The word ar was the Quenya word for “and” for much of Tolkien’s life. It was related to (and originally identical with) Q. ar(a) “beside” (PE17/70). The word ar was always used between sentences, but in sets of items sometimes yo and ta were used instead.
Conceptual Development: In the Qenya Lexicon of the 1910s ᴱQ. ar(a) was glossed “but” under the early root ᴱ√ƷARA “spread, extend sideways” (QL/32). In this earliest period the word for “and” was ᴱQ. ya(n) (QL/104). By the end of the 1920s when Tolkien composed the Nieninqe and Earendil poems, he consistently translated ar as “and” (MC/216; PE16/100). The translation “but” reappeared in a few phrases from the 1940s (PE22/124; PE23/74), but it is not clear if these were genuine shifts in meaning or loose translations.
By the time Tolkien wrote The Etymologies of the 1930s, ᴹQ. ar “and” was derived from the root ᴹ√AR “beside, outside” (Ety/AR²). This seems to have remained the case up through most of the 1950s, with the possible exception of a couple phrases in the 1930s where Tolkien used a “and” instead (LR/61, 72). In this period the usual Noldorin/Sindarin word for “and” was also ar.
At some point while writing drafts of The Lord of the Rings, Tolkien decided that the Sindarin word for “and” should a in the famous phrase pedo mellon a minno “speak, friend, and enter”. His motivations for this change are unclear, but he noticed the problem in notes written between the first and second edition of The Lord of the Rings, saying “a·Berhael. ‘And’ cannot therefore be [derived from] arĭ!” (PE17/102). From this point forward, Tolkien considered two possible roots serving as the basis for “and”: √AD(A) and √AS, both meaning “beside” (PE17/145; VT48/25). Of the two, Tolkien appears to have settled on √AS, which appeared in a few different notes from 1968 (VT47/31; VT48/25).
Assimilations: In the notes written between both the 1st and 2nd edition of The Lord of the Rings, Tolkien consider what kind of assimilations ar might have before consonants if it were derived from √AS or √AD (PE17/41). For √AS it became a before the consonants f, h, hw, hy, became as before t, k, p, q, s, and became al before l. For √AD it became a before the consonants n, m, became as before s, and became al before l. In notes from around 1964 Tolkien said:
> It is not necessary here to specify all the assimilations that could have occurred at these different stages, since in fact few have left traces in the forms of “and” ... Later after development to ar, only as survived as an occasional form before t, and as the usual form before s (of any origin); while al appeared before l. But in written Quenya ar was usually written in all cases, though the pronunciation of ar-s, ar-l as as-s, al-l remained usual (PE17/71).
In this particular discussion, ar as derived from √AD. However, the system Tolkien described was that all the older assimilations were abandoned, and the only ones that survived were based on later assimilations involved r of any origin: rs > ss and rl > ll. These sound shifts only affected pronunciation, not spelling. Thus the same arguments would be apply if ar was derived from √AS.
Neo-Quenya: For purposes of Neo-Quenya, I would write ar “and” in all cases, and would assume it was derived from derived from √AS, but would further assume that the Tarquesta pronunciations before s and l were as-s, al-l.
Ara-, ar- a prefixed form of the stem Ara- "noble" (PM:344). In the masc. names Aracáno "high chieftain", mothername (amilessë, q.v.) of Fingolfin (PM:360, cf. 344), Arafinwë "Finarfin" (MR:230)