as, preposition with. As we argued in the entry ar above, it seems that Tolkien eventually decided that ar and is to be derived from a stem *as rather than ar as in earlier sources; the Sindarin cognates a, ah suggest this. The conjunction and and the preposition with could descend from the same stem; the semantic gap between them is not too wide for this to be plausible. All that remains to be explained is why the s of *as becomes r in ar and, but stays s in as with. The development s > z > r is a well-known phenomenon in Quenya, but Tolkien seems to have entertained various ideas about what precise environment triggers this development. In the Etymologies, s normally becomes (z >) r if it follows a vowel and there is no unvoiced consonant following it. Hence we have primitive besnô husband > Q verno (LR:352 s.v. bes-), and the stem ólos- produces Q olor dream (LR:379). In this scenario, Quenya ar and could come from primitive *as, while Q as with would have to represent a form where the s was originally followed by another, unvoiced consonant so that it could not be voiced to z (later > r). Probably this consonant would simply be another s; double ss is common and cannot become voiced (e.g. primitive bessê > Q vesse, LR:352 s.v. bes-; a form **vezze > **verre did not arise even though the group ss immediately followed a vowel). As with could then represent earlier *assa (or conceivably *asse or *asso), later shortened to as. Compare nisse woman having the shorter form nis, LR:375 s.v. ndis-: Quenya does not permit double consonants finally, so when the final vowel is omitted, ss had to be simplified to s. – However, Tolkien later decided that for s to be voiced to z (in turn becoming r), it is not enough that it follows a vowel; it has to be intervocalic, a second vowel following after it as well (presumably a voiced consonant following would also do the trick, so that we would still see primitive besnô > Q verno rather than **vesno). Above we quoted olor dream from the stem ólos as evidence for the change s > r. A later source (UT:396) similarly quotes the stem as olo-s, but now the Quenya word for dream is given as olos with the final s unchanged, and only in the plural does the change s > z > r occur: The plural form is cited as olozi/olori. Here the original s was intervocalic because of the plural ending -i that followed it. According to this new system (final -s being unchanged) it would be possible to derive as with directly from primitive *as. Now it is rather ar and that is the mysterious word; since the change to r has occurred, the original s must here have been intervocalic at an earlier stage. Perhaps we are to assume that ar < *az is shortened from *aza < *asa? If so, the Quenya preposition ara outside, beside could be re-explained as the same word with the final vowel intact (this vowel persisting when the word was used as a preposition, but disappearing when it was used as a conjunction and shortened).
Átaremma, noun with pronominal ending: our Father. Concerning the ending -mmaour (here following a connecting vowel -e- to avoid an impossible consonant cluster) see separate entry. The word for father would here seem to be #átar; sources both earlier and later than this text have atar with a short initial vowel instead (QL:33, LR:349 s.v. ata-, WJ:402). Conceivably the vocative particle a (concerning which see aia above) is included here: *a Ataremma o Father of us being contracted to Átaremma. But if so, the integrated particle cannot be obligatory: The word atarinya "my father" in LR:70 is another vocative (Herendil talking to his father Elendil), even with a pronominal ending as in Átaremma, but this seemingly completely parallel example still does not show #átar- with a long vowel. Of course, this is a human son talking to a human father; it could be that #Átar with a lengthened vowel is rather a special strengthened form used when the title Father is applied to God. If so it may parallel Héru Lord in the Hail Mary-text; this word elsewhere appears as heru with a short e. – The word atar father, as it appears elsewhere, is in the Etymologies derived from a stem ata- that is likewise defined father (LR:349). The primitive form (one of the few ancestral forms that are explicitly identified as "PQ", Primitive Quendian) is also said to be atar, which Tolkien at this stage probably thought of as representing simply an extended form of the stem ata itself (*ata-r). This, according to Etym, produced Quenya atar pl. atari. Yet the plural #atári occurs as part of the compound Atanatári Fathers of Men which is attested several places, such as WJ:39 (also genitive plural Atanatárion in WJ:175). Tolkien repeatedly changed his mind about the precise shape of this word; the variants Atanatardi and Atanatarni are also found (WJ:174, 166/174). If atar father was to have the stem #atár- this would require a primitive form *atâr(-) with or without some short final vowel (cf. Anar sun being derived from primitive anâr, LR:348 s.v. anár-, and therefore having the plural form #anári – attested as part of a compound in PM:126 – instead of **anari). The stem-forms #atard- and #atarn- that Tolkien experimented with elsewhere (the first of which is hinted at already in the Qenya Lexicon, QL:33) would likewise require primitive forms including the "extra" consonant, probably *atardo and *atarno, respectively. Compare Quenya halatir kingsfisher becoming halatirn- before endings because the word descends from Primitive Quendian khalatirno (LR:394 s.v. tir-). Yet in the text before us, Tolkien wrote neither *Atáremma, *Atardemma, nor *Atarnemma when translating our Father, so this version of the Lord’s Prayer cannot be contemporaneous with any of these other experiments. Átaremma itself may rather represent yet another experiment with the precise form and behavior of the Quenya word for father.