418 Commentary §§41-2
§ 42. Byzantine Empire
I.Kh., 100-13, Qudāma,
vii, 323, and Historiae, ii. 171-8; I.R., 119-30; I. Faqīh,
136-56 (see also under Yāqūt);
8, 45 (Kharshana), 68-71; I.Ḥ.,
128-37 (an interesting and independent report based chiefly on Abul-Ḥasan
Muḥammad b. 'Abd al-Wahhāb
Maq., 147-8, 150; Yāqūt,
ii, 861-6, who quotes a long description of the provinces which he attributes
to I. Faqīh, though
it is not found in BGA, v; Idrīsī,
ii, 209-304 sq. Constantine Porphyrogenitus, De thematibus, ed.
Becker, Bonn, 1840 (cf. idem, De administrando imperio: additional
remarks on Charsianon,
, &c.)  ; W. R. Ramsay, Asia Minor;
Tomaschek, Zur hist. Topographie von Kleinasien, in Sitz. WAW,
1891, cxxiv, pp. 1-106; Gelzer, Die Genesis d. byzant. Themenver-
1. I have also used the commented Russian translation by G. Laskin, Moscow 1899.
§ 42 Byzantine Empire 419
fassung in Abh. Sächs. Gesell. d. Wiss., 1899, xviii, No. 5, pp. 1-134 (a Map); Brooks, Arabic Lists of the Bysantine Themes, in Jour. of Hellenic Studies, xxi, 1901, pp. 67-77 (I.Kh., Qudāma, Mas'ūdī, and I. Faqīh as transmitted in Yāqūt); Le Strange, The Lands, 127-58; Honigmann, Ostgrenze, passim.
In addition to the present chapter some interesting details on the Byzantine Empire are found in § 3, n. 12. (the lakes); § 5, 18., 20., 21., 25., 28. (the mountains); § 6, 58.-60., 66. (the rivers). On the northern and eastern frontiers of the Empire see under each of the countries mentioned.
As Marquart, Streifzüge, xxxiii, 28, 207, &c., has shown, the principal sources from which the earlier Muslim geographers derived their information on the Byzantine Empire were Muslim b. Abī Muslim al-Jarmī and Hārūn b. Yaḥyā. On the former we possess a notice in Mas'ūdi's Tanbīh, 190, according to which he lived on the Arabo-Byzantine frontier (thughūr) and wrote "on the history of the Byzantines and their kings and dignitaries, on their land and its roads and routes, the times (favourable) for the raids into their territory,  the campaigns therein, on the neighbouring kingdoms of the Burjān, Abar, Burghar, Ṣaqāliba, and Khazar". Mas'ūdī also gives the exact date (231/845-6) at which Muslim was redeemed from Byzantine captivity. I.Kh. expressis verbis quotes Muslim as his source.
Much less is known of Hārūn who also was a prisoner of war and taken from 'Asqalān (§38, 15.) to Constantinople whence at a later date he may have travelled to Rome. He wrote towards the very end of the ninth century (v.i. 17.) and his writings are known to us through the important excerpt in I.R., 119-32, and some items in Zakariyā Qazwīnī, ii, 406-7 and 397-9. I believe that some traces of his account can also be discovered in the Ḥ.-'Ā. and Gardīzī. 
On the whole our author follows I.Kh. (< Muslim). One point is particularly
characteristic in this respect. In the introduction of the present chapter
he says that the northern Byzantine frontier ran along "some parts of the Ṣaqlāb
and *Burjān countries
and some parts of the Khazar sea".
1. This part of Muslim's writings has the survived in Qudāma, 259.
2. See now an English translation of Hārūn's report by A. A. Vasiliev, with extremely apposite additions by G. Ostrogorsky, in Seminarium Kondakovianum, Prague 1932, v, 149-64 and 251-7; critical review by H. Gregoire, Byzantion, Brussels 1932, pp. 666-73. The upshot of the conclusions of these Byzantine scholars is that Hārūn, captured probably towards the end of Leo's reign describes Constantinople under the brief reign of the Emperor Alexander (11 May 912-6 June 913). [The most striking of Ostrogorsky's arguments is Hārūn's silence about the presence at ceremony either of the Empress or of the Emperor's co-regent, which only suits Alexander's reign. However, it appears from p. 252 that, between the years 893 and 894, 896 and 899, and finally 900 and 906, his predecessor Leo VI lived as a widower. This leaves a gap for my tentative dating of Hārūn b. Yaḥyā's report circa 900. The absence of a co-regent may be due to some temporary circumstances, or to Hārūn's oversight. The date 912, even admitting that it is not too late for I.R., may be too late for Jayhānī, if the latter, as is quite probable, was I.R.'s direct source on this point.]
420 Commentary § 42
These indications are directly borrowed from I.Kh., 105, who uses the same very uncommon term for the Black Sea. Through his blind imitation our author falls here into contradiction with his own terminology, cf. § 3, 5. and 6. and § 22,14. The influence of Hārūn's data is apparent in our 15. and 17., to say nothing of the general conception of §§ 22, 46, and 53.
I.Kh.'s (< Muslim's) description of Byzantine provinces is very much to the credit of the Arab intelligence service. Gelzer calls I.Kh. "eine hochst zuverlässige zeitgenössische Quelle ersten Ranges". The data refer to the times of the Amorian dynasty (820-67)  for which no similar systematic descriptions in Greek are available.
I.Kh., 105, quotes 3 themes in Europe and 11 in Asia ()
and the same number is preserved in Qudāma
and our author, though the order of enumeration is different in each of
the sources as appears from the following table [in each column the numbers
refer to the place of the theme within the respective list].
Qudāma's order of enumeration  is perhaps geographically the best (see Gelzer's map) but our author's system is very curious as indicating that he had a map before him, for starting three times in the south (4., 7., n.) he each time moves straight towards the north!
The spelling on the whole is nearer to Qudāma
than to I.Kh.
1. See now the French edition of A. Vasiliev, Byzance et les Arabes, t. I, La dynastie d'Amorium, Brussels, 1935.
2. With which that of I. Faqīh (in Yāqūt) totally agrees.
§ 42 Byzantine Empire 421
The name 1. Ṭāblān (numerous variants in different sources: ) has been explained by de Goeje (I.Kh., 105, note i) as (). Suidas, Lexicon, ed. 1853, p. 1053: . Consequently the popular name would refer to the Great Wall (, I. Faqīh in Yāqūt, ii, 863, ) or rather to the moat () round it. Gelzer, o.c., 86, accepts the interpretation and compares it with the terms and provincia suburbicaria. Bury, A History of the Eastern Roman Empire, 1912, p. 224, thinks that "the solution (of I.Kh.'s term) has not been discovered". He starts, however, from the form *Ṭalāyā, whereas the better attested forms are *Ṭāflā, Ṭāblān.
Our author takes no notice of the changes which had taken place between
Muslim's times and his own. Constantine Porphyrogenitus, writing half a
century before him (a.d. 932), enumerates 12 themes in Europe and 18 in
Asia; of the latter, 4 are islands and the rest is as follows:
Constantine Porphyrogenitus does not separately mention our 11. and 12. He speaks of Cappadocia under Armeniacon and adds that is the middle part of Cappadocia, De them., pp. 18-20. Cf. now E. Honigmann, Charsianon kastron in Byzantion, x, 1935, pp. 129-60. On the other hand Constantine mentions i, k, l, m, o, unknown to our three authors. Mas'ūdī, Tanbīh, going his own way, mentions 5 provinces in Europe (inclusive of Salonika and Peloponnesus) but only 9 in Asia, viz. our 8., 5., 4., 1., 9., 6., 13., 10. plus Decapolis (mentioned between 4. and 11.).
As regards the number of troops in the provinces our text is certainly out of order. According to Qudāma, 258 (< Muslim) the number varied from 15,000 (in Nāṭlīq) to 4,000 (Kharshana, Cappadocia, Khaldia).
15. The paragraph on the Gurz ("Georgians") is one of the most confused in the book. Gurz is a parallel Iranian form of Gurj, modern Persian (and Turkish) Gurjī, Russian Gruz-in. The element -z (-j) is a suffix of origin, see Marquart, ZDMG, 49, p. 664; cf. also § 36, 36.: Layzān and § 50, 3.: *Lakz. The older form of Gur-z is attested in Armenian Vir-k', pointing to Middle Persian *Vr-kān. The earlier Arabic transcription is (i.e. *Gurz-ān), Balādhurī, 202, but already Ya'qūbī, Historiae, ii, 519, gives *Gurj-ān and the later authors write *Gurj, see Ibn al-Athīr, passim, Yāqūt, ii, 219. See now Markwart (Marquart), Iberer und
422 Commentary § 42
Hyrkanier, in Caucasica, viii, 1931, p. 78. The variation of the forms Gurz/Gurj must be due to some dialect distinctions, cf. -nz/-nj in the name of Ganja: earlier Arabic , I.Kh., 119, and later Arabic , cf. Minorsky, Jour. As., July 1930, p. 72.
Some of the information contained in this paragraph undoubtedly refers to Western Georgia drained by the rivers flowing to the Black Sea. With the exception of the early Balādhurī, p. 202 (conquest of "Armenia"), the Arab authors know nothing of Western Georgia, whereas they usually include Eastern Georgia (watered by the Kur) in Armenia, as also does our author, cf. § 36, 28. (Tiflis). The inclusion of (Western) Georgia in the Byzantine Empire, which may be explained by the fact that the Georgians belonged to the Greek Orthodox Church, is responsible for the statement (§ 49) on the Byzantine Empire being conterminous with the Sarīr. Under §5, 21. our author speaks of a Kūh-i Gurz, perhaps identical with § 5, 17 B. These details refer to the real Georgia.
On the other hand, the designation of the Black Sea as daryā-yi Gurziyān is entirely without a parallel, and it is astonishing to see the Pontos baptized after a people never known as navigators. Still more amazing is the representation of the Gurz as living "on small islands", whereas there are no islands in the eastern part of the Black Sea. 
As already stated in the note to § 3, 6., this part of the information may be due to a confusion of *Warang and *Gurz, not impossible in Arabic script. Warang, very rarely met in Muslim sources, is another appellation of the Norman Rūs (§ 44) who according to some earlier source lived on an island. A series of errors with regard to the whereabouts of the real Maeotis (§ 3, 8.) may have caused our author to dissociate the rare Warang from Rūs and finally to misread it into another rare name Gurz. Bīrūnī is the earliest known author mentioning but he must have found it in some literary source. 
The other source of confusion may have been I.Kh., 105 (< Muslim
to whom the themes of Ṭāflā
(Constantinople) and Trāqiya
(Thrace) bordered in the north on .
This unusual appellation of the Black Sea crept into the introductory paragraph
of our § 42, though under § 42, 3. Thrace is said to lie by the Daryā-yi Gurz. In
our author's terminology the Khazarian Sea is the Caspian. With
regard to the general frontiers of the Byzantine Empire (extended up to
the Sarīr!) I.Kh.'s
strange term may have passed unobserved, but in the particular case of
Thrace our author could not help noticing that this province does not lie
by his Khazarian Sea (i.e. Caspian). Therefore he may have
1. By some mistake Idrīsī, ii, 396, mentions an island on the way between Taman (on the Azov Sea) and Trebizond. [Const. Porph., De adm. imp., ch. 42, mentions an island near Tamatarkha (evidently a part of the Taman peninsula) and several islands off the coast of (Circassia), near the estuary of the Kuban, but none of them has any relation to the Georgians.]
2. On some curious points of contact between our author and Bīrūnī cf. § 10, 55., § 11, 9., § 26, 13., &c.
§ 42 Byzantine Empire 423
supposed that I.Kh.'s referred not to the Khazar but to the Jurz (a classical confusion in Arabic script). It only remained, then, for our author, who shows a notable predilection for Iranian terminology, to substitute Persian Gurs for Arabic Jurz. Cf. supra, p. 182.
16. and 18. These peoples are no more distinct than the Spartans and Lacedaemonians. Both names refer to the Danubian Bulghars who, moreover, seem to be described as "Inner Bulghars" (§ 45) and "V.n.nd.r" (§ 53). I.Kh., 92, 105, 109, and Qudāma, 257, systematically following Muslim al-Jarmī, call the Danubian Bulghars Burjān.  On the other hand, Hārūn (I.R., 130) applies this term to the Burgundians whereas he calls the Danubian Bulghars Bulghar.  Hārūn, too, is most probably responsible for the term V.n.nd.r (= Onoghundur-Bulghars). As it was impossible to unravel such complications in a compilation, our author's source (Jayhānī ?) must have solved the difficulty by incorporating all these names as if they referred to separate entities. Consequently the Burjān and Bulgharī were differentiated artificially: the former being imagined to be more submissive plain-dwellers, and the latter highlanders  "perpetually at war with the Rūmīs" (as in I.R., 12622). The Bulgharī are called Rūmī because they were christianized from Byzantium in A.D. 864. The qualification kāfir is rather strange. One may remember that Bakrī, 4520, calls the Burjān "Magians" (majūsiya) and this term is constantly applied to the Normans as well, cf. Lévi-Provençal, Madjūs in EI and Idrīsī-Tallgren, pp. 80 and 140.
17. This short paragraph is of great importance as indicating our author's
sources. I.Kh., 105, quoting by name Muslim al-Jarmī,
thus describes the boundaries of Macedonia: in the east the walls (v.s.
1.) stretching between the Black Sea (Baḥr
al-Khazar) and the Syrian Sea (Baḥr
al-Sha'm, here evidently "Marmora Sea"); in the south, the Mediterranean;
in the west, the lands of the Ṣaqāliba;
in the north, the Burjān.
This quotation leaves no doubt that by the "Ṣaqaliba
lands" the Serbian territory is meant. However, much more decisive is I.R.,
127, who quotes Hārūn
description of a road from Constantinople, over Salūqiya
(read: Salonica)  and Venice (B.nd.qīs)
to Rome. The text is out of order, but Marquart, Streifzüge,
237-59, has suggested a series of very ingenious corrections of it. At
3 days' distance to the west of *Salonica lies Mutr.n (Marquart:
*Qutron < );
"beyond it you travel through wooded lowlands (ghiyāḍ
min al-shajar) among the Ṣaqāliba
who live in wooden
1. It is true that Muslim, v.s., is also said to have written of the Burghar but this term could possibly refer to the Volga, or Azov Sea, Bulghars.
2. The name (I.R., 12618) stands first erroneously for Belgrad (v.s., § 6, 66.) but after this passage on the water conduct comes (I.R., 12622) the remark on the perpetual war going on between the real Bulghars and Byzantines. This last item looks like an interpolation but it is found both in I.R. and our source (18.). Therefore, if it is an interpolation, it must belong to the two authors' common source (Jayhānī?).
3. Cf. § 5, 28. on their mountain and § 6, 66. on their river.
4. Mas'ūdī, Murūj, ii, 318, also has Salūqiya for Salonica.
424 Commentary §§ 42-53
houses. They are Christians; they were (gradually) converted (kānū yatanaṣṣirūna) in the time of the king (*Basil) and to-day they hold the Christian faith. Among them you travel for a month across their woods until you reach the town of B.lātīs (*Spalato, Const. Porph. )." In our text al-Ṣaqāliba al-mutanaṣṣira, standing in the Arabic garb, without any doubt reflects Hārūn's account (through Jayhānī's medium ?). According to Marquart, ibid., 207, Hārūn must have drawn up his report between A.D. 880 and 890, but the text seems to indicate that the Emperor Basil I's time (a.D. 866-86) was regarded as past; therefore we may bring Hārūn's date down to the years 890-900. The exact date of the conversion of the Serbs cannot be established. In the years 867 and 870 Basil I subjugated the Serbs (Narentani, Croati) on the Dalmatian coast, and in 879 for the first time the bishop of Moravia (i.e. probably of the Serbian region lying along the southern affluent of the Danube, Morava) is mentioned, cf. F. Dvorník, Les Slaves, Byzance et Rome au IXe siècle, Paris, 1926, p. 239 (where it is assumed that Basil I converted the Slavs between a.d. 879 and 882). To sum up: the mention of "the Christianized Slavs" is a clear indication that besides al-Jarmī our author knew also Hārūn's report. If so, we may assume, contrary to Marquart, o.c., 28, that in other chapters too, particularly those on the Magyars, V.n.nd.r, and Mirvāt , our author's source was Hārūn, who was recording the situation towards the very end of the ninth century, and not al-Jarmi, who belonged to the earlier part of that century. [This admission has a considerable importance for the history of Magyar migrations, v.s., § 22.]
19.-23. are a drastic epitome not devoid of misunderstandings. Rūm (Byzantine Empire) is supposed here to comprise all the countries lying by the Rūm Sea (Mediterranean). The original authority seems to be Iṣṭ., 43 (and also, 68-71), who says: "and among the different classes of infidels who adjoin Andalus the most numerous are the Ifranja whose king is called Qārula (*Carolus), but the Ifranja conterminous with the Muslims are less numerous than the other classes of infidels on account of the fact that the Ifranja protrude into the sea (dukhūluhum fil-baḥr) and on account of the buffer (ḥājiz) which other countries of polytheism constitute between the Muslims and the Ifranja. Next in numbers (after the Ifranja) are the Jalāliqa, and less numerous still the Baskunas (though they) are more warlike (ashaddu shaukatan). The places on the Andalus border neighbouring the Baskunas are Saraquṣṭa, Tuṭayla (Tudela), and Lerida. Then follows a Christian people called Ghalijaskas who are less harmful than the Baskunas (aqalluhum ghā'ilatan); they constitute a buffer between them (the Baskunas) and the Ifranja." I.Ḥ., 43, says that the frontier line following the eastern coast of Andalus joins on the sea the Ifranja country, and on the west that of the Ghalijaskas "who are a tribe of al-Ankubarda (Lombards?)", then the Baskunas country, then that of the Jalāliqa, then the sea.
Our author entirely omits the important Jalāliqa (Galicians). The Baskunas are Vascones (Basques). The Ghalijaskas are the inhabitants of the
§§ 42-53 Byzantine Empire 425
Jacá region, south of the important pass in the Pyrenees (at present Jacá is connected by a railway tunnel with Oloron on the French side). The Jacá people were already known to the classical authors under the names of Jaccetani, (Ptolemy, ii, ch. 5), Lacetani (Livy, xxi, 60-1). Ya'qūbī, BGA, vii, 355, says that north of Saragossa lies the town of Tudela, situated towards the land of the unbelievers called Baskunas, and that to the north of Tudela lies Huesca (Washqa) situated towards a tribe of the Ifranj called (the latter name has a variant corrected in a different hand into , read: al-Jāqiya; cf. also Ibn al-'Adhārī, ii, 302: ). Ya'qūbī's passage leaves no doubt on the identity of the people. On the different forms of the name Codera says: "la confusión pudo quizá originarse por la semejanza de nombres entre Jacetanos y Lacetanos de los autores antiguos", but more probably it is attributable to the usual vagaries of Arabic script. I. Faqīh, 87, is wrong in placing (variant ) "near the sea". Cf. de Goeje, Specimen exhibens descriptionem al-Magribi sumtam e Libra Regionum al-Jaqubii, Leiden, 1860, pp. 112-13, and F. Codera, Límites probables de la conquista árabe en la Cordillera pirenaica, in Estudios críticos de Historia Arabe espan~ola, vii-ix, Madrid, 1917, pp. 235-76. I owe the indication of the last work to the kindness of Prof. A. Gonzalez Palencia. Bīrūnī, Canon, places Lerida over against the Gh.l.j.sk, see § 41, 4.
About 19. Rūmiya
(Rome) I.Kh., 10, says that it was [in turn] the seat of 29 Roman kings
(emperors). 21. Britannia is not in I.Kh., Iṣṭ.,
or I.Ḥ., but I.R.,
130 (after Hārūn
(note the spelling of our source, too!) as a large town on the coast of
the Western Sea. On the confusion about Yūnān
see the original text of Iṣṭ.,
is the seat of learning of the Ionians (Yūnāniyūn)
and there their sciences and learning are preserved", cf. Barthold, Preface,
pp. 21 and 41.
[Back to Index]