to salvage early “butter” words.
√MŌ “labour, be afflicted”
This root first appeared as unglossed ᴹ√MŌ in The Etymologies of the 1930s with derivatives like ᴹQ. mól/N. mûl “slave, thrall” and ᴹQ. móta-/N. muda- “labour, toil” (Ety/MŌ). The root √mō was mentioned again in notes associated with 1950s Quenya prayers, where it was glossed “labour, be afflicted” along with verb Q. moia- of similar meaning and noun (Q. or primitive) mōl “slave” < mō-l; Tolkien went on to suggest √mol might be a better root, though he marked it with a “?” (VT43/31).
In notes on “large vs. small” from the late 1950s or early 1960s Tolkien had (root or verb?) mol- “labour” and (noun?) mōle, followed by MBOL/BOL with derivative Q. molda “big, large”, but this note was crossed through and √(M)BOL “*large” did not appear in the next list of the large/small roots (PE17/115). Thus I think √MO(L) reverted back to the meaning “labour”.
√MOR “black, dark, darkness”
This root was connected to “black” and “darkness” for all of Tolkien’s life. It first appeared as unglossed ᴱ√MORO in the Qenya Lexicon of the 1910s with derivatives like ᴱQ. móre “night” and ᴱQ. morna “black” (QL/62). In the contemporaneous Gnomish Lexicon it had derivatives like G. morn “dark, black” and G. morth “darkness” (GL/58). It appeared as ᴹ√MOR in The Etymologies of the 1930s with a similar set of derivatives (Ety/MOR) and √MOR was mentioned regularly in Tolkien’s later writings with glosses like “black, dark, darkness” (Let/308, 382; PE17/73).
ᴹ√MOROK “*bear”
A root in The Etymologies of the 1930s serving as the basis for “bear” words: ᴹQ. morko, N. brôg, and Ilk. broga (Ety/MORÓK), replacing rejected ᴹ√MOROG (EtyAC/LIS). The primitive form ✶morokō “bear” reappeared in the Common Eldarin: Noun Structure from the early 1950s, but Tolkien wrote an “X” above it (PE21/82 and note #55). In Tolkien’s later writings it may have been replaced by ✶grā “bear” which appeared in notes from the late 1960s, but in those notes Tolkien changed the gloss of ✶grā to “dog” (VT47/12, 35). This leaves ᴹ√MOROK as the best available root for “bear”.
√MOT “fen, marsh”
A root in Quenya Notes (QN) from 1957 glossed “fen, marsh” with derivatives like [Q?] motto and [S?] both or moth, apparently of similar meaning (PE17/165). A likely precursor is the unglossed root ᴹ√MBOTH in The Etymologies of the 1930s serving as the basis for words like ᴹQ. motto “blot”, N. both “puddle, small pool”, and Ilk. umboth “large pool”, the last of these being the basis for Ilk. Umboth Muilin “Veiled Pool” (Ety/MBOTH, MUY).
In the Gnomish of the 1910s, the first element of G. Umboth-muilin “Pools of Twilight” was derived from ᴱ√mbaþ-, a strengthened form of ᴱ√MAÞA “dusk”, with the second element muilin meaning “pools”, the plural G. muil (LT2/225; GL/58, 75). Later still this name became S. Aelin-uial, and it seems √M(B)OT was repurposed for swamp words.
ᴹ√MOY “*dear; [ᴱ√] *soft, gentle”
An unglossed root in The Etymologies of the 1930s serving as the basis for ᴹQ. moina and N. muin “dear” (Ety/MOY). An early root *ᴱ√MOYO “*soft, gentle” is strongly implied by words appearing in the Gnomish Lexicon of the 1910s: G. muig/ᴱQ. moika and “soft, gentle” and G. muis “softness, tenderness, gentleness” (GL/58). The words G. muin/ᴱQ. moina “safe, secure” and G. muil/ᴱQ. moile “tarn [mountain lake]” make also be related. Given the semantic variation of all these words, it’s hard to know the common meaning (if any) of their root, but for purposes of Neo-Eldarin, “soft, gentle” is probably a good foundation.
ᴱ√MṚÐṚ “grind”
A root in the Qenya Lexicon of the 1910s given as ᴱ√MṚŘṚ with derivatives like ᴱQ. mar (mard-) “grit, course grain or powder”, ᴱQ. mard- “grind, pulverize” and ᴱQ. marma “sand” (QL/63). In a list of roots at the end of the M-section, ᴱ√MṚÐṚ was gloss “grind” (QL/63). Early in the M-section Tolkien had the root ᴱ√MARA as a dialectical variation or byform of ᴱ√MṚŘṚ, with the derivative ᴱQ. marilla “pearl” (QL/59). Its cognate G. brithla “pearl” is given in the contemporaneous Gnomish Lexicon (GL/24), and in that document these words served as the base for ᴱQ. Silmaril and its Gnomish equivalent G. Silubrilt (GL/67). There are no signs of this root in Tolkien’s later writing, and Q. Silmaril was given different etymologies later on.
ᴱ√MUKU “cacare [Latin], *defecate”
The root ᴱ√MUKU “cacare [Latin = defecate]” appeared in the Qenya Lexicon of the 1910s with derivatives like ᴱQ. múko “dung” and ᴱQ. muqa “filthy” (QL/63). As there are no other roots of similar meaning in Tolkien’s later writing, I think it is worth retaining a Neo-Eldarin root ᴺ√MUK for this purpose.
ᴱ√MULU “grind (fine)”
The root ᴱ√MULU “grind fine” appeared in the Qenya Lexicon of the 1910s with derivatives like ᴱQ. mulda “powdery” and ᴱQ. mulma “fine flour” (QL/63). In the contemporaneous Gnomish Lexicon, G. bloss “wheat” was derived from mol- (GL/25). This root probably as served as the basis for ᴹQ. mulo “dust” from the early 1930s (PE21/10-11). The root √MUL may have been restored later on, as evidenced by the word Q. mŭle “meal [ground grain]” from a list of words having to do with “big” and “small” from the late 1950s or early 1960s, replacing Q. polë of similar meaning (PE17/115).
ᴹ√MUY “*secret”
An unglossed root in The Etymologies of the 1930s with derivatives like ᴹQ. muile “secrecy” and Ilk. muil “twilight, shadow, vagueness”, as well as Ilk. muilin “secret, veiled” in Ilk. Umboth Muilin “Veiled Pool” (Ety/MUY). Tolkien also gave N. muin “secret”, but he crossed that out and said the root was “not in N because it became identical with moina [MOY]” (Ety/MUY; EtyAC/MUY). Tolkien later abandoned the name Umboth Muilin, but S. muil reappeared in Emyn Muil “Drear Hills” (LotR/373; RC/334), hinting that this root may have remained valid.