¹ I use the symbols b, d, and, by implication, g and gw, not for true voiced stops but for voiceless lenes; p, t, k, kw then represent the corresponding voiceless fortes. It seems unfortunate that Sturtevant transliterates both -d-, -t-, for instance, and -dd-, -tt- as -t-, instead of -d- and -t- respectively, as strict phonologic and phonemic method would have required, -d- : -t- seems preferable to -t- : -tt-, in spite of Hittite usage, as it is most improbable that the phonetic difference expressed by the Hittite symbols -d-, -t- as against -dd-, -tt- was primarily one of length in view of the fact that they respectively correspond to IE -d- or -dh- and to IE -t-. Similarly, the single -s- of easy as contrasted with the doubled -ss- of messy is not, for English today, indicative of a quantitative difference but of a difference of voicing. I use ḥ in ḥabadi- to warn the reader that, while h is a serviceable transliteration for Hittite, the consonant in question sounded more like a voiceless velar spirant (Akkadian ḥ = ch of German ach) than like a simple aspiration, h. This does not prove, however, that Hittite ḥ was pronounced exactly like Akkadian ḥ.

² There are, in all probability, fewer exceptions than Sturtevant admits. Thus, Hittite mi-ta-iz-zi (77) must probably be interpreted as meda(e)-, with IE *med-(Goth. mitan, Gk. , Lat. modus), not, as Sturtevant suggests, as meta(e)-, directly related to Lat. mētor. Again, the fact that Hittite wak- ‘bite’ is consistently written with -k-, -q-, -g-, never with -kk-, -gg-, merely suggests that it goes back to IE *woĝ(h)-, *waĝ(h)- or *wog(h)-, *wag(h)-, not that, being related to Skr. aś-‘eat’, it has an irregular orthography for wakk-; the difference of meaning and the Hittite w- already warn us that Hittite wak- (read wag-) and Skr. aś- (IE *e- or *a-) have probably nothing to do with each other. Again, that Hittite sakwa ‘eyes’ (81), even if analyzed into s-akwa, has always -k-, not -kk-, presumably means that it cannot be connected with Lat. oculus, nor can it be directly equated with Hittite sek-, sak- ‘know’, which has both -kk-, -gg- and -k-, -g-; this latter difficulty is confirmed by the fact that ‘eye’ always has a labialized single -k-, while ‘to know’ has an unlabialized k-sound.

³ For which, as Sturtevant observes, ‘Indo-European’ may be substituted by those who, like myself, are not convinced of the necessity of contrasting Hittite with all the other languages of the ‘Indo-Hittite’ family.

⁴ I hope to show in another paper that the most common value of Hittite ḥ is x (voiceless velar spirant) or ḥ (Arabic ḥā), and that another value was (voiced velar spirant) or ‘ (Arabic ‘ain).

⁵ See examples in Bauer and Leander, Historische Grammatik der hebräischen Sprache, p. 14 (old Canaanite *‘aparu “dust”, Hebrew ‘āfar, e.g., is transliterated as ḥaparu), p. 22, and elsewhere.

⁶ Both the spirantizing of d to đ and the lengthening of accented a to ā are later Hebrew sound changes.

⁷ For a discussion of this identification see Franz Böhl, Das Zeitalter Abrahams, Der alte Orient, 29. 1, 14, l5. The Egyptian transliteration of the Hittite king's name also shows the '.

⁸ It is perfectly possible, of course, that the borrowing was not directly from Semitic but from some Anatolian language, say Proto-hattic, which had in turn borrowed it from a northern Semitic language.

⁹ Ras Shamra Canaanite has been proved to have masculine plurals in -īm and ‘segholate’ nouns (type qatl-) with plurals in type qatal-.

¹⁰ See Otto Proksch, Die Völker Altpalästinas, 25-8. In the latter part of the Hittite rule the Amorites (Amurru) of northern Syria were the most important vassal state of the Hittite monarch; see A. Götze, Das Hethiter-Reich, Der alte Orient, 27. 2, 36, 40.

¹¹ It should be borne in mind, however, as Bolling warns me, that these formations are far more archaic than -o - denominatives certainly used at the time that όπάζω was formed. The -o- denominative of an *όπαδι might reasonably be expected to yield *όπαδίω.