DOI: https://doi.org/10.3318/ERIU.2017.67.4 Ériu LXVII (2017) 219–226 Royal Irish Academy VARIA II The Origin of Time* One of the words for‘time’in the Celtic languages is the etymon represented by Old Irish aimser (ā, f), Welsh amser (masc.), Old Cornish anser (n for m probably a scribal error), Old Breton amser, Modern Breton amzer (fem.). In Old Welsh, amser occurs as a temporal and causal conjunction ‘when; because’ (Falileyev 2000, 6). In accordance with a cross-linguistic tendency, these words for ‘time’ can also refer to the ‘weather’,1 thereby evidencing a remarkable semantic shift of reference from a primary abstract2 concept to a phenomenon of daily practical experience. The Irish and British words manifestly go back to a common ancestor which is typically mechanically reconstructed as *ammesterā (see, for example, De Bernardo Stempel 1999, 273, 426) or, disregarding the historical phonology and morphology of Celtic, as *amstero/ā “join, even”’ . To avoid the circularity of an explanation obscurum per obscurius that besets all previous suggestions, a new solution is proposed in this article . It will be argued that a more satisfactory etymological and morphological analysis is possible, and that this group of words can be connected with semantically more proximate words outside Celtic. The Proto-Indo-European root *√h2 et- ‘to walk, wander’ (LIV 273), which occurs as a verb only in the rare Vedic at- ‘to go, walk’, underlies Italic and Germanic words for ‘year’, that is Latin annus; Oscan loc.sg. akeneí, acenei, gen.sg. aceneis, gen.pl. acunum; Umbrian acc. sg./pl. acnu; and Gothic aþn(s)* and ataþni*, all ultimately from a preform *h2 et-no- (De Vaan 2008, 43–4; Kroonen 2013, 40).The semantic motivation behind the occurrence of this root in a word for ‘year’ is probably the idea of the ‘cycle/course/perambulation’ that the sun performs across the sky from one solstice until it reaches the same position again. Although super ficially only the initial a- seems to relate the words in these two language families with the above-mentioned Celtic lexemes, on a deeper level a preform can be set up from which all forms can be derived in equal measure. Starting from this premise, two slightly diverging reconstructions for the immediate precursors of Old Irish amm and Gaulish amman are possible, both meaning approximately ‘the going/course (of time)’ . 3 Kloekhorst (2008, 280–1) derives hamesha(nt)- from the root *√h2 meh1 - ‘to mow’ (LIV 279), since spring, which could also be written with the sumerogramme Ú.BAR8 ‘harvest ’, was the time of harvest in Anatolia and in the Near East. The three months of April, May and June are the time of bloom and of rain in the Anatolian highlands (Tischler 1983, 143–4). This semantic link with mowing or harvesting is ruled out for Celtic by agricultural facts. Kloekhorst accordingly does not mention the alleged connection with Celtic at all. 4 RIG III, 422 refers to the dictionary entry for ḫamešḫ(a)- in Tischler (1983, 143–4) who, however, makes no mention of the Celtic words. In the discussion of the etymology of the Hittite word,Tischler critically reports a suggestion by Čop according to which it derives from an Indo-European root *Hem-/Hom- ‘hot’ . Other Indo-European scholars have not accepted the existence of such a root. VARIA II 221 1. The first one starts from pre-Celtic *h2 et-mo- > *atmo- > Proto-Celtic *ammo- for Old Irish amm, and mutatis mutandis from *h2 et-mn ̥ for Gaulish amman. This etymology implies an ad hoc rule that Proto-Indo- European *‑tm- resulted in *‑mm- in Celtic. There is, to my knowledge, no other example in support of this change, nor is there a counter-example. Original *-dm- behaved differently. In Old Irish ammus ‘attempt’ Old Irish attach ‘refuge’, or *ad-bati̯ on- > *abbatii̯ on- > apthu ‘death’, this behaviour of preverbal ad could also have been extended to the position before m. Otherwise the retention of the cluster *dm can be observed. In formations such as Old Irish maidm ‘breaking’ *asmo- > *ammo- (for the treatment of Proto-Celtic clusters involving s, see Stifter (2017, 1191–2)). In...