¹ I follow Sturtevant in using h and s without the diacritical marks which are superfluous from the purely Hittite point of view.

² Charact. Indo-Européen de la Langue Hittite 144, 160.

³ Some New Indo-European Coincidences in Hittite 15.

⁴ Lang. 3. 109ff.

⁵ Act. du pr. congr. intern, de ling. 159.

⁶ AJP 53. 199.

⁷ Lang. 9. 3f.

⁸ It is true that Kurylowicz and Cuny as well as Sturtevant assume certain results of combinatory changes with h (see below), but these instances can all be explained otherwise, so that they are no more than corollaries to their theories, if accepted, but they in no way constitute a proof.

⁹ While the s of Hitt. pahs- is ultimately a formative (see below), yet it must have been very old in view of the correspondence of pahs- with the Latin participle pāstus. If we refuse to accept this conclusion, we must refer the -s- of pāstus and pāstor to IE t as in Gr. πατέоμα, Goth, fōdjan ‘feed’, etc.

¹⁰ To assume, e.g., * ₃or- for Hitt. haras ‘eagle’, and *or- for Gr. őρν ‘bird’ (Cuny, Rev. Hitt. et As. 14. 204) is a makeshift that cannot be convincing in the absence of a larger number of examples which conform to Kurylowicz's system without doing violence to such obvious correspondences.

¹¹ In fact all four Semitic laryngals are as old as Semitic itself. Cf. Brockelmann, Grundriss d. vgl. Gram. d. sem. Spr. 1. 42f.

¹² It is possible, of course, in spite of this that the IE laryngal is ultimately identical with those of Semitic. If so, the latter languages at one time passed through the same process as Hittite later, and the originally non-phonemic laryngals became phonemic at such an early age that it is hopeless to try to fathom its cause.

¹³ Since a non-phonemic laryngal stop would not be recognized in writing, the loss of pre-IE ' may be more apparent than real. It would be very hard indeed to judge where and when that loss had taken place. Even from languages of the present day as actually heard, we could only draw conclusions as to the final results.

¹⁴ The various Semitic laryngals and even gutturals are apt to shade into each other at various times and places, cf. Brockelmann 1. 128ff. The precise change of ' to h would belong to a period so much earlier than the specific changes there mentioned, that we could not find immediate documentary evidence for its ultimate source.

¹⁵ Cf. note 13.

¹⁶ In Sturtevant's Grammar are found some transcriptions with initial hw. Of these hwrtais ‘curse’ (with alleged syllabic r, p. 60) is a wrong interpretation of ḥu-wa-ar-ta-aš beside ḫur-ta-aš. The former must represent an actual by-form of hurtas (cf. also the noun ḫur-ta-a-iš, i.e., hurtais 'curse') which is probably a phonetic variant of some kind. If hwek- is a correct transcription, its h came from the by-form huk-. However, since no other combinations of initial h with consonants occur, we may conclude that, even where w was consonantal originally, it became vocalized so as to allow a preceding laryngal stop, e.g., huwantis 'wind, storm' (:Lat. ventus), although with original , must have been at one time actually pronounced 'uwantis with three syllables. If u was a consonant, there would be scant reason for the second u as it appeared in ḫu-u-wa-an-te-eš.

In view of the absence of h after other initial consonants Sturtevant is certainly wrong in interpreting iš-ha-a-i as shai instead of ishai.

¹⁷ Kurylowicz cites Recueil des publications scientifiques 127.

¹⁸ At any rate because of being obscure etymologically they cannot be used in any way in solving a question which concerns pre-Hittite conditions and thus depends on the behavior of h in words known to be old.

¹⁹ Cf., e.g., Theocr. 2. 11 άλλά, σ∈λάνα, φαϊνε καλόν ‘shine beautifully’ with φαίν∈ιν ∈voιav ‘display one's goodwill’ (Herodot. 3. 36).

²⁰ Cf. Brockelmann 1. 51.

²¹ AJP 51. 199.

²² Sturtevant, Lang. 5. 9f., derives verbs of the type handami from IE verbs in -āį-; the present writer from those in -ā- (Lang. 9. 291ff.). In either case, e.g., handaizzi must at one time have been tetrasyllabic.

²³ So Sturtevant and Cuny in passages already quoted (Lang. 9. 3f.; RHA 14. 202).

²⁴ So last in Hitt. Gram. 257.

²⁵ Cf. Brockelmann 50.

²⁶ Strictly speaking, of course, the of the Skt. iṣ-aorist is the final of a dissyllabic root, but it was extended analogically and then could appear secondarily as a variant of s. Cf. Thumb, Handb. d. Skt. 381f.

²⁷ This is the less probable of the alternatives because of the apparent age of the s of pahs- and Lat. pāstus, but cf. note 9.

²⁸ The change of radical e to a was a case of vowel assimilation. Cf. JAOS 54. 161ff.

²⁹ Hitt. Gram. 241.

³⁰ In article quoted above.

³¹ Such forms are not extant because of the almost universal spread of the κα-perfect in verb-stems ending with a vowel.

³² I am citing only present forms of the verbs in -ah-, because that tense of the hi-conjugation was the direct descendant of the IE perfect. However, the participle *newahanz(a) and the imperative 3. pl. newahantu may have been old enough to assist in the development of denominatives in -ah-.

³³ The assumed contacts between Hittites and Semites could have taken place only before the former settled in Asia Minor; not so much because the existence of large bodies of Semites in this territory is generally considered dubious, as because the analogical changes which caused Hittite h to get into ante-consonantal position probably required more time than could have elapsed between their settlement at Bhogazkewi and the time of the earliest Hittite documents. Undoubtedly the raid of the Hittites to Babylon about 2000 B.C. was not the only contact between Hittites and Semites. While the earlier movements of the former are obscure, it seems most reasonable that, as Indo-European newcomers from the East, their route took them through Assyrian and Babylonian territory. In fact the history of Hittite h as unfolded above is a distinct point in favor of such an assumption.

³⁴ It must be remembered that the phonetic value of h assumed above can be no more than tentative. Whatever it was, it does not invalidate the conclusions drawn about its original nature, and its change to phonemic value under Semitic influence. Compare Sapir, Lang. 10. 276, who maintains that although the main value of h is that of a voiceless velar spirant, it was also sometimes voiced, corresponding to Arabic 'ain.