¹ Abandoning further resistance I shall henceforth adopt the common practice of transliterating, in Christian Sogdian, ṭēth by plain t, and tau by θ.

² In the light of later-published material one wonders whether this is not misreading of by'[n]ypšqty (y and n resembling each other in that position).

³ BSOAS, xi, 3, 1945, 486 sq.Google Scholar

⁴ See recently E. Benveniste, Hittite et Indo-Europeen, 33–40.

⁵ Spelt wydwtk'n in Pahlavi. In the horoscope Gr. Bd. 51.10 it corresponds with femmes, JA, 1915, I, 17 (Ikhwān al-Ṣafa) = nisā', Mafātīḥ al-'ulum 227.15, with ‘wives, concubines, matrimony and nuptials’ in al-Beruni's Tafhīm, see Taqizadeh, Gāh-šumārī, 328 sq. (in the Persian Tafhīm: zanan u surrīyatān va zanāšū'ī u ‘arusi…, p. 429, ed. J. Huma'i).

⁶ Berliner soghdische Texte, II, 884.

⁷ nwyštyt in the edition is probably misprinted.

⁸ Tr. Phil. Soc., 1954,133 sqq., 154; Annali 1st. Univ. Or. Napoli, Sez. Ling., 1,2,1959,133–44.

⁹ The prosthetic vowel (ekt-), which is commonly marked in Man. orthography, is expressed here only because the scribe began a fresh line with the final part of the compound; it is naturally absent when the final is joined with the main body.

¹⁰ Annali, I, 2, 1959, 137. As arises from the preceding annotation, -šp'- does not exist.

¹¹ H. W. Bailey regarded Sogd. ‘sp’nč- (MPers. aspinj, Pers. sipanj etc.) as a distant relative. However, that may be a derivative of MPers. asp- (Man. hasp-) ‘to rest’, aspīn (Man. hspyn) sbst. ‘rest’ (in Pahl. e.g. Dd., 36.17; DkM., 161.3; in Pazend ŠGV), so that sipanj would mean ‘resthouse’ even by etymology. The Mugh material has now shown that spnčyrspn consists of spnčy + (')rspn; the alternative spelling 's, spynš'rspn-t'masterly paper ‘Madu’ (Silver jubilee volume of the Zinbun-Kagaku-Kenkyusyo, KyotoUniversity, 1954).Google Scholar

¹³ BSOAS, XVII, 3, 1955, 603 sq.Google Scholar

¹⁴ The other passage discussed there (PT, 113.4) is too involved in various difficulties to be relied on.

¹⁵ ‘Les gens de ces diverses principautēs [of Sogdiane] aiment le vin’, T'ang-shu ap. Chavannes, Doc., 134. The Mugh documents abundantly confirm that observation, see e.g. Mugh B2 (Freiman, Problemī Vostokovedeniya, 1, 1959, 120 sq.).

¹⁶ See Laufer, Sino-Iranica; Bailey, ‘Madu’ E. H. Schafer, Golden peaches, 141 sqq. Laufer was the first to suggest that Chin, p'u-t'ao represented ‘a dialectic form of Avestan maóav ’ (p. 225); I thought the dialect was Sogdian, BSOS, x, 1, 1939, p. 98, n. 3.

¹⁷ BSOS, x, 1, 1939, 98.Google Scholar

¹⁸ S.v. ārūy, p. 57.13, ed. Horn; see his preface p. 13, n.

¹⁹ K. al-abniyah, 189 (where Rhazes is cited).

²⁰ BSOAS, XI, 4, 1946, 719 sq. It owes its frequency to having been employed as translationequivalent of Skt. madya-.Google Scholar

²¹ It is impossible to construct an Old Iranian form from which it could descend.

²² According to Bailey, Annali, I, 2, 1959, 136, both are primarily loan-words in Ossetic, but ultimately of Iranian origin.

²³ Annali, I, 2, 1959, 133 sqq.

²⁴ Elsewhere, haurā is the ordinary word for ‘gift’ in general.

²⁵ G. R. Driver, Aram, doc., No. 6.

²⁶ Great inscription of Shapur, Pahl. lines 25, 27; cf. Jackson mem. vol., 42 sq.

²⁷ Bailey, H. W.assumed ‘barley-yeast’ instead, later (BSOAS, XXIII, 1, 1960, 28) ‘a drink made of fermented barley’.Google Scholar

²⁸ Pahl. texts, p. 30, şş 33–5, and p. 33, ş 66. Persian xāmīz (used also in Arabic, in Tha‘ālibī’s Ghurar, 707.2). For other languages see Hübschmann, Arm. Gr., 96, s.v. amič Talm. ‘wms’, explained as ‘ raw meat’ Levy I, 41, should be added. The persistent definition as‘raw meat pickled in vinegar etc.’ speaks for the derivationfrom āma- (xām) ‘raw’.

²⁹ A special study on the Khotanese preverbs is an urgent desideratum. The distribution of vi, abi, ava, apa, upa, and pati in particular is in need of clarification. Etymologists increasingly operate withOlr. pa-, a counsel of despair. We have seen that in Sogdian, e.g., every apparent pareflects antecedent apa-orupa- (or, rarely, pati-). It is an error to rely here on Armenian, where pa- was first recognized as a separable element, in Iranian loan-words, by A.Meillet, REA, II, 1922, 6; cf. E. Benveniste, Ét.langue oss., 99 sqq., who has several fresh examples but takes a different view from mine. That surely reflects Iranian upa-, with loss of initial u- in accord with a well-established rule, from which case-forms and derivatives of monosyllables alone are exempt. Naturally this applies only to the oldest layer of loan-words; for later on initial u- was replaced by a- in Western Middle Iranian (hence, e.g., apart'an). A telling example is Arm. pačar-k' ‘device’ where u in the Ir. original is directly attested through Aram. ‘wpčr (see my ‘Mitteliranisch’, Handbuch der Orienlalistik, iv, 1, 1958, p. 39, n. 4). Arm. pēt-k’ ‘need’ is identical with Parth. 'byd, both from upēt- (cf. MPers. 'b'y-, ever derived from upa-i-). Further paran ‘lassoo’ from upa-dāna (cf. abi-ddna ‘bridle’ and pari-dāna ‘saddle’, JRAS, 1944, p. 140, n. 1); paxrē ‘ xpnma', Sogd. prxyy, BS0A8, XI, 3, 1945, p.468, n. 5, from xrī-‘buy’ pastaṙ ‘blanket’, evidently = Av. upastarana- (rejected by Hūbschmann, Arm. Gr., 222); parēt from upadēt- (cf. Arm. dēt); the words adduced by Meillet and Benveniste; and several uncertain ones. Clear examples of loss of initial u- are zgoiš ‘circumspect’ from uzgauša-, and zraw- ‘terminate’, from uz + raw- (as in Pers. raflan), cf. Man. MPers. 'zrw- ‘go out, end’ (e.g. k' cr'h… 'zrwyd ‘when the lamp goes out’ dwswx 'dwryn 'y ny 'zrunvyd ‘hell-fire that never comes to an end’); the last-named was judged differently by H. W. Bailey, Tr. Phil. Soc., 1953, 37.

³⁰ See Gershevitch, Gramm., ş ş 314, 441.

³¹ This does occur in other Iranian languages.

³² -θu- must be struck off the list; for γyδrp- is now confirmed through the Mugh letters, and pš'βr, in spite of occasional pyš-, had originally-a-as first vowel, therefore cannot belong to Av. piθwā-: that is proved by Arm. pašar, which reflects a contracted form, ★pašār.

³³ BSOAS, XI, 3, 1945, p. 486, n. 1. The word for ‘bridegroom’ mentioned there is far more likely to read ρys'k (than pys'k), thus hardly enters into consideration (its resemblance to Arm. p'esay may be coincidental). The wording of t he Middle Persian verse (above, p. 242) may yet be held to favour retention of the connexion with puBra: pwsryn (r'y) wdtvdg'n qyrd (hence lodwdg'n = fiyny, pwsr = ps, qyrd = 'krty). 34 On Sogd. š from sr see Gershevitch, Gramm., ş 371.

³⁵ On 'krty (etc.) see above, pp. 243–4. The modes of spelling the word-final (Man. -ty, -t'k; Chr.-ty,-t) are self-contradictory; they both demand and exclude -ē from -ak. One may question the reading in Bin. sogd. T, II, 884, 17, which would necessitate regarding -y as a case-ending and -'k as a mistake. The precise form of kar- cannot be determined securely. Apparently kṛta-ka-; kṛti- (Sogd. 'krtyh, Man. 'kty', Chr. qty' Benveniste, JA, CCXLVII, 1, 1959,131 sq.) should have fem.-h in Sogd. script and never appears as -kt, see Gershevitch, Tr. Phil. Soc., 1945, 141.

³⁶ M. G. Dumézil sees the distinction between Aryaman and Bhaga as parallel to that between personae and res: ‘ sous Mitra … Aryaman s'occupe de maintenir la société des hommes arya … et Bhaga … assure la distribution et la jouissance régulierès des biens des Arya ' (L'idéologie tripartie des Indo-Européens, 1958,68; in greater detail Les dieux des Indo-Européens, 1952,47 sqq., esp. p. 54). This view does not sufficiently account for Bhaga's interest in marriage, the very foundation of society.

³⁷ Thus rather than ‘hostage’ np'k- has both meanings, as has Pers. navā, which should be added (Khwar. nb'k = Arab. rahn).

³⁸ The remaining words are not wholly clear.

³⁹ See Livshitz, pp. 41, 81.

⁴⁰ Passages where that meaning is applicable are: A 17, line 1; V 15, lines 6 and 7 (cf. Livshitz, pp. 162, 169). Less clear is βγy in A 13, line 2 (Livshitz, p. 69; cf.M.N. Bogolyubov and O. I. Smirnova, Xozyaystvennïe dohumentï, 1963, p. 71). Neither can ZY βγy be the same as ZYβγy, nor can βγy function as an accusative after kw (should be βγw). The word following βγy has been variously read as pr'w, pβ'w, and pryw; of these, the second and third may be discarded as misreadings (‘beloved lord’ in the address of a letter would also be an oddity), but the first, producing an impossible sequence, pr'w rty c'nkw, is no better (rty is not = ZY). The only way out of this desperate situation is to propose a fourth reading: prnw (which is technically unobjectionable), and take βγy-prnw (distributed on successive lines) as the acc. of a personal name, βγy-prn, which occurs precisely so in the Mahrnāmag, line 87 (in Manich. script naturally βγyfrn). The letter was thus addressed to Trγ'n and βγyprn, two persons, as required in any case to account for the plural form of the verbs. Tarxān and Bagēfarn were bankers, and the letter is a bank draft. I do not recall expressing the opinion attributed to me by Livshitz, p. 69, n. 29, concerning the word read enkw by Freiman (in line 5).

⁴¹ So also in a fragmentary letter, Livshitz, p. 221.

⁴² cf. H. W. Bailey, BSOAS, xiv, 3, 1952, 422; H. Lüders, Mathurā inscriptions, ed. K. L. Janert, 1961, 95.

⁴³ I. M. Diakonoff and V. A. Livshitz, Dohumenti iz Nisī, 1960, 24.

⁴⁴ cf. BSOAS, xxiv, 2, 1961, 191, where I suggested ‘(trusting in) a contract of alliance with Tīr(ī)’ (which could be applied also to ✶Bag(a)mihr).

⁴⁵ JRAS, 1944, 133 sqq.

⁴⁶ Orientalia, NS, VIII, 1939, 95.

⁴⁷ Livshitz, loc. cit., p. 63, n. 37.

⁴⁸ As arises from the enumeration of the lunar mansions, see JRAS, 1942, 242.

⁴⁹ Used by Grigor Magistros (Nor baṙgirk'); hitherto overlooked.

⁵⁰ BSOAS, XI, 1, 1943, 68 (line 26).

⁵¹ Not counting their appearance as day-names.

⁵² Sogdica, p. 16, line 16, and p. 19, where Stackelberg, WZKM, XII, 1898, 244, should have been cited. On ' rwtprn- cf. BSOAS, XI, 4, 1946, 737; now 'riwtprnc also at Mugh.

⁶³ Agathangelos, p. 325, line 13, ed. Tiflis, 1914.

⁵⁴ smyr γry 'ivy' sp'nl'rmδz'y cyntr tyst‘he enters into the earth of Mount Sumeru’ P 16, 36; cf. Khwar. spnd'rmd = Arab, arḋ ‘earth’.

⁶⁵ Bailey, H. W., ‘Languages of the Saka’, Handbuch d-er Orientalistik, iv, 1, 1958, 134.Google Scholar

⁵⁶ The problem is twofold: (a) why did the Manichaeans not use rwxšny'k ‘light’ here, as they did elsewhere (the answer supplied by Waldschmidt-Lentz, Munich. Dogm., 564 sq., does not carry conviction); and (b) if they wished to avoid rwxšny'k, why did they choose 'rtxwšt, which, if it was regarded as ‘Genius of the Fire’ in Sogdiane, too, was particularly unfitted for service as ‘light’ in a list of elements that included fire

⁵⁷ cf. I. Gerahevitch, The Av. hymn to Mithra, 19.

⁵⁸ Al-Beruni has two separate series (Chron., 47.1–3), both corrupt. In the Mugh material a single name occurs, 'rtγwst rwe A 4, R 2, 4 (the edition by Bogolyubov-Smirnova, loc. cit., p. 60, follows the reading proposed in Orientalia, NS, VII, 1939, 90, without acknowledgment). Freiman originally identified it with the fifth day of Series A, which I accepted. However, the resemblance to al-Beruni's form is too slight altogether. I would now assume that at Mugh the ordinary day-names (1–5) were used also for the epagomenae, so that 'rtywšt rwc of the ‘month’ MNwp'nc'k was the 363rd day of the Sogdian year. A name allegedly belonging to Series B (No. 3), srδycrt, has been read by Bogolyubφv-Smirnova, p. 43, whose edition of Mugh B 1 has been justly criticized by Livshitz, 219 sq., who reads myδyckt (as a personal name). Such a mixture of calendar systems would be highly unlikely in any case.

⁵⁹ Other points deserving attention include: the absence of yazata- (yzt' BSOAS, XI, 4, 1946, 737, on 185/6, is doubtful); the function of fravaši- (prwrty in a Manich. text as ‘soul’ or part of a soul, JBAS, 1944, p. 137, n. 7; but in Christian Sogd. frwrt = ‘grave’, frwrt-qty = ‘tomb’ Manich. 'rt'wfrwrtyy as element merely reflects Parth. 'rd'w frwrdyn); ✶bagadānaka- ‘altar’ žwšy δr'wšyh as ‘(bloody) sacrifice’ (like Arm. zoh) against Av. zaoθrā, see Gershevitch, JRAS, 1946, 183; etc.

⁶⁰ cf. W. Barthold's remark on the frequency of place-names ending in -faγn (Turkestan, p. 120, n. 6) = Sogd. βγn-. One of them, xšw-fyn ‘the six temples’, is apposite (see Gershevitch, Gramm., 64). Arm. bagin excludes ✶bagana-, ✶baganya-, etc., and demands ✶bagina-, which meets all requirements (incl those of Mathurā bakanapati, recognized by H. W. Bailey, BSOAS, XIV, 3, 1952, 420 sq., which reflects a form with reduced vowels, ✶eq1, ✶βaγna-); it has been approved by Wackernagel-Debrunner, II, 2, 352.

⁶¹ 'rtyxw-flntk, ‘Ancient Letters’, II, 35, BSOAS, XII, 3–4, 1948, p. 607, n. 5, = ‘the slave of A.’. A misunderstanding has been caused by βntk elsewhere. Livshitz, p. 54, n. 6, claimed that in an inscription he had published in Izv. Otd. Obšč. Nauk AN Tadž. SSR, 14, 1957, 101 sq., a word which was ‘clearly’ (HCHO) βntk had been wrongly read βn-snk by me in ‘Mitteliranisch ’, p. 130. However, I have never even seen the cited article and have no access to the periodical in question. The inscription I referred to had been published, three years earlier, by A. M. Belenitzkiy, whose photograph shows a plain -s-, markedly different from -t- in the preceding line; in shape it resembles the -s- in the SCE

⁶² δrw'sp-βntk ‘the slave of D.’, ‘ Anc. Lett.’. Cf. Arm. Druasp (shortened name, like Hormizd, Bahrāin, etc.).

⁶³ γwmδ't (Mugh); on the plant (xwm) see ‘Mitteliranisch ’, 85.

⁶⁴ Tīšfarn (Mahrnāmag), Tīšδat (Mugh), and others, cf. Livshitz, p. 63.

⁶⁵ BSOAS, XI, 1, 1943, 74.

⁶⁶ BSOAS, xi, 3, 1945, p. 482, n. 3.

⁶⁷ In a Manichaean fragment published in JRAS, 1944, 142–4 (cf. also p. 137), Nnδβ'mbn ‘Nana the Lady’ mourns somebody's death on a bridge, in spectacular fashion. An undated Chinese memorandum (about A.D. 600 ?) on the customs of the people of K'ang, translated by Chavannes, Doc., p. 133, n., may be called upon to identify the cause of her sorrow: it was the death of Adonis-Tammūz. ' Ils ont coutume de rendre un culte au dieu céleste et l'honorent extrêmement. Us disent que l'enfant divin est mort le septiéme mois et qu'on a perdu son corps … les hommes chargés de rendre un culte au dieu … revêtent tous des habits noirs … vont pieds nus, se frappant la poitrine et selamentant; … des hommes et des femmes … se dispersent dans la campagne pour réchercher le corps de l'enfant céleste; le septième jour, (cétte cérēmonie) prend fin '. This is a fair description of Adonia, which took place at the height of the summer: the seventh (Chinese) month began at the end of July or in August. The possibility that ‘Kūγūne the son of Ahriman ’ was a Sogdian version of Adonis should now be considered. Sacrificing on a bridge: cf. Herod., VII, 54; Ammian-Marc, xvm, 7, 1; cutting off ears: cf. Sūrah, IV, 118; laceration of faces: al-Beruni, Chron., 235.10, at the Sogd. equivalent of the Frawardīgān.

⁶⁸ Pncy nnδβ'mpnh on coins, see O. I. Smimova, Katalog monet s gorodišča Pendžikent, 1963, Nos. 356–463. This reading (which inevitably is obvious to the editor of the Manichaean fragment mentioned above) has also been recognized by O. I. Smimova, ace. to a hand-written note in her Katalog. It is a pity that the discovery came too late to help in the attribution of the coins; they may belong to Δēwāštīč, whose absence from the coinage of his country it would be difficult to understand. The reading of several other names appearing on the coins (‘Amogyan’, ‘Vidyan’, etc.) will also have to be revised.

⁶⁹ Rēwaxš-yān ‘gift of R.’ (Mahrnāmag, Mugh).

⁷⁰ Chavannes, Doc., p. 139, n. 3 (cf. p. 312); ‘le p'o-lo d'or’ is a golden banner (paiā).

⁷¹ Txs'yc-flntk ‘the slave of T.’, II, 53, 59. Not, of course, ‘T., the slave’. The author of the letter, Nny-βntk (‘the slave of Nanai’), is speaking of his own son. One could draw the inference that Nanai and Taxsič were associates (but T., a rapacious god, whose statue required daily sacrifices of 5 camels, 10 horses, and 100 sheep, was no Adonis). There is no reason, other than partial surface resemblance, for connecting this Sogdian god with the Turkish tribal name Tuxsī (V. Minorsky, ḥudūd, 300). His famed sanctuary may be sought at Taxsī(č) in the district of Abghar (cf. Barthold, Turkestan, 132), some 20 miles to the north-west of Samarkand; it may have been regarded as part of nearby Ištīxan = ‘Western Ts'ao’.

⁷² cf. Sogdica, 7. The resemblance of δ'p'tšyrh to nny-šyrh etc. suggested that δ'p't was the name of a divinity (cf. now also δ'p'tšyr, Mugh). Yet, as Chr. d'p't (ST, I) has been confirmed since (‘Passion of St. George ’, 266, see I. Gershevitch, JRAS, 1946, 183), it becomes possible to understand the name as ‘truly good’. The other name formed with δ'p't (δ'p'tsyγ[h], cf. BSOAS, XI, 4, 1946, p. 737, n. 1) may favour that opinion. The exclamation could be analysed as meaning ‘with (the word of applause) Spat’.

⁷³ E. Benveniste, Textes sogdiens, 68 sq. (P 3, 203–19); cf. also BSOAS, XI, 4, 1946, 714, 729.

⁷⁴ cf. ‘Mitteliranisch’, 85.

⁷⁵ cf. Livshitz, 125.

⁷⁶ cf. Hübschmann, Arm. Gr., 37, 506; Justi, s.v.

⁷⁷ In Mazendaran, in similar conditions, the ‘White Dēv’ maintained himself as a god, see Noldeke, Archiv für Religionstvissenschaft, XVIII, 1915, 597–600. The history of daiwa- in Iranian has been correctly seen and clearly represented by H. Lominel, Die Religion Zarathustras, 90 sq.