¹ KZ 37.277 ff.

² This is, he says, because roots beginning with a vowel are few.

³ Sturtevant, A comparative grammar of the Hittite language, rev. ed. 1.60 (New Haven, 1951).

⁴ Austin, Lg. 18.22 ff.

⁵ The guttural is of course difficult. If the far more numerous palatal forms are admitted on an equal footing (on the ground that there may have been only one guttural series in IH anyhow), Arm. asełn ‘needle’ should be particularly mentioned as included in the material; see fn. 11.

⁶ Schrijnen, Wörter und Sachen 5.195 and elsewhere. Schwartz, Lg. Diss. 40.19 ff.

⁷ Hitt. e; see Sturtevant, The Indo-Hittite laryngeals 55, 60 (Baltimore, 1942). Since it is not assumed here that the different laryngeals as set up by Sturtevant had different histories after initial s-, our results are not affected if one of the other versions of the laryngeal hypothesis is preferred.

⁸ Sturtevant, IHL 40.

⁹ Sturtevant, CGHL² 1.39, proposes'.

¹⁰ KZ 14.341.

¹¹ If Austin is right in thinking that the third laryngeal gave Arm. h initially, that laryngeal would be excluded for sets 6, 7, and 13. But as we remarked concerning set 9, a vocalic initial in Armenian can also represent IE/IH s-; so that, in our examples, Arm. a- may stand for sъ- where reduced grade is possible.

¹² Sturtevant, IHL 51. For the following, see the entries in Sturtevant's Hittite glossary² (Baltimore, 1936) and Supplement² (Baltimore, 1939).

¹³ Lg. 23.8 fn. 8.

¹⁴ Benveniste, Origines de la formation des noms en indo-européen 1.163 ff. (Paris, 1935), points out that amplifications of roots with s movable are normally built on the s-less variant only (ker/sker ‘cut’: ker-t, kr-et); examples like sperg (: (s)per?) ‘scatter’ are exceptional. In the light of this it is remarkable that skēi and sreu seem to have some relation with sek and ser (sets 6 and 2). Perhaps root amplification belongs to a period subsequent to the loss of a laryngeal after initial s- and before a vowel.