The root √ANGWA “snake” with variant √ANGU appeared in The Etymologies of the 1930s as the basis for the words ᴹQ. ango “snake” (Ety/ANGWA) and angulóke “dragon” (Ety/LOK). The Noldorin equivalent am- seems to have survived only as a prefix (Ety/ANGWA), and is a good example of how [[on|[ŋgw] > [mb]]] in that language. There are a variety of other words for “snake” in Tolkien’s later writings, so whether this root remained valid is unclear.
Middle Primitive Elvish
(g)lingi
root. hang
lingi
root. hang
angwa
root. snake
auluta-
verb. [unglossed]
bay
root. [unglossed]
iw
root. [unglossed], [ᴱ√] *fish
kaltwa
?. [unglossed]
khlip
root. [unglossed]
An unglossed root in a rejected paragraph from the Quenya Verbal System (QVS) written in 1948 (PE22/112 note #78).
kōmā
noun. [unglossed]
lingikā̆la
proper name. Hanging-flame
maiga
root. [unglossed]
An unglossed root appearing in the first version of Tengwesta Qenderinwa from the 1930s (TQ1) to illustrate certain patterns of root formation (PE18/66). It may have serving as the basis for ᴹQ. Maia, though this word was given different derivations later.
phan
root. [unglossed]
A deleted root in The Etymologies of the 1930s with no glosses or derivatives and no clear function (EtyAC/PHAN).
skil
root. [unglossed]
A root mentioned in passing in as a variant of ᴹ√KIL “divide” in The Etymologies of the 1930s, but it had no derivatives and appeared nowhere else (Ety/KIL).
stin
root. [unglossed]
stā
root. [unglossed]
torōmā
noun. [unglossed]
uruk
root. [unglossed]
A root in The Etymologies of the 1930s with no glosses or derivatives and no clear function (EtyAC/URUK).
us
root. [unglossed]
A root in The Etymologies of the 1930s with no glosses or derivatives and no clear function (EtyAC/US).
A root in The Etymologies of the 1930s glossed “hang” and acting as an alternate explanation of ᴹQ. Laurelin, usually interpreted as “Song of Gold” but also meaning “Hanging Flame” as reflected in its Noldorin name Glingal (Ety/GLING, LIN², LING). The continued appearance of S. Glingal in later versions of The Silmarillion indicates this root likely remained valid (S/126).
A pair of forms ᴱQ. kinka- “to hang (intr.)” and ᴱQ. kinkata- “hang (tr.)” in the Qenya Lexicon of the 1910s imply the existence of a root ᴱ√KINIKI, which may be a precursor to ᴹ√(G)LINGI.