^{page 67 note 2} There are also some notices relating to the themes in the De Adm. Imp. (Const. Porph. iii. pp. 220–231).

^{page 67 note 3} De Caer. 2, 52 (ed. Bonn. i. pp. 713–715, 727, 728).

^{page 67 note 4} Op. cit. 2. 50.

^{page 67 note 5} Edited and translated by De Goeje, (Bibl. Geog. Arab. vi. p. 77 ff.)Google Scholar. The date was probably 845–8 (id. p. xix. ff.).

^{page 67 note 6} Transl. Jaubert ii. p. 299 ff. The full Arabic text remains unpublished. Al Idrisi gives only the Asiatic themes.

^{page 67 note 7} Ed. Wüstenfeld, ii. p. 863 ff. The description of Macedonia is also in iv. pp. 602, 603, where it is cited as from ‘Ibn Al Fakih in the account of the districts of the Romans.’ An epitome of Ibn Al Fakih's work has been edited by De Goeje (op. cit. pars v.), but it does not contain the account of the themes.

^{page 67 note 8} Edited and translated by De Goeje (op. cit. vi. p. 197, ff.).

^{page 67 note 9} Edited by De Goeje (op. cit. viii. p. 176, ff.); translated by Carra de Vaux (Paris 1896).

^{page 69 note 1} ‘Al Anti Mati [Optimatoi],…and that is the army of Al Natalik [Anatolikoi].’ The Optimates are however mentioned later, and the description here following is clearly that of the Anatolikoi.

^{page 69 note 2} ‘Nantiliya (v. l. ‘Nantuliliya.’), and that is Dakabuli [Dekapolis].’ De Goeje supposes this to stand for Pamphylia; but it seems rather to represent Anatolikoi, though the description can hardly apply to any other theme than Kibyrrhaiotai.

^{page 69 note 3} The last clearly by error, since he says himself that the Armeniac theme reached to the sea. As to Seleukeia see Gelzer, p. 93, note, and below, p.71, note 10.

^{page 69 note 4} The theme of Peloponnesos existed in 811 (Anon, , de Leon, Arm. in Bonn. Corpus, xxx. p. 336)Google Scholar, and a seal of a ατρατηγός is ascribed by Schlumberger, (Sigillographie de ľ Empire Byzantin, p. 179Google Scholar) to the eighth century. The passage adduced by Gelzer, from Const. De Adm. Imp. (ed. Bonn, , iii. p. 221, l. 3–10)Google Scholar to show that Peloponnesos was made a theme in the time of Michael III. is insufficient to prove this. The omission of European themes cannot however be used to fix the dates of the Arabic lists, since all omit Hellas, which existed in 695 (Theoph. A M 6187). This passage is neglected by Gelzer, who ascribes the institution of this theme also to the time of Michael III.

^{page 69 note 5} Const. Porph. iii. pp. 31, 32. Lykandos however was not made a στρατηγίς till the regency of Zoe (912–919); id. p. 228.

^{page 69 note 6} Const. Porph. iii. p. 36.

^{page 69 note 7} The list in De Caer. 2. 50. differs from that in De Them. by omitting Optimatoi and Cyprus and adding Leontokomis and Dalmatia. As this list gives the salaries of the generals, the omission of Optimatoi is no doubt due to its being under a δομέστικος. The three lists in De Caer. 2. 52 include the δομέστικος τῶν ὀπτιμάτων, but omit Mesopotamia, Sebasteia, Lykandos, Seleukeia, Leontokomis, and Lombardy, probably because they were under officers of lower rank.

^{page 70 note 1} Unless indeed we are to bring this into connexion with the omission of Chaldia in Theoph. Cont. p. 81 (Gelzer p. 99), and suppose that the theme of Chaldia was temporarily suppressed or its territory temporarily lost to the Empire.

^{page 70 note 2} Theoph. Cont. l.c.

^{page 70 note 3} Similarly Charsianon, which in Al Mas‘udi is a κλεισοῦρα appears in 873 as a στρατηγίς (Genesius, p. 122), and Sebasteia, not mentioned by him, was a κλεισοῦρα under Leo VI. (Const. Porph. i. p. 697, iii. p. 227).

^{page 70 note 4} Or brought up for exchange. As he denied the creation of the Kuran, it is not clear whether he was actually exchanged at this time.

^{page 70 note 5} I cannot make anything else of ‘dha mahal.’ de Vaux, Carra and de Meynard, Barbier (Prairies ď Or, ix. p. 357Google Scholar) omit the expression in translation.

^{page 70 note 6} Tanbih, p. 190; Transl. p. 257.

^{page 70 note 7} Ibn Al Fakih's statement that Amorion was in the author's time waste need not necessarily be derived from Al Garmi, but may b an insertion either of Ibn Khurd. (see p. 71, note 4), or of Ibn Al Fakih himself.

^{page 70 note 8} Theoph. A M 6281.

^{page 70 note 9} It existed however in 802 (id. A M 6294), and a seal of Sergius, στρατηγός of Macedonia, is ascribed by Schlumberger, (Sigillographic de ľEmpire Byzantin, p. 111Google Scholar). to the eighth century. It is not unlikely that its institution was a consequence of the disaster of 789.

^{page 71 note 1} Const. Porph. iii. p. 178; Theoph. Cont. p. 123.

^{page 71 note 2} Const. Porph. i. p. 788.

^{page 71 note 3} He is called στρατηγός in 863 (Theoph. Cont. p. 181), and we find Paphlagonia described as a θέμα as early as the time of Michael, II (Mich. Mon. vit. Theod. Stud. 54)Google Scholar.

^{page 71 note 4} Our present text is incomplete (De Goeje p. xv. ff. )

^{page 71 note 5} Id. p. xxii. His account of the raiding-seasons (p. 199) clearly comes from Al Garmi (see above p. 70), but may have been in the full text of Ibn Khurd.

^{page 71 note 6} De Goeje B.G.A. v. p. xii.

^{page 71 note 7} In the work of Philotheos (Const. i. p. 715) we find Kibyrrhaiotai among the western themes. See also Gelzer p. 105.

^{page 71 note 8} Const. iii. pp. 35, 38.

^{page 71 note 9} Id. pp. 25, 43, 44.

^{page 71 note 10} It may have been this fact which led Al Mas‘udi to make the mistake of making Seleukeia part of Kibyrrhaiotai. It is however possible, but not likely, that before 863 a territory had been assigned to the commander of the Kibyrrhaiotai and the κλεισουράρχης of Seleukeia placed under him. The expression τῷ ἀκρωτηρίῳ τῶν Κιβυῥῥαιωτῶν τῷ λεγομένῳ Χελιδονία (Geo. Mon. p. 720) does not prove territorial jurisdiction, since George is speaking of naval affairs and therefore writing from the naval point of view.

^{page 72 note 1} Gelzer (pp. 87, 88) believes that they formed an actual theme under the prefect of the city and that this was suppressed by Leo VI. This is plausible, but the evidence for the military authority of the prefect is very weak.

^{page 72 note 2} See De Boor's index to Theophanes s.v. τάγμα Talaya might also represent τὴν αὐλήν or τὰ ἐν τῇ αὐλῇ (τἀγματα) It is a tempting but somewhat too daring conjecture that it stands for τὴν Ελλάδα and that Al Garmi through some blunder supposed this to be Constantinople. This would explain the apparent omission of Hellas.

^{page 72 note 3} Charsianon is mentioned as a κλϵισοῦρα in 863 (Theoph. Cont. p. 181). See also p. 70, note 3.

^{page 72 note 4} It was a στρατηγίς in 863 (Theoph. Cont. l.c.) and appears as such in Al Mas‘udi.

^{page 72 note 5} I.c. Ibn Al Fakih.

^{page 72 note 6} See J.H.S. xviii. p. 194, note 5; xix. p. 23.

^{page 73 note 1} It is not clear whether this apology is to be ascribed to Ibn Al Fakih or to Yakut.

^{page 73 note 2} From the omission of the ‘Khalig’ among the boundaries and the statement below that the E. boundary of Macedonia was the long wall it is clear that the Thracian theme did not reach to the Propontis.

^{page 73 note 3} Al Garmi seems to have been in some confusion as to the points of the compass, since Bulgaria was clearly the N. and the Euxine the E. boundary. Similar errors are often found in Kudama, who, unlike Ibn Al Fakih, gives the points of the compass for the Asiatic themes also.

^{page 73 note 4} The sentence, as it stands in the text, can hardly be translated, and a comparison with Ibn Khurdadhbah and Kudama shows that these words have fallen out.

^{page 73 note 5} The explanation is perhaps due to Yakut.

^{page 73 note 6} The length has perhaps fallen out.

^{page 73 note 7} Clearly an insertion of Yakut.

^{page 73 note 8} Wüstenfeld suggests Abydos; if this is right, there must be some confusion. Possibly Kassandreia is meant, but more probably the author has taken βάνδον for a proper name; cf. Al. Mas. p. 176; transl, p. 239 and note.

^{page 73 note 9} This shows that Paphlagonia reached much farther west than in Constantine's time, when it stopped at the Billaios, the intervening space being occupied by the Buccellarii (Const. iii. pp. 28, 29). That this is not a mere slip appears from the fact that our author places Optimatoi ‘by the side’ of Paphlagonia. See also p. 76, note 3. Al Mas. however makes Buccellarii extend to the sea, and the change had therefore been made before 863. Kudama in describing the boundaries of Optimatoi includes Paphlagonia and omits Buccellarii.

^{page 73 note 10} Neither Nikopolis nor Sebasteia can ever have been in Paphlagonia, and at the end Yakut says that Sebasteia is not mentioned by Ibn Al Fakih. Moreover Siwas seems to be a form of Turkish origin, and the name is here spelt differently. For ‘Naikus’ we might by a change of points read ‘Biyufus’ or ‘Babufus,’ which might be a shortened form of Pompeioupolis, or ‘Nifus’ (= Sinope (?)). For ‘Siwas’ there is a variant ‘Sulas.’

^{page 74 note 1} The other boundaries and the seat of the στρατηγός have perhaps fallen out.

^{page 74 note 2} τὸ καλούμενον . . . . θέμα ᾿Οπτίματος οὐδεμίαν ἔχει κοινωνίαν πρὸς θέματα εἰς γὰρ δουλείαν μόνην προσείληπται διὰ τὸ εἶναι αὐτὸ οἰκτρότατον καὶ μήτε τούρμαις μήτε δρούγγοις τετιμημένον . . . εἰς γἀρ ὐπηρεσίαν ἐτέτακτο τῶν στρατιωτῶν Const. Porph. iii. p. 26.

^{page 74 note 3} The omission oi Buccellarii (so also Kudama) shows that at least at this time the Anatolic theme reached farther north than is usually supposed (see also p. 76, note 3), the boundary being probably the northern portion of the Sangarios. Jaubert's identification of Marg Al Shahm, the seat of the Anatolic στρατηγός, with Germa, (see J.H.S. xix. p. 31)Google Scholar is therefore not impossible.

^{page 74 note 4} With an alteration of points we may read ‘Nitaya,’ which closely resembles Nikaia. Kotyaion may also be suggested.

^{page 74 note 5} The author can hardly mean to make Thrakesioi reach to the Hellespont, and Kudama distinctly makes the Khalig the W. and the Syrian Sea the S. boundary. The ‘Khalig’ must therefore here include the Aegean, which in the descriptions of Macedonia and Talaya is included in ‘the Sea of Al Sham.’

^{page 74 note 6} The omission of Seleukeia (so also Kud.) is apparently an oversight of Al Garmi, since Thrakesioi is given among the boundaries of Seleukeia.

^{page 74 note 7} Territorially it would appear that Thrakesioi was larger, but the author is probably thinking of the size of the army.

^{page 74 note 8} The other boundaries have perhaps fallen out.

^{page 74 note 9} See J.H.S. xviii, p. 190 note 1; xix. p. 31 ad fin. The reading of Ibn Khurd. however, which is unpointed, seems to be meant for ‘Burg Al Takhm’ (tower of the boundary), for which ‘Marg Al Shahm’ should no doubt be restored with De Goeje from Al Idrisi.

^{page 74 note 10} The occurrence of these three names in this place is very puzzling. Hierapolis was not Byzantine till 968, and Barbalissos cannot have been so earlier, nor can any of the three have boen in the Anatolic theme. Moreover Ibn Khurd. places Burghuth in the Anatolic theme, but without identifying it with Mar‘ash, which was not in his time Byzantine. Nor can this be an insertion of Yakut, in whose time these places had long been lost to the Empire. Probably therefore there is some corruption and the words belong to another context.

^{page 75 note 1} See Tomaschek, in Sitzungsber. d. Wiener Akad. cxxiv. VIII. p. 67Google Scholar.

^{page 75 note 2} Either the river or the town may be meant. Between Seleukeia and Al Lamis Al Mas‘udi mentions a fort which De Goeje prints as ‘Bukiya.’ There is however a variant ‘Brakiya’ or ‘Brakana,’ and no doubt Prakana is meant (cf. Tomaschek p. 60).

^{page 75 note 3} The translation following shows this to be the title meant, and Seleukeia is in fact called a κλϵισοῦρα by Const. Porph. (iii p. 35); cf. Theoph. Cont. p. 181. Wüstenfeld however corrects the word to ‘khiliyarg’ = χιλίαρχος. This is no doubt also the meaning of Ibn Khurd's expression, ‘and its wali is the ruler of the passes,’ which is obscured in De Goeje's translation.

^{page 75 note 4} This no doubt refers to the title ἐκ προσώπου of which several examples are found on seals (Schlumberger, , Sigillographie de ľEmpire Byzantin, p. 576Google Scholar ff.); cf. also Const. Porph. iii. p. 230; i. pp. 715, 729.

^{page 75 note 5} The obscure Antioch in Isauria must apparently be meant, but even this is eighty miles from Seleukeia. Moreover the preposition should rather mean ‘as far as,’and the omission of the strength of the army perhaps points to some words having fallen out. Bead perhaps, ‘<and its army consists of 5,000 men (Kud.), and it reaches from Seleukeia> to Antakhiya.’

^{page 75 note 6} So Kudama. This shows that, as we should expect, the κλϵισουραρχίαι of Seleukeia and Cappadocia now adjoined one another and were not divided by a piece of Anatolikoi, as in the maps of Menke and Gelzer. Al Mas. however extends Anatolikoi to the frontier and places Herakleia in it, so that a change had been made before 863.

^{page 75 note 7} V.l. ‘Tulifus,’ which differs only by a point from Tulikus.

^{page 75 note 8} ‘φρούριον τὸ καλούμϵνον Κόρον’ (Const. iii. p. 21). See also Eng. Hist. Rev. xv. p. 742 an Ramsay, H.G. p. 355Google Scholar.

^{page 75 note 9} Ikonion was in the Anatolic theme (Const. iii. p. 16), and Yakut at the end states that it was not mentioned by Ibn Al Fakih. Probably therefore the name is corrupt. Ibn Khurd. has ‘Karniya’ or ‘Kutiya’ and Al Mas. ‘Karniya’ or ‘Kunana.’ Kanna may perhaps be suggested.

^{page 76 note 1} V.l. ‘Sariga.’

^{page 76 note 2} See Ramsay, H. G. p. 312Google Scholar. There is some difficulty about the occurrence of this name here, since in Theoph. Cont. p. 369 we read of τὸν ἐν τῷ Σιριχᾷ σταυρόν and from Const. iii. p. 225 we learn that ἡ τοποτηρησία τοῦ τιμίου Σταυροῦ was transferred from Buccellarii to Charsianon by Leo. Either there were two σταυροί or the place had been at an earlier time transferred from Charsianon to Buccellarii.

^{page 76 note 3} The omission of the sea among the boundaries shows that the statement above as to the Paphlagonian theme (see p. 73, note 9), is not a mere slip. A similar deduction as to the extension of Anatolikoi may be made from the omission of Opsikion (see p. 74, note 3). Optimatoi, which is given among the boundaries by Kudama, has probably been omitted by an oversight.

^{page 76 note 4} Clearly an insertion of Yakut, in whose work the story of Amru'l Kais is recorded under ‘Ankyra.’ It is not unlikely that the whole reference to Amru'l Kais is due to him.

^{page 76 note 5} From these boundaries we might at first sight infer that the Armeniae theme did not at this time extend to the frontier but was separated from it by Chaldia and Charsianon. As however Chaldia is not mentioned among the boundaries of Charsianon or vice versa, the omission of Thughur Al Gazira among the boundaries of Armeniakoi seems to be only an oversight.

^{page 76 note 6} The Paulicians arc perhaps meant; but, if so, the passage can hardly date earlier than 843. Moreover their chief centre was in the Armeniae theme.

^{page 76 note 7} See Eng. Hist. Rev. xv. p. 740.

^{page 77 note 1} The author has also applied this title to the governor of Cappadoeia. He omits to note that the commander of Opsikion was called count.

^{page 77 note 2} Possibly κομμϵρκιάριος.

^{page 77 note 3} See p. 75, note 9.

^{page 77 note 4} See Ramsay, , H. G. p. 285Google Scholar.

^{page 77 note 5} I.e. Antioch in Pisidia.

^{page 77 note 6} See p. 73, note 10.