¹ E. Boisacq, Dictionnaire étymologique de la langue grecque², 1923, s.v. ἀτύζω. Also, doubtfully, in Walde-Pokorny 2. 616.

² In accordance with Sturtevant's usage I write medial stopped (and affricative) consonants as single voiceless stops (-p-, -t-, -k-, -z-) if they regularly occur single in cuneiform syllabic structure, regardless of whether they appear as surds (-p-, -t-, -k-) or sonants (-b-, -d-, -g-) or both; and as doubled voiceless stops (-pp-, -tt-, -kk-, -zz-) if they are written double part or all of the time (whether as type -pp-, or type -bb-, or both). The former set is probably to be interpreted as voiceless lenis, the latter as voiceless fortis. The former set corresponds to the IE voiced and aspirated voiced stops (type -b- and type -bh-), the latter to the IE unvoiced stops (type -p-). Initially only simple p-, t-, k-, z- are used for both type p- and type b- of Hittite, because it seems impossible to prove that the Hittite orthographic distinctions are phonetically significant.

³ See Sturtevant, A Comparative Grammar of the Hittite Language (references given in Hittite index sub hatukes- and hatukis); also hatugātar ‘terror’ listed in his Hittite Glossary. See, for ḫatuka, also A. Götze and H. Pedersen, Muršilis Sprachlähmung, ein hethitischer Text (Det Kgl. Danske Videnskabernes Selskab., Historisk-filologiske Meddelelser 21. 4, 10, 17–20, 50.)

⁴ See Sapir, Language 10. 274–79 (1934).

⁵ As the references in Götze and Pedersen's paper (4, 10) suggest. See particularly 18 note 1.

⁶ See Iliad, Z 38; Σ 7. Terror in movement seems to be expressed in the other Homeric examples as well (Iliad: Z 41; Θ 183; Ο 90; Φ 4, 554; X 474; particularly Odyssey: λ 606).

⁷ Götze and Pedersen, op. cit. 50.

⁸ See Thurneysen, Handbuch des Alt-Irischen §199, a.

⁹ See Walde-Pokorny 2. 111, sub 2. bel- ‘stark.‘

¹⁰ See Ernout et Meillet's interesting comment in Dictionnaire étymologique de la langue latine, s.v. āter: 'Implique souvent une idée morale de terreur, de malheur, de mort, et ce caractère affectif de l'adj. explique qu'il soit particulièrement usité en poésie.' With regard to the ātrī diēs they quote from Gellius, ap. Macr.: 'pontífices . . . statuisse postridie omnes Kalendas Nonas Idus atros dies habendos, ut hi dies neque proeliares neque puri neque comitiales essent', remarking of these ‘black’ (originally 'dreadful, ill-omened'?) days: 'Bien que l'origine en soit très contestée, l'expression provient peut-être de l'habitude de considérer comme des jours “noirs” ceux que viennent après les ides, c.-à-d. après la pleine lune par opposition aux jours “clairs” de la lune croissante.' (Cf. also quīnqu-ātrūs, sex-ātrūs, septim-ātrūs '5, 6, 7 dreadful, unlucky [?] days following the Ides', quoted from Wackernagel.) But the original meaning of āter may have been ‘dreadful, inauspicious’ with particular reference to ‘darkening’ storm clouds, so that we are perhaps once more brought back to the storm connotation noted by Götze for Hittite. Aside from Umbrian atru, adro (neut. pl.) there are no cognates of Latin āter which connect with ‘black’, though attempts have been made to relate it to IE words referring to fire, hearth, burning, ash (see Walde-Pokorny 1.42; A. Walde, Lateinisches Etymologisches Wörterbuch, s.v. āter; Ernout et Meillet, op. cit., s.v. āter).

¹¹ See F. Sommer, Handbuch der Lateinischen Laut- und Formenlehre §127, 1; 3, b; R. G. Kent, The Sounds of Latin §181, V; §141, VIII.

¹² Sommer, op. cit. §138, 1.

¹³ According to the de Saussure-Möller ‘laryngeal’ hypothesis, which is gradually making its way in Indo-European phonologic theory and which seems altogether inescapable to me, the x which colors a theoretical 'e' to a is absorbed into a preceding a < 'e' with lengthening of this a to ā (full grade of one of three types of 'heavy base' syllable).

¹⁴ I propose ọ for this vowel (Brugmann's å) to distinguish it from the more common o, which ablauts with e. Brugmann was misled by Armenian akn ‘eye’ into the belief that the high-grade ọ was nearer to a than was the normal o. Exactly the reverse is the case, as is shown by the fact that Toch. has o- for IE ọ-(really ọ-), e.g. okät and opäś-śi: Lat. octō and opus, while IE o- appears as Toch. a-, e.g. an-ac ‘to him’ < IE *ono-, Hittite ana-.

¹⁵ There could be no other method of indicating this sound in cuneiform orthography; see Sapir, loc. cit. For comparative purposes, and for comparative purposes only, it might be well to distinguish Hittite ḫ ₁ (= IE x) and Hittite ḫ ₂ (= IE , at least in certain cases), leaving ḫ for ambiguous cases and for transliterations of cuneiform orthography. We know from the reflexes of these two consonants in both Greek and Canaanite (Hebrew) that they must have differed in Hittite itself and these reflexes powerfully support the comparative Indo-European evidence. Hittite ḫ ₁ appears in Hebrew as ḥ (e.g. ḥitt : Hit. ḫatti-), while ḫ₂ appears in Hebrew as ‘(e.g. tid'al): Hit. tudḫaliyas). ḫ₁a-, i.e. xa-, appears in borrowed Greek words as ἀ-, as we have seen, while ḫ₂a-, i.e. γa-, appears as ό- (e.g. όπᾱδός: Hit. ḫapatis < Sem. ‘abad-).

¹⁶ That Toch. m- < IE y- is no figment of the imagination here is proved by the cognate Toch. mokc- (in abl. mokc-äṣ) ‘bowstring’ < *yow(n)k-ter- (cf. Gk. ζϵυκ-τ ρ-ϵς ‘straps attached to the yoke’, Skr. yoktár-). -c- results from -t-palatalized by following -e-.

¹⁷ See SSS 248.

¹⁸ See SSS 221–8, for many examples of such synonym compounds, omäs-keṃ belongs to a specialized set of such compounds, in which the analysis is no longer obvious. A more transparent example is ñom-klyu (dialect B ñem-kälywe) ‘fame’ < ‘name-fame’ (IE* nōm_en-klewos); here too -klyu seems not to occur alone.