^{page 19 note 1} Ed. de Goeje, and de Jong, (Fragmenta Historicorum Arabicorum Vol. I., Leiden 1869)Google Scholar.

^{page 19 note 2} The mere fact that no exact dates are given tends to show that the chronology of the siege was unknown to the Arab writers.

^{page 19 note 3} Arabic translation in Brit. Mus. MS. Or. 4402.

^{page 20 note 1} The divergences can hardly be explained by supposing that Theophanes drew throughout from his western source, for the long narrative under AM 6208 down to Leo's accession is not in Nikephoros, and can scarcely have been wholly omitted by him, if he found it in his authority.

^{page 20 note 2} Michael makes certain messages pass between ‘Umar and the army before its retreat (see p. 29 note 7), so that the expression of Theophanes (p. 29 note 5) is perhaps a loose one.

^{page 20 note 3} Owing to the variation between Theophanes and Nikephoros the exact date at which the siege began cannot be determined.

^{page 20 note 4} Theoph. AM 6232.

^{page 20 note 5} I.e. without omissions. I do not think it necessary to repeat over again the opening and concluding sections, which were given in full in the previous article.

^{page 21 note 1} According to Theoph. AM 6206 the preparations for the expedition began before the death of Al Walid.

^{page 21 note 2} For the use of naphtha in sieges see Dio Cass. 36, 3, 1; 75, 11, 4; Proc. de Bell. Goth. 4, 11. In all these cases however it was used by the defenders, and I do not know another instance of its use in attack.

^{page 21 note 3} Here the city of the Slavs appears in its right place; hence the note in J.H.S. xviii. p. 194 may be corrected.

^{page 21 note 4} As there are no points over the last letter, it might also be read ‘Afik,’ and so de Goeje prints; but Aphaka in Phoenicia is absurdly out of place. Even Epiphaneia (in Cilicia) seems to be too far back.

^{page 21 note 5} This must be the winter of 715–6. Solomon's accession was in Feb. 715, while in 716 we know from Theophanes and Michael that Maslama was in Asia Minor.

^{page 21 note 6} I.e. General of the Anatolikoi. Ibn Khurdadhbah (ed. de Goeje p. 109) also calls the Anatolic general ‘patrician of Amorion.’

^{page 21 note 7} This name is probably only an inference from the fact that the name of Leo's son was Constantine. Similarly our author calls Leo V. ‘son of Constantine,’ though we know that his father's name was Vard.

^{page 21 note 8} ‘And, when the Arab army reached the city of Amorion, Leo met them and made an agreement with them with regard to the capture of the city’ Mich. fol. 264 v; ‘promised to help them to take the royal city’ Greg. p. 116.

^{page 21 note 9} The reason for negociating with Leo was his enmity to Theodosius ῾ἀκούων δὲ ὁ Μασαλ μᾶς τὴν ἔχθραν, ἤν εἶχε Θεοδ ὁ βασιλεὺς πρὸς τὸν στρατηγόν βουλόμενος τοῦτον δελεάσαι κἀ εἰρηνεῦσαι μετ᾿ αὐτοῦ καὶ δι᾿ αὐτοῦ τὴν ῾Ρωμανίαν ὑποτάξαι (Theoph. AM 6208).

^{page 21 note 10} ἐκ τὴς Γερμανικέων καταγόμενος Theoph. AM 6209.

^{page 21 note 11} Perhaps in 698.

^{page 22 note 1} This is a striking confirmation of the conjecture of Prof.Bury, (History of the later Roman Empire, vol. 2, p. 380)Google Scholar that Leo could speak Arabic. If he remained in Germanikeia after the Arabic occupation, which was probably in 695 (J.H.S. xviii. pp. 189, 207), it is easily explained without adopting the suggestion put by our author into the mouths of the Amorians that he was a Nabataean. Theoph. (AM 6209) makes him remove to Thrace before 695; but, even so, Germanikeia as a frontier-town must have, been in frequent intercourse with the Arabs for fifty years before that time.

^{page 22 note 2} This was in 705 (Theoph. l.c.) ῾ἐρχο μένου αὐτοῦ μετὰ τῶν Βουλγάρων ὐπήντησεν αὐτῷ μετὰ δώρων προβάτων φ᾿ . θεραπευθεὶς δὲ ὁ ᾿ Ιου στινιαὸς σπαθάριον αὐτὸν εὐθέως πεποίηκεν

^{page 22 note 3} The appointment was made by Anastasius (713–715). Theoph. l.c.

^{page 22 note 4} According to Theoph. (AM 6208) the quarrel between Leo [and the Amorians was owing to the fact that he supported Anastasius, while they supported Theodosius (τὸ ᾿ Αμώριον . . . πρὸς τὸν στρατηγὸν ἐν ἔχθρᾳ διακει μένον διὰ τὴν πρὸς ᾿ Αρτέμιον συμμαχίαν) According to Michael (fol. 264 v) it was owing to his dealings with the Arabs. It is not likely to have happened on his first appointment to the office of general.

^{page 22 note 5} The attack on Amorion was made by Solomon according to Theoph. (l.c.), who does not bring Leo into the presence of Maslama at all.

^{page 22 note 6} It is here clear, as de Goeje points out, that some words introducing the revolt against Anastasius have dropped out.

^{page 22 note 7} ῾ ἀπράγμονά τε καὶ ἰδιώτην᾿ Theoph. AM 6207; cf. Nikeph. p. 51. ἦν δὲ Θεοδ. . . . ἀπράγμων. . .ἀνὴρ καὶ πρὸς πραγμἁτων διοίκησιν καὶ ταῦτα βασιλείας σφὁδρα γε ἀποπεφυκώς᾿ Zon. (ed. Bonn), 14, 28, 1.

^{page 22 note 8} Theoph. AM 6209. Nikeph. p. 52.

^{page 23 note 1} Theoph. l.c.; cf. Nikeph. p. 53.

^{page 23 note 2} There is some confusion here, since it is clear that this account is correct only if the description begins from the south.

^{page 23 note 3} It is clear that these words must be inserted with de Goeje from Ibn Khurdadhbah, who in his ‘Book of Roads’ has an account of Constantinople similar to this (ed. de Goeje p. 104).

^{page 23 note 4} Or, during his (Leo's or Maslama's) continuance at ‘Ammuriya.’ If the rendering given above is the right one, we have here one of the instances of confusion in the narrative.

^{page 23 note 5} An error for ‘Tarakiya’ (de Goeje).

^{page 23 note 6} This is in all probability taken from either Al Wakidi or Al Madaini, and ‘at the present day’ therefore means soon after 800.

^{page 24 note 1} These negotiations are probably the same as those recorded in the Arabic Gregory: see p. 28 note 5. Our author by combining several different versions has produced a somewhat inconsequent narrative.

^{page 24 note 2} This sentence seems quite out of place here and must have come in from some other account: see last note.

^{page 24 note 3} Cf. Mich. fol. 264 v ‘Maslama told Leo that, when he took Constantinople, he would make him king over the Romans.’ This was while Maslama was still at Amorion. Theoph. (AM 6208) makes the writers of the letter Solomon and Bakhara:

^{page 24 note 4} The idea is that Leo by submitting to Maslama had put himself in the position of a freedman with regard to him.

^{page 24 note 5} Michael (l.c.) represents this as taking place at Amorion: ‘The people of the city (Amorion) were afraid of Leo; and, when Leo approached the wall, he told them that he was dealing treacherously with the Arabs.’

^{page 24 note 6} This Solomon is not mentioned by any other Arab writer, but only by Theoph., who makes him conduct the negotiations with Leo at Amorion (AM 6208).

^{page 24 note 7} Cf. J.H.S. xviii. pp. 200, 202; Theoph. AM 6231.

^{page 25 note 1} This is de Goeje's correction. The MS. has ‘Maslama went out‘and said to them.’

^{page 25 note 2} Even this absurd story is not altogether without basis. According to Theoph. (AM 6208) Maslama avoided ravaging the territory under Leo's governorship in the belief that he was a friend to the Arabs, and Leo was careful to protract the negotiations until Maslama had passed beyond his territory. This would of course limit the amount of provisions in the army. All authorities agree that Leo in some way tricked the Arabs.

^{page 25 note 3} This really refers to his proclamation at Amorion in 716; (Theoph. l.c.).

^{page 26 note 1} Solomon's death is not mentioned by Theoph., but he has nothing inconsistent with it, for the Solomon who brought the fleet to Constantinople in Sept. 717 (Theoph. AM 6209; Nikeph. p. 53) must be a different person from the Solomon who commanded the army before Amorion. By the eastern writer followed by Theoph. and Michael the second Solomon seems to have been confused with the Caliph, for Theoph. makes Maslama summon Solomon a word which in Theoph. always stands for the Caliph, and Michael (fol. 264 v.) makes Solomon ‘the king’ come and encamp at Chalkedon. Much confusion in the narratives is probably due to the existence of these three Solomons; see p. 30, note 6. The death of Solomon the son of Mu‘adh probably happened before the siege began, and may be attributed to his having allowed Leo to slip through his hands at Amorion (Theoph. AM 6208).

^{page 26 note 2} Tessarakontapechys. In the Acts of the 7th Synod (Mansi 13, pp. 197–200) a Jew of this name is stated to have advised Yazid II. (720–724) to issue his decree against images and to have been put to death by Al Walid II. (743–744). From this passage it seems not improbable that the Synod was mistaken, and that he was an adviser not of Yazid but of Leo. There is however nothing against supposing that he was by origin a Jew of Tiberias, as the Synod states, which would explain his being chosen to negotiate with the Arabs. Constantine Serantapechos, brother-in-law of the Empress Eirene (Theoph. AM 6291), was probably a descendant, for the unwieldy and ill-sounding name would naturally be shortened.

^{page 27 note 1} It seems clear that this proposal must have been made at an earlier stage than that mentioned above (p. 26). Gregory, (Chron. Arab. ed. Salihani, p. 196Google Scholar) in fact places it before the negotiations of the patricians with Leo.

^{page 27 note 2} This is perhaps a reference to the storm which according to Theoph. AM 6210, Nikeph. p. 55 attacked the Arabs on their retreat.

^{page 28 note 1} The first winter (716–7) according to Theoph. (AM 6208) was spent in Asia Cf. AM 6209 Theoph. nowhere states what these were, and the Arabic narratives therefore form a useful supplement.

^{page 28 note 2} Gregory, (Chron. Arab. p. 197)Google Scholar makes the siege last 30 months, while the Spanish Chronicle of 741 (Mommsen, , Chron. Min. vol. 2, p. 355Google Scholar) makes it last two years. Such differences may be due to varying interpretations of the term ‘siege.’

^{page 28 note 3} Theoph. AM 6209; cf. Nikeph. p. 53.

^{page 28 note 4} The text is here corrupt: I adopt de Goeje's second suggestion and insert this verb.

^{page 28 note 5} Al Madaini (ap. Al Tab., see below) places this after Leo's accession; and that this was the original account appears from the fact that the narrative there goes straight on, ‘This was done in the night, and in the morning Leo fought. The same words occur in our author, but with a sentence taken from Al Wakidi in between, and the literal meaning of the verb, ‘to do a thing in the morning,’ is therefore lost. Gregory, (Chron. Arab. pp. 196, 197Google Scholar) makes Leo induce Maslama to absent himself for a time and relax the siege on the understanding that he would surrender the city. He then gets himself made Emperor and carries off the corn during Maslama's absence.

^{page 28 note 6} There is some corruption in this sentence, but the meaning is clear.

^{page 28 note 7} Mich. (fol. 265 r) ‘they ate dead bodies and dung.’ Chron. of 846 ‘they ate the flesh and the dung of their draught-animals.’ Chron. of 775 ‘their cattle and horses.’

^{page 29 note 1} This is perhaps a confusion with Anastasius, who was beheaded after a rebellion in 719 (Theoph, AM 6211; Nikeph. p. 55).

^{page 29 note 2} We should perhaps, as de Goeje suggests, read ‘Salunik,’ which is not a great departure from the text. Anastasius was banished to Thessalonike by Theodosius (Theoph. AM 6207; Nikeph. p. 52).

^{page 29 note 3} Theoph. l.c.; cf. Nikeph. p. 51.

^{page 29 note 4} Theoph. AM 6209.

^{page 29 note 5} So Theoph. (AM 6210) This is however quite inconsistent with Theophanes' own chronology, for he makes the siege last from Aug. 15, 717 to Aug. 15, 718 and places the death of Solomon on Oct. 8, 717. Nikephoros also makes the siege end on Aug. 15, 718 (p. 55), but states that it lasted 13 months (p. 53), and therefore supposed it to have begun in Jul. 717.

^{page 29 note 6} The subject of this and the following sentence must be not the Caliph but the governor.

^{page 29 note 7} According to Mich. (fol. 265 r; cf. Greg. p. 117) ‘Umar sent to ask for news of the army, and Maslama falsely answered that he was on the point of taking the city. ‘Umar however heard the truth from the messengers and sent an order to Maslama to return, and, if he did not obey, the messengers were to order the troops to return.

^{page 29 note 8} Al Wakidi b. 747 d. 822. For the beginning of his narrative see J.H.S. xviii. p. 195 1. 18. The narrative here given follows upon the introductory sentence of Al Tab. in J.H.S. xviii. p. 195 1. 24–28.

^{page 29 note 9} The ‘mud’ is variously reckoned as 1⅓ pints and 2 pints.

^{page 30 note 1} Mahomet the Khuwarizmi (circ. 833) ap. El. Nis. (Abh.für die Kunde des Morgenlandes 8, 3, p. 122) ‘The Arabs sowed fields and reaped them and ate of what they had sown.’

^{page 30 note 2} For the conclusion of Al Wakidi, narrative see J.H.S. xviii. p. 196Google Scholar 1. 6–15.

^{page 30 note 3} Al Madaini b. 753 d. circ. 840.

^{page 30 note 4} This conversation is unintelligible as it stands, and is clearly an unreasoning summary of a longer account, perhaps the same as that from which the author of the Khitab got his narrative of the conversation between Ibn Hubaira and Tessarakontapechys.

^{page 30 note 5} Here again the abrupt transition cannot have been in the original narrative.

^{page 30 note 6} This, though also quoted from Al Madaini, is clearly a different account altogether from the preceding. There is a somewhat similar story in Mich. fol. 264 v, Greg. p. 116, where it is stated that the Caliph Solomon encamped at Chalkedon with 12,000 men, and Leo, hearing that Theodosius had arrested some of his followers, came to him, whereupon Solomon gave him 6,000 men and sent him to Amorion. The origin of this story is probably Leo's visit to the general Solomon before Amorion (Theoph. AM 6208).