^{page 501 note 1} Central Asian forms: Bailey, , JRAS., 1939, 90Google Scholar. The list of ‘Oriental variants’ given by Moravesik, G., Byzantinoturcica, ii, 280Google Scholar, under χαγάνος, is strangely incomplete; not even Arabic and Persian xāqān is mentioned; while one and the same Armenian form (xak'an) is quoted in two different transliterations.

^{page 501 note 2} So-called Pseudo-Avars.

^{page 501 note 3} Nöldeke, , Tabari, 99Google Scholar; Taārib al-Umam, i, 153 sqq.Google Scholar

^{page 501 note 4} The collocation with ‘Turks’ suffices to discredit the Xāqān; Marquart, , Ērānšahr, 52 sqq.Google Scholar, judged differently.

^{page 501 note 5} D. Sinor, in his paper Autour d'une migration de peuples au v⁶ siècle, J.A., 1946–1947, 34 sqq.Google Scholar, has thrown doubt upon the identity of the Žuan-žuan with the (true) Avars, which I continue to regard as firmly established. The determining factor, it seems to me, is the story of the defeat suffered by the true Avars at the hands of the Turks, Theophylactus Simocatta, vii, 7. According to it, the remnants of the Avars flee to Τανγάστ and the Μουκρί: we know now that these are the names of (Northern) China and Korea. This fits well with the history of the Žuan-žuan, but excludes the possibility of placing the Avars in the neighbourhood of the Ural.—On Μουκρί = Skt. Mukuri = Tibetan Muglig see Pelliot apud Bagchi, , Deux Lexiques, ii, 348Google Scholar, cf. my Sogdica, p. 7Google Scholar. It has not so far been recognized that this name is found also in the Orkhon inscriptions, in the form Bōkli or rather Bükli: the list of the nations that came to mourn the passing of Bumīn qaγan and Istämi qaγan (i E 4 = ii E 5) opens with Bükli čölig il (Thomsen ‘das ferne Bökli Volk’) and continues with Tabγač; the list began in the east (ōṅrä kün toγusīqda); the Bükli qaγan (i E 8 = ii E 8) had been the easternmost of the enemies with whom the Turks had had to fight, in former times, at the behest of the Chinese (on this passage see Schaeder, , Iranica, 39, n. 6)Google Scholar. Thus when the Turkish report reproduced by Theophylactus declared that the Avars fled to Ταυγάστ/Ταbγαč and Μουκρί/Βüκlι, we should understand ‘to our eastern (mainly hostile) neighbours’.

^{page 502 note 1} See Parker, , Thousand years, 139, 153, 161Google Scholar; China Review, 24 (1899), p. 34Google Scholar ( about A.d. 265). Cf. Marquart, , Ērānšahr, 53, n. 2.Google Scholar

^{page 502 note 2} See BSOAS., xii, 54.Google Scholar

^{page 502 note 3} ibidem, 601 sqq.

^{page 502 note 4} The obvious but often challenged inference: that they were Huns, has rightly been defended by Thompson, E. A., Attila and the Huns, 10 sqGoogle Scholar. Harmatta, J., Acta Archaeologica Academiae Scientiarum Hungaricae, i, 1951, 137 sq.Google Scholar, again attributes a loose use of the name ‘Huns’ to Priseus, on the ground that those writing ‘immediately after Priscus’, Malalas, Procopius, Agathias, etc., used the term as vaguely as he used ‘Scythian’. He misses the point of Thompson's argument; which is precisely that these authors, in fact, wrote a long time, two and three generations, after Priscus.

^{page 502 note 5} Re-examination of the photograph of the Sogdian, NāfnāmakGoogle Scholar (see my Sogdica, p. 8)Google Scholar has shown that the name in line 17 ends in -yry and that the preceding letter is -p- rather than -k-, The resulting (s)[.]pyry can hardly be restored otherwise than as s[']pyry = Sabir-ē (-ē is Sogdian ending). There were thus remnants of the Sabirs in the neighbourhood of Turfan long after the migration of the 5th century; it is scarcely accidental that it is precisely near Turfan that many scholars of rank have placed their home-country (cf. Sinor, D., loc. cit., 15 sqq.Google Scholar, who disagrees).

^{page 502 note 6} This migration has recently been discussed at length by D. Sinor in the article quoted above, p. 501, n. 5.

^{page 503 note 1} The names of the crossing-places are confused, here as almost always. See Marquart, , Ērānšahr, 99Google Scholar sqq., who reversed the sequence of the ‘Gates’.

^{page 503 note 2} Sinor, D., loc. cit., p. 2Google Scholar, boldly asserts that Jordanès … les [= Acatziros] situe à l'est des Estoniens.

^{page 503 note 3} Outright rejection has at least been envisaged by Thompson, E. A., op. cit., p. 96.Google Scholar

^{page 503 note 4} Where precisely one is to imagine their seats is not by any means clear.

^{page 504 note 1} Thompson, E. A., op. cit., p. 13.Google Scholar

^{page 504 note 2} This title is also mentioned by Gaujakec'i, Kirakos, Venice, 1865, p. 98Google Scholar; the translation of the passage quoted (from Brosset) by Chavannes, , Documents, 253, n. 7Google Scholar, is somewhat inaccurate. ‘[List of the bishops of Albania] Tēr Viroy, 33 years. He had spent many years imprisoned at the court of Xosrov the king of Persia, but after his death was freed and returned to his country. He redeemed the Armenians, Iberians, and Albanians made captives by Šat’ the Xazir, the son of abu-xak'ani, who had enslaved our land; he founded five [sic] towns in the name of Šat': Šat'ar, Šamk'or, Šak'i, Širuan, Šamaxi, Šaporan.' That Viroy returned only after the death of Khosrau ii is stated also ibidem, p. 30, where the last Sassanian kings are enumerated in these terms: ‘After Xosrov, the king of Persia, Kawat took the kingship; he released from captivity Viroy, the Catholicus of Albania, whom his father had imprisoned. After Kawat, Artašir; then Xoṙeam by decree of Heraclius; then Xosrov, and after him Born and Zarmanduxt—all these were short-lived; and then Yazkert.’

^{page 504 note 3} The contrary passage in Sebēos has been made to refer to the Khazars only by arbitrary emendation: by changing i cafayut'iwn 'into the servitude (of the great Xak'an, etc.) 'into i caṙayut'enē ‘out of the servitude’ (Marquart, , WZKM., xii, 1898, p. 191).Google Scholar

^{page 504 note 4} The corresponding national name in pseudo-Zachariah (see presently), B'GRSYQ, has been restored as B'RSYLQ by Marquart. An even easier correction would be B'RSYGQ, directly = the Armenian Nom. Barsiłk' (Barsełk'), which occurs beside Ba(r)silk'; Syriac G = γ perfectly answers to Armenian ł. If this correction is accepted, Armenian intermission will have to be assumed; this is in any case recommended by the final -Q, which is best explained as the ending of the Armenian nominative.—In view of the remarks made by Sinor, D., loc. cit., p. 63Google Scholar, it must be stressed that the -k' of such Armenian names as Barsiłk', Sawirk', belongs solely to the nominative; the form in which Barsiłk' (Bargełk') appears most frequently in texts is in fact Barslac'.—The Syriac form BRS'LY' (Michael and from him Barhebræus) does not go back to John of Ephesus, as Menges, K. H., Byzantion, xvii, 276 claimsGoogle Scholar; see below.

^{page 504 note 5} Streifzüge, 489 sq.Google Scholar, cf. p. 485.

^{page 504 note 6} Nöldeke, , Tabari, 270Google Scholar = Taārib al-Umam, i, 219⁷Google Scholar. Marquart's attempt to prove that Hormizd's mother had been the daughter of a Khagan of the Khazars, (WZKM., xii, 199 sq.)Google Scholar carries little conviction.—It may not be superfluous to warn students against accepting at its face value Marquart, 's phrase die Chazaren welche in den gleichzeitigen Urkunden zuerst in der Kirchengeschichte des Johannes von Ephesus a. 585/86 genannt werden (Streifzüge, 46)Google Scholar. In this form this assertion was never correct (it should have read wurden in the place of werden); it has been disproved by none other than Marquart himself. The matter stands thus: Barhebræus, in a passage about the Avars, mentions the Khazars twice, firstly in an allusion to a ‘Khāqān, king of the Khazars’, secondly as having been named ‘Khazars’ after the eldest of three brothers (in an eponymic story). The importance of this text was recognized by Marquart, Chronologie der alttürkischen Inschriften, 1898, 82 sqq. On reading his book, Nöldeke informed Marquart that Barhebræus' story probably derived from the lost chapters of the Ecclesiastical History by John of Ephesus, who wrote in A.d. 585–6; the headings of the lost chapters in question (3rd part, book 6, chaps. 45 sqq.) are preserved and give some indication of their contents; the name Khazar is not mentioned in them. Barhebræus, however, used John not directly, but at second hand through Michael the Syrian (end of 12th century); the latter was not then accessible to Marquart (see WZKM., xii, 1898, 198 sq.)Google Scholar. Later Marquart secured the relevant text of Michael and discussed it fully (Streifzüge, 1903, Addenda, 479 sqq.)Google Scholar. It now emerged that in the first passage, which he in fact had copied from John, Michael had ‘Khāgān the king of the Abārīs (Ἄβαρεις)’, as was to be expected; and that the second passage (about the three brothers, two of whom came to Alān = BKS'LY', the eldest being named Xazarīg) was not derived from John at all, but from a much later source (not earlier than A.d. 678 in Marquart's judgement).

^{page 505 note 1} Nöldeke, , Tabari, 155, 166, 167Google Scholar; Taārib al-Umam, i, 181¹², 182⁶, 183², 192¹⁰, 198⁷.Google Scholar

^{page 505 note 2} Disputable are all passages that contain interesting details (Ṭabarī, , i 895¹¹⁶Google Scholar, 896⁴ = Nöldeke, , Tabari, 157Google Scholar, 159): here Marquart imported the name into the text by an emendation (WZKM., xii (1898), 169Google Scholar, n. 6 = Chronologie der Alttürkischen Inschriften, 96, n. 2)Google Scholar, which, since Taārib al-Umam, i, 179³Google Scholar () agrees with the MSS. of Ṭabarī, is only acceptable if it is referred back to the Pahlavi text, in which *'PGL and *ḤČYL (= Xazir as in Arm.) were indistinguishable.—There is no certain occurrence of the name Xazar in an existing Pahlavi text; the curious 'TWL in Bahman Yašt, iv, 58Google Scholar, ed. Anklesaria, was emended by H. W. Bailey first into ḤPTL = Hephthalites (BSOS., vi, 1932, 946)Google Scholar, later into ḤČL = Khazars, (BSOAS., xi, 1943, 1 sq.).Google Scholar

^{page 505 note 3} Translated by Hamilton, F. J. and Brooks, E. W., p. 328.Google Scholar

^{page 505 note 4} ibidem, p. 327.

^{page 505 note 5} ibidem, p. 329. The story is extraordinarily confused, but nevertheless there is no reason to doubt the author's truthfulness; on the Albanian mission see Marquart, , Streifzüge, 489Google Scholar. That the list was not derived from Greek sources is assured by the forms of the names (names transliterated from Greek into Syriac are invariably recognizable as such); to describe it as eine aus griechischer Ueberlieferung … zusammengestellte syrische Völkerliste (Schaeder, , Iranica, p. 40)Google Scholar does it no justice. Attention should be paid also to the names in the lines immediately preceding the list, Gurzān, Arrān, Sīsagān, etc., none of which admits Greek transmission. Incidentally, Marquart was so thoroughly convinced of the independence of the passage from Greek influence that he used forms found in it to argue for Syriac transmission of a report on Turkish affairs preserved by Theophylactus Simocatta (see WZKM., xii, 189 sq.).Google Scholar

^{page 505 note 6} cf. Müller, F. W. K., Ostasiatische Zeitschrift, 8, 312.Google Scholar

^{page 505 note 7} Streifzüge, 356, n. 1.Google Scholar

^{page 506 note 1} e.g. by Zeuss, Kaspar, Die Deutschen und die Nachbarstämme (1837), 714 sq.Google Scholar

^{page 506 note 2} Streifzüge, 41, n. 2, 43.Google Scholar

^{page 506 note 3} To meet Marquart's principal objection.

^{page 506 note 4} On this form see Moravcsik, , op. cit., ii, 289 sq.Google Scholar (with references).—No weight, perhaps, should be attached to the fact that in the fragments of Prisons the name appears as Κάτζιροι in one place (Κατζίρων frg. 8 Müller, p. 83^a10 = Excerpta de legationibus 130²⁴ de Boor).Google Scholar

^{page 506 note 5} Full value should be given to the Armenian form; of the nations whose historical records we possess none was so close to the Khazars as the Armenians. Xazir suffices to exclude *Qasar (so, e.g. Pelliot, , T'oung Pao, xxxvii, 1944, 98, n. 1)Google Scholar from serious consideration as the original form.

^{page 506 note 6} τζ is used for c as often as for č, cf. Moravcsik, op. cit., ii, 44. Marquart's assumption that Syriac s may have been used to reproduce foreign č (Ērānšahr, 253, n. 5) is unacceptable.Google Scholar

^{page 506 note 7} In the heat of arguing against Howorth's theory, Marquart, who otherwise admitted no connexion between Acatziri and Khazars, went so far as to say that, on the contrary, die Ἀκάτζιροι entsprechen vielmehr den Schwarz-Chazaren (op. cit., 41, n. 2, at the end).

^{page 506 note 8} Notwithstanding the existence of such a word, both as appellative and as tribal name, 800 years and more after the time of Attila: in Houtsma, 's Glossar, p. 30²Google Scholar, and apparently in Rašid ad-Dīn, see Marquart, loc. cit. (I have not found the passage). As a tribal name it exists even nowadays, in Khūzistān, where a conglomeration of Turkish, Tāīk, and Lur tribes is called = Aya-eri, Fārsnāme-yi Nāṣirī, ii, 270Google Scholar; Mann, O., Mundarten der Lur-Stämme, p. xviiGoogle Scholar; the oilfield situated within the tribal area has recently often been mentioned in the newspapers, in the guise of Aghā Jarī (this is the spelling one finds also on maps). As I notice belatedly, the scholar to whom these pages are dedicated has already drawn attention to this name (The Tribes of Western Iran. Journal of the Anthropological Institute, IXXV, 1945, 77).Google Scholar

^{page 506 note 9} Jordanes' description of the Acatziri, gens … frugum ignara quae pecoribus et venationibus victitat, lets one envisage them as typical nomads of the steppe, certainly not as ‘woodmen’. Strangely enough, Marquart found support in it for his suggestion that they were ein echtfinnisches Fischer- und Jägervolk (op. cit., p. 40).Google Scholar

^{page 507 note 1} For reasons that will become clear presently, no attempt is made here to discuss the form of the Pahlavi name, 'KKTL'N.

^{page 507 note 2} It is curious to note that not one of the many scholars who have discussed the problem of the Acatziri since the publication of Paikuli (1924)Google Scholar ever so much as adverted to the form in the t Pahlavi inscription; the times of a Marquart are indeed past.

^{page 507 note 3} The glossary unfortunately refers only to the block numb ers, not to the lines of the printed text, in which the blocks are not marked.

^{page 508 note 1} Here, as always, I am using square brackets for [restored letters], round for (uncertain or damaged letters), a dot within square brackets for a missing letter, an unbracketed dot for a letter of which a small part is visible, but too little to determine its nature with reasonable certainty. Where Herzfeld's readings are quoted, his overdotted letters are replaced by letters in round brackets.

^{page 508 note 2} The observations made here are valid for the style of Pahlavi writing employed in the inscription of Paikuli, but not necessarily for the script of other inscriptions, even those from the same period.

^{page 508 note 3} Or ]dktl'(p).

a Herzfeld's restoration of the passage (Warhrānīkān Xwatāy) has not convinced me.

b cf. also Herzfeld, , AMI., vii, p. 61.Google Scholar The equation Pahlavi zwl'dčyn = Parthian zwrdtšn is rather doubtful.—For zwl'd- one might consider Armenian Jor(ay), the region around Bitlis, see Hübschmann, , Indogerm. Forsch., vol. xvi, p. 447.Google Scholar

^{page 509 note 1} I have been assured by friends that it has been set up in print; efforts to obtain a proof or a specimen copy have proved unavailing.

^{page 510 note 1} It is essential for the understanding of the lists to realize that Warāz is a family name, and that all persons preceding the Warāz are members of the royal house.

^{page 510 note 2} Ohrmizd the Warāz: line 7′ = A'2 ]zd wr'z; cf. line 16 = C 11. Herzfeld had a different explanation.

^{page 510 note 3} His absence may well be due to a gap in our material.

^{page 510 note 4} The name itself is old enough (see Justi s.v.), but the ‘Great Family’ of that name came to prominence only under the Sassanians; of. Nöldeke, , Tdbari, 139 sq.Google Scholar

^{page 510 note 5} Known as a name in use among the Mihrān, cf. Justi s.v.

^{page 510 note 6} ZY MN ldy, only in the Pahlavi version (line 34). Ray is often mentioned as the seat of this family.

^{page 510 note 7} The seat of the family is not known; perhaps Hindiyān/'Hindī ān (and Binduwān by popular etymology) at the head of the Persian Gulf.

^{page 511 note 1} Unless it is to be dismissed as mechanical transposition of a name whose true pronunciation was unknown to the translator (-rd- = -l-, but why -t for -d ?).

^{page 511 note 2} The ending -āy- may be due to an intervening Syriac form (*Bēθ Zrāyē), as it clearly was in the ease of mwks'dčyn.

^{page 511 note 3} -gd- has disappeared, though not without trace, also in MPers. mādiyān (Arm. matean) from mātagdān (originally mātak-dān).

^{page 511 note 4} cf. also Inscr. of Shapur Greek 67 Βαδον = Parth. (28) and Pahl. (35) bgdt. An interesting case of an inverse spelling is found in the inscr. of Kartīr at the Ka'be, line 7, 'L bgd'n g'sy 'ZLWNt against lines 3/4 and 5 'L 'RḤY'n g'sy 'ZLWN ‘to depart to the throne of the Gods’ = ‘to die’; here bgd'n = actual bayān or ba'ān ( < bagān). The forms have been misread and misunderstood.

^{page 511 note 5} Neubauer, A., La Géographie du Talmud, 327, 346 sq.Google Scholar

^{page 511 note 6} There are quite a few names and titles in the final list of Paikuli that require further consideration. Instead of 'štbwyn G'll (line 43′) I read 'štb(wy)n; in Altpers. Inschr., 77Google Scholar, Herzfeld quoted 'štbrnn, which appears to be intended as a new reading of this name. In G'1 (Une 43′), Herzfeld's yppt may have to be replaced by nppt[y] = nāfapati (arm. nahapet). In H 5 (line 45), ḥlw'nyk is probably ‘of Holwān’, but the reading of the title is uncertain; the name ends in ]wsydy, which calls to mind such names as Pūsai ([p]wsydy ? one would expect [p]wsdy).

^{page 512 note 1} That was recognized by Herzfeld, who at first copied mwšk (plate 198).

^{page 512 note 2} This is admittedly uncertain. The first letter is awkward: all the requisite traces for both S and M appear to be present; here, as often, it is difficult to distinguish cracks from incisions.

^{page 512 note 3} See Marquart, , Ērānšahr, 120 sqq.Google Scholar

^{page 512 note 4} The name of Albania in Parthian is spelt not 'rr'n (as I thought), but in fact 'rd'n, with -rd-as inverse spelling of -l-. It thus accords with the Pahlavi forms, 'ld'n and 'l'ny, which latter was liable to confusion with the name of the Alans. Cf. inscr. of Kartīr at the Ka'be, line 12, 'lmny štry W wlwč'n W 'ny W bl'sk'n 'D pl'č 'L 'l'n'n BB'; in the corresponding passage in Sar-Mašhad we have 'ld'n.—I cannot let this opportunity go by without drawing attention to the Pahlavi name of Peshawar, which I have now found in the inscr. of Sar-Maˇhad: it is Pškpwly; cf. BSOAS., xii, 53 sq.Google Scholar

^{page 512 note 5} The correct explanation of Siwnik', which is now confirmed, was found by de Lagarde. See Marquart, , Ērānšahr, 120, n. 3.Google Scholar

^{page 512 note 6} Annuaire de l'Institut de philologie et d'histoire Orientates et Slaves, xi (1951)Google Scholar, Mélanges Grégoire, 253 n.Google Scholar The author has made two further interesting points in that annotation. (1) That pryšḥwr ṬWR' = τὸ Πρεσσοναρ ὄρος n'est pas le Paryadres, mais bien l'Elburz. This is correct, and was already indicated by me in BSOAS., xii, 54, n. 2, perhaps in too obscure a fashion. ‘The whole Parišxwār mountain’ introduces and sums up the provinces of Media, Hyrcania, Margiane, and Areia, in the same way in which ‘the whole upper countries’ (a phrase that is heir to the ‘upper satrapies’ of Seleucid times) introduces and sums up the next following group (Kermān, Sekastān, etc.). Accordingly it would be more adequate to say that pryšḥwr equalled the Strabonian Παραχοάθρας, which it was believed stretched from Armenia and the Zagros to the Tajand. Incidentally, the inscription spells Πρεσεοναρ, not Πρεσσουαρ.—(2) That the town renamed (by Shapur) Prgwz-Šḥypwḥr = Πηρωσσαβουρ is the well-known Pērōz-Šābūr = Anbār on the Euphrates. This I have always assumed; one can only share the author's surprise at finding so often explained a term as Asūristān misunderstood even now. However, the final proof of the identity of the town (and district) with Anbār has not been mentioned by Maricq: it lies in its ancient name, which Pērōz-Šābūr was to replace, Parthian Mšyk (twice), Greek MHCIXICH and MICIXH; this corresponds to Βεσήχανα πόλις ἐν ἱερὸν Ἀτάργατι, Characenus, Isidorus, G.G.M., i, 249, 6.Google Scholar

^{page 513 note 1} viz. under this name; whether certain other names (Μαχλυηνή, Μάκρωνες etc.) appertain to this nation is a question that does not concern us here.

^{page 513 note 2} This is shown, e.g., by the case of Anazarbos. In BSOS., ix (1938), 840Google Scholar, having only a small photograph of the defective Pahlavi version at my disposal, I rejected Sprengling's reading 'ks ‥ sy and substituted 'n. l ‥ sy, which I then restored as 'n[z]l[bw]sy = Anazarbos, because in the region indicated by the context that is the only town whose name fits the legible letters, 'n. l ‥ sy. Now, having inspected the original inscription, I can confidently say that the stone actually shows 'nzl(pw)sy. It is thus certain that Anazarbos was meant; the fact that the other versions have something else makes no difference. The Parthian text offers 'nglpws: the corruption can only be understood on the assumption that the translator had a written copy before his eyes and misread 'nzlpwsy as 'nglpwsy; in the Pahlavi script of the 3rd century Z and G are almost indistinguishable, and frequently have been confused with each other. They were confused in this case also by the Greek translator, who, unable to account for 'nglp-, invented his Ἀγριππιάς (ΑΓΡΙΠΠΑΝ corrected into ΑΓΡΙΠΠΙΑΔΑ). Ensslin, W., Zu den Kriegen des Sassaniden Schapur i. (Sb. Bayer. Ak. Wiss., 1947, 5)Google Scholar, 1949, 110 sq., came close to appreciating the precedence of the Pahlavi text, at least in this case.—Every line of the Parthian version shows its dependence on the Pahlavi, e.g. in the constant confusion of R and L, the mistranscription of Pahlavi W by R (e.g. in krm'n'dy' for Pahl. kwm'n'dy = Comana), the occasional reproduction of inverse Pahlavi spellings (as of -d- in krm'n'dy' or spstyd'y'), etc.

^{page 514 note 1} That it was not made from the Parthian is evident by several passages where the Parthian is wrong but the Greek right. For example, Pahl. kwm'n'dy, Κομανα, but Parth. krm'n'dy' (see preceding annotation); ‘Ρεφάνεαν correctly, but Parth. rnypws (from *rpnyws); Ἀλεάνδριαν τὴν κατ᾽ Ἴσον correctly, but Parth. 'lyḥsndry' W ktyswsy; Ἐπιφάνιαν correctly, but Parth. 'ypyn'y (for *'pypny'y); Φλανιάδα correctly, but Parth. pr'ny's (confusion of Pahl. W and N; mistaking Pahl. L for R where it stood for actual L; the Pahlavi text had *pl'wy's); etc.

^{page 514 note 2} He was learned enough to replace Mnbwk by Ἱεράπολις, Knšr'y by Χαλκίς.

^{page 514 note 3} His acquaintance with Persia and Persian affairs was very good; in his renderings of Persian names he gave their actual pronunciation (no mean achievement).

^{page 515 note 1} The other repetitions were caused by similar misunderstandings; ‘Amastria’ is merely another blunder of the translator's.

^{page 515 note 2} In both oases the letter N is as clear as it can well be.

^{page 515 note 3} Some further titles were mentioned by Herzfeld, , A.M.I., vii, 60Google Scholar, from the unpublished blocks.

^{page 515 note 4} Paikuli, i, 244.Google Scholar

^{page 515 note 5} Herzfeld znkp[lyk] in the text, žandaf[rík] in the transcription and the glossary; later čndp[lyk], A.M.I., vii, 60.Google Scholar

^{page 515 note 6} Possibly a descendant of the Indo-Parthian king Gondophares. The name does not occur elsewhere with final -ak (-ag). A Gondophar is mentioned also in the inscr. of Shapur, line 32, gwndply ZT 'dwk'n = Parth. wyndprn 'bkn = Γυνδιφερ Αβγαν. Cf. the seal inscriptions Mordtmann, , ZDMG., xxix, p. 207Google Scholar, No. 17 gwndply ZY dpywr ZY ḥdywy BR, ibid., p. 210, No. 33, gwndply ZY dpywr ZY ḥdwny BR (cf. Horn apud Justi, 369Google Scholar; Justi, 248, No. 23, and additions opp. p. xxvi); probably both ḥdywy and ḥdwny are misreadings of 'dwky, so that the seals belonged to the man mentioned in the inscr. of Shapur.

^{page 516 note 1} See Sogdica, 34Google Scholar, and cf. Parthian sn'c- (Boyce, M., above, p. 441, n. 4).Google Scholar

^{page 516 note 2} cf. Müller, apudGoogle Scholar Salemann Man. Stud., 27 n.Google Scholar

^{page 516 note 3} Müller 'd; the lower part of the letter z is destroyed.

^{page 516 note 4} Müller wrongly phryzyn[d].

^{page 517 note 1} cf. Homilies, 32^6–7, Wehe euch, dass ihr gestorben und hinausgegangen seid …

^{page 517 note 2} hn'r- ‘to lift (up)’, which I have restored from ]'r-, is ordinarily construed with 'wl 'wl 'w (‘up to’); it would probably be unjustified to change 'wl 'z (‘up from’) here.

^{page 517 note 3} cf. Homilies, 32^5–6, Wer wird euch aufwecken, dass ihr das Ausmass der Freude sähet, in der wir heute sind.

^{page 517 note 4} Not ‘Great king of Armenia’ (as has often been translated); the title equals Arm. t'agawor Hayoc' Mecac'. Similarly, ‘king of Great Kušān’ is preferable to ‘Great king of K.’.

^{page 517 note 5} wḥyč-, past stem wḥ[s]ty (D 2/3, line 19); Pahlavi wḥyč-, Man. MPers. whyz-; Parth. past stem uryḥšt, Shapur, line 4 = ὁρμάομαι; not, therefore, ‘move upwards’; ‘wihēčakīk’ simply = ‘movable’.

^{page 518 note 1} I assume now that the imperfect still existed in the earlier forms of Western Middle Iranian; there are many cases of it in the inscriptions; note 'kylydy (with augment) = ‘was made’.

^{page 518 note 2} = viśve devāḥ.

^{page 518 note 3} pad … nām has often been misunderstood.

^{page 518 note 4} On the position of the town see Minorsky, , BSOAS., xi, 243 sqq.Google Scholar

^{page 518 note 5} Heraclius (who may well have been responsible for the destruction of the monument of Paikuli) covered the distance from Siarzūr to Ganzaca in ten days (not counting days of rest), see Minorsky, , loc. cit., 251.Google Scholar

^{page 518 note 6} Herzfeld failed to recognize that this word, Pahl. plky = Parth. plk, is an appellative descriptive of the type of monument of which Paikuli is a specimen. We cannot translate it properly until we know the precise shape which the monument of Paikuli once possessed. Herzfeld reconstructed it as almost a cube on a square base, adorned with battlements; the reconstruction may have to be econsidered in the light of the excavations carried out by him in 1923.—Plk(y) is no doubt the same word as Man. MPers. and Parth. pylg, on which see BSOAS., xi, 725 n.Google Scholar, where I hesitated between ‘altar’ and ‘steps, staircase’ (Pers. pille). Professor G. Morgenstierne kindly suggested to me that pylg might be a loanword from a Middle Indian form of Skt. pīṭha; one of its meanings, ‘pedestal of an idol,’ would fit excellently several of the passages in which pylg occurs. All these meanings, ‘altar’ or ‘pedestal’, ‘a monument of the type of Paikuli’, and ‘steps, staircase’, do not necessarily exclude each other; a ‘stepped altar’ (as e.g. at Tang-i Sarvak, see Asia Major, ii, 159)Google Scholar may have been meant, and the monument of Paikuli, even if it was not in the shape of a ziggurat, may have possessed a great staircase.

^{page 519 note 1} Or similarly.

^{page 519 note 2} Parth. ḥmy-'bdyn (so to be read) = Arm. hamaurēn; Pahlavi, h'mwdynGoogle Scholar, Pahl. Psalter h'mdwyn.

^{page 519 note 3} ptylky = Pers. paδīre; = istiqbāl.

^{page 519 note 4} TNH ‘here’ is certain; not PNH ‘hither’.

^{page 519 note 5} Which was no doubt followed by Heraclius in February, A.d. 628, on his march from Dastagerd to Siarzūr.

^{page 519 note 6} cf. above, p. 509.

^{page 520 note 1} The last letter—which does not affect our argument—is doubtful. The apparent traces favour Y, but may be a hole in the shape of Y. Final -ā (as -'y is to be pronounced) would be unexpected in this name, which in any case appears without it in Syriac; though foreign names are sometimes adorned with it in Pahlavi without good reason. If Herzfeld's restoration, [N], is accepted, we shall regard Nydktl'n as an adjective of reference (commonly formed with -ān from proper names); hence ‘the Nicatorian mansion (etc.)’.

^{page 520 note 2} So also in Persian words, chiefly at the end of words (cf. BSOAS., xii, 64 sq.)Google Scholar, but sometimes even in their interior, cf. e.g. Pahl. Psalter nšydm- = nišēm- according to K. Barr (beside nšdmy, Ka'be Kartīr, line 10).

^{page 520 note 3} The ending -ān, in place of the expected -ăn, made that explanation not too attractive; it was merely in the nature of a working hypothesis.

^{page 520 note 4} de Lagarde, , Gesammelte Abhandlungen (1866), 148 sq.Google Scholar

^{page 520 note 5} See Burrow, T., BSOS., vii, 779.Google Scholar

^{page 520 note 6} Levy, J., i, 41a.Google Scholar

^{page 520 note 7} Nöldeke, , Mand. Gramm., 136Google Scholar, with n. 1. Nöldeke rejected the derivation from Iranian, which has now been placed beyond doubt thanks to the appearance of the word also on the eastern fringe of the Iranian language territory.

^{page 520 note 8} So always in Kharoṣṭhi documents (see Burrow, loc. cit.), often in Armenian, sometimes in Syriac, and once in the Talmud (Levy's passage, , invites emendation to = 'Ukbarā and Awānā [differently Neubauer, Géographie du Talmud, 331 sq.; the MS. he used had 'ngr' w-'wwn']; these two towns lay on the Tigris, very close by each other).

^{page 520 note 9} Brockelmann, hospitium ubi guis noctem agitGoogle Scholar; Hoffmann, (occasionally) Post-stationGoogle Scholar; Nöldeke, (for Mand.) QuartierGoogle Scholar; Levy, Station, Nachtherberge.Google Scholar

^{page 521 note 1} Especially the otherwise troublesome text in which hāyān refers to the grave; a word connoting temporariness well accords with Manichæan ideas about it.

^{page 521 note 2} See the passages collected in the Thesaurus, col. 491.

^{page 521 note 3} See e.g. Sachau, E., Die Chronik von Arbela (Abh. Preuss. Ak. Wiss., 1915), p. 61.Google Scholar

^{page 521 note 4} Sachau, , Die Chronik von Arbela, p. 21Google Scholar, identified Bēθ Nīqātōr with Qaṭrabbul. This is inadmissible. He relied on G. Hoffmann's etymology of Qaṭrabbul: * Νικατορόπολις (Märtyrerakten, 41Google Scholar, n. 343). The acceptance of Hoffmann's etymology (which in itself is attractive enough) almost suffices to exclude Sachau's identification; for if the Greek name persisted up to Muslim times, one would expect to find it being used in earlier centuries, and not a Syriac translation of it. The principal objection, however, is that Bēθ Nīqātōr lay in Bēθ Garmai, while Qaṭrabbul (the district immediately to the NW of Baghdad, on the western bank of the Tigris) is solidly placed in the heart of Bēθ Armāyē.—Sachau did not refer to this erroneous identification in his later work Zur Ausbreitung des Christentums in Asien (Abh. Preuss. Ak. Wiss., 1919)Google Scholar, in which he listed all known bishoprics of Bēθ Armāyē, pp. 26–38.

^{page 521 note 5} Die Aufnahme des sasanidischen Denkmals von Paikūli (Abh. Preuss. Ak. Wiss., 1914), p. 6.Google Scholar The name is marked on the map attached to that paper.

^{page 521 note 6} The consonantism is exceptionally well preserved; the ending, -a(h), may hardly be used to defend the reading N[y]dktl'[y], cf. above, p. 520, n. 1. Herzfeld justly recognized that Binkudrah represented an old name compounded with Bēθ; however, the original form he posed (‘Bā-Nuhadrā’) is unsupported and leaves -k- unaccounted for.—The form developed probably in this way: Bēθ Nīqāṭōr—Bēnikátur—Bēnkátur—Bēnkútrə—Bīnkúdra.

^{page 522 note 1} cf. BSOAS., x, 941 sq.Google Scholar

^{page 522 note 2} [Further inquiry, in which Mr. C. J. Edmonds kindly lent me his help, has shown that the name of Binkudrah now chiefly appertains to the village that forms the centre of the district. It lies on the left bank of the Diyālah (Sirwān), at a distance of about half a mile from the river; measured on the excellent map in Herzfeld's Paikuli its position is 34° 31′ 30″ N., 45° 14′ E., its air distance from Paikuli 43½ miles. Jones, James Felix, Memoirs. Selections from the, Records of the Bombay Government, No. 43, 1857, mapopp. p. 136Google Scholar, showed the plain of B. as running several miles along the left bank of the Sirwān, almost down to the confluence with the Holwān. C. J. Rich, who visited the village, spelt its name Binkudreh, Narrative of a Residence in Koordistan, ii, 271 sqq., 378–380.Google Scholar However, Mr. Edmonds points out that according to his recollection the name is pronounced with -q-, and this agrees with the spelling adopted by H. Rawlinson, Bíṅ-ḳudrah (i.e. Bīηqudra^h), Notes on a March from Zoháb, JRGS., ix, 1839, 29Google Scholar, and, incidentally, with the original form posited here. There are many Sassanian ruins in the neighbourhood of the village; those now named Kattar Tepesi (Rich, , ii, 274)Google Scholar may well conceal the old Station of Nicator.]