¹ De Goeje, iii. 1236 sqq. I have used the Russian translation of Vasil'ev, Vizantiia i Araby, i. Prilozhenie, 30 sqq. and my references are to it. So far as I know, Vasil'ev is the only modern critic who has worked up the relation of Tabari.

² He left Samarra early in April, Yakubi, in Vas, . Pril. 9Google Scholar; cp. Masudi, , Golden Meadows, ib. 68Google Scholar (Barbier de Meynard, vii. 135).

³ He started from Saruja (Tabari 31) = Surghi.

⁴ The most direct route to Ancyra was by Caesarea. But that Afshin marched by Sebastea must be inferred from the fact that his battle with the army of Theophilus was fought in the neighbourhood of Dazimon (Tokat). Vasil'ev is probably right in supposing (op. cit. 121) that it was part of the plan that Afshin should join another army, from Armenia and Melitene (perhaps τῆς ἐξ ᾿Αρμενίων στρατιᾶς of Genesios 67). The westward roads from Melitene and Armenia met at Sebastea. The fertile plain of Dazimon (Kaz-ova) lay on the right (north) of the route Sebastea—Sebastopolis—Ancyra. Cp. Anderson's map of Asia Minor (1903).

⁵ Tabari, ib.

⁶ Lulon and al-Safsaf are names for the same fortress as Ramsay has shown. For the identification of the fortress and a full description of the Pass, see Ramsay, , Geographical Journal, Oct. 903Google Scholar.

⁷ Yakubi, , in Vas., Pril. 8Google Scholar; Tabari, ib. 25.

⁸ Tabari says that Mutasim sent the advance guard of his own army in the steps of Ashnas, and started himself on June 21. Masudi (ib. 68) says that Mutasim marched by the Pass of Podandos (Darb as-Salam), Afshin by the Pass of Hadath, and other armies by other routes. The last words must refer not to Ashnas, but to the forces from Melitene and Armenia.

⁹ Historical Geography, 293; 356 (‘the plain of Venasa, about Sasima and Malokopaia; great underground residences are a special and peculiar feature of this plain, which lies on the direct road north from the Gates’).

¹⁰ Translated by De Goeje, Bibl. Geog. Arab. vii. It is probable that the routes given by Ibn Khuidadhbah were derived from the work of Al-Garmi, who wrote books on the history and geography of the Roman Empire. He had been a captive among the Romans and was released on the occasion of the exchange of prisoners in A.D. 845. We know from Masudi that he wrote about the roads (transi. by Carra de Vaux, 257), and we know that Ibn Khuidadhbah made use of his works elsewhere (De Goeje, 77).

¹¹ Translation by Jaubert.

¹² Jahreshefte dos österreichischen archäolo gischen Institutes, vii. 123–125 (1904).

¹³ Some of Idrisi's stations are different, but both lists agree in Marj al-Uskuf.

¹⁴ This station is given by Idrisi as well as Ibn Khuidadhbah.

¹⁵ Tabari, 31–2. Tabari seeras to mean four parasangs from the place where Amr captured the Greek. This is impossible, as Hassan Dagh is much further from the nearest point on the Halys. I will return to this point further on.

¹⁶ Historical Geog. 355, Zycaonia 127. The second beacon station from Lulou was, as Ramsay has shown, on this Mt. Argaios.

¹⁷ Following Ramsay's, conjecture, Hist. Geog. 308Google Scholar. Cp. Strabo, 12. 2. 5, ὄ τε ᾿´Αργος ἔρυμα ὐψηλὸν πρὸς τῷ Ταύρῳ καὶ τὰ Νῶρα ὄ νῦν καλεῖται Νηροασσός, ὲν ᾦ Εὐμενὴς πολιορκούμενος ἀντέσχε πολὺν χρὁνον. He adds that Sisinas kept his treasures in it. This passage shows that Nora was a fort, and suggests that it was not far from Argos. Argos, as Ramsay says (353), must be associated with Mt. Argaios.

¹⁸ Hist. Geography 77 (He is mistaken in the date which he gives as A.D. 860). The author ities are: (1) Arabic, Yakubi (11); Tabari, 62 (cp. Abu 'l-Fida, Annals, ed. Reiske, ii. 208); (2) Greek, the Logothete (= George Mou. ed. Bonn 825, ed. Muralt, 733–4; Theodosios Mel. 167, Leo Gramm. 238; cp. Pseudo-Symeon, 666); Genesios, 94–7; Cont. Th. 179–83. Skylitzes (Cedrenus 163–5) abbreviates from Cont. Th. (but 165, he adds πρὸς Μελιτηνή as the goal of the Emir's son, perhaps suo marte; and he interchanges the names of the river and the meadow). Zonaras, xvi. 3, 16–29 (ed. Büttner-Wobst, 396–7) depends on Skylitzes.

¹⁹ George Mon. ed. Bonn, 824, c. 16. The notice is separated from the account of the battle of Lalakaon by a notice of Michael's expedition against the Bulgarians.

²⁰ Poson is the only one of the local names mentioned by both Genesios and the Continuer. Lalakaon (εἰς τὸν Λαλακάωνα) is mentioned by the Logothete. None of the names are found, so far as I know, in other contexts. Ramsay refers to a correspondent of Photios, , Θεόδοτος καγὰ (Ep. 63, ed. Valettas, p. 367)Google Scholar. This means that Theodotos belonged to the Lalakon family, members of which we meet elsewhere in history (e.g. Const. Porph, . De adm. imp. c. 45, p. 199Google Scholar; Nicetas, Vita Ignatii, in Harduin, , Conc. v. p. 964Google Scholar). Of course a local connexion of the name is possible.

²¹ This pun shows that Πόσων is the correct form of the name, not Πόσων as it is written in the text of Genesios.

²² Byzantinische Studien, 157.

^22a διοδεὑσας ἀπὸ ᾿Αμινσοῦ

²³ Hist. Geog. Map between pp. 196 and 197.

²⁴ George Mon. ed. Bonn, 825.

²⁵ I measured on Anderson's map of Asia Minor. The distance comes to 450 Roman miles, but if we allow for twistings of the roads and gradients, it must be considerably more.

²⁶ Baron Rosen (Tabari, ed. De Goeje iii. 1509) thought that the three letters may be read k, r, n; and Vasil'ev suggests (202, n. 5) that the name may correspond to Γῦριν—We are unable to say how Petronas reached the scene of the battle. N”o doubt, he set out from his own Thracesian Theme (the anecdote in Cont. Th. 180 makes him visit Mt. Latros just before he started), and the most direct way to intercept Omar would have been by Archelais. The Thracian, Macedonian, and tagmatic troops, marching by Dorylaion and Amorion, might have joined him, e.g. at Tyriaion or Laodicea Kekaumene.

²⁷ Tabari, 32, and confirmed, 35.

²⁸ If we may assume that there had been about the same distance between the camps, then, if we suppose that Mutasim was encamped somewhere near Malakopaia, in Matamir, Ashnas would have been in the neighbourhood of Soandos.

²⁹ Tabari, 36.

³⁰ Cp. Ramsay, , Hist. Geog. 329–30Google Scholar.