Hudud al-'Alam, The Regions of the World
V. Minorsky

§ 50. The Khazar.

Marquart, Streifzüge, Index; Barthold, Khazar in EI; H. v. Kutschera, Die Chasaren, Wien 1910 (unimportant); J. N. Simchowitsch, Studien z. d. Berichten arab. Historiker über d. Chazaren, Berliner Dissertation 1920, still unpublished; the author's résumé in Jahrbuch d. Dissert, der Philol. Fakultät . . . zu Berlin, 1919-20, pp. 248-52, is reviewed by M. Palló in Ungar. Jahrbücher, ii, 1922, pp. 157-60 (with a list of Hungarian literature on the subject): Simchowitsch studies the earlier history of the Khazars down to Hārūn al-Rashīd's times; M. Kmoskó, Die Quellen Istahri's in seinem Berichte über die Chasaren, Kőrösi Csoma-Archivum, 1/2, 1921, pp. 141-8; M. Kmoskó, Araber und Chasaren, ibid., 1/4, 1924, pp. 280-92 and 1/5, 1925, pp. 356-68. The Jewish sources on the Khazars have been recently republished by P. K. Kokovtsov, Yevreysko-khasarskaya perepiska v X veke, Leningrad 1932 (exhaustive bibliography and very detailed commentary). Some Byzantine sources are quoted in Dietrich, o.c., Index; see also Constantine Porphyr., De admin. imperio, chap. 10, 12, 42 (scarce details).

The principal Muslim source on the Khazars is Ibn Faḍlān  (in Yāqūt, ii, 436-40), many of whose data are found also in Iṣṭ., 220-5, though each of the two sources has a good many independent details. Since Frähn it has been admitted that Ibn Faḍlān  (who travelled in 309-10/921-2) was the

§ 50   The Khazar   451

source of Iṣṭakhrī, but lately Kmoskó has advanced a new theory of their common dependence on some previous report drawn up towards A.D. 800.

Another source is that utilized by I.R., 119-20, Bakrī, Gardīzī, and 'Aufī.

Our author's very condensed report reflects both groups of sources and shows a knowledge of I.Kh. For the items of the Ātil town, the seven judges (governors?) communicating with the king, and the maritime customs Iṣṭ., 220-5, is undoubtedly responsible, but our author cuts down even such characteristic features as the existence of a dual political system under which the supreme chief only appointed the head of the executive power who was the real ruler. The system is mentioned in all the sources :
Const. Porphyr. cap. 42, (cf. Marquart, o.c., 27)
Ibn Rusta Khazar Khāqān Ayshā
Mas'ūdī, Murūj, ii, 12 Khāqān Malik
Iṣṭakhrī Malik Khazar [1] Khāqān Khazar, [1] or Bek
Ibn Ḥauqal Khāqān Khazar Malik Khazar
Gardīzī Khazar Khāqān Abshād

In the Ḥ.-'Ā. the two persons are run into one and the king (pādshāh) is called Ṭarkhān Khāqān, from the children of Ansā (cf. Ayshā, Abshād). The latter name was borrowed from the source common also to I.R. and Gardīzī, whereas the addition to the title khāqān of a further title ṭarkhān finds an explanation in the story of the interpreter Sallām's journey to the wall of Gog and Magog, I.Kh., 163, where Ṭarkhān malik al-Khazar is mentioned, though at another place I.Kh., 41, says that ṭarkhān was the title of lesser Turkish kings. [A Khwārazmian mercenary Rās-Ṭarkhān commanded the Khazar forces which invaded Transcaucasia in 147/764. Marquart, Ungar. Jahrbücher, 1924, p. 271, explains by this person's name that of the later town Astrakhan. Cf. Ṭabarī, iii, 328, s-tarkhān?.]

Our author equally omits the important statement regarding the outward appearance of the Khazars: being of two distinct types (one very dark, the other fair-skinned and handsome) they did not resemble the Turks. Their language was also different from Turkish, but resembled that of the (Volga) Bulghārs, Iṣṭ., 225. According to this description the Khazar language of which no texts have come down to us, belonged to the aberrant branch of Turkish languages of which the only living representative is now the Chuvash language.

Prima facie our enumeration of the Khazar towns presents great difficulties. In fact these towns were only four, of which two were divided by the Volga near its estuary, and the other two lay in the Caucasian region (Balanjar and Samandar). Our compiler mentions the two Volga towns under five different names and thus his total rises to seven, to say nothing of the five additional names wrongly quoted under Khazar.

1. The places of the rulers are wrong.

452   Commentary   § 50

The complication with the Volga towns will be best presented in the following table:
Iṣṭ. Western Ātil  Eastern Ātil
 Ḥ.-'Ā. [Western] Ātil, , and   [Eastern] Ātil and 

The three traditions, namely: A (Iṣṭ. < I. Faḍlān ), B (I.Kh.), C (I.R. and Gardīzī), are all side by side incorporated in Ḥ.-'Ā. The order of enumeration fully confirms this conclusion.

1. and 2. are evidently borrowed from Iṣṭ., 220-3, who in the Khazar land knows only these two towns of which (read Ātil > Etil) was a double town for it was divided by the Ātil river (§ 6, 43.) into a western and an eastern part, the former being the residence of the king and his army, and the latter the commercial centre. The two towns lay probably near the estuary of the Volga. On their different names see the table above.

Mas'ūdī, Murūj, ii, 7, reckons from Darband to Samandar 8 days and thence toĀtil (so instead of Āmul) 7 days. According to Iṣṭ., 219, 227, the respective distances are 4 and 7 days, the distance between Samandar and the Sarīr boundary being only 2 farsakhs. These data indicate for Samandar a place somewhere between Kizlar (on the Terek) and Petrovsk (now Makhach-qal'a) on the Caspian sea. It is usually (Dorn, Marquart) accepted that Samandar [1] corresponds to Tarqu/Tarkhu, situated at a few Kms. to the south-west of Petrovsk and in favour of this opinion could be quoted the short distance between Samandar and the Sarīr, and our author's indication that Samandar lay near the sea-coast. Iṣṭ., 222, mentions extensive gardens and vineyards between Samandar, Darband, and the Sarīr, which detail is also not contradicted by the situation of Tarqu. Finally, in the letter of the Khazar king (though in the more extensive and still suspect version B, cf. Kokovtsov, p. 100) Samandar is placed "at the end of  T.dlū'' which may easily stand for  Tarkhu!

3. In this paragraph the names borrowed from different authorities are jumbled together into a long list. We shall treat them in three groups.

3a. The first three names come exactly as I.Kh., 12412, enumerates them: wa mudun al-Khazar: *Khamlīkh wa Balanjar wal-Bayḍā. Of these Balanjar lay certainly in the Caucasian region. During his campaign of 119/737 Marwān penetrated into Khazaria, as it seems, through the Alān gate, i.e. the Darial pass in the Central Caucasus, and then (I. Athīr, v, 160) marched eastwards to Balanjar, Samandar, and al-Bayḍā. Consequently Balanjar is to be sought between the Darial and Samandar. The only other geographical detail referring to it is the existence of a river called nahr al-Balanjar, Balādhurī, 204, Ya'qūbī, Historiae, 194 (in the account of

1. The reading of the name may be *Sumundur, *Samundur, &c.

§ 50   The Khazar   453

Salmān's campaign). The Khazar king's letter mentions a river V.r.shan situated at 20 farsakhs from the capital, Kokovtsov, pp. 86 and 102. [1] Marquart, o.c., 16-19, compares this name with Balanjar [2] and tentatively identifies the Balanjar river with the Qoy-su "the Sheep river" (Abul-Fidā, 204: nahr al-aghnām flowing through the Sarīr). It is true that according to the Khazar letter the river ought to be placed much more to the north (Kokovtsov: Kuma river?) but then it would be difficult to understand how Marwān could march to Samandar via Balanjar. So besides the Qoy-su (Sulaq), only some of the right affluents of the Terek, or the Khasav-yurt river could eventually be taken into consideration with regard to the still doubtful situation of Balanjar. As Samandar lay by the sea, al-Bayḍā, whither Marwān [3] marched from Samandar, could lie either to the south, or, more probably, to the north of Tarqu. As I. Athīr, v, 160, definitely says that al-Bayḍā was the Khāqān's residence it must be identical with one of the two Volga towns, and more particularly with that which I.R., 139, calls . The first element of the name *Sarīgh-sh.n is evidently Turkish sargh "yellow", a colour of which the Arabic al-Bayḍā"white" might be an approximate rendering, perhaps even more suitable for the original Khazar meaning. [4] Marquart, o.c., i, arbitrarily restores the second element  as shar < shahr, but I am strongly inclined to think that the name *Sargh-shin is the original form of the still enigmatic Saqsin, as the geographers of Mongol times call a town situated by a mighty river and usually quoted along with the Volga Bulghār, cf. Barthold, Sasīn in EI. [5] To sum up: Bayḍā may be only an Arabic name for the first of the two Ātil towns already mentioned under I. As according to Iṣṭ. the Khāqān lived in the western town, al-Bayḍā taken by Marwān, must be the latter. There is no record of the Arabs having crossed the Volga and in principle it would have been a most difficult feat.

3b. The following two names are borrowed from the source common

1. It is curious that in the account of Maslama's campaign Ya'qūbī, Historiae, ii, 381, says that he was met by the khāqān of the Khazars in , which here is an entirely different place from Varthān in Ādharbayjān and evidently refers to northern Daghestan. [It is very probable, however, that the name refers here to Barshliya, v.s., p. 449, note i, in Armenian Varač'an, see Moses Kałankatvats'i, book ii, ch. xxix, Russian transl. by Patkanov, SPb., 1861, p 192.]

2. Marquart, ibid., 166, identifies Balanjar with Varač'an or Varajan by which name the Armenians call the capital of the Caucasian "Huns", but  he withdraws this suggestion, ibid., 492.

3. He was coming from the west.

4. The Khazar fortress on the lower Don  = Russian Bělaveža "White tent" is called in the Khazar king's letter (version B) Sharkīl, cf. Chuvash shura "white" and kil "house", as suggested by Poppe in Kokovtsov, o.c., 105. [In Chuvash u < old a.]

5. The geographical identity of Saqsīn with the Itil town was recognized by Westberg, o.c., 1908 (March), p. 40; I think that even phonetically Saqsīn < Sarigh-shin (or –sin?). [As a parallel cf. tne name of Tsaritsn "Queen's town" (now Stalingrad), important centre situated on the Volga above Astrakhan, which is said to be a popular Russian etymology for the original "Sarchin (?) supposed to mean "yellow island" (?). *Sar-sn would mean "yellow tomb".]

454   Commentary   § 50

also to I.R., Bakrī, and Gardīzī (see the table above). The unusual form of the first name  may have been influenced by the two towns Shāvghar in Transoxiana, cf. Barthold, Turkestan, 174. The second name, as spelt by our author, would be *Khutlugh "happy" but to judge by I.R.'s variants it looks like a compound with the Turkish word -balgh "town". Very probably another form of the same name is I.Kh.'s  (so instead of  chosen by de Goeje) Khamlikh, possibly with a contraction from < Khammalikh < Kham-balikh < Kham-baligh. The first element still offers a difficulty. Marquart, Komanen, 71, rightly criticized M. Hartmann's restoration *Khan-balgh, but his own reading *Qapgh-balgh is still more improbable. That this town stood on the eastern bank of the Volga may be indirectly concluded from the fact that I.Kh., 124, quotes it as the terminus of the road from Jurjān, i.e. along the eastern coast of the Caspian.

3c. The last three names are found in I.Kh., 124, who following on the enumeration of the three Khazar towns says: "and outside al-Bāb (Darband) are a. the Malik of Suwar, b. the Malik of al-Lakz, c. the Malik of al-Lān, d. the Malik of Fīlān, e. the Malik of al-M.sq.t, f. the Master of the Sarīr, and g. the town of Samandar". In this list a. corresponds to our Swr; b. to our Lkn, c. to our M.s.ṭ; g. was already mentioned under 2., and c. and f. are treated in separate chapters (§§ 48 and 49). Only d. Fīlan has been left out of consideration. [1]

Neither of the three names *Lakz, Swr, and M.sq.ṭ could be quoted in the tenth century under the heading Khazar. Even Darband-i Khazarān (§ 36, 40.) is a purely conventional historical term pointing to the fact that Darband (which from circa A.D. 800 remained in the hands of the Muslims) was a "frontier post" (thaghr) directed against the Khazars and their successors. In our author's times Darband and consequently the lands lying to the south of it belonged to the Shirvān-shāh, v.s., notes to § 36, 36. and Iṣṭ., 219. A remote reason for the inclusion of *Lakz, *Suwar, and

1. Perhaps because our author, like Mas'ūdī, Murūj, ii, 42, took Fīlān-shāh for the hereditary title of the Sarīr kings. This, however, is inexact, for Balādhurī, 196, names separately ṣāḥib al-Sarīr and malik-Fīlān. Nothing practically is known of this prince and his people. In Yāqūt's very valuable passage on the peoples of Daghestan, i, 438 (cf. BGA, i, 184) immediately after Ṭabarsarān (on the Rubas river) is mentioned umma ilā janbihim tu'raf bi-Fīlān which suggests that the Fīlān lived quite close to the Ṭabarsarān. After the Fīlān come the Lakz, al-Līrān, and Sharvān (sic). Balādhurī, 194, speaks of  "the wall of the L.b.n" which the Sasanian Qubādh built between Shirvān and Bāb al-Lān (Darial), cf. I.Kh., 123, bāb L.bān-shāh. This L.b.n probably corresponds to Lip'in-k' of the Armenian authors and Lupenii of Pliny, n. h., vi, 29, and it is not impossible to connect Lip'in, &c. with Fīl- by admitting a metathesis *Līf/Fīl. It is true that Balādhurī, 196, specially mentions Malik Fīlān but the different sources may account for the difference L.b.n/Fīlān. [In the eastern part of Shakkī near the sources of the Turiyan-chay several places are found with such names as Filifli, perhaps < Fīl-i Fīlān (cf. the royal title of Gīl-i Gīlān). This is only a hint to the future investigators on the spot.]

§ 50   The Khazar   455

*Masqut in the Khazar chapter may be the fact that Marwān is said to have brought from his famous expedition (of 119/737) a number of Khazars whom he settled between the Samūr river and Shābarān in the lower parts of the Lakz lands (fi sahl arḍ al-Lakz), see Balādhurī, 207. On the middle course of the Samūr there is still a village Khazri < *Khazarī.

stands undoubtedly for  Lakz. As mentioned above (§ 36, 36.) the Lakz, or a part of them, seem to be identical with the *Khursān (Balādhurī, 196: ). According to Mas'ūdī, Murūj, ii, 6, the Shirvān-shāh Muḥammad b. Yazīd annexed the possessions of  (read:  and this agrees with the threefold title of the Shirvān-shāh in our source (v.s.). Mas'ūdī, ii, 5, even adds that the Lakz kingdom (mamlaka) was the bulwark (mu'awwal) of the Shirvān kingdom. Balādhurī, 209, mentions a fortress of the Shirvān-shāh named . The original extent of the Lakz territory is uncertain but they appear as the immediate neighbours of the Layzān (v.s.). According to Abul-Fidā, trans, ii 2, p. 299, the Samūr river flowed across the Lakz territory and Balādhurī's passage, 207, indicates that in the region between the Samūr and Belbela rivers the Lakz originally occupied even the plains. The name Lak-z as shown by Marquart, ZDMG, 49, p. 666, is formed with the Iranian suffix of origin -z and the stem of the name is *Lak. This is now the appellation of the Daghestanian Qazi-Qumuqs (Arab. ), living on the eastern branch of the Qoy-su. The linguistic evidence shows that the Lak once occupied a much larger area (Prince N. S. Trubetskoy's lecture at the School of Oriental Studies, 21.iii.1934), but the connexion of the Lakz with the present-day Lak is still uncertain. By metathesis Lakz became Lazg, which form was further used by Persians with the addition of the usual suffix of origin Lazg-ī (in Russian Lezg-in, with the Russian "singulative" suffix -in). This later Perso-Turkish term came to denote indiscriminately all the Daghestanian mountaineers, but more especially those of the southern part of Daghestan, cf. Barthold, Dāghestān in EI. See Map xi.

On  vocalized Suwar in I.Kh., 124, nothing is known and de Goeje's annotation: " vulgo " (cf. § 51) remains on his responsibility. As a guess one might connect the name Swr (*Sawir ?) with that of the people Sabir who were defeated by the Avars in A.D. 461; a part of them was settled by the Romans south of the Kur. Mas'ūdī, Tanbīh, 83, pretends that the "Turkish" name of the Khazar was . [1] [V.s., p. 401.]  (cf. also § 49, 3.) vocalized in I.Kh. al-Masqat most probably must be read *Masqut. [2] Marquart, Kulturanalekten in Ungar. Jahrbücher, ix/1, 1929, p. 78, quotes as its parallels Armenian Mazk'ut'k', Maskut', &c., and ingeniously takes the present-day Mushkur for a later avatar of the old

1. The Swār and Swwr whom the Khazar king mentions in the list of his neighbours, Kokovtsov, 98, do not seem to be connected with Daghestān.

2. Cf. a mountain south of Ganja called Maskhut on Russian maps.

456   Commentary   § 50

name (the passage t > d > r is characteristic for the Iranian Tātī dialects, v.s., note to § 36, 36.). The Mushkur district is situated south of the Samūr river, between the southern branch of the latter, Yalama, and the river Belbela, see Butkov, Nov. istor. Kavkaza, i, 94, cf. Abul-Fidā, transl. ii/2, p. 229. In Balādhurī's time (p. 196) Masqut had already ceased to exist as a kingdom.

4. These names [omitted in Gardīzī] occur in the following writers (cf. Marquart, Streifzüge, 173, and v.s., p. 445):
I.R., 139.  Ṭūlās  Lūgh.r
'Aufī  Ṭūlās  Kūgh.r
Shukrullāh  Ṭūlās   K.rgh.ra

The earliest and clearest text on these peoples is found in I.R., 139, who says that "on one side" the Khazar lands adjoin "a huge mountain at the farthest end of which (fī aqṣāhu) live the Ṭūlās and Lūgh.r and which stretches to the land of Tiftīs". To Marquart, Streifzüge, 31, 164-76, is due the ingenious explanation of the two names. He interpretsṬūlās as *Ṭūl-Ās in which Ās represents the well-known alternative name of the Alāns: old Russian Yas; Georgian Ous-i and, with the suffix denoting the country Ous-et'i > modern Russian Oset-in. In Muslim literature Ās replaces Alān (§48) in Mongol times, cf. Juwaynī, GMS, i, 214, 222: ; Ibn Baṭṭūta, ii, 448: . Bakrī's  could then be easily improved into , and, as a compound, Ṭūl-Ās would be paralleled by the name of the principal clan of the Alān as given by I.R., 148: D.ḥs-Ās. [1]

The second name , Bakrī's  is restored by Marquart as Aughaz, i.e. Abkhaz, Arabic , Greek , [Contarini, ed. Hakluyt Society, p. 144: Avogasia]. This people occupies, on the Black Sea coast, the south-westernmost slopes of the Caucasian range, which quite well suits I.R.'s mention of the "farthest end of the mountain" and Bakrī's, p. 45, clear indication that the people in question lived "below that mountain on the sea-coast". In the tenth century all the western Georgians (of the Rion basin) were usually called Abkhaz after the dynasty which ruled them. Mas'ūdī, ii, 65, seems to distinguish between the  (Eastern Georgians) and the Abkhāz, whereas our author quotes Eastern Georgian lands under Armenia but extends (§ 3, 6.) the name Gurz (Western Georgians) even to the Black Sea. Therefore, following our text  would refer only to the real Abkhaz. [2]

1. It is indeed possible that the name of one of the clans was substituted to that of the Alān in general. Abul-Fidā, p. 203, who wrote at the epoch when the terminology was changing, says that the Ās are a Turkish (?) people living near the Alān, being of the same origin as the latter (!) and professing the same religion. [V.i., p. 481, 3.]

2. 'Aufī and Shukrullāh consider the Ṭūlās and Kūgh.r (K.rgh.ra) as "two kinds of Turks" [cf. also Abul Fidā, quoted above in note 1]. The term Turk is here applied in a loose sense: not only the Magyars but the Rūs as well were considered Turks by Muslim writers.  could even have been mistaken for  (v.s., notes to § 14).

§ 50   The Khazar   457

So far, so good, but Marquart in his Streifzüge, 173, 495, overreached the goal by further identifying the Ṭūlās and *Aughaz respectively with the N.nd.r and M.rdāt mentioned in Gardīzī. This part of his theory is undoubtedly wrong and Marquart himself later hinted at the proper explanation of the term N.nd.r (see §§ 46 and 53). [1]

Summing up the situation, we should:

(1.) distinguish between the two pairs of peoples (see notes to §§ 42 and 53);
(2.) locate the T
̣ūlās and Lūgh.r in the western Caucasus;
(3.) provisionally maintain the first part of Marquart's hypothesis: T
̣ūlās = some tribe of Ās, and Lūgh.r = Abkhaz.

Our additional remarks will be as follows:

(4.) I.R., 139, only says that at the farthest end of the mountain near which lay the Khazar land, lived the Ṭūlās and Lūgh.r, whereas our author makes of the latter "two districts of the Khazar". It is true that in the seventh century the Khazars penetrated down to Tiflis through the central Caucasian pass but the western Caucasus was hardly ever under Khazar sway. Our author's mistake may be somehow connected with the frequent confusion of  Khazar with  Jurz "Georgians". I.R.'s detail on the mountains "stretching to the land (bilād) of Tiflis" is perhaps a hint of some mention of the Jurz in the original source.

(5.) The first element of Ṭūl-Ās is confronted by Marquart, ibid., 172, with the name of the Alān prince Dula, known from Magyar sources. [2] It is much simpler, however, to identify it with the Osset Tual-tä, i.e. the Tual, or Southern Ossets, in Georgian Dvali, who on the map annexed to Brosset's edition of Prince Vakhusht's Geography, St. Petersburg 1842, are shown (1) north of the Caucasian range on the upper course of the Ardon which is the left tributary of the Terek, and (2) in the upper valley of the Great Liakhvi which, south of the range, flows into the Kur. The Tuals living in the heart of the Caucasus would very well suit the requirement of our case including the remark on the warlike character of the people. The name  would then be read *Ṭuwal-Ās. [3] See Additional Note to § 48.

(6.) As regards Bakrī's report on the  and , here is a com-

1. However, it remains possible that a similar confusion of the two pairs of names had already occurred in Muslim authors themselves and there may lie the explanation of some puzzling characteristics of the Mirvāt in our author and Gardīzī (v.s., § 46).

2. Even in Ungar. Jahrb., ix/1, 1929, p. 86, Marquart repeats: "ich bin . . . nach wie vor der Ansicht, dass jener Name [Dula] mil dem des Stammes Ṭūl-ās zusammenhangt."

3. There exists a Georgian family Tulasdze but I am unable to ascertain their origins. Brosset, Histoire de la Géorgie, ii/2, p. 151, mentions a locality T'ula which does not seem to be connected with the Ossets. In any case, the attested Georgian form of the name Tual is Dval-i (from which the family name of Dvalishvili is derived). [The imaginary name  which Niẓāmī in his Iskandar-nāma gives to the Abkhāz king may reproduce Dvali.]

458   Commentary   § 50

parative table of the relevant passages in I.R.'s and Bakrī's chapters on the Khazars:
Description of the road from the Pechenegs to the Khazars. ditto
"The Khazar country is a vast land one of whose sides adjoins a huge mountain" ditto
"and this is the mountain at the farthest end of which live the Ṭūlās and Lūgh.r" left out, v.i.
"and this mountain stretches to the lands of Tiflīs" "then [you go tasīru] to the lands of Tiflīs, the latter (Tiflīs) being the beginning of the frontier of Armenia"

Instead of the sentence on the two peoples left out in the chapter on the Khazars, Bakrī, in the chapter on the Majgharī, says:

"a frontier of their country adjoins the Rūm country whereas another frontier of theirs, on the steppe side, adjoins a mountain inhabited by the people called  who possess horses, cattle, and fields; under that mountain on the sea-coast lives the people called Aughūna; they are Christians and are conterminous with the Islamic lands belonging to the country of Tiflīs which is the beginning of the frontier of Armenia. [1] This mountain continues down to Bāb al-abwāb and joins the Khazar country."
Bakrī's information on the one hand contains some independent traits and on the other reflects his own arrangement of the principal source. The description in I.R. moves from east to west (the Pechenegs [in their Ural seats], the Khazars, the mountain stretching to Tiflīs, the peoples at its farther end). Bakrī proceeds in an opposite direction (the Majgharī [in some of their seats on the Black Sea coast], the [Caucasian] mountains, the *Ās and Aughūna, then Tiflīs, Bāb al-abwāb, and the Khazar). The form of Bakrī's names is peculiar. If  for  is due to the general use of forms in -iya (Bajānākiya, Majgharīya),  presents more difficulty. Marquart, o.c., 167, restored it as Ās which is a later appellation of the Alān (§ 48), the latter name not appearing in the known fragments of Bakrī. Although the forms *D.khs-Ās and Ṭwl-Ās occur already in I. Rusta as the names of special tribes, the pure form Ās as referring to the Alāns in general appears only in Mongol times. Moreover Bakrī's description of the  lacks the characteristic features of the Alān. Even the combination of  with the *Aughaziya suggests that Bakrī has in view the particular clan corresponding toṬwlās. [2] The disclosure of the identity of Bakrī's  (i.e. whether it stands for Alān or Ṭwlās) is important

1. The passage in italics is a repetition of what had been said under the Khazar.

2. As the separation from the name Ṭwlās of the basic element Ās is not at all an obvious matter we are perhaps entitled to suppose that Bakrī has been inspired by some later source. Under Pecheneg he quotes the evidence of Muslim captives in Constantinople for the events after A.D. 1009.

§ 50 The Khazar 459

for in the former case Bakrī possessed some more detailed knowledge of the early Magyar seats near the Caucasus than is found in the more complete text of the earlier I. Rusta. In the second eventuality the vicinity of the Magyars to the  must be merely a guess on Bakrī's part.

This author's information on the Magyars [who over a century before had settled beyond the Carpathians] is certainly traditional and derived from the same source as that utilized by I.R., Gardīzī, and 'Aufī. This group of authors definitely says that the Magyar country reaches down to the Rūm sea (bahr, daryā) instead of which Bakrī mentions "Rūm country" (bilād al-Rūm), thus considerably modifying the situation. This procedure does not give us much confidence as to the eastern frontier of the Magyars with regard to which Bakrī quotes a detail not found in I.R., Ḥ.-'Ā., Gardīzī, or 'Aufī. We must remember that according to I.R., 143 [1], the Khazars "some time ago" entrenched themselves against the Magyars and other peoples (yuqālu anna-'l-Khazar fīmā taqaddama kānat qad khandaqat 'alā nafsi-hā ittiqā'a 'l-Majghariya wa ghayrihim min al-umam al-mutākhima li-bilādihim). Assuming then that the Magyars were the neighbours of the Khazars, Bakrī could logically infer that, more precisely, they bordered on the peoples who were said to live at the westernmost limit of the mountain mentioned on the confines of the Khazars. Such then may be an explanation of Bakrī's mysterious passage.

This hypothesis may be objected to on the ground that according to our § 47 the Khazarian Pechenegs neighboured in the south on the Alān and a similar view is suggested by Mas'ūdi's embroiled passage on the W.l.nd.riya (v.i. § 53). Both indications are supported by the well-known passage in Const. Porph., ed. Bonn, p. 166, according to which the Pechenegs lived at 6 days' distance from the Alāns. As the Pechenegs ousted the Magyars from their Lebedia seats it could have been inferred that the latter as well had bordered on the Alāns. However, the fact is that Muslim authors knew nothing of what we ourselves, thanks to Const. Porph., know about the events, cf. Iṣṭ., 10, and our § 47. Therefore a retrospective conclusion is highly improbable for a Muslim author. Only the arrival of the Pechenegs seems to have cleared up for Muslims the situation near the Azov sea but for Bakrī the Pechenegs were still in the north and, living a century later than I. Rusta and depending on the same source as I.R., he could hardly have improved on the latter's data. Therefore I am inclined to maintain the view that (a) Bakrī's  refers not to the Alāns as a whole but to the little-known tribe of Ṭwlās, and (b) that the idea that the Magyars and  were neighbours is a result of Bakrī's personal surmise. As a matter of fact even at the time when the Magyars lived near the Caucasus the Ṭwlās mountaineers must have been separated from them by the other Alān tribes living in the plains. [Cf. p. 458, 1. 18? .]

(7.) During his expedition to the north-eastern Caucasus Timur operated against the , see Ẓafar-nāma, i, 766, 788. Further Kūlā and

1. Marquart, Streifzüge, 28, connects this report with the construction of the Khazar fortress of Sarkel (on the Don) after A.D. 833.

460   Commentary   §§ 50-1

 Ṭā'ūs appear as the names of two local chiefs, [1] though they may represent hereditary titles. The fortress of Ṭā'ūs which was particularly strong, lay at the third range of mountains counting from the north, probably near the sources of the Terek and the Kuban for, immediately after, Timur marched to Balqān (Balqar ? at the sources of the Terek). Both the name , which could be easily restored as , and the geographical details make it possible to see in our passage an echo of the tenth-century terminology. [Ḥājjī-Khalīfā, p. 402, repeats the statement of the Ẓ.-nāma.]

1. Rashīd al-dīn, ed. Blochet, p. 45, mentions an Ās chief executed by Ögedey:  (note the final ).

[Previous] [Next]
[Back to Index]