^{page 289 note 1} Sitzungsberichie der Königlich Preussischen Akademie der Wissenschaften, xliv, 1904.Google Scholar

^{page 289 note 2} There is now perhaps a possibility of Dialect A of the Tocharisehe Grammatilc being recognized as the language of the kingdom of Agni, see Pelliot, T'oung Pao, 1936, A propos du “tolcharien”, Sieg, SBAW, 1937, Und dennoch “Tocharisch”, Haloun, ZDMQ, 1937, 257.

^{page 289 note 3} The dash through the g in g is intended to range the aksara with the other new Brāhmī aksaras , etc.

^{page 290 note 1} Two unclear akijaras.

^{page 290 note 2} Added below the line.

^{page 290 note 3} A small blur before dhā, not an akgara.

^{page 291 note 1} kadeparāni = kalevarāni.

^{page 291 note 2} hi.

^{page 291 note 3} sahitau.

^{page 291 note 4} citāyām.

^{page 291 note 6} labdhvā.

^{page 291 note 6} -tād (Turkish renders by ablative).

^{page 291 note 7} vistīrna. I am indebted to M. de la Vallée Ponssin for the reading vetana.

^{page 291 note 8} -kād (Turkish renders by ablative).

^{page 291 note 9} tā for tu ?

^{page 291 note 10} Turkish renders by instrumental.

^{page 291 note 11} sutavat.

^{page 291 note 12} bhūsana.

^{page 297 note 1} g may have been fricative.

^{page 297 note 2} Thomsen, Samlede Skrifter, iii, 47, note 1, considered that both and were probably fricative.

^{page 298 note 1} In Iranian loanwords used also for Sogd. ‘moon’. Hence q was evidently not Von Le Coq gave two modified forms, SBAW 1909.

^{page 299 note 1} The value kk of this aksara has been adopted from its use in Sanskrit words, to avoid an interpreted transcription. It occurs in the Sanskrit word savitarkkam, Pischel, SBAW, 1905, Tafel x, 1. 4.

^{page 300 note 1} It is now clear that ha: represents Turkish fricative y, and initially also x, as shown by an (unpublished) Khotanese ha:ham:na, ha:ham:ni xayan, beside khaham:ni. The conjecture, BSOS, VIII, 918, that ha: might also represent q is wrong. For q, k is used. A renewed consideration of the word uha: has shown that it is actually, as first suggested BSOS, VIII, 884, note 5, a title.

^{page 301 note 1} If the Turks adopted g from another (presumably Iranian) language, it may have represented a there and have been adapted to their nearest equivalent in the palatal series.

^{page 301 note 2} Turkish g is interpreted by gy in Khot. digyirnna ‘tigin’.

^{page 301 note 3} The replacement of is perhaps part of a larger problem— which can only be treated adequately when all texts are available—of palatalization in later Khotanese, where we find ky-, py-, -ly-, sty- (see the published lists of Khotanese words). The Barčuq dialect seems to have reached a similar stage.

^{page 301 note 4} Henning has perhaps prematurely adopted the value velar y for g in ABAW, Ein manichäisches Set- und Beichtbuch, 58, note 2 (as earlier ZDMG, 90 (1936)), in comparing gi with Sogd. yy. It is better to compare first the more closely related Khotanese, which has ji.

^{page 302 note 1} We may see in this the name of a people, perhaps the Khoca of E 16⁹ (maṃkuya rro īndā heinā khoca u huna cimgga supīya), with the suffix -ana, which appears also in su ana, if this is, as Henning has suggested, ZDMG, 90 (1936), 13, and as seems probable, ‘Sogdians’. It is evidently an oversight, apart from the g, to compare Turkish xayan. Chinese (Karlgren, 414 and 296) shows that the word did not have o. We know the word also in Tib. kha–gan (JRAS, 1927, 58), and, as noted above, in Khotanese khaham:ni with a. In Arabic script we have , and in script ) with the q used with a.

^{page 302 note 2} If it is desired to see in this word a Turkish word, the -trk'n of the Mahrnāmay 34 is comparable, if k is not a mistake for x (cf. 37), -if a form existed.