This root and ones like it were connected to flowers for all of Tolkien’s life. The earliest manifestation of this root was ᴱ√LOHO or ᴱ√LO’O from the Qenya Lexicon of the 1910s; the entry for ᴱ√LOHO appears immediately below ᴱ√LO’O, and Tolkien indicates they are related roots, both extended from ᴱ√OLO “tip” (QL/55). These roots include derivatives like ᴱQ. lōte “flower”, ᴱQ. lotōrea “flourishing” and ᴱQ. lokta- “sprout, bud, put forth leaves or flowers”. There are also derivatives of these roots in the contemporaneous Gnomish Lexicon: G. lost “blossom, bloom”, G. lothli “floret”, G. luitha- “to bloom” (GL/54-55), though G. lôs “flower” was said to be unrelated, connected to G. lass “leaf” instead (GL/55). ᴱQ. losse “rose” probably had a similar derivation (QL/56).
This confusion of √LOT(H) and √LOS carried forward into Tolkien’s later writings. In The Etymologies of the 1930s ᴹ√LOT(H) was given as the root for “flower” (Ety/LOT(H)), but this entry originally included a variant ᴹ√LOS (EtyAC/LOT(H)). Tolkien then said ᴹQ. losse “blossom” (< ᴹ√LOT(H)) was “usually, owing to association with olosse snow, only used of white blossom” (Ety/LOT(H)), where ᴹQ. olosse was derived from ᴹ√GOLOS “✱snow, white” (Ety/GOLÓS). This intermingling carried forward into etymological notes from the late 1950s or early 1960s, where Tolkien said (PE17/26):
> The stems √LŎS, √LOTH, √LOT are much entangled both for formal reasons, and because of actual associations of meaning (probably from beginning of Primitive Quendian and explaining the approach of the forms). Quenya word for “flower, a single bloom” is lóte, but S loth (< lotho/a), but Quenya also has lōs. Q. for snow is losse (S los).
These associations were also mentioned in etymological notes on roots for flowers from this same period, where Tolkien clarified that √LOT, √LOTH were the roots for “flower” and √LOS for “snow” (PE17/160-161). These roots were mentioned again in notes associated with The Rivers and Beacon-hills of Gondor from 1967-69 (VT42/18):
> S. loss is a derivative of (G)LOS “white”; but loth is from LOT. Sindarin used loss as a noun, but the strengthened form gloss as an adjective “(dazzling) white”. loth was the only derivative of LOT that it retained, probably because other forms of the stem assumed a phonetic shape that seemed inappropriate, or were confusible with other stems (such as LUT “float”), e.g. ✱lod, ✱lûd. loth is from a diminutive lotse and probably also from derivative lotta-.
In this last note, Tolkien seems to have abandoned √LOTH, explaining S. loth “flower” as derived from √LOT via ✱lotse. In any case, starting in the 1930s Tolkien was consistent that the roots for “flower” and “snow” were distinct but often confused, and that snow-words were derived from roots like √(G)LOS and flower words from roots like √LOT(H), though he waffled a bit on the exact details.
This was the root for Elvish words meaning “yellow” for much of Tolkien’s life, though with some minor variations. It appeared as ᴱ√MALA “yellow” (usually mali-) in the Qenya Lexicon of the 1910s with derivatives like ᴱQ. malina “yellow” and ᴱQ. malikon “amber” (QL/58). It also appeared in a list of M-roots at the end of that section (QL/63). It had derivatives in the contemporaneous Gnomish Lexicon such as G. malon “yellow” and G. malthos “butter cup” (GL/56).
In The Etymologies of the 1930s it seems Tolkien first gave this root as ᴹ√MAL (EtyAC/MAL) but rejected this and replaced it with ᴹ√SMAL “yellow” (Ety/SMAL). It had derivatives like ᴹQ. malina/N. malen “yellow”, ᴹQ. malta/N. malt “gold (as metal)” and ᴹQ. malo/N. hmâl “pollen, yellow powder” (< ᴹ✶smalu), with some revisions in Noldorin forms as Tolkien vacillated on whether or not primitive sm- resulted in voiceless nasal hm- or a voiced nasal m-.
This √SMAL vs. √MAL variation seems to have continued into Tolkien’s later writings, as seen in Common Eldarin: Noun Structure from the early 1950s where ✶malu >> ✶smalu “dust, grit” (PE21/80), probably a later iteration of ᴹ✶smalu “pollen, yellow powder” from The Etymologies. But it seems Tolkien settled on √MAL as evidenced by the extended root √MALAT “gold” from The Shibboleth of Fëanor from 1968 (PM/366).