^{page 601 note 1} Their claim to such eminence has already been widely admitted.

^{page 601 note 2} It is not as if the document of A.D. 137 were one of a series; it is isolated by a long gap (forty–three years–nearly a life–time) from the last preceding.

^{page 602 note 1} Cf. Barthold, Turkestan, p. 181

^{page 602 note 2} Reichelts list (p. 2) can be added to: MYh mother, iii, 12; TMTw when, several times in Letter No. ii, regarded as an Iranian word by Reichelt (e.g. 21/2 [s]t iii srδ YMTw MN cnt[ compared with 6/7 st iii srδ YKZY MN entry sr; in spite of the complementary –w it should correspond to kδ, which does not occur in the long Letter ii; were it not for this notable absence of kδ, one would of course think of cnkw; kaδ > lcai is = YMT also in Pahlavi); KL all in iv, 3 (Reichelt kδ), 8 (Reichelt mδ), distinct from kδ (in KL the tails of the letters are crossed); on MRY see below, p. 611

^{page 602 note 3} Proved by the personal names and the βγnpt–, i, 10 (cf. BS08., VIII, 583 sq.), who may well have been a priest at a local Nanai temple (Letter No. i was written in Tun–huang). Heathenish personal names, however, were only slowly abandoned by the Sogdians when they became Buddhists or Manichseans; the monks only of either Church had to take religious names. See e.g. the colophon of P 8 (Buddhists); the Mahrnāmag and Sogdica, 6–7 (Manichseans).

^{page 603 note 1} Sogdica, 7; JBAS., 1944, 137; BSOAS., XI, 737. Bānūk is said of Anahita, not only in Paikuli, but also in the inscription translated by Sprengling, AJ8LL., lvii (1940), 219 (in both instances the ideogram). Cf. further Horn–Steindorff, Sassanidische Siegelsteine, plate vi, No. 1621, Altindische Gemme, inscribed bnwky; the figure (a lady in a somewhat extravagant posture) may represent Anahita/Nanai.

^{page 603 note 2} Two such are known so far, n = Ān and γn = Xān in, see BSOAS., XI, 736. They clearly correspond to and respectively. Cf. Boodberg, Harvard Journal of Asiatic Studies, i, 1936, 291, n. 32. Professor Haloun points out that in Tang times the Chinese names must be understood to mean man from Bukhara (An) and man from Samarkand (Kang).

^{page 603 note 3} The Indians and Sogdians are mentioned in one breath in ii, 37 (in Lo–yang). Lou–Ian occurs once. The womans name;Čtysh, i, 1, is found also in the Niya documents, Catisa (–sa, –sae). Indian loan–words in the Letters were probably borrowed from the Indians of Lou–Ian. So e.g. Sykh = lekha, which was not used in Sogdiana (see the Mugh documents, 24, 25, 38, where nmk instead). In prst(k) Reichelt recognized prastha, cf. Niya 721, 6 mepoga prasta i. Here mepoga (left untranslated by Burrow) is prob. Pers. maipuxte (Syr. maipuxt;ā, Ar. maifuxlaj, –buxtai), and masu potīa (see Burrows note on 225) is the same, half–translated; maipuxte is a sickly–sweet decoction of grape–juice or must with spices added. Cf. Athenæus, 31d–e (i, 57); Yule–Cordier, Marco Polo, i, 84, 153, and notes; Kempfer, Amoen. Exot., 380.

^{page 604 note 1} I am indebted to Professor Haloun, who was ever ready to give of his time to help with the problems presented by the Sogdian Letters, for supplying precise dates and data for the years before and after 311.

^{page 604 note 2} [Restored], (incomplete) or (uncertain) letters.–Reading and precise meaning of xwtynβ/xwtyzβ are unknown. Reichelts translation is no doubt approximately correct; the derivation proposed by him is unacceptable.

^{page 604 note 3} Cf. ZDMQ., 90, 1936, 197 sq.

^{page 604 note 4} The only other restoration to be considered here is [cy]nnw.

^{page 604 note 5} Or zyzy(c)h. The third and seventh letters (possibly ––) are doubtful. The word may be incomplete at the beginning.

^{page 604 note 6} A scrap of paper which belongs somewhere else is stuck over the first half of the –t– and the lower part of Lin line 17.

^{page 604 note 7} The lines must be slightly rearranged.

^{page 604 note 8} knx{n)s. ? knx(w)s. ?

^{page 604 note 9} Or [n.. Perhaps [δn].

^{page 604 note 10} Cf. [δβ]npnwh vi, 4 (so to be restored).

^{page 604 note 11} Reichelt has kcmyδ instead. The third letter is certainly n/z, the fifth probably –n.

^{page 605 note 1} So probably, cf. r–sc, γwtm–sc and Av. daēnō.sāč.

^{page 605 note 2} In yncy(y) we can now recognize an older form of Man.Sogd. ycy (BBB., 96, 6 78; the etymology there proposed must be withdrawn). Reichelts translation ( woman ) not merely makes strange sense, but involves a grammatical error (δrwk in the place of Srwch).

^{page 605 note 3} Or from him. The correct translation is in Reichelts glossary, not however in his text. Instead of cn one can also say –nc (in enclisis), see line 34. This –n– is an enclitic pronoun of the 3rd pers. sing, and pi.; it is often devoid of meaning (as dativus ethicus). Cf. here cn from it 31; –mnc by me from them or him 32; kwyδt–n 9, kwyδwt–n 34 ktmw ZYn 10, lcCrwZYnYl in sentences in all of which the self–interest is involved, from the point of view of the subject. When it is doubled, the first is an ethical dative, while the second has the force of a full pronoun; ef. here WZYn–n wysynt or took (for themselves) their 18.

^{page 605 note 4} With some hesitation I have accepted Reichelts explanation of rδnk as from drang–. One could refer to Skt. draiiga frontier–watch station (ace. to Stein), Niya drarpga which, however, is office ace. to Burrow, BSOS., vii, 509 sq.

^{page 605 note 5} Cf. Parthian Hrwjft in the inscr. of Shapur, lines 5 and 12 (in Greek Kaiw, in Pahlavi trswhty, BSOS., IX, 836). Sogd. we–, wγt– emit, let; cf. P 13, 23

^{page 605 note 6} The second letter of the word I have restored as wy[kn] destroyed is not quite certain. Should one read w[ and restore a finite form of the stem occurring in line 15 ? Hence, pillaged ?

^{page 605 note 7} Moreover then the [Emperor was taken prisoner and led into captivity] by the [Hū]ns or some such words may well have stood in this line.

^{page 606 note 1} The restoration of ~}rnt kr[ as [wjrnt kr[t] is merely in the nature of a trial balloon. Unfortunately [w]rnt does not occur elsewhere. It might belong to the stem of Av. varata–, varaiBya (and poss. paiti–, pairi–varaya–, Sogd. ptw(y)rtk, BBB., 78, Sogdica 33 sq. ?). One could compare wrnt hrt– with Pahl. wUyc hrty pillaged in the inscr. of Shapur, line 14 (the explanation given in BS08., IX, 836, has now been fully confirmed by the T?arthian and Greek translations discovered since). The meaning of these words is to appropriate and take away movable goods as.well as prisoners from enemy territory; they take the name of the ravaged country as direct object; they cannot mean to occupy an enemy country. Pillage and despoil are fairly adequate equivalents (on the understanding that human beings are included in the spoils). Contrary to the views put forward by Sprengling, AJSLL., Mi, 1940, 368 sq., countries are not, as a rule, carried into captivity.

^{page 606 note 2} A place–name. The reading is very uncertain. See above, p. 604, n. 5.

^{page 606 note 3} my = mē from Old Iran. imai. Not to be confused with later Sogdian my, my, BSOAS., XI, 476, line 19, 736.

^{page 606 note 4} zyy myδ (Reichelt nyy) = azyi mē (as Pers. dī–rūz). A classic example of Sogdian –i from Olr. –ah.

^{page 606 note 5} Sogdian xēpaO, originally = own, belonging, presumably conveys the idea of vassalage here.

^{page 606 note 6} Wypryt– (vi + apa + riā–) scarcely differed much in meaning from pryt– (apa –f– ri) left behind, remaining.

^{page 606 note 7} Činastan here seems to mean the region around Khumdan, or at least chiefly that region, but not China. That the original sense of that term (the lands of Chin) should have been preserved till so late a date is surprising.–The remark added above the line is in the nature of a self–correction, could drive the Huns out of Khumdan–nay, even out of dinastan.

^{page 606 note 8} My first translation was or the other nations possessed themselves of it (= Khumdān). It may, however, be better to keep the remnant Chinese as the subject also of the second half of the clause, and to give full value to the ending of npw which is probably meant to indicate the accusative case. The chief difficulty lies in the meaning of np– which ordinarily is tribe, nation, people. For instance, the Nāf–nāmak list of nāfs (Sogdica, 8) enumerates names of nations (e.g. parsik Persian ), not countries (pars Persia ). Still, it must be admitted that by translating np– as country (as was suggested by Beichelt) the sense is considerably improved here (as well as in line 10).

^{page 606 note 9} Or freemen, lit. son(s) of a noble or freeman, āzat–pi$rak reflecting Olr. *āzātapuBraka–. Another hitherto unrecognized compound with āzāta– is ztkr noble(s) Fragm. iio, 25 (Reichelt, i, 59).–The passage indicates unexpectedly great strength in the Sogdian colonies in China. If there were one hundred freemen alone in Tun–huang–to whose number one has to add that of their families and slaves–the total Sogdian population of that town can have been scarcely below one thousand souls.

^{page 606 note 10} It cannot be made out whether Nanai–vandak himself returned with his wife or one of his agents. In the passage not translated here there are several verbal forms in the first person plural; their existence could be adduced as arguing for the assumption that Nanai–vandak undertook those journeys. Those sentences, however, seem to contain a general complaint of the bad times–of the windy and rainy weather in Reiehelts translation–and do not necessarily refer to the travels mentioned in lines 21–3. On balance it is more likely that the traveller of these lines is the one referred to in lines 6/7. Nanai–vandak would hardly have said it is three years since a Sogdian came from inside, if he had been that Sbgdian himself. The question is important for determining the place where N. wrote the letter.–Lines 23, 2nd half, to 29, 1st half, have been omitted here chiefly because I do not understand them; also, they do not seem to contribute to the solving of our problem.

^{page 607 note 1} The implication is: what happened to the China trade.

^{page 607 note 2} pār = Av. pāra guilt, Pashto por debt. It occurs also as prh. So in iii, 35, where Reiehelt read wrh; through the guilt of Farnxund, or through his debts, we have become the servants of Chinamen, I as well as my mother. Further in i, 9, I live in the deepest distress, without clothes, without food. I try to obtain a loan (xpsmprh, R. xwnmpnh), but every one refuses to give me one nmt, R. Filz, is finite verb, from nm– to agree, engage to do something, submit (lit. bend), with V to refuse, cf. V nmw i, 6, I shall refuse, R. Prugd; nmt δrt iii, 8; nm[(] iii, 11; further nm BSOAS., XI, 472, line 8).

^{page 607 note 3} Possibly = Buddh. zrmh, VJ., 399; zrmwnt, 213.

^{page 607 note 4} Probably = the China branch of your business is completely ruined; your (former) wealth from it = the China trade is no more. Scarcely = you would have no profit from my giving you a fuller account of recent Chinese history.

^{page 607 note 5} The name means slave of (Av.) Ašiš–vayuhi. The name of the goddess is spelt rtγw, rtwγ in the later Sogdian, see Orientalia, viii (1939), 9.

^{page 607 note 6} A gap makes it impossible to decide which of the following two meanings should be attributed to the sentence. (1) A. travelled with a caravan up to Kacan, but when the caravan went on, he had to stay there = in Kacan for six months, because he was ill or imprisoned. (2) A. was on the road there = with the caravan for six months–an abnormally long time for the distance from K. to.Saraγ; it demonstrates the state China was in four years ago.

^{page 607 note 7} A. and his party, or the caravan of line 35 with which A. travelled ?

^{page 608 note 1} Cf. BSOAS., XI, 726.

^{page 608 note 2} See e.g. Schseder, Iranica, 48 sq. The spelling γtvmδn, quoted by him from the Sogdian inscr. of Karabalgasun (p. 45) does not in fact exist. In his edition of that inscription Hansen gave mtvmSk (p. 21, fragra. 7, line 6), but being at a loss to account for such a word suggested timidly (pp. 36 sq.) that one could perhaps read ytvm–S–.–Schaeder quotes neither the Kumudana of the Fan–yil–tsa–ming, fol. 396 (Bagchi, Deux lex., i, 1929, 78, 295) nor, of course, the xitmtn of our Letter.

^{page 611 note 1} Cf. Schseder, Ung. Jahrb., xv, 575, n. 5. Schaeder thinks cyrδswn is a personal name, but gives no reason for this view.

^{page 611 note 2} Professor Haloun (who disapproves of my explanation) points out that a prefect of Chiuchflan, Chang Chen by name, is mentioned as having been involved in intrigues against Chang Kuei in 308. He was probably removed from his office at that time, and so whoever was prefect of Su–chou in 311 or a year or so later cannot have been in bis 13th year of office

^{page 612 note 1} No good alternative explanation is at hand. uHry is conceivably an Iranian word for number (it does not occur), and cyrδs– could be the (unattested) Sogdian word for 14 (this would leave –wn unaccounted for). However, to operate with the 13th year of the 14th number = period (?) is useless. The nearest 13th year of the sexagesimal cycle is 316, but the number of the cycle in the most widely used system of chronology is fifty (and in none fourteen). Cyrδswn fits no known name of a ruler or nien–hao.

^{page 612 note 2} R. Grousset, LEmpire des Steppes, 96. Cf. Sogdian Letter, line 16

^{page 612 note 3} Grousset, loc. cit.

^{page 612 note 4} Cf. Sogdian Letter, lines 37–8.

^{page 612 note 5} A. des Michcls, Hist. Geogr. des Seize Boyaumes, notes du comm., p. iv. Cf. Sogd. Lett., line 11. The verb, pršt, means fled, not weggegangen, cf. iii, 33, Farnxund absconded: the Chinese sought him (xwyzynt), but did not find him (βyrynt). Chr. Sogd. pryšty [read prěty ?], S.T., ii; pryi, George, 284; Yaghnobi apirēž, apiraiž, GrlrPhil, i, 2, 335, all = flee. From apa–raz– (Skt. rah–, OPers. rad–, MPers. raštan, NPers. rastan).

^{page 614 note 1} The first day of Tymyc is the 271st day of the Sogdian year; the first day of the Sogdian year fell on 10th September in 310 and 311, on 9th September in 312–15.

^{page 615 note 1} The other Letters bear no dates or no complete dates. It is taken for granted that those months that are designated by ordinal numbers are Chinese months.

^{page 615 note 2} Some of the passages have been translated in this article, so line 34, p. 612, n. 5; line 35, p. 607, n. 2. cy[nf]y [yh] also in iii, 16; cyn in line 17 is not yet clear. line 20 I shall (have to) learn how to be polite to the Chinese –plainly, to the writers mind, a fate worse than death. Reichelt has not well understood this letter, which must be read in conjunction with Letter No. i; both are written by the same woman (No. i to her mother, No. iii to her husband) and deal with the same matters. Her husband, Nanai–Sat, had left her without money in Tun–huang, and now she has been reduced to the ultimate ignominy of having to go into service with Chinese people. While writing (or rather, dictating) the letter she becomes more and more enraged, and ends up with the words Id rather be a dogs or a pigs wife than yours!

^{page 615 note 3} Huns and Hsiung–nu and The Legend of the Origin of the Huns (with full references), Byzantion, xvii, 1944–5, 222–251.

^{page 615 note 4} Found only in hūn–zādek slave (lit. a Huns son or a Hun–boy ) and hūnān slave–girl (lit. a she–Hun; formed with a Khwar. femin. suffix –ān), the equivalents of Arabic abd and jāriyah, Persian bande and kanīzak respectively.

^{page 615 note 5} xvi, 9, Huna Cirjigga may mean Hsiung–nu and Chinese, but this cannot be proved (cf. St. Konow, NTS., xi, 1938, 35).