Another word analysis, this time for a single word: the conjunction “and”. We know the net result: Q. ar and S. a. Knowing its development is nevertheless very important, because it effects a number of secondary words and its history is very complex.
In the Quenya Lexicon of the 1910s, the word for “and” is ᴱQ. ya(n) from the root YA, which is possibly a variant of (D)ẎṆTṆ having to do with “(en)large” (QL/104, 106). Meanwhile, the word ᴱQ. ar(a) is glossed “but”, derived from (Ʒ)ARA “spread” (QL/32). In the Gnomish lexicon, however, the word for “and, too” is G. ar² (GL/20, 49), and in the Gnomish Lexicon Slips ar is associated with “besides” (PE13/110). Tolkien began using ᴱQ. ar for “and” in the Early Markirya poem of the 1920s (MC/216) and it remained the basic Quenya word for “and” thereafter.
This scenario is firm in the Etymologies of the 1930s, where ᴹQ. ar “and” is derived from the root AR² whose basic sense is “beside, outside” (Ety/AR²). The related prefix ar- meant “outside, beside” in both Quenya and Noldorin. In Noldorin it developed an additional privative sense “without”, and so appeared in N. arnediad “numberless”, which appears in the name of the great battle: Nirnaith or Nírnaeth Arnediad “Unnumbered Tears” (LR/147, 310).
The word ᴹQ. ar is used pretty consistently as “and” in the 1930s and 1940s, although there is one example where it glossed “but” (PE22/124). In notes on Thrór’s Map from “The Hobbit” from 1936, N. ar is also used as “and”, but in the Moria Gate inscription in LotR drafts of the 1940s it has become a as in: pedo mellon a minno. In the published LotR, we have Q. ar and S. a.
Tolkien revisited the derivations of these words in the 1950s. There are signs he wanted to split up the basic meanings of the root AR² “beside/outside, without”: in a list of prepositions from the 1950s, he had ARA “along side” vs. ADA “against, opposite” (PE43/33). In notes appearing in PE17, Tolkien seems to have flipped this, making AR mean “beyond, further than” (PE17/147) and AD mean “along side, beside” (PE17/145). This new sense “beside” for AD may have replaced its sense from the Etymologies, where ᴹ√AD was glossed “entrance, gate”. The two main derivatives of ᴹ√AD, Q. ando and S. annon “(great) gate”, were both transfered to ANA “to, towards”, having now the original sense “†entrances, approaches” (PE17/40).
With AD now meaning “beside”, Tolkien devised a new etymology for “and”. The Quenya word remained essentially unchanged, but the Sindarin word a now took the form adh before vowels (PE17/71, 145). To further cement this change, Tolkien seems to have decided that the prefix ar- only meant “beside” in Quenya, not Sindarin, explaining Arnen “beside the water” as a muddled combination of both languages (VT42/17). Presumably the Sindarin prefix ar- retained its privative sense “without” based on its new derivation from AR “beyond”, since Tolkien retained that prefix in S. arnoediad/arnediad “unnumbered”.
Later still Tolkien considered yet another scenario for “and”, this time using the root AS for “beside”. This still produced Q. ar, S. a, but in Sindarin the prevocalic form was now ah (VT43/30, VT47/31). This last scenario appears in notes from the late 1960s, but Tolkien may have considered it even earlier, since prevocalic S. ah appears in the title Athrabeth Finrod ah Andreth from the late 1950s (MR/329). In this scenario, Q. ar was derived from primitive asa “beside”, as asa > aza > ara > ar, with the loss of the final vowel probably due frequent elision.
This gives us three possible roots for Q. ar, S. a “and”: AR, AD and AS. The main distinctions are in the prevocalic Sindarin form of the word: S. ar, adh or ah. As is typical, Tolkien vacillated a great deal in his later writings, and it is difficult to determine which scenario (if any) he settled on. I think AR “beside” was likely discarded, since S ar doesn’t appear after the early 1950s (its last appearance may be in Túrin’s Wrapper: VT50/5), but the choice between AD and AS is less clear. Of the two, AS seems to be the “later” scenario (1960s vs 1950s). Normally I don’t give this too much weight, but there are a few words, most notably Q. armaro/asambar(o) and S. sammar/ahamar “neighbor, (lit.) beside-inhabit-er”, that are best explained by AS and not AD.
I have yet to find any derivatives of AD that cannot be explained equally well by derivation for AS. Therefore, I think it is best to assume Q. ar, S. a < AS, with prevocalic ah in Sindarin. I am not sure what to do with the root AD, since it seems to be no longer necessary, have been replaced by AN (for “gate”) and AS (for “beside”). Unless/until some new form requiring AD is published, I am assuming this root faded into obscurity.