^{page 755 note 1} Apart from this, however, -'k could represent -ak. So 3 'ngr'k GrBd 95⁴ (see Herzfeld, AMI i, 142, note 3) representing an unattested Av. θri-aюura-, as the following gloss ‘claw’ indicates. In IndBd. 29⁸ is in 30³IndBd 18¹⁷ is GrBd 66⁴rēšak ‘root’, GrBd 96⁷ TD 2 DH zūzak ‘hedgehog’. An extreme case is K 20 92, v. 9 for bang ‘narcotic’, in GrBd 43¹²mang explained as *bang-ič It is not an otherwise unmentioned plant, as Christensen thought, Le premier homme et le premier roi, p. 16 bīnāy, p. 48 note bīnay. Cf. also Frah. Pahl. 25⁷m ‘not’, and .

^{page 756 note 1} Dāt Dēn 36⁹. ¹⁸hzngrvkzym, corrupted 36²⁴hzgrvkm'n and K 35 197, r 14 (= SBE 18, 258).

^{page 757 note 1} This development nh > , not nh > ng, makes the explanation of Turfan Mid. Iran, 'rdhng, “rdhng as *arta-θanga- equivalent to the Arabic bušrā al-haqq, by Schaeder, Gnomon 9, 1933, p. 347, note 3 (which seems likely to impose itself), very uncertain. For *sanha- or *θanha- either *sah (*sāh) or tah (*teh) must be expected, while -hang could represent only Old Iran. *θanga- or hanga-, in association with ard-.

^{page 757 note 2} Arm. gan (stem gani-) ‘striking’, ganem ‘I strike’, could as well be an Iranian *gāni- (cf. Old Pers. bāji- to bag-), as a genuine Arm. word, HAG 431.

^{page 760 note 1} Continued consideration of the problem of , etc., now inclines me to understand AP- as Aramaic, to be read ut (uδ, u), just as 1 Aramaic V is ut (uδ, u). For similar double spelling cf. (perhaps = Z for Aramaic ZY, as L = man is for LY) = ī beside ZY-m = ī-m.

^{page 760 note 2} For the gloss in this position, cf. GrBd 95⁴ P angurak panč <ak> ‘claw, that is pančak (claw)’.

^{page 764 note 1} A different reading may here be proposed for GrBd 197⁵ TD 2 has … kangdiz ī bāmīk aγrēraθ, IndBd (ed. Justi) , K 20, 122 v 19 has perhaps P omits. It is probably better to read pus <ī> aγrēraθ than būm with IndBd. [Corr. Examination with magnifying glass makes. almost certain inK20.] 1 take the opportunity to correct the reading of GrBd 231⁶ (BSOS vi, 951). The Paris MS. has . Hence read andar gar ī patišxvār-gar griftār kart.

^{page 764 note 2} Since my own attempt to solve the problem in BSOS vi, 945, Herzfeld, AMI vi, 58 ff., has re-examined the matter. In neither discussion was all the evidence cited. Benveniste has touched upon the problem in BSOS vii (1934), 271 ff.

^{page 764 note 3} Sām is immortal, but asleep (GrBd 198²). A difficulty has arisen over the description of his resting-place. Mēnōk ī Xrat 61²² (ed. Anklesaria) states that Pišīn is full of wormwood: ut-aš zar-gōnīhfrahist *dramnak (, Pāz. dramna, Skt. damanakah). This is the lord of non-medicinal plants: GrBd 121¹²*dramnak (TD 2 P , K 20, 116 v 12 , Pāz. darmanah) ī daštīk abičakān urvarān rat. It is the NPers. diramna, a word I found used in Yazd indefinitely for ‘fuel’. These passages provide the explanation of GrBd 198^12–13 (of Pišīn) ōδ *dramnak (TD2, P DH , K20, 123 v 7 ) pat bālā δ bulandtar rōδēt ‘there wormwood grows higher in height’. and GrBd 198³ (of Sām) miyān ī *dramnak (TD 2 , DH K20, 123 v 15 ut-š vafr hačapar nišast ēstēt ‘he lies in the midst of the wormwood, and over him snow has settled’. Other translations are given by West, SBE v, 119, and Herzfeld, AMI ii, 60.

* The reading of the Iranian equivalent of this Aramaic word has not yet been. satisfactorily settled. [See the supplementary note.]

^{page 766 note 1} The MS. evidence of P makes an original reading impossible, hence the reading suβδastān adopted in BSOS vi, 950, must be rejected.

^{page 766 note 2} Cf. K 20, 106 r 11 and v 7 syāmak, GrBd 106², 107²

^{page 766 note 3} Cf. the various spellings of dāitē (Av. dāityā) with yā > ē: K 20, 113 r 19 d'tyk, Vid. 19², GrBd 86³, GrBd 87⁹

^{page 768 note 1} Cf. the transcription K 20, 123 r 17 mechanically from harzak ‘free’. It is not necessary to suppose that an allusion to Gōpat was intended.

^{page 768 note 2} Cf. the converse agōβāk (Zātspram 31, 1; I am able to quote this through the courtesy of Mr. Anklesaria).

^{page 771 note 1} -ist from -ast is probably to be recognized frequently. This explains best Turfan Mid. Iran, ‘vbyst *βist “fell” < ava-pasta-, but also the -ist- of NPers. gulistān attested in some Pahl. spellings with -yst'n), beside Armen. burastan. It may be seen also in 'vyšt'd *ava-stāta (rather than abi-štata, as Henning, ZII 9, 195, with obi-). Georgian ostat-i “expert” may presuppose an Armenian *ostat, Pahl.’ vst't, NPers. ustād, cf. Old Pers. avastaym.