^{page 10 note 1} Seiler, l.c., explains the regular long quantity of the second a in as probably metrical. I shall return below to this question.

^{page 11 note 1} See my Studies in the Greek Negatives, pp. 47–48; Lejeune, M., ‘Observations sur les composés privatifs’, Rev. Phil, xxxii (1958), 198–205Google Scholar. The evidence of Mycenaean, with which Lejeune mainly deals, is unfortunately not clear on the questions whether it consistently had (1) a- before h plus vowel, and (2) an- before vowel without h. The most difficult forms are three with an- where his postulated, contrasted with seven showing an- without h (including such clear forms as a-na-pu-ke and a-no-we, a-no-wo-to): they are a-na-mo-to, a-no-pa²-si-ja, a-no-po. Of these the first is probably the most securely established, as either (Ventris–Chadwick, etc.) or Palmer). But even here the truth is tantalizingly fugitive, since the base of both and and the aspirate is not original, although regularly appearing in these derived forms. So for Lejeune suggests the possibility of a Mycenaean form in opposed to the Classical However, the point in doubt for Mycenaean is purely concerned with psilosis, with the question of the form of prefix before stems which originally had h. There is no sure example quoted from it of h- before a vowel in stems which never had h: the single case of a-e-ti-to is insecure evidence, as Lejeune grants. To this extent, therefore, there is agreement with Homer, and that is what is relevant to the present discussion.

^{page 11 note 2} There are two examples of the negative prefix before h in Homer, apart from (see above), ‘insatiable’ and ouwo? supno-, weak grade of *sueopno-.

^{page 11 note 3} So in Lex. frühgriech. Epos (Erbse and Laser), following Wackernagel and Bechtel.

^{page 11 note 4} Glotta xxxiv (1955), 292–4.Google Scholar

^{page 11 note 5} The etymology of is not certain. Hofmann, Et. Wb., connects with Skt. yabhati, so giving as the source of Greek h.

^{page 12 note 1} See my Studies …, pp. 49–50.

^{page 12 note 2} The latter sense wrongly listed as active in L.S.J.

^{page 12 note 3} Fraenkel, ad Aesch. Ag. 1211, remarks that it is inappropriate in the case of (as in that of ) to speak of an active or passive force, since the word means ‘devoid of hurt’. This disagrees with his own note on v. 238 (discussing ), where he grant! that of course active and passive force are common in verbal adjectives in and quotes as one such, taking v. 1211 as showing the passive sense (vol. ii, p. 137, fn. 1). The confusion arises from doubt whether is nominally derived (possessive compound) or verbally: it may be either. The further question arises, whether the distinction is worth retaining, that we should speak of active and passive force only for adjectives of verbal derivation. For this presupposes that active or passive sense is formally expressed if we have a verbal adjective in yet that is not the case. The formation is ai adjective, not a participle integrated in thi conjugational system; what is expressed i no more than association with the idea contained in the verb (see Wackernagel, , Syntax i. 136). Hence if we have a nominal derivative such as where the original nour describes an action or the product of an action, it seems proper to establish the sam categories of meaning lexically as for a verba derivative. So and may be described in parallel: botfi have (a), activ sense, ‘not speaking’ (which meaning is shared by English ‘speechless’); (b) passive sense, ‘not to be spoken (of)’. In short active and passive are here categories established by the lexicographer for our convenience of understanding: they are logically not formally based.Google Scholar

^{page 14 note 1} ‘Methodology in Linear B interpretations’, Die Sprache v (1959), 131–6.Google Scholar

^{page 14 note 2} The laryngeal evidence is supplied by H. Hitt. hasas ‘satiety’, Palaic aš(š) ‘drink one's fill’.

^{page 14 note 3} Palmer, , op. cit., p. 136, n. 21, would take directly from *ṇ-sã-tos, with the verb stem showing zero plus normal grades.Google Scholar

^{page 14 note 4} Lejeune, , Traite de phonétique grecque, § 75.Google Scholar

^{page 14 note 5} For in other forms of the verb, see Palmer, , op. cit., pp. 135–6.Google Scholar

^{page 14 note 6} So Risch, , Wortbildung der homerischen Sprache, p. 17Google Scholar; cf. Hirt, , Handbuch der gr. Lautund Formenkhre, p. 97.Google Scholar

^{page 14 note 7} The ablaut is not of course confined to such verbal adjectives: see Schwyzer, , Gr. Gr. i. 359 ff. for examples and discussion.Google Scholar

^{page 14 note 8} Risch, , Wortbildung, p. 18Google Scholar; Schwyzer, , Gr. Gr. i. 503.Google Scholar

^{page 14 note 9} I have discussed these in A.J.P. lxx (1949), 159Google Scholar ff., and lxxiii (1952), 298 ff. See further Specht, F., K.Z. lxiii (1936), 207 ff., ‘Zur Vermeidung von Wörtern mit drei kurzen Silben’.Google Scholar

^{page 14 note 10} So Risch, , Wortbildung, p. 22, n. 1 (with modification wohl).Google Scholar

^{page 14 note 11} So Wackernagel, , Kleine Schriften, ii. 1157, n. 1, while accepting . as nominal, is puzzled that a verbal was not used instead.Google Scholar

^{page 15 note 1} Risch, , Wortbildung, p. 18Google Scholar, Chantraine, , Formation des noms, p. 305Google Scholar. But Buck-Petersen, , Reverse Index of Greek Nouns and Adjectives, p. 470, regard derivation in from nouns as post-Homeric.Google Scholar

^{page 15 note 2} Lehmann, W. P., Proto-Indo-European Phonology (1955), pp. 88 ff.Google Scholar

^{page 15 note 3} In Handbuch …, pp. 82, 95 ff. Specht, , K.Z. lix (1932), 89–119Google Scholar, has a useful collection of material, without himself giving a convincing solution. In Latin it does not seem always possible to start from a reduced, monosyllabic form of the stem in cognate words. (in) domitus shows the same grade as the present stem of domo, while there is no trace here of the reduced *dmã-; cf. Skt. damita-, dãihta-. In genitus, opposed to gnãtus, there is again a strong form of the stem (i.e. normal plus reduced grades), with cognates in O. Genetaí and in Celtic. Walde-Hofmann, Lat. Et. Wb., s.vv. domo, gigno, regard the strong forms as secondary; similarly Ernout-Meillet, Diet, Étym., s.v. domo. But Kent, , Forms of Latin, z 420. 1, regards this as a regular Latin process, comparing lautus from *laui-tosə, and so in the second conj. tacitus from tacẽcirc;-re, etc.Google Scholar

^{page 16 note 1} In Homer and Hesiod there are four formulae, differing in metrical value, (I) II. 14. 271 (2) ibid. 8. 369 (3) ibid. 15. 37 (= Od. 5. 185, H. Ap. 85) (not equivalent to (2): it is used at the start of the line, with caesura after ); (4) Theog. 805 In the Hymns two of these are duplicated: (5) H. Herm. 519 (= (4)); (6) H. Dem. 259 (= (I)).

^{page 16 note 2} Cf. Lehmann, , op. cit., p. 15 for the phonology.Google Scholar

^{page 16 note 3} See the account of by Risch (Lex. frühgriech. Epos): he lists the many explanations offered in antiquity, but finds none very satisfying. Of these he prefers connexion with (cf. II. 21. 386 ), as does Verdenius, , Mnemos., 1957, p. 248Google Scholar. But in the sense of the verb is ‘waver’ (in doubt and fear): this is not a very promising parallel for (Il. 18. 410 of Hephaestus) does not fit into the picture as a compound of either for form (because of -ι-) or sense. It is attractive to follow the suggestion of Palmer, (Minos v [1957], 61) to connect it with Mycen. a-ja-me-no and a verb ‘do, make’, used in the special sense of craftsmanship.Google Scholar

^{page 17 note 1} I am obliged to Mr. J. Chadwick for criticism, especially on some Mycenaean aspects in this article; it is of course not implied that he agrees with my remarks thereon.