^{page 80 note 1} I do not share Thurneysen's feeling (Archiv 13, 27) that annōna, if a derivative of annus, must first have meant ‘Jahresgöttin,’ any more than that matrōna must first have meant ‘muttersgöttin’; nor do I suppose that our tale of Latin words enables us to satisfy such delicate questions of relation between primitive and derivative. But the earliest usage of annona is that gathered in the Thesaurus under rubric ii., with the defintion ' frumenti pretium et mercatus ': which leads me to derive annona from *anno + vosnd (: Skr. vasndm, ' Kauffreis,' with the vowel colour and gender of ὠν⋯, ‘price, purchase-money’, ‘price, ase-money’), taking anno-, of course, in the sense ‘anni proventus’ (cf. Thesaurus s.v. 120, 124), like Gr. ὧρα=τ⋯ ὡραῖα (Xenophon, cf. Liddell and Scott, s.v. iii.), and supposing annus to have been originally sort of harvest-⋯γών, celebrating the completed cursus of the year.

^{page 80 note 2} See below, p. 87.

^{page 80 note 3} If not derived from *tembno:σημβω.

^{page 81 note 1} We can hardly question but guminasium was a pronunciation in vogue somewhere or in some social circle, cf. Catullus, 63, 60 guminasiis )( 63, 64 gymnasi.

^{page 81 note 2} I advanced this etymology for dōnec in Trans. Am. Phil. Assoc, 29. 7-, but it has never, so far as I know, been noticed. I bring it up again only to give a better explanation of -nec. The demonstrative dum was originally used in pairs (cf. Richardson, de dum particulae …usu, p. 6, and Lane's Latin-Gr.,2§ 1992), and when the second dum became relative its demonstrative correlate fell away. Illustrative examples may be made as follows: haud, desinam (i.e., *dom nec d.) dōnec perfecero hoc (Terence, Ph., 419), ‘I shall not stop < the while >, the while I shall not have finished this’; Tacitus, Ann., 3, 20, neque (i.e. *dom nec) proelium omisit donec caderet, ‘he did not stop fighting < the while >, the while he did not fall.’ In justification of the negative in donec, in addition to the French instance originally cited, cf. Spanish no amé nunca hasta que no Ví a V., ‘dom nec amaví umquam donec videram te’; es un loco mientras no (=quasi ‘donee’ (se llegue a los céntimos (Valdés); cf. ‘me parecía que trascurrería un siglo desde que no [ =cum+a ‘superfluous’ non] hablaba con Pastora (Pardo-Bazán).

^{page 81 note 3} Thurneysen (KZ., 30, 493) also denied the retention in Latin of primitive -mn-.

^{page 81 note 4} With *agmnos cf. *g^whermnos as written by Brugmann, Grund., ii.,² § 174; *pelno- ( *pelmno-ib., § 181, p. 261); *leuqsmno- (§ 172, p. 244). According to Joh. Schmidt, Son. Theor., p. 147, *aĝmnς would yield *aĝmς and *aĝgmnoς áĝnoς, but with the stem *aĝmen vigorously alive, we may admit the interruption of the normal development of *agmnoς.

^{page 81 note 5} These definitions and descriptions are, of course, extracted from Grassmann's and the Petersburg lexica.

^{page 82 note 1} Leo Meyer's Gr. Etym., i. 516, can be consulted with great advantage on the etymology of ὂγμoς. Brugmann (Grund., ii.,² p. 246) thinks that ὂγμoς cannot be from a -mno- stem ‘wegen der o-Abtönung,’ which assumes a knowledge of all the accentual conditions under which *aĝgmno-might have fallen in composition far beyond our possibility of finding out. Cf. the doublet ἄκρις (ib. § 275 and fn.).

^{page 82 note 2} But the m may have got into sollemnis by some association with the himuς-group; cf. Skr. Héman.:hieumς.

^{page 82 note 3} The compound amb(i)-egnus, probably correctly defined by Huschke (ap. Walde, s.v.) as a <‘victima> circumacta’ (cf. Cato, R.R. 141. 2, agrum terram fundumque meum suovetaurilia circumagi iussi), is more likely to contain a no-participle stem to ago than to be a reduction of ag(m)no-ς, according to the phonetics maintained for the group -mn- by Joh. Schmidt (l.c., p. IOI, sq.), but amb(i)egnus from -ag(m)nos would bear the sense of ‘anniculus’; cf. μφιετηρς, of a yearly festival, μφιες | μφιτηoς, ‘annuus.’

^{page 82 note 4} Connected with the festival are agonalia and agonium =‘dies appellabatur quo rex <sacrificulus> hostiam immolabat’ (Paulus-Festus, p. 7, 31); further cf. agone ? =sacruficone ? (Thesaurus, i. 1390, 37; Varro, L.L. 6, 12), whence the later idiom agere lege=occidere (Thesaurus, i. 1388, 17).

^{page 82 note 5} In view of Hesychian τλεια ἱερ· τ νιασα (cf. Odyssey, ii. 454, σ***ν ἰερεσαντες νιασιoν. ‘suem sacruficantes anniculum’), we may realize how a temporal sense might have been reached for perakni-, sevakni-, though they started with the sense ‘κμα***oς maturus.’

^{page 83 note 1} See note 5, p. 82

^{page 84 note 1} Has anyone noted that Walde, s.v. acnua, credits Thurneysen with the rendering ‘der nächsten sechs Tage’ (lege Jahren) ? This made me, before looking up the passage cited, ask myself if we had to recognize a weak stem aĝhn-varying, thanks to loss of aspiration before nasals, with aĝn.: Skr. áhan., ‘day.’

^{page 84 note 2} Unless the ara igniaria had, each in a compartment of its own, two fires (cf. Fay, Class. Rev. 13, 397; and for the fact Stengel in Iw. Müller's Handbuch, v. 3, p. 16).

^{page 85 note 1} Cf. also Skr. āçā, Raum, Gegend. Himmelsgegend, perhaps cognate with Skr. áçman- in the sense of ‘Himmel’; cf. Hesychius, ἄκμων, κρνoςoὐρανς.

^{page 86 note 1} Greek χρνoς, for which no convincing etymon is offered, may well be a secondary form of κρνoς, with χ- due to association with some lost member of the χιών group; cf. χμαια: Lat. -himus in bimus, ‘two-year-old.’ The identification of χρνoς with κρνoς by the Greeks themselves is well attested, not only by literary instances, but by an inscription of the fifth century B.C. (cf. citations in Roscher's Lexicon, ii., col. 1546).

^{page 86 note 2} It is not certain that ‘year’ rather than ‘young animal’ is the primary signification of WETOS. The year may have been named from the ‘calf-time’ or ‘lamb-time ’; or, conversely, the young animal was named ‘anniculus.’

^{page 86 note 3} The religious rite had its popular application also; cf. Petronius, 135, 8, 8–9: at paries circa palea satiatus inani fortuitoque luto clavos numerabat agrestes where ‘numerabat clavos’ can hardly mean merely ‘clavos frequentes habebat.’

^{page 87 note 1} A reason for ōn instead of ŏnn is yet to seek.

^{page 88 note 1} It is curious that -mnos lends itself to comparison with Skr. Mánus, ‘man.’

^{page 88 note 2} Mommsen, Corp. v. 1685, reads alumn < a > bus; and Corp. xii. 4563 add. has a dat. sg. Fem < a > lumini (? with erroneous i in the penult), which makes one wonder if an *alomen, something like flāmen and Skr. Brahmän-, ever existed.

^{page 88 note 3} Lindsay's text reads at Poen. 703 balineator, without critical note; whereas Goetz and Schoell, edit, minor, report only balniator for A., and balneator for the Palatini.

^{page 89 note 1} I refer here to CLASSICAL QUARTERLY, 1,281 ff., where I have connected concinnat and its kin with the root of caedit. The semantic connection there effected I believe to be sound, but an etymology dependent on a morphological reconstruction can never satisfy as thoroughly as an etymology supported by a verbal identity in two different languages.