constructing the seventh century ORIENT ET MÉDITERRANÉE (UMR 8167) – BYZANCE COllègE DE FRANCE – INsTITUT D’ÉTUDEs BYZANTINEs TRAVAUX ET MÉMOIRES Fondés par Paul lemerle Continués par gilbert Dagron Comité de rédaction : Jean-Claude Cheynet, Vincent Déroche, Denis Feissel, Bernard Flusin, Constantin Zuckerman secrétariat de rédaction, relecture et composition : Emmanuelle Capet, Artyom Ter-Markosyan Vardanyan ©Association des Amis du Centre d’Histoire et Civilisation de Byzance – 2013 IsBN 978-2-916716-45-9 IssN 0577-1471 COllègE DE FRANCE – CNRs CENTRE DE RECHERCHE D’HIsTOIRE ET CIVIlIsATION DE BYZANCE TRAVAUx ET MÉMOIREs 17 constructing the seventh century edited by Constantin Zuckerman Ouvrage publié avec le concours de la fondation Ebersolt du Collège de France et de l’université Paris-Sorbonne Association des Amis du Centre d’Histoire et Civilisation de Byzance 52, rue du Cardinal-lemoine – 75005 Paris 2013 TAblE Of cOnTEnTS Abbreviations ............................................................................................................................I Constantin Zuckerman, Preface ..............................................................................................1 Dialogica polymorpha antiiudaica (CPG 7796, olim Dialogus Papisci et Philonis iudaeorum cum monacho) 5 Patrick Andrist (avec le concours de Vincent Déroche), Questions ouvertes autour des Dialogica polymorpha antiiudaica ........................................................................................ 9 Dmitry Afinogenov, Patrick Andrist, Vincent Déroche, la recension γ des Dialogica polymorpha antiiudaica et sa version slavonne, Disputatio in Hierosolymis sub Sophronio Patriarcha : une première approche ................................................................................... 27 Patrick Andrist, Essai sur la famille γ des Dialogica polymorpha antiiudaica et de ses sources : une composition d’époque iconoclaste ? .............................................. 105 Claudio schiano, les Dialogica polymorpha antiiudaica dans le Paris. Coisl. 193 et dans les manuscrits de la famille β................................................................................................ 139 Wars and disturbances 171 georges Kiourtzian, l’incident de Cnossos (fin septembre/début octobre 610) ................ Constantin Zuckerman, Heraclius and the return of the Holy Cross ................................. Denis Feissel, Jean de soloi, un évêque chypriote au milieu du viie siècle ........................... Marek Jankowiak, The first Arab siege of Constantinople .................................................. Offices, titles, and office-holders 321 Constantin Zuckerman, silk “made in Byzantium”: a study of economic policies of Emperor Justinian ...................................................................................................... Federico Montinaro, les premiers commerciaires byzantins ............................................. georges sidéris, sur l’origine des anges eunuques à Byzance .............................................. Christian settipani, The seventh-century Bagratids between Armenia and Byzantium....... Mikaël Nichanian, la distinction à Byzance : société de cour et hiérarchie des dignités à Constantinople (vie-ixe s.) ............................................................................................ The beginnings of Arab Egypt 637 Phil Booth, The Muslim conquest of Egypt reconsidered .................................................. Jean gascou, Arabic Taxation in the mid-seventh-century greek papyri ............................ Youssef Ragheb, les premiers documents arabes de l’ère musulmane ................................. Annexe : Annie Pralong, l’inscription arabe de la basilique de la plage de Kourion ...... Frédéric Imbert, graffiti arabes de Cnide et de Kos : premières traces épigraphiques de la conquête musulmane en mer Égée .......................................................................... 173 197 219 237 323 351 539 559 579 639 671 679 727 731 The forest and the steppe 759 Étienne de la Vaissière, Ziebel Qaghan identified ............................................................. 761 Michel Kazanski, The Middle Dnieper area in the seventh century: an archaeological survey .................................................................................................. 769 Rimma D. goldina, Igor Ju. Pastushenko, Elizaveta M. Chernykh, The Nevolino culture in the context of the 7th-century East-West trade: the finds from Bartym ............ 865 AbbREVIATIOnS AASS ACO ACO, ser. sec. ADAJ AnBoll AnIsl AnTard ArchOr BB BCH BgA BgU BHG BMGS BNC 1 et 2 BSA BSAC BSl. BSOAS Bull. ép. Byz. Byz. Forsch. BZ CArch CCsg CFHB CJ CPG CPR CsCO CTh Dig. DChAE Acta sanctorum. Acta conciliorum oecumenicorum, ed. instituit E. schwartz, continuavit J. straub, Berlin 1914-1940. Acta conciliorum oecumenicorum. Series secunda, ed. R. Riedinger, Berlin 1984-. Annual of the Department of antiquities of Jordan. Amman. Analecta Bollandiana. Bruxelles. Annales islamologiques. le Caire. Antiquité tardive. Turnhout Archiv Orientální. Praha. Византийский временник. Москва. Bulletin de correspondance hellénique. Paris. Bibliotheca geographorum Arabicorum. leiden. Aegyptische Urkunden aus den Königlichen (Staatlichen) Museen zu Berlin, Griechische Urkunden, Berlin 1895-. Bibliotheca hagiographica Graeca, 3e éd. mise à jour et considérablement augmentée, Bruxelles 1957. Byzantine and modern Greek studies. leeds. C. Morrisson, Catalogue des monnaies byzantines de la Bibliothèque nationale. 1, D’Anastase I er à Justinien II (491-711) ; 2, De Philippicus à Alexis III (711-1204), Paris 1970. The annual of the British School at Athens. london. Bulletin de la Société d’archéologie copte. le Caire. Byzantinoslavica : revue internationale des études byzantines. Praha. Bulletin of the School of Oriental and African studies. Cambridge. Bulletin épigraphique de la REG. Byzantion : revue internationale des études byzantines. Wetteren. Byzantinische Forschungen : internationale Zeitschrift für Byzantinistik. Amsterdam. Byzantinische Zeitschrift. Berlin. Cahiers archéologiques. Paris. Corpus christianorum. series graeca. Turnhout. Corpus fontium historiae Byzantinae. Corpus iuris civilis. 2, Codex Justinianus, recognovit P. Krüger, Berolini 1877. Clavis patrum Graecorum. Turnhout 1974-2003. Corpus Papyrorum Raineri. Wien 1895-. Corpus scriptorum christianorum orientalium. louvain. Codex Theodosianus. Corpus iuris civilis. 1, Digesta, recognovit Th. Mommsen, retractavit P. Krüger, Berolini 1908. Δελτίον τῆς Χριστιανικῆς ἀρχαιολογικῆς ἑταιρείας. Athènes. ABBREVIATIONs VII DOC I A. R. Bellinger, Catalogue of the Byzantine coins in the Dumbarton Oaks collection and in the Whittemore collection. 1, Anastasius I to Maurice 491-602, Washington DC 1966. DOC II, 1 et 2 Ph. grierson, Catalogue of the Byzantine coins in the Dumbarton Oaks collection and in the Whittemore collection. 2, Phocas to Theodosius III, 602-717. 1, Phocas and Heraclius, 602-641 ; 2, Heraclius Constantine to Theodosius III, 641-717, Washington DC 1968. DOC III, 1 Ph. grierson, Catalogue of the Byzantine coins in the Dumbarton Oaks collection and in the Whittemore collection. 3, Leo III to Nicephorus III, 717-1081. 1, Leo III to Michael III, 717-867, Washington DC 1973. DOC III, 2 Ph. grierson, Catalogue of the Byzantine coins in the Dumbarton Oaks collection and in the Whittemore collection. 3, Leo III to Nicephorus III, 717-1081. 2, Basil I to Nicephorus III, 867-1081, Washington DC 1993. DOP Dumbarton Oaks papers. Washington DOSeals 1-6 Catalogue of Byzantine seals at Dumbarton Oaks and in the Fogg Museum of Art. 1, Italy, North of the Balkans, North of the Black Sea, ed. by J. Nesbitt and N. Oikonomides, Washington DC 1991 ; 2, South of the Balkans, the Islands, South of Asia Minor, ed. by J. Nesbitt and N. Oikonomides, Washington DC 1994 ; 3, West, Northwest, and Central Asia Minor and the Orient, ed. by J. Nesbitt and N. Oikonomides, Washington DC 1996 ; 4, he East, ed. by E. Mcgeer, J. Nesbitt and N. Oikonomides, Washington DC 2001 ; 5, he East (continued), Constantinople and environs, unknown locations, addenda, uncertain readings, ed. by E. Mcgeer, J. Nesbitt and N. Oikonomides, Washington DC 2005 ; 6, Emperors, patriarchs of Constantinople, addenda, ed. by J. Nesbitt, Washington DC 2009. EEBS Ἐπετηρὶς Ἑταιρείας Βυζαντινῶν σπουδῶν. Athènes. EHB The economic history of Byzantium : from the seventh through the fifteenth century, A. E. laiou, ed.-in-chief (Dumbarton Oaks studies 39), Washington DC 2002. EI Encyclopédie de l’Islam, leiden – Paris 1913-1938. Encyclopédie de l’Islam, nouvelle édition, leiden – Paris 1954-2009. EI2 ÉtudPap Études de papyrologie. le Caire. ΕΦΣ Ἑλληνικὸς Φιλολογικὸς Σύλλογος Κωνσταντινουπόλεως. FHG GRBS IG IGLS Fragmenta historicorum Graecorum, Paris 1841-1872. Greek, Roman and Byzantine studies. Durham. Inscriptiones Graecae. Berlin 1903-. Inscriptions grecques et latines de la Syrie, Beyrouth – Paris 1929-. ИРАИК Известия Русского археологического института в Константинополе. Одесса, София. IslCult JHS JJP JNES JÖB JÖBG JournAs JRAS Mansi Islamic culture : an English quarterly. Hyderabad. The journal of Hellenic studies. london. Journal of juristic papyrology. Warszawa. Journal of Near Eastern studies. Chicago. Jahrbuch der österreichischen Byzantinistik. Wien. Jahrbuch der österreichischen Byzantinischen Gesellschaft. Wien. Journal asiatique. Paris. The journal of the Royal Asiatic Society. Cambridge. Sacrorum conciliorum nova et amplissima collectio, Florentiae – Venetiis 17591798. [réimpr. Paris 1901 et graz 1960]. VIII MEFRM MgH MIB III MIBE I MIBE II MIFAO MUSJ NC Nov. OCP ODB PBE PERF Pg PLRE PmbZ PO RE REB REG RN sB SBS sC SEG sPP Syn. CP TIB TM Zepos ZPE ZRVI ABBREVIATIONs Mélanges de l’École française de Rome. Moyen Âge. Rome – Paris. Monumenta germaniae historica. Berlin. W. Hahn, Moneta Imperii Byzantini. 3, Von Heraclius bis Leo III./ Alleinregierung (610-720) (Veröffentlichungen der numismatischen Kommission 10), Wien 1981. M. A. Metlich, Money of the incipient Byzantine Empire. Anastasius I – Justinian I, 491-565 (Veröffentlichungen des Instituts für Numismatik und geldgeschichte der Universität Wien 6), Wien 2000. W. Hahn et M. A. Metlich, Money of the incipient Byzantine Empire. Justin II – Revolt of the Heraclii, 565-610 (Veröffentlichungen des Instituts für Numismatik und geldgeschichte der Universität Wien 13), Wien 2009. Mémoires publiés par les membres de l’Institut français d’archéologie orientale du Caire. le Caire. Mélanges de l’Université Saint-Joseph. Beyrouth. The numismatic chronicle. london. Corpus iuris civilis. 3, Novellae, recognovit R. schoell, absolvit g. Kroll, Berolini 1895. Orientalia Christiana periodica : commentarii de re orientali aetatis christianae sacra et profana. Roma. Oxford dictionary of Byzantium, A. P. Kazhdan ed. in chief, New York 1991. Prosopography of the Byzantine Empire. 1, 614–867, ed. by J. R. Martindale, Aldershot 2001. Papyrus Erzherzog Rainer, Führer durch die Ausstellung, Wien 1894. Patrologiae cursus completus. series graeca, accur. J.-P. Migne, Paris 1856-1866. The prosopography of the later Roman Empire, by A. H. M. Jones, J. R. Martindale & J. Morris, Cambridge 1971-1992. Prosopographie der mittelbyzantinischen Zeit. Berlin 1998-. Patrologia Orientalis. Paris. Paulys Realencyclopädie der classischen Altertumswissenschaft, stuttgart – München 1894-1997. Revue des études byzantines. Paris. Revue des études grecques. Paris. Revue numismatique. Paris. Sammelbuch griechischer Urkunden aus Ägypten. Wiesbaden 1913-. Studies in Byzantine sigillography. sources chrétiennes. Paris. Supplementum epigraphicum Graecum. Studien zur Palaeographie und Papyruskunde, hrsg. von C. Wessely, leipzig 1901-1924. Synaxarium Ecclesiae Constantinopolitanae e codice Sirmondiano nunc Berolinensi, adiectis synaxariis selectis, Propylaeum ad Acta Sanctorum Novembris, éd. H. Delehaye, Bruxelles 1902 [réimpr. louvain 1954]. Tabula Imperii Byzantini. Wien. Travaux et mémoires. Paris. Jus Graecoromanum, cur. J. et P. Zepos, Athenis 1931. Zeitschrift für Papyrologie und Epigraphik. Bonn. Зборник радова Византолошког института. Београд. ThE fIRST ARAb SIEgE Of cOnSTAnTInOplE* by Marek Jankowiak ‫لتفتحن القسطنطينية فلنعم اأَمير أميرها ولنعم الجيش ذلك الجيش‬ Verily you shall conquer Constantinople. What a commander this will be, and what an army this will be! Saying of the Prophet Muḥammad1 No handbook of Byzantine history fails to mention the first Arab siege of Constantinople, a symbolic event that marks the turning of the tide in the first wave of Arab conquests. Usually thought to have lasted four years, from around 674 to 678, it has assumed in the eyes of Byzantinists an epochal significance. After 40 years of continuous defeats by victorious armies of Islam, the successful defence of the city of Constantine saved in extremis the Empire and, for some scholars, the European civilisation. To Georgije Ostrogorsky, “the significance of the Byzantine victory of 678 cannot be overestimated. For the first time the Arab advance was really checked and the invasion which had swept forward as irresistibly as an avalanche was now halted. In the defence of Europe against the Arab onslaught this triumph of Constantine IV was a turning point of world-wide historical importance. […] The fact that it [Constantinople] held saved not only the Byzantine Empire, but the whole of the European civilization.”2 Few historians would employ today the same language, but the narrative has been universally accepted in Western historiography. The details of the siege remain, however, shrouded in mystery: its exact dates (670–7 or 674–8?) and length (4 or 7 years?) are a matter of controversy; it is disputed whether the Arabs subjected Constantinople to a * I am grateful to Phil Booth, Denis Feissel, James Howard-Johnston, Robert Hoyland, Federico Montinaro, Cécile Morrisson, Harry Munt, Vivien Prigent, Ewa Wipszycka and the participants of her seminar, Adam Ziółkowski and Constantin Zuckerman for their help and comments. It does not follow that they subscribe to the views expressed below. All mistakes are mine. 1. Ibn Ḥanbal (d. 241/855–6), Musnad, Cairo AH 1313, iv.335; other references in S. Bashear, Apocalyptic and other materials on early Muslim-Byzantine wars : a review of Arabic sources, Journal of the Royal Asiatic Society 3rd series, 1, 1991, pp. 173–207, at p. 191 n. 123. 2. G. Ostrogorsky, History of the Byzantine state, rev. ed., New Brunswick NJ 1969, p. 125. Constructing the seventh century, ed. by C. Zuckerman (Travaux et mémoires 17), Paris 2013, pp. 237–320. 238 MAREk JANkOWIAk regular siege or only to a naval blockade; and the overall logic of events is far from clear.3 These problems can be traced back to the single source upon which the modern accounts of the siege are based, the Chronicle of Theophanes the Confessor. His narrative is beset by internal contradictions which, as I will try to show, result from Theophanes’ daring, but unsuccessful attempt at reconciling two sources he had at his disposal. I will argue that these contradictions cannot be satisfactorily solved. However, when considered separately, Theophanes’ two sources tell a consistent story of the Byzantine-Arab warfare in the 660s and 670s, arguably the most obscure decades in Byzantine history. In the second part of this paper, we will see that the picture emerging from their reappraisal is confirmed by independent sources—Greek, Latin, Syriac, and Arabic. All these accounts are, unfortunately, fragmentary, but even though none of them tells the full story of the first Arab siege of Constantinople, they all speak of important events that took place around the year 668. We will see that the early Islamic historical tradition, insofar as it can be accessed in a form preceding its classical codification by historians such as al-Ṭabarī, preserves a trustworthy account of the events. But the key to solving the riddle of the first Arab siege of Constantinople is offered by a contemporary and hitherto unnoticed Byzantine text with which it is fitting to start. Proem: the declaration of the CHARTOPHYLAX George The only explicit mention of the first Arab siege of Constantinople in a contemporary Byzantine source comes from the Acts of the Sixth Ecumenical Council held in 680–1 in Constantinople. On 28 March 681, at the thirteenth session of the council, George, the chartophylax of the patriarchate of Constantinople, was asked to search in the patriarchal archive for documents that would cast light on the theological positions of three ephemeral patriarchs who presided over the church of Constantinople in the first decade of Constantine IV’s reign (668–85), in particular with respect to the central question of the will of Christ. On his return from the chartophylakion, in front of a hundred of bishops and dignitaries, he made the following deposition: I have got here the codex, my lord, that is the register containing copies of the synodical letters written by Thomas [667–9], John [669–75] and Constantine [675–7] of holy memory, late patriarchs of this imperial city. In addition, I have also the original sealed synodical letter made by Thomas of holy memory, the late patriarch, to Vitalian of holy memory, the late pope of Rome, which, when he wanted to send it to him, he was unable 3. The study of A. N. Stratos, Siège ou blocus de Constantinople sous Constantin IV, JÖB 33, 1983, pp. 89–107, was rightly judged by C. Mango as “not entirely satisfactory”: Nikephoros, patriarch of Constantinople, Short history, text, transl., and commentary by C. Mango (CFHB 13), Washington 1990, p. 194. Virtually all recent accounts of Byzantine history refer to a siege of Constantinople in the 670s (see below, note 31); a rare example of questioning the traditional account is J. HowardJohnston, Witnesses to a world crisis : historians and histories of the Middle East in the seventh century, Oxford 2010, esp. p. 304, and now also J. Howard-Johnston, The Mardaites, in Arab-Byzantine coins and history : papers presented at the 13th seventh century Syrian numismatic round table held at Corpus Christi College, Oxford on 10th and 11th September 2011, ed. by T. Goodwin, London 2012, pp. 27–38. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 239 to do so because, as you know, of the long-lasting incursion by the godless Saracens and their presence throughout the two years when he was a bishop.4 The sealed synodical letter of Thomas was then opened in front of the assembly, read out, compared with the copy in the patriarchal register, and confirmed as authentic.5 Having been judged by the council to be in accordance with the newly endorsed Dyothelete creed, it has not been included in the Acts and is lost, except for the address and the incipit.6 According to the chartophylax George, the reason why the sealed, ready-for-despatch letter of the patriarch Thomas to Pope Vitalian was still kept, 14 years later, in the patriarchal archive was an Arab incursion that targeted Constantinople and lasted for the two years of the patriarchate of Thomas, from 667 to 669. George refers to this attack as a fact well-known to the audience of bishops and top-ranking officials (ὡς ἐπίστασθε). It does not follow from his words that there was no other siege of Constantinople at a later date, but he would probably not have chosen the word “long-lasting” (ἐπίμονος) to describe the Arab attack during the patriarchate of Thomas, if it was followed by an even longer siege of Constantinople in the 670s. The chartophylax implies that his audience remembered the threat posed to the Byzantine capital by an Arab army between 667 and 669 as the main Arab attempt to conquer Constantinople. The declaration of the chartophylax George raises some textual problems. The key sentence—διὰ τὴν γενομένην, ὡς ἐπίστασθε, τῶν ἀθέων Σαρακηνῶν ἐπίμονον καταδρομὴν καὶ παράστασιν εἰς τὸν διετῆ χρόνον, ὃν ἐπεσκόπησεν—has elicited comments and conjectures on the part of modern editors and interpreters. On the one hand, ἐπίμονον (“protracted, long-lasting”7) has been emended by Rudolf Riedinger, the editor of the Acts of the Sixth Ecumenical Council, to ἐπικείμενον (“imminent”), even though taken together with καταδρομή (“incursion, inroad”) it gives the satisfactory meaning of a “long-lasting incursion”. Riedinger was probably influenced not only by the traditional 4. ACO, ser. sec. II, pp. 61220–6145: τὸ κωδίκιον, δέσποτα, ἤτοι ῥέγεστρον τὸ ἔχον τὰ ἰσότυπα τῶν γενομένων συνοδικῶν ἀπό τε Θωμᾶ, Ἰωάννου καὶ Κωνσταντίνου τῶν ἐν ἁγίᾳ τῇ μνήμῃ γενομένων πατριαρχῶν τῆς βασιλίδος ταύτης πόλεως ὕπεστί μοι ἐνταῦθα, πρὸς τούτοις καὶ τὰ αὐθεντικὰ συνοδικὰ βεβουλλωμένα τὰ γενόμενα ὑπὸ Θωμᾶ τοῦ ἐν ἁγίᾳ τῇ μνήμῃ γενομένου πατριάρχου πρὸς Βιταλιανὸν τὸν ἐν ἁγίᾳ τῇ μνήμῃ γενόμενον πάπαν Ῥώμης, ἅτινα καὶ ἀποστεῖλαι πρὸς αὐτὸν βουληθέντος αὐτοῦ διὰ τὴν γενομένην, ὡς ἐπίστασθε, τῶν ἀθέων Σαρακηνῶν ἐπίμονον καταδρομὴν καὶ παράστασιν εἰς τὸν διετῆ χρόνον, ὃν ἐπεσκόπησεν, ἐκπέμψαι οὐκ ἴσχυσε. 5. ACO, ser. sec. II, pp. 61417–19 and 6161–3: λαβὼν Ἀγάθων ὁ εὐλαβέστατος ἀναγνώστης καὶ νοτάριος τοῦ ἁγιωτάτου πατριάρχου Κωνσταντινουπόλεως τὸ αὐθεντικὸν συνοδικὸν γράμμα τοῦ ἐν ἁγίᾳ τῇ μνήμῃ Θωμᾶ βεβουλλωμένον, καὶ τῆς ἐν αὐτῷ ἐπικειμένης βούλλας ἐπὶ πάντων ἀφαιρεθείσης ἀνέγνω. […] Ὅπερ συνοδικὸν ἀντεβλήθη πρὸς τὸ ῥέγεστρον τὸ προκομισθὲν παρὰ Γεωργίου τοῦ θεοσεβεστάτου διακόνου καὶ χαρτοφύλακος, καὶ ἐστοίχησεν ἀπὸ ἀρχῆς ἕως τέλους ἀπαραλείπτως. 6. ACO, ser. sec. II, p. 61421–23: τῷ τὰ πάντα ἁγιωτάτῳ καὶ μακαριωτάτῳ ἀδελφῷ καὶ συλλειτουργῷ Βιταλιανῷ Θωμᾶς ἀνάξιος ἐπίσκοπος ἐν κυρίῳ χαίρειν. οὗ ἡ ἀρχή· Ὁ μὲν ἐξ οὐκ ὄντων εἰς τὸ εἶναι παραγαγὼν τὰ σύμπαντα. 7. For a similar contemporary usage of the substantive τὸ ἐπίμονον, see P. Lemerle, Les plus anciens recueils des miracles de saint Démétrius. 1, Le texte, Paris 1979 (henceforth Mirac. Dem.), ii.4.247, pp. 21231–2131: τὸ ὕδωρ στενωθέντας διὰ τὸ ἐπίμονον τῶν βαρβάρων, paraphrased by Lemerle: “l’eau vint à manquer, du fait de la pression ennemie” (p. 201); see also ii.4.278, p. 22012–13. 240 MAREk JANkOWIAk chronology of the siege of Constantinople, but also by the very early Latin translation of the Acts which reads: propter impiorum Sarracinorum incursum assidue imminentem, ut nostis, et obsessionem quae facta est per biennium, quo gessit episcopatum.8 But ἐπίμονον is paralleled in the Latin text not merely by imminentem, but by assidue imminentem (“continually threatening”), a syntagm underscoring the length of the threat posed by the Saracen incursus. This nuance is absent from the Greek adjective ἐπικείμενος, but not from ἐπίμονος—Riedinger’s emendation is thus unnecessary. The second difficulty concerns the word παράστασις (“proximity, presence, being beside”), rendered in Latin as obsessio (“siege”). The meaning of “siege” is not attested for the Greek παράστασις,9 even if this word implies a physical presence rather than a mere “threat”, as suggested by one commentator.10 In the otherwise literal Latin translation of the Acts the somewhat inadequate rendering of παράστασις by obsessio may be due to the fact that the Latin translator, at work only several years after the council, made his text more precise than the original, based on an independent knowledge of the events. It is not necessary, at any rate, to tamper with the Greek text of the declaration of the chartophylax George, who spoke in unambiguous terms of an Arab blockade of Constantinople throughout the two years of the patriarchate of Thomas. Thomas was patriarch from 17 April 667 to his death on 14 or 15 November 669. The starting day of his patriarchate, the Holy Saturday, is given by the catalogues of the patriarchs of Constantinople; the year results from the solidly established chronology of the patriarchs of Constantinople in the 7th century.11 The catalogues add that Thomas was consecrated after an unusual vacancy of over 6 months, no doubt necessary to obtain the approval of Constans II, who then resided in Syracuse, for the new patriarch. All the catalogues assign to Thomas a patriarchate of 2 years and 7 months;12 this is 8. ACO, ser. sec. II, p. 6152–4. The very accurate Latin translation of the Acts dates from the first years after the council, see Riedinger’s introduction to ACO, ser. sec. II, vol. 1, pp. ix–xii. 9. H. G. Liddell, R. Scott, Greek-English lexicon, with a revised supplement, Oxford 1996, p. 1325; G. W. H. Lampe, A patristic Greek lexicon, Oxford 1961, pp. 1025–6. 10. J. L. van Dieten, Geschichte der Patriarchen. 4, Von Sergios I. bis Johannes VI. (610–715), Amsterdam 1972, p. 117 and n. 4, translates παράστασις by “Bedrohung”. 11. F. Fischer, De patriarcharum Constantinopolitanorum catalogis, Lipsiae 1884, p. 2896–10: Θωμᾶς διάκονος τῆς αὐτῆς ἐκκλησίας Κπόλεως, νοτάριος, καγκελλάριος, ῥαιφερενδάριος, χαρτοφύλαξ, σκευοφύλαξ, γηρωκόμος τοῦ γηρωκομείου τῆς Σκάλας καὶ πτωχοτρόφος τοῦ ἐν Νεαπόλει πτωχείου ἐχειροτονήθη τῷ μεγάλῷ σαββάτῷ καὶ ἐπεσκόπησεν ἔτη βʹ μῆνας ζʹ. The chronology has been established by E. W. Brooks, On the lists of the patriarchs of Constantinople from 638 to 715, BZ 6, 1897, pp. 33–54 (see pp. 47–8 on Thomas), and corrected for the second patriarchate of Pyrrhus (in 654, not 655) by van Dieten, Patriarchen (quoted n. 10) [see pp. 117–20 on Thomas]. Pace HowardJohnston, Witnesses (quoted n. 3), pp. 492–3 n. 13, and Id., The Mardaites (quoted n. 3), pp. 31–2, the patriarchal chronology in the 7th c. is guaranteed by many synchronisms with securely dated events; it is thus impossible to shift the patriarchate of Thomas to 669–71. 12. On the vacancy, see Fischer, De catalogis (quoted n. 11), p. 2895: καὶ δία μέσου ἐχήρευσεν ὁ θρόνος μῆνας ςʹ ἡμέρας δʹ. The length of Thomas’ patriarchate is also given by Nikephoros kallistos, De episcopis Byzantii, PG 147, col. 457B (who has a vacancy of 6 months and 16 days); Nikephoros, Chronographicon syntomon, p. 11819–20 de Boor; Zonaras XIV 19.27 (p. 22016–18 BüttnerWobst); Synopsis chronike, ed. C. Sathas, Bibliotheca graeca medii aevi, vol. 7, Paris 1894, p. 11125–28. Theophanes (p. 34821 de Boor) rounded his tenure up to three years. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 241 independently confirmed by the day of his death attested by the synaxaria.13 It follows that George correctly stated the length of Thomas’ patriarchate. Although the testimony of the chartophylax George on the Arab attack against Constantinople in 667–9 comes from an ocular witness and was recorded a little more than a decade after the events, it has found little favour with the historians. Before it was largely forgotten in the last century,14 it had been contested on the grounds of its incompatibility with the canonical account of Theophanes. Historians easily found the reason that incited the chartophylax to invent an inexistent siege: the usual reluctance of the patriarchs of Constantinople to communicate with the Roman popes. E. W. Brooks thought that the chartophylax George rounded the 2 years and 7 months of Thomas’ patriarchate down to two years, rather than up to three years, in order to “make the episcopate appear as short as possible”, and thus “to excuse Thomas for not communicating with the Pope”. He concluded that “the statement that during the whole of his episcopate the Saracen blockade prevented him from entering into communication with Rome is of course not true: the receipt of the news of the Emperor’s death and the expedition of his successor to Sicily in the latter half of 668 is proof to the contrary”.15 Even though Brooks later came to doubt the reality of the expedition of Constantine IV to Sicily and allowed for a kernel of historical truth in the statement of the chartophylax George, his initial opinions have never been challenged.16 Brooks did not realize the importance of ecclesiastical communion between the patriarchal sees of the Empire for preserving the cohesion of the imperial church. Synodical letters addressed by newly consecrated patriarchs to their colleagues made known their accession and their definition of faith.17 Their acceptance or refusal signified unity or schism within the Chalcedonian church. The evidence for regular exchanges of synodical letters between the patriarchates is particularly rich during the Monothelete controversy in the 7th century. In 635, the bishops of Rome and Constantinople rejected the lengthy synodical letter of Sophronius of Jerusalem because of its transparent allusions to the doctrine of two activities in Christ.18 Twenty years later, in the context of severely strained relations with Rome, Peter, the new patriarch of Constantinople, did not hesitate 13. The synaxaria disagree on the precise day, 14 or 15 November, see Syn. CP, col. 22341–42, 227–846–47, and the note at col. 966. 14. The most recent commentary on this text I am aware of is that of E. W. Brooks, The Sicilian expedition of Constantine IV, BZ 17, 1908, pp. 455–9, at p. 457, quoted below, note 16. 15. Brooks, On the lists of the patriarchs (quoted n. 11), p. 48. The idea was voiced already by C. Baronius, Annales ecclesiastici una cum critica historico-chronologica P. Antonii Pagii, vol. 11, Lucae 1742, s.a. 656, p. 490, who based it, however, on erroneous chronology, as seen already by G. Cuperus, Tractatus historico-chronologicus de patriarchis Constantinopolitanis, Venetiis 1751, p. 82. 16. Brooks, The Sicilian expedition (quoted n. 14), p. 457: “it was stated at the Synod of 680 that the patriarch Thomas […] was prevented from communicating with the Pope by the Saracen blockade; and, though this need not be strictly true, the statement could not have been made if the attack had not taken place during his episcopacy”. Brooks did not draw the implications of his new position. 17. See, for instance, P. Allen, Sophronius of Jerusalem and seventh-century heresy, Oxford 2009, pp. 47–51. 18. This results from the recourse of Sophronius to the intermediation of Arcadius of Cyprus when contacting Sergius of Constantinople and Honorius of Rome, see S. Brock, An early Syriac Life of Maximus the Confessor, AnBoll 91, 1973, pp. 299–346, at p. 315 § 8. 242 MAREk JANkOWIAk to “send a synodical letter to the Apostolic See according to the custom”, but “the people and clergy” of Rome forced the pope to reject it.19 The next pope, Vitalian, marked the reconciliation with the church of Constantinople by sending a synodical letter to Peter, which was enthusiastically accepted by the latter.20 When Rome reconsidered its theological stance once again in the 670s, the patriarchs of Constantinople did not desist from sending their synodical letters to the popes. Only after they had been twice rejected, a new patriarch of Constantinople decided to address a less formal letter to the pope—but even then he still conformed to the custom of establishing epistolary contact with his Roman colleague.21 It is thus unlikely that Patriarch Thomas was the only 7th-century bishop of Constantinople to have neglected the customary sending of the synodical letter to Rome: if he did not despatch the notification of his consecration to Rome, it was because of a force majeure. The testimony of the chartophylax George can be trusted: the Byzantine capital came under Arab attack in 667–9. We will see that many other sources confirm the reality of a lengthy Arab incursion at that time. Why, then, do modern historians disagree and almost unanimously situate the siege of Constantinople in the 670s? I. The siege of Constantinople in the 670s: a historiographical misunderstanding 1. Theophanes’ mistake Modern accounts of warfare between the Byzantines and the Arabs during the reign of Constantine IV do little more than paraphrase the early 9th-century Chronicle of Theophanes, the only Greek source to provide a chronological backbone to the history of Byzantium in the 7th century. Theophanes’ account is composite: it has long been acknowledged that, in the section on the last third of the 7th century, he weaves together two sources that can be identified and to a large extent reconstructed thanks to parallel passages in other extant sources. heophanes’ two sources One of Theophanes’ sources, for which I accept the tentative attribution to the “patrician Trajan” (which I put in inverted commas to emphasize the hypothetical character of this attribution), was also used by the patriarch Nicephorus in his Short history; 19. LP, p. 3415–10: Petrus, patriarcha Constantinopolitanus, direxit synodicum ad sedem apostolicam iuxta consuetudinem […] et accensus populus uel clerus eo quod talem synodicam direxisset, minime est suscepta, sed cum maiore strepitu est a sancta Dei ecclesia proiecta; ut etiam nec eundem papam demitteret populus uel clerus missas caelebrare […] nisi promisisset his ipse pontifex minime eam aliquando suscipere. English translation:The Book of Pontiffs (Liber pontificalis) : the ancient biographies of first ninety Roman bishops to AD 715, transl. with an introd. and notes by R. Davis (Translated texts for historians 6), rev. 3rd ed., Liverpool 2010, p. 69. 20. Maximus, epistula ad Anastasium Monachum, lines 3–8, in Scripta saeculi VII Vitam Maximi Confessoris illustrantia, ed. P. Allen et B. Neil, Turnhout – Leuven 1999 (CCSG 39), p. 161; LP, p. 3433–5; ACO, ser. sec. II, p. 6101–4. 21. See below, pp. 283–4. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 243 the other, the “Oriental source” often ascribed to Theophilus of Edessa, transmitted a significant amount of material to several Syriac and Arabic chronicles. Almost every sentence in Theophanes’ description of the siege of Constantinople is paralleled in these texts, the two sources can thus be readily separated, as shown below. Passages copied from the “Oriental source” are printed in italics, and parallel texts are specified after each item.22 AM 6162 = Constantine 2 [AD 670/1]23 In this year there was a severe cold, and many men as well as beasts suffered hardship. [Agap. p. 491; Mich. Syr. xi.13, p. 456] Phadalas wintered at Kyzikos. [no parallel24] AM 6163 = Constantine 3 [AD 671/2] In this year Bousour made an expedition and, after taking many captives, returned home. [Agap. p. 491] AM 6164 = Constantine 4 [AD 672/3] In this year, in the month of Dystros (March), a rainbow appeared in the sky, and all men shuddered and said it was the end of the world.25 [Agap. p. 491; Mich. Syr. xi.13, p. 456; Chron. 1234 § 141] In this year the deniers of Christ equipped a great fleet, and after they had sailed past Cilicia, Mouamed, son of Abdelas, wintered at Smyrna, while Kaisos wintered in Cilicia and Lycia.26 [Agap. p. 492] A plague occurred in Egypt. [Agap. p. 492] He also sent the emir Chale, inasmuch as he was a competent and bold warrior, with another fleet27 to assist them. The aforesaid Constantine, on being informed of so great an expedition of God’s enemies against Constantinople, built large biremes bearing cauldrons of fire and dromones equipped with siphons, and ordered them to be stationed at the Proclianesian harbour of Caesarius. [Nikeph. § 34.2–6, 8–9] AM 6165 = Constantine 5 [AD 673/4] In this year the aforesaid fleet of God’s enemies set sail and came to anchor in the region of Thrace, between the western point of the Hebdomon, that is the Magnaura, as it is called, and the eastern promontory, named 22. Theoph. AM 6162–6169, pp. 3531–3562. Unless specified otherwise, I use the translation of Cyril Mango in C. Mango, R. Scott, with the assistance of G. Greatrex, The Chronicle of Theophanes Confessor : Byzantine and Near Eastern history AD 284–813, Oxford 1997, pp. 492–6. I also follow, with minor modifications explained in the footnotes, his attribution of individual passages to the “Oriental source”. See p. 319 for abbreviations. 23. See below, p. 258–9, on the chronology of Theophanes. “Constantine 2” stands for the 2nd regnal year of Constantine IV, “AM”—for annus mundi i.e. year from Creation. 24. Although unattested in other sources, this notice comes in all likelihood from the “Oriental source”, Theophanes’ source for the Arab raids. 25. Theophanes condensed this notice to the point of making it unintelligible; parallel sources make it clear that the “rainbow” appeared at the third or fourth watch of the night and was probably an aurora borealis. 26. Mango – Scott, Theophanes (quoted n. 22), p. 493, do not attribute this sentence to the Oriental source, but Agapios (p. 492) also mentions a great Arab fleet between the notices on the rainbow and the plague in Egypt. 27. Mango – Scott, Theophanes (quoted n. 22), p. 493, omit the words μετὰ καὶ ἑτέρου στόλου (p. 35318 de Boor). I also modify their translation of the first part of the sentence to render the sudden change of the subject: Theophanes did not specify who sent Chale against Constantinople. 244 MAREk JANkOWIAk kyklobion. Every day there was a military engagement from morning until evening, between the brachialion of the Golden Gate and the kyklobion, with thrust and counter-thrust. The enemy kept this up from the month of April until September. Then, turning back, they went to kyzikos, which they captured, and wintered there. And in the spring they set out and, in similar fashion, made war on sea against the Christians. After doing the same for seven years and being put to shame with the help of God and His Mother; having, furthermore, lost a multitude of warriors and had a great many wounded, they turned back with much sorrow. And as this fleet (which was to be sunk by God) put out to sea, it was overtaken by a wintry storm and the squalls of a hurricane in the area of Syllaion. It was dashed to pieces and perished entirely. [Nikeph. § 34.6–8, 9–21] Now Souphian, son of Auph, the second brother,28 joined battle with Florus, Petronas, and Cyprian, who were at the head of a Roman force, and 30,000 Arabs were killed. At that time Kallinikos, an architect from Helioupolis in Syria, took refuge with the Romans and manufactured a naval fire with which he kindled the ships of the Arabs and burnt them with their crews. In this way the Romans came back in victory and acquired the naval fire. [Agap. p. 492; Mich. Syr. xi.13, p. 455] AM 6166 = Constantine 6 [AD 674/5] In this year Abdelas, the son of Kais, and Phadalas wintered in Crete. [no parallel29] AM 6167 = Constantine 7 [AD 675/6] In this year a sign appeared in the sky on a Saturday. [Mich. Syr. xi.13, p. 456] AM 6168 = Constantine 8 [AD 676/7] In this year there was a great plague of locusts in Syria and Mesopotamia. [Agap. p. 492(?);30 Mich. Syr. xi.13, p. 457] AM 6169 = Constantine 9 [AD 677/8] In this year the Mardaites entered the Lebanon range and made themselves masters from the Black Mountain as far as the Holy City and captured the peaks of Lebanon. Many slaves, captives, and natives took refuge with them, so that in a short time they grew to many thousands. [Agap. pp. 492–3; Mich. Syr. xi.13, p. 455; Chron. 1234 § 141] When Mauias and his advisers had learnt of this, they were much afraid, realizing that the Roman Empire was guarded by God. So he sent ambassadors to the emperor Constantine, asking for peace and promising to pay yearly tribute to the emperor. Upon receiving these ambassadors and hearing their request, the emperor dispatched with them to Syria the patrician John, surnamed Pitzigaudes, a man of ancient lineage in the state and possessed of much experience and excellent judgement, that he might parley suitably with the Arabs and conclude a treaty of peace. When this man had arrived in Syria, Mauias gathered a group of emirs and korasenoi and received him with great honour. After exchanging many conciliatory speeches, they mutually 28. See below, p. 248, on these words. I slightly modify the translation of Mango – Scott. 29. See above, note 24. 30. Agapios speaks of a plague of rats; the “Oriental source” seems to have mentioned two plagues, of rats and of locusts, in two consecutive years, see Mich. Syr. xi.13, p. 457. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 245 agreed to draw up a written treaty of peace under oath, stipulating a yearly tribute of 3,000 [pieces] of gold, fifty captives, and fifty thoroughbred horses to be paid by the Hagarenes to the Roman state. These conditions having been approved by both sides for a duration of thirty years, complete peace prevailed between the Romans and the Arabs. After these two written treaties had been mutually sworn and exchanged, the oft-mentioned illustrious man returned to the emperor with many gifts. [Nikeph. § 34.21–31] Theophanes tells the familiar story of the many-year Arab siege of Constantinople. Since the beginning of Constantine IV’s reign, Arab raiding parties were drawing close to Constantinople. In 670/1 a general named Phadalas wintered in kyzikos, little more than 100 km from the Byzantine capital. Two years later, the Arabs “equipped a great fleet” that wintered, on its way to Constantinople, at Smyrna and in Cilicia and Lycia. The assault against the Byzantine capital began the next spring. During seven years, fighting raged at the city walls next to the Golden Gate. From April to September, there were daily encounters “from morning until evening” “with thrust and counter-thrust”. Eventually, worn by Byzantine resistance, the Arabs lifted the siege and repaired home, only to be surprised at Syllaion by a storm. At the same time, three Byzantine patricians won a resounding victory on land and—thanks to the newly invented Greek fire—on sea. The Byzantines pressed home their advantage with the invasion of the Mardaites who “entered Lebanon”. Jerusalem itself came under threat, and the caliph Muʿāwiyah was forced to sue for peace which was concluded on terms favourable to the Empire. Half a century after the beginning of the Arab invasion, Byzantium secured its first permanent peace treaty with the Caliphate. This account, accepted by modern historians,31 appears upon closer inspection to be beset with contradictions. The “Oriental source” did not provide Theophanes with the chronological space for the seven years of siege: he describes both its beginning and the end in the same year, AM 6165. In assuming that Constantinople was besieged until AM 6168 or 6169, when the Byzantines invaded Lebanon and forced the Arabs to sue for peace, most historians shorten the siege to four or five years, from 673/4 to 676/7 or 677/8 inclusive. Others sacrifice the absolute chronology of Theophanes to preserve the length of the siege, with Phadalas’ wintering in kyzikos in AM 6162 (AD 670/1) as a convenient starting point: it falls exactly seven years before the peace treaty dated by Theophanes to AM 6169 (AD 677/8), and seems to fit with the description of kyzikos as the winter base of the Arab fleet during the siege of the Byzantine capital. But the two mentions of kyzikos come from two different sources and nothing indicates that they refer to the same event: Theophanes, at least, did not make the link. 31. Perhaps the most influential remains the account of Ostrogorsky, History of the Byzantine state (quoted n. 2), pp. 123–5. See also, to quote a handful of examples, R.-J. Lilie, Die byzantinische Reaktion auf die Ausbreitung der Araber : Studien zur Strukturwandlung des byzantinischen Staates im 7. und 8. Jahrhundert, München 1976, pp. 76–82; J. J. Norwich, Byzantium. The early centuries, London 1990, pp. 323–4; Le monde byzantin. 2, L’Empire byzantin (641-1204), sous la dir. de J.-C. Cheynet, Paris 2006, pp. 7–8; A. Louth, Byzantium transforming (600–700), in The Cambridge history of the Byzantine Empire c. 500–1492, ed. J. Shepard, Cambridge 2008, pp. 221–48, at pp. 232–3. 246 MAREk JANkOWIAk If the chronological problem can be removed by discarding either the duration of the siege or its starting date, the logic of the events reported by Theophanes is even more elusive. It is not clear how the resounding victory of the patricians Florus, Petronas and Cyprian over Sufyān b. ʿAwf relates to the siege of Constantinople. The sequence of events in the Chronicle suggests that the battle took place after the siege. But if such a powerful Arab army was retreating from the Byzantine capital, why did it lift the siege? If, however, Sufyān was leading reinforcements, should we assume that an important fraction of the Byzantine army was operating outside of Constantinople during the siege? Or was the Arab force sent to attempt another siege after the previous attack had come to naught? Theophanes provides no answer. Furthermore, was there any connection between the wreckage of the Arab fleet at Syllaion and the defeat of Sufyān on land followed by the destruction of (another?) Arab fleet by Greek fire? The conjunction δέ32 and the arrangement of notices suggest that Theophanes drew such a link, perhaps because—judging from the notice in Michael the Syrian—the “Oriental source” located the land battle next to a coastal city in Lycia, not far from Syllaion, and mentioned damage suffered by the Arab fleet in a storm,33 or perhaps because Theophanes merely needed to close his description of the siege with an important Arab defeat. At any rate, he did not succeed in connecting the two Arab disasters into a logical sequence. Other incoherences are not difficult to spot. Theophanes has Constantine IV equip the Byzantine fleet with Greek fire one year before its invention by kallinikos of Helioupolis in AM 6165. He also makes us believe that, during the siege, warfare concentrated at the walls of Constantinople, but nonetheless mentions other theatres of military operations, such as Crete, where Abdelas, the son of kais, and Phadalas wintered in AM 6166, the reported second year of the siege. If ʿAbdallāh b. Qays was one of the commanders of the “big fleet” that, according to Theophanes, laid siege to Constantinople,34 it remains to be explained why he left the winter base in kyzikos. Warfare in the 670s was apparently not restricted to the siege of the Byzantine capital. Several solutions have been proposed to this conundrum. In a rather confused study of the first Arab siege of Constantinople, Andreas Stratos followed Theophanes’ chronology, but argued for a loose blockade rather than a regular siege.35 His analysis of Theophanes’ account is, however, superficial and his wholesale dismissal of the Islamic sources, allegedly too confused to be of any use and too reluctant to acknowledge Arab defeats, flaws his argument. More recently, James Howard-Johnston came to the conclusion that “the blockade of Constantinople on the 670s is a myth which has been allowed to mask the very real success achieved by the Byzantines in the last decade of Muʿāwiyah’s caliphate”. His solution consists in dating the Arab attack on the Byzantine capital to 670–1, in association with the notice on the wintering of Phadalas in kyzikos. Despite significant Arab pressure in the following years, the Byzantines were able to gain the initiative after a 32. Theoph. AM 6165, p. 35411–13: Σουφιᾶν δέ, ὁ υἱὸς τοῦ Ἀΰφ, ὁ δεύτερος ἀδελφός, συνέβαλε πόλεμον μετὰ Φλώρου καὶ Πετρωνᾶ καὶ Κυπριανοῦ ἐχόντων δύναμιν Ῥωμαϊκήν. 33. Mich. Syr. xi.13, p. 455, quoted below, note 95. 34. On the confused names of the Arab commanders of the “big fleet”, see Mango – Scott AM 6164 n. 1, and below, pp. 271–2. 35. Stratos, Siège ou blocus (quoted n. 3); a similar argument earlier in A. N. Stratos, Byzantium in the seventh century. 4, 668–685, Amsterdam 1978, pp. 29–39. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 247 major victory in 674, and to counter-attack in the second half of the 670s.36 The general logic of this argument is enticing, but many details—in particular the circumstances of the siege of the Byzantine capital—can be reinterpreted on a wider source basis. A balanced appraisal of Theophanes’ editorial technique suggests that the story of the siege of Constantinople is neither a factual report nor a “myth”.37 Having adopted the ambitious annalistic format more systematically than his mentor George Synkellos, Theophanes was constantly confronted with the challenge, rarely faced by his fellow Byzantine historians, of establishing the absolute date of each event. For much of the 7th century, he was helped by the “Oriental source”, conveniently arranged by annual entries.38 Into this chronological framework, he inserted notices from other sources at places that he had to figure out himself. He thus placed the description of the siege of Constantinople, borrowed from the “patrician Trajan”—where it was a single narrative, as it appears in Nicephorus—in two annual entries AM 6164 and 6165, separating the preparations for the siege from the siege itself. Several years later, in AM 6169, Theophanes inserted the account of peace negotiations with Muʿāwiyah. Given that the chronicle of “patrician Trajan” lacked dates, Theophanes’ choice of these years was based on associations with dated notices from the “Oriental source”. This technique resulted in a narrative substantially modified with respect to the original texts: by arranging notices in a given order and connecting them by means of conjunctions, Theophanes created causal links absent from his sources. Theophanes, who did not strive for stylistic consistency, excerpted his multiple sources without modifying their original style:39 changes in style and clumsy transitions between individual sources mark thus the junction points. In the account of the Byzantine-Arab warfare in the reign of Constantine IV, Theophanes switched between his two sources thrice, in AM 6164, 6165 and 6169 (the fourth transition at the end of the second passage copied from the “patrician Trajan” corresponds to the end of the notice for AM 6169). In AM 6164, he associated the “great fleet” that wintered at Smyrna and in Cilicia and Lycia with the “many ships” sent under the command of the emir Chale. The transition between the sources is clumsy: Theophanes intercalated the mention of a plague in Egypt between the descriptions of the two fleets, and forgot to copy the subject of the first sentence from the new source (ἀπέστειλε δὲ καὶ Χαλὲ τὸν ἀμηραῖον μετὰ καὶ ἑτέρου στόλου πρὸς βοήθειαν αὐτῶν, ὡς ἱκανώτατον καὶ τολμηρὸν πρὸς μάχην). The context and the parallel passage in Nicephorus make it clear that the missing subject of ἀπέστειλε was “the leader of the Saracens” i.e., the caliph Muʿāwiyah.40 Nonetheless, Theophanes made 36. Howard-Johnston, Witnesses (quoted n. 3), pp. 302–4 (quotation p. 304), followed by P. Sarris, Empires of faith : the fall of Rome to the rise of Islam, 500–700, Oxford 2011, p. 292. See now, in more detail, Howard-Johnston, The Mardaites (quoted n. 3). Various points raised in that paper are addressed throughout this article. 37. On Theophanes’ use of his sources in the section on the 7th century, see my paper forthcoming in the acts of the conference on the Chronicle of Theophanes, held in Paris in September 2012. A similar interpretation of Theophanes’ editing of his two sources in Howard-Johnston, Witnesses (quoted n. 3), pp. 302–4, and Id., The Mardaites (quoted n. 3), p. 29. 38. See below, p. 258, on the format of the work of Theophilus of Edessa. 39. See below, p. 250, on the distinctive jargon of the “patrician Trajan” preserved by Theophanes. 40. Compare Theoph. AM 6164, p. 35317–19: ἀπέστειλε δὲ καὶ Χαλὲ τὸν ἀμηραῖον μετὰ καὶ ἑτέρου στόλου πρὸς βοήθειαν αὐτῶν, ὡς ἱκανώτατον καὶ τολμηρὸν πρὸς μάχην, and Nikeph. § 342–6: 248 MAREk JANkOWIAk an effort to tie together his two sources. A comparison with the parallel text of Nicephorus shows that he retouched the common source in order to create the illusion that the fleet led by Chale belonged to the same expedition as the fleet that wintered at Smyrna and in Cilicia and Lycia: he qualified the fleet commanded by Chale as “another”, and added that it was sent “in order to assist” the first fleet (πρὸς βοήθειαν αὐτῶν). Theophanes used a similar technique in AM 6165, where he associated the wreckage of the Arab fleet at Syllaion, copied from the “patrician Trajan”, with the land defeat of an Arab army, described in the “Oriental source”, although the two texts place the Arab disaster in different circumstances. Again, the transition is not smooth. Theophanes started the second item with the words: “Souphian, son of Auph, the second brother” (Σουφιᾶν δέ, ὁ υἱὸς τοῦ Ἀΰφ, ὁ δεύτερος ἀδελφός), but he did not specify whose brother Souphian was, and why this information was relevant.41 Finally, the entry concluding the story of Arab-Byzantine warfare under Constantine IV (AM 6169) represents another daring attempt at historical synthesis by Theophanes. In the parallel excerpt of the same passage in Nicephorus, the anonymous “king of the Saracens” opens peace negotiations on the news of the wreckage of the Arab fleet at Syllaion. Theophanes, however, in striving to integrate the Mardaite invasion of Lebanon from the “Oriental source”,42 concludes that it preceded the peace treaty—of which the “Oriental source” made no mention—and prompted the caliph to sue for peace. Thus, the words used by Theophanes to introduce the peace negotiations—“when Mauias and his advisers had learnt of this”43—no longer refer to the wreckage, but to the Mardaite invasion. This seemingly logical supposition was, as we shall see, probably false. Overall, despite crude editing, Theophanes skilfully pieced together his story from the two accounts at his disposal, achieving an illusion of coherence that seduced generations of scholars. He plausibly associated the big Arab fleet described in the “Oriental source” with the fleet of the emir Chale that laid siege to Constantinople. He conjectured, probably correctly, that the two descriptions of the disaster of the Arab fleet refer to the ὁ τῶν Σαρακηνῶν ἡγούμενος ναῦς πλείστας κατασκευάσας κατὰ τοῦ Βυζαντίου ἐκπέμπει, ἡγεμόνα τούτων ἐπιστήσας, ἅτε πιστότατον καὶ τὰ πολέμια ἔμπειρον, κατὰ τὴν ἑαυτῶν διάλεκτον Χαλὲβ ὀνομαζόμενον. The fact that the caliph Muʿāwiyah is never mentioned by name by Nicephorus suggests that he was anonymous in the work of the “patrician Trajan”. 41. Other sources based on Theophilus of Edessa do not help to understand the information provided by Theophanes. Islamic sources do not seem to mention a brother of Sufyān b. ʿAwf, see W. Caskel, Ǧamharat an-nasab : das genealogische Werk des Hišām Ibn Muḥammad al-Kalbī, Leiden 1966, table 218, generation 29, and Sufyān’s biography in Tārīkh madīnat Dimashq of Ibn ʿAsākir referred to below, note 174. Perhaps the words δεύτερος ἀδελφός are a mistranslation from Syriac or Arabic. 42. Mango, Nikephoros (quoted n. 3), pp. 193–4, suggested that Nicephorus connected the peace treaty with the failure of the Arab attack on Constantinople, rather than with the invasion of the Mardaites, as a result of abridgement of the source he shared with Theophanes. But the “patrician Trajan” does not seem to have mentioned the Mardaite invasion; at least the only mention of the Mardaites in Nicephorus (§ 3815–16) does not betray any knowledge of the circumstances of their appearance in Lebanon. 43. Theoph. AM 6169, p. 35510–14: ταῦτα μαθὼν Μαυΐας […] ἀποστέλλει πρέσβεις πρὸς τὸν αὐτοκράτορα Κωνσταντῖνον ζητῶν εἰρήνην, ὑποσχόμενος καὶ ἐτήσια τῷ βασιλεῖ παρέχειν πάκτα, paralleled by Nikeph. § 3421–23: ὁ δὲ τῶν Σαρακηνῶν βασιλεὺς τὸ τοῦ στόλου ἀκούσας δυστύχημα πρέσβεις ἀποστέλλει πρὸς Κωνσταντῖνον ὡς σπεισόμενος πρὸς αὐτὸν ἐπὶ τελέσμασιν ἐνιαυσίοις. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 249 same event; and he cleverly surmised that Mardaite invasion of Lebanon had triggered the caliph’s plea for peace. But the overall logic of events created by Theophanes is vague and difficult to follow: his absolute chronology contradicts the given duration of the siege; the connection between the siege and the victory of the three patricians over Sufyān b. ʿAwf is unclear; and the two versions of destruction of a major Arab force reported side by side are mutually exclusive. All these difficulties result from one mistake: Theophanes inserted the account of the siege of Constantinople copied from the “patrician Trajan” in the wrong place in the “Oriental source”. There is nothing to suggest that the fleet led by the emir Chale, which laid siege to Constantinople, was part of the same expedition as the “big fleet” that wintered on the Anatolian coast. On the contrary, when read separately, the two sources tell different stories: the “patrician Trajan” that of the siege; the “Oriental source”—that of the destruction of an invading Arab army in a battle fought possibly in Lycia. Since the latter source does not seem to have mentioned any other “big fleet”—nor, for that matter, a siege of Constantinople—Theophanes associated the two Arab expeditions faute de mieux. It is true that he could have connected the beginning of the siege and the establishment of a winter base in kyzikos with the wintering of Phadalas in kyzikos in AM 6162, or with the expedition of Yazīd, the son of the caliph, in AM 6159. But the former, merely mentioned in the “Oriental source”, was probably not prominent enough to justify a connection with the siege of Constantinople, while Yazīd was credited by the “Oriental source” only with an advance to Chalcedon. Having at his disposal only two sources, Theophanes could hardly avoid his mistake. We are in a more fortunate position, but in order to reconstruct the events it is necessary to first undo Theophanes’ work and to study his two sources separately, before recomposing them, if possible, in a more coherent way. he “patrician Trajan” The classical description of the siege of Constantinople, with its daily “thrusts and counter-thrusts” at the walls of the city “between the brachialion of the Golden Gate and the kyklobion” comes from a source used, besides by Theophanes, also by Nicephorus in his Breviarium, or Short history. The Breviarium, an œuvre de jeunesse of the patriarch composed perhaps in the 780s, appears to be a compilation of three sources, copied one after another.44 The reign of Constans II (641–68) is not described, probably because of a gap between sources that Nicephorus had at his disposal.45 Consequently, our source started with Constantine IV’s accession in 668;46 as we will see, it probably ended a little after 44. Mango, Nikephoros (quoted n. 3), pp. 8–18; Howard-Johnston, Witnesses (quoted n. 3), p. 241. 45. Mango, Nikephoros (quoted n. 3), pp. 14–5; Howard-Johnston, Witnesses (quoted n. 3), pp. 256–7. 46. Pace W. Treadgold, Trajan the Patrician, Nicephorus, and Theophanes, in Bibel, Byzanz und Christlicher Orient : Festschrift für Stephen Gerö zum 65. Geburtstag, hrsg. von D. Bumazhnov et al., Leuven 2011, pp. 589–621, who proposed to attribute to the “patrician Trajan” all the notices of the years 629–68 in Theophanes that cannot be assigned to the “Oriental source”. This hypothesis fails to convince, as it requires complex speculations to account for the gap in Nicephorus, and does not take into account the stylistic coherence of the passages Theophanes excerpted from the “patrician 250 MAREk JANkOWIAk the description of the second siege of Constantinople in 717–8. Nicephorus’ Breviarium is thus of crucial importance for understanding the source it shared with Theophanes, as it was probably based on this single source for the period between 668 and ca 718, i.e. for the reigns of Constantine IV, Justinian II and the ephemeral predecessors of Leo III. The nature, chronological scope and thrust of this source have recently been subject to debate.47 Most notices of Nicephorus are paralleled, often in more detail, in the Chronicle of Theophanes, who seems to preserve the style of the original narrative, rewritten by Nicephorus in classicizing prose. The stylistic analysis of Theophanes’ entries covering the years 685–717 performed by Dmitry Afinogenov has revealed a distinctive technical jargon, stuffed with Latinisms and military terms, such as τὸ καβαλλαρικόν (“cavalry”) or διήρεις κακκαβοπυρφόροι (“biremes equipped with cauldrons of fire”).48 Afinogenov observed a significantly lower density of such terms in the preceding section on the years 641–85. His analysis can, however, be refined and extended. On the one hand, the choice of 685 as the cut-off date blurs the limits of the sources given that, as we have seen, the common source of Theophanes and Nicephorus probably started in 668; more occurrences of technical terms can, therefore, be expected after than before that date. Indeed, out of the seven Latinisms identified by Afinogenov in the section covering the years 641–85, six belong to the reign of Constantine IV: τὸ καβαλλαρικόν (p. 3594), πάκτον (pp. 35514, 23, 35917, 20), πόρτα (p. 35330), to which one can add such technical terms as θέμα (p. 35214, usually τὰ καβαλλαρικὰ θέματα: pp. 3648, 3665, 37111, 37614, 38330; but also without τὰ καβαλλαρικὰ: pp. 35816, 38314, 3858, 28), κακκαβοπυρφόρος (p. 35321) and σιφωνοφόρος (p. 35322). This points to a new source having been used by Theophanes starting in 668, as no such terms occur in the preceding section of the Chronicle. On the other hand, technical and Latin terms turn up with a similar frequency in the description of the siege of 717–8, also excluded by Afinogenov from his analysis, e.g. φῶσα (p. 39520), περιτείχισμα (p. 39521), κατῆναι (pp. 39524, 3965, 28, 3973, 7, a very rare word for “transport ship”), λωρικάτοι (p. 3965) and ἀρμαμέντον (p. 3973). These distinctive linguistic features indicate a single source for the years 668–718. Passages excerpted by Theophanes and Nicephorus have also other features in common: a penchant for story-telling, a matter-of-fact and lucid narrative, focus on political and military history, at the expense of ecclesiastical events49—and a general bias. The author did not disguise his sympathy for Leo III and Artemios-Anastasios II, and his Trajan” between 668 and 718 (see below), as well as Theophanes’ use of other sources, in particular on the Monothelete controversy, on which see my paper forthcoming in the acts of the Theophanes conference held in Paris in September 2012. 47. Mango, Nikephoros (quoted n. 3), pp. 15–7; Mango – Scott, Theophanes (quoted n. 22), p. lxxxvii; D. Afinogenov, The source of Theophanes’ Chronography and Nikephoros’ Breviarium for the years 685–717, Hristianskij Vostok n.s. 4, 2002, pp. 11–22; Howard-Johnston, Witnesses (quoted n. 3), pp. 256–64 and 299–307; Treadgold, Trajan the Patrician (quoted n. 46). A reconstruction, covering only the years 685–717, was attempted by D. Afinogenov, The History of Justinian and Leo, in La Crimée entre Byzance et le Khaganat khazar, éd. par C. Zuckerman, Paris 2006 (Centre de recherche d’histoire et civilisation de Byzance, Monographies 25), pp. 181–200. 48. Afinogenov, The source (quoted n. 47). 49. Church history is, however, not entirely absent from his narrative, see the reports on the Sixth Ecumenical Council (Theoph. AM 6171, pp. 35926–3607 = Nikeph. § 371–10), and on Philippikos’ revival of Monotheletism (Theoph. AM 6204, p. 38210–21 = Nikeph. § 46). THE FIRST ARAB SIEGE OF CONSTANTINOPLE 251 antipathy for most of the other emperors. His main villain is the emperor Justinian II. “On assuming power, [he] undid the measures which his father had taken for the sake of peace and the good order of the state” (§ 382–4)—this introductory sentence gives the tone of the narrative of his two reigns, pictured as an unbroken sequence of disasters brought upon the Empire by the foolish emperor. His successors, although less tyrannical, were even less apt to govern: Philippikos “appeared to administer the Empire in an indecorous and negligent manner”, while Theodosius III, “an idle and ordinary fellow”, was forced to abdicate because of his incapacity. “On account of the frequent assumptions of imperial power and the prevalence of usurpation, the affairs of the Empire and of the City were being neglected and declined.”50 The incompetence of the emperors and a crisis of legitimacy are thus identified as the prime reasons for the deep crisis of the Empire which culminated with the Arab attack on Constantinople in 717–8. Legitimacy is indeed the central theme of our source. Whether the two long biographical passages demonstrating the personal ability of Leo III and clarifying the motives behind his tactical alliance with the invading Arab army belong to our source or not,51 the victorious defence of Constantinople is represented as the proof of Leo’s capacity to rule and protect the Empire. More importantly, our source delegitimizes Leo’s predecessors: the ephemeral emperors were usurpers, and Justinian II, although a scion of the legitimate dynasty, was a tyrant.52 Every episode of Justinian II’s two reigns serves to demonstrate that his calamitous management of public affairs brought the Empire to the brink of disaster. By contrast, Justinian’s father, despite his disastrous campaign against the Bulgars, benefits from a far more sympathetic treatment: having dealt with all the threats faced by the Empire—the Arabs, the Bulgars and the Monotheletes—he is said to have “spent the rest of his life in tranquillity and peace”.53 The work of our author—who clearly belonged to the military establishment of the Empire and enjoyed privileged access to the court—was thus a polemical and subjective portrayal of the recent history of the Empire, perhaps a memoir of a retired civil or military officer.54 It has been recently attributed to the elusive patrician Trajan,55 known from a single mention in the Chronicle of Theophanes56 and an entry in the Suda: 50. Philippikos: Nikeph. § 461–2, cf. Theoph. AM 6203, p. 38129–30; Theodosius III: Theoph. AM 6207, p. 38521–22, cf. Nikeph. § 5021; his abdication: Nikeph. § 52; decline of the Empire: Nikeph. § 521–3. 51. Theoph. AM 6208 and 6209; neither of them is to be found in Nicephorus. Afinogenov, The source (quoted n. 47), pp. 14–7, and, more hesitatingly, Howard-Johnston, Witnesses (quoted n. 3), p. 300, are in favour of including them in our source; contra Mango – Scott, Theophanes (quoted n. 22), pp. lxxxvii–lxxxviii. 52. A different analysis in Afinogenov, The source (quoted n. 47), pp. 17–20. 53. Nikeph. § 3710–11. “Peace and tranquillity” is a slogan used by our author in relation to good emperors, see also § 3436 (Constantine IV) and § 5521 (Leo III). 54. Howard-Johnston, Witnesses (quoted n. 3), pp. 306–7. 55. Howard-Johnston, Witnesses (quoted n. 3), pp. 306–7; Treadgold, Trajan the Patrician (quoted n. 46), pp. 590–4. See the next note on Cyril Mango’s reservations. Afinogenov, The source (quoted n. 47), p. 21, thought of Leo III as the author of our source, but there is nothing to commend this hypothesis. 56. Theoph. AM 5870, p. 663. Pace C. de Boor, Der Historiker Traianus, Hermes 17, 1882, pp. 489–92, followed by Mango – Scott, Theophanes (quoted n. 22), p. lxxxviii–xc, and Mango, Nikephoros (quoted n. 3), pp. 16–7, nothing suggests that Theophanes confused our πατρίκιος Trajan 252 MAREk JANkOWIAk “Trajan, patrician. He flourished under Justinian whose nose was cut. He wrote a most excellent short chronicle. He was very Christian and most orthodox.”57 This description tallies well with our text: we have seen the emphasis on Justinian’s reign, his chronicle is concise, vivid and of a respectable standard, and its author was a staunch opponent of the Monothelete restoration under Philippikos; but, needless to say, this attribution remains hypothetical. Rather than on research and written documentation, the “patrician Trajan” based his work on personal memory, vague and unspecific in the first part of his work, but increasingly detailed with time, and culminating with the circumstantial account of the turmoil following the fall of Justinian II in 711.58 He wrote in the first two years of Leo’s reign: the positive picture of the former emperor Artemios (713–5) suggests a date before his abortive revolt against Leo III in 719,59 while the description of the second siege of Constantinople in 717–8 was in all likelihood one of the last episodes covered by our source.60 The half century separating the “patrician Trajan” from the siege of Constantinople under Constantine IV and the lack of written sources account for the imprecision of his description of the siege. Although details, such as place names, seem to abound—the Arab fleet moored between the Magnaura and the kyklobion in the Hebdomon; fighting took place between the brachialion of the Golden Gate and the kyklobion; the Arab fleet wintered in kyzikos and perished in a storm near Syllaion—his account is rather vague. The daily “thrusts and counter-thrusts” re-enacted in the same sector of the Theodosian Wall over seven consecutive summers bring to mind the Trojan War rather than a real siege. No reason is given for the retreat of the Arab fleet, and its providential destruction on its way back home sounds suspect. The Arab commander is given a puzzling name, perhaps to be read Chaleph, which cannot be associated with any of the numerous Arab generals known from the Islamic sources, but sounds similar to the title of “caliph”.61 Moreover, with the homonymous στρατηγός of Valens, mentioned in AM 5867, p. 6210, as observed also by Treadgold, Trajan the Patrician (quoted n. 46), pp. 591–2. 57. Suda Τ 901 (iv.582): Τ ρ α ϊ α ν ό ς , πατρίκιος. ἤκμαζε δὲ ἐπὶ Ἰουστινιανοῦ τοῦ Ῥινοτμήτου. οὗτος ἔγραψε χρονικὸν σύντομον, πάνυ θαυμάσιον. ἦν δὲ χριστιανικώτατος καὶ ὀρθοδοξότατος (transl. Mango – Scott, Theophanes [quoted n. 22], p. lxxxviii). 58. Howard-Johnston, Witnesses (quoted n. 3), p. 307. 59. It is remarkable that although Nikeph. § 57 and Theoph. AM 6211, pp. 40018–4013, provide the same details, including an identical list of the six plotters, they present the revolt of Artemios from different perspectives, with Nicephorus putting the blame on Artemios, and Theophanes—on the magistros Niketas Xylinites. It is unlikely that they used different sources: either Theophanes rewrote the story in line with his “anti-Bulgarian bias”, as suggested by Mango – Scott, Theophanes (quoted n. 22), p. 553 n. 8, or perhaps the “patrician Trajan” modified the end of his work in the light of recent events. 60. Contra D. Afinogenov, A lost 8th-century pamphlet against Leo III and Constantine V?, Eranos 100, 2002, pp. 1–17, but see above on the technical military vocabulary used in the description of the siege of 717–8. 61. Theoph. AM 6164, p. 35317: Χαλέ or Χάλε; Nikeph. § 346: Χαλέβ or Χαλέφ. Anastasius Bibliothecarius has the words saeuus amiraeus in the place of the name of the Arab commander, he therefore read χαλεπὸν ἀμηραῖον, see Theophanis chronographia. 2, rec. C. de Boor, Lipsiae 1885, p. 2233. Mango – Scott, Theophanes (quoted n. 22), AM 6164, p. 493 n. 2, proposed to read khālid, but neither this name nor, for instance, khalīfah can be found in the lists of Arab commanders, see THE FIRST ARAB SIEGE OF CONSTANTINOPLE 253 the account of the “patrician Trajan” lacked dates.62 The only chronological indication preserved by Nicephorus placed the Arab attack against Constantinople “immediately after the accession” of Constantine IV.63 Taken literally, this would date the beginning of the siege of Constantinople to 668, but, as we have seen, Theophanes discarded this indication, and preferred to tie the narrative of “Trajan” to a different event. The date provided by Nicephorus is indeed not absolute, and it does not look more precise than the length of the siege, seven years, probably only a synonym of a long time.64 These features—lack of chronological precision and focus on secondary topographical detail, rather than on the course of events—are signs of a report written from memory long time after the events. In this case, however, the faded memories of the “patrician Trajan” may have been revived by recent events. It is perhaps not coincidental that, writing perhaps only several months after the failure of the second Arab attempt to conquer the Byzantine capital, he chose to start and to end his account with a siege of Constantinople: the two Arab sieges thus played a structural role in his account, spanning half century of the Empire’s calamities. Moreover, the recent siege probably orientated the description of the first Arab assault. Even if there is a marked contrast between the dramatic and circumstantial description of the siege of 717–8 and the naive account of the seven-year siege half a century earlier, a parallel reading of both accounts reveals similarities. In both cases, the Arabs first moored between the Magnaura and the kyklobion, west from the city; both emperors prepared dromones and biremes equipped with Greek fire to defend Constantinople; both Arab fleets departed from Constantinople “with much sorrow” or “shame”, only to be wrecked on their way home by violent storms.65 These parallels suggest that there is not below, pp. 264–9. On the early instances of the title of caliph, see R. Hoyland, New documentary texts and the early Islamic state, BSOAS 69, 2006, pp. 395–416, at p. 405: the only contemporary example from the first two centuries of Islam are Arab-Sassanian drachms of ʿAbd al-Malik issued between 75/694–5 and probably 77/696–7, on which see L. Treadwell, ʿAbd al-Malik’s coinage reforms : the role of the Damascus mint, RN 165, 2009, pp. 357–82, at pp. 364 and 370–2. The references to the use of the title khalīfat Allāh by ʿAbd al-Malik’s predecessors quoted by P. Crone, M. Hinds, God’s caliph : religious authority in the first centuries of Islam, Cambridge 1986, pp. 6–7, come from later literary sources. The first Greek rendering of this title (χαλιφᾶς) known to me comes from the late 11th-century continuation of the chronicle of John Skylitzes (in Georgius Cedrenus Joannis Scylitzae ope, ed. I. Bekker, Bonn 1838–9, vol. 2, pp. 73215 and 7335); earlier possibly in an astrological treaty ascribed to Theophilus of Edessa, Catalogus codicum astrologorum Graecorum. 11, 1, Codices Hispanienses, descripsit C. O. Zuretti, Bruxellis 1932, p. 2295–6, but the attribution is uncertain. 62. Pace Afinogenov, The History (quoted n. 47), p. 199. 63. Nikeph. § 342–4: καὶ τούτου [sc. Κωνσταντίνου] ἀρξαμένου εὐθὺς ὁ τῶν Σαρακηνῶν ἡγούμενος ναῦς πλείστας κατασκευάσας κατὰ τοῦ Βυζαντίου ἐκπέμπει. 64. In the same vein, Islamic sources describe the seven-year occupation of Rhodes by an Arab garrison in the 670s, see below, p. 275. 65. Magnaura and kyklobion: Theoph. AM 6165, p. 35326–28, and AM 6209, p. 39525; fire-bearing ships: Theoph. AM 6164, p. 35321–22: κατεσκεύασε […] διήρεις εὐμεγέθεις κακκαβοπυρφόρους καὶ δρόμωνας σιφωνοφόρους, and AM 6209, p. 39710–11: σίφωνας πυρσοφόρους κατασκευάσας εἰς δρόμωνάς τε καὶ διήρεις τούτους ἐμβαλὼν; sorrow and shame: Theoph. AM 6165, p. 3548: ἀνθυπέστρεψαν μετὰ μεγάλης λύπης, and AM 6210, p. 3996–7: ἀποκινησάντων δὲ τῶν Ἀγαρηνῶν […] μετὰ πολλῆς αἰσχύνης; final disaster: Theoph. AM 6165, p. 3548–11, and AM 6210, p. 3997–19— note that in both these places, and only there, Theophanes uses the rare verb ἐκπορίζω, “to sail away from Constantinople through the straits”, on which see Mango – Scott, Theophanes (quoted n. 22), p. 551 n. 7. 254 MAREk JANkOWIAk much original material in the account of the first siege, apart perhaps from the mentions of the winter base at kyzikos and of the disaster next to Syllaion. If the description of the first siege of Constantinople was probably contaminated by fresh memories of the second Arab attack, the narrative of the peace concluded after the failure of the siege has been distorted for other reasons. We have seen that the tendency of the “patrician Trajan” to denigrate Justinian II resulted in idealising his father, Constantine IV. Peace negotiations between Muʿāwiyah and Constantine IV are a case in point. Unknown to the “Oriental source” and to Muslim chronographers,66 they are alluded to only by al-Masʿūdī, who refers to “exchanges of letters and truce negotiations” (murāsalāt wa muhādanāt) between Muʿāwiyah and an emperor called Fukunaṭ b. Fūq by the intermediary of a certain Yanāq or Fanāq al-Rūmī.67 Even if Fukunaṭ can be identified with Constantine IV, sometimes called Pogonatos, this information comes from a very confused section on Byzantine history within the Golden Meadows. Al-Masʿūdī later corrected it in his Book of instruction and revision, where he speaks of “exchanges of letters and courteous treatment” (al-murāsalāt wa l-mulāṭafāt), and not “truce negotiations”, intermediated by the same Fanāq, but this time between Heraclius and Muʿāwiyah, then still governor of Syria.68 Both these mentions serve merely to introduce a story of how a Roman emperor predicted Muʿāwiyah’s caliphate. It would be hazardous to base any historical reconstruction on them. More worryingly, the emperor Constantine IV himself was unaware of the peace he is thought to have concluded with the caliph. His sacra to the pope of Rome, dated 12 August 678, repeatedly stresses the impossibility of convoking a general council because “the times do not allow it”.69 On the contrary, the emperor hoped that the reconciliation of the churches “will put an end to the turmoil that is being inflicted on 66. There is no mention of the treaty in Agapios, Michael the Syrian and the Chronicle of 1234. For Islamic sources, see A. D. Beihammer, Nachrichten zum byzantinischen Urkundenwesen in arabischen Quellen (565-811), Bonn 2000, no. 284, pp. 330–2. The silence of the Islamic sources is unusual: other 7th-century peace treaties and truces are attested both by Byzantine and Islamic sources, see for instance Beihammer, Nachrichten, nos. 266 (the peace of 657/8) and 295 (peace between Justinian II and ʿAbd al-Malik). 67. Al-Masʿūdī, Les prairies d’or, éd. C. Barbier de Meynard et A. Pavet de Courteille, revue et corrigée par Ch. Pellat, vol. 2, Beyrouth 1966, § 756, p. 54; French translation: al-Masʿūdī, Les prairies d’or, trad. de C. Barbier de Meynard et A. Pavet de Courteille, vol. 2, Paris 1965, p. 284. The name of the emperor (‫ )فُكُنط بن فوق‬is a modern conjecture, the manuscripts have ‫ فلنط بن فوق‬or ‫فلقط بن مورق‬. 68. Al-Masʿūdī, Kitāb al-tanbīh wa-al-ishrāf, ed. M. J. de Goeje, Lugduni Batavorum 1894 (Bibliotheca geographorum Arabicorum 8), p. 1583–5; French transl.: al-Masʿūdī, Le livre de l’avertissement et de la revision, trad. B. Carra de Vaux, Paris 1896, p. 216. 69. ACO, ser. sec. II, pp. 215 (τοῦ καιροῦ μὴ ἐπιδεχομένου), 221–26, 67 (ἐπὰν οὖν ὁ χρόνος οὐκ ἐπιδέχεται τελείαν συνάθροισιν γενέσθαι), and 631–81 (εἰ γὰρ ἐπεδέχετο ὁ χρόνος, ὡς προείρηται, καὶ τελείαν συνάθροισιν εἴχομεν παρασκευάσαι γενέσθαι, ἀλλ’ ἐπειδὰν ὁ χρόνος οὐκ ἐπιδέχεται…). This letter cannot be adduced as evidence for a recent peace with the Arabs, see A. kaplony, Konstantinopel und Damaskus : Gesandschaften und Verträge zwischen Kaisern und Kalifen 639–750, Berlin 1996, p. 84 (but he still keeps the date of 677/8 for the peace), pace F. Dölger, Regesten der Kaiserurkunden des oströmischen Reiches von 565–1453, 1, 1, Regesten 565–867, zweite Auflage, ed. A. E. Müller, München 2009, no. 239, p. 113, and Beihammer, Nachrichten (quoted n. 66), no. 284, p. 332. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 255 us by the tribes”:70 ecclesiastical peace was a prerequisite to the military pacification, not the opposite. When two years later, on the 10 September 680, the emperor eventually convoked the council, he did so prompted by the arrival of the long-awaited delegation from Rome, despite the continuing war, as is made clear in the first sentence of the convocation: “even though Our Serenity is ceaselessly afflicted by military and political anxieties, nevertheless we place all the affairs of our Christ-loving state second to our most Christian faith, which in war is a defender both of ourselves and of our Christloving armies.”71 For chronological reasons, it is unlikely that the war with the Bulgars is meant;72 Arab raids, attested by Islamic sources for the last years of Muʿāwiyah and possibly the first year of Yazīd (680/1), are a more plausible candidate. The only source for a peace concluded between the Arabs and Constantine IV before the death of Muʿāwiyah in 680 is thus the “patrician Trajan”. The terms of the treaty, concluded for 30 years with yearly payments of 3,000 gold coins, 50 captives and 50 horses,73 resemble those of the treaties negotiated by ʿAbd al-Malik first with Constantine IV in 685 and then with Justinian II in the first years of his reign, which both stipulated, in addition to territorial clauses, payment of 1,000 gold pieces, a captive and a horse per day.74 Neither of these treaties was known to the “patrician Trajan”, in whose account, moreover, the “leader of the Saracens” remains anonymous.75 Nothing indicates, therefore, that the treaty he remembered was distinct from the only securely 70. ACO, ser. sec. II, p. 827–29 : ἐλπίδα ἔχομεν εἰς τὸν στέψαντα ἡμᾶς θεόν, ὅτι ἡνίκα θεωρήσῃ θερμὴν περὶ τούτου ἐξ ἡμῶν πρόθεσιν, ἡ αὐτοῦ ἀγαθότης καὶ τοὺς ταράχους τοὺς ἐπερχομένους ἡμῖν τῶν ἐθνῶν ἀποπαύει (transl. R. Price). 71. ACO, ser. sec. II, p. 1021–24: εἰ καὶ μερίμναις στρατιωτικαῖς τε καὶ πολιτικαῖς ἀνενδότως ἡ ἡμετέρα συνέχεται γαλήνη, ἀλλ’ οὖν ἅπαντα τῆς φιλοχρίστου ἡμῶν πολιτείας τὰ πράγματα δεύτερα τῆς χριστιανικωτάτης ἡμῶν πίστεως τιθέμενοι, ἣν καὶ ἐν πολέμοις ὑπέρμαχον ἡμῶν τε καὶ τῶν ἡμετέρων φιλοχρίστων ἐκστρατευμάτων κεκτήμεθα (transl. R. Price). 72. Pace G. De Gregorio, O. kresten, Ἐφέτος—“in diesem Jahr” : zur Datierung des Bulgarenfeldzugs des kaisers konstantinos IV. (Sommer/Herbst 680), Rivista di studi bizantini e neoellenici 43, 2006, pp. 21–56. See my arguments supporting the traditional date of the war with the Bulgars (summer 681) in M. Jankowiak, The Third Ecumenical Council in Constantinople (680–681) : from the council to the acts and back, U schyłku starȯytnóci. Studia ́ródłoznawcze 10, 2011, pp. 9–88, at pp. 68–71 (in Polish), and in the forthcoming commented translation of the Acts of the Sixth Ecumenical Council in the Translated texts for historians series. 73. On the differing lessons of the manuscripts of Theophanes, see Mango – Scott, Theophanes (quoted n. 22), AM 6169 n. 3. 74. Treaty of ʿAbd al-Malik with Constantine IV: Theoph. AM 6176, p. 3618–13; Agap. p. 497; Chron. 1234 § 145; no mention in Michael the Syrian; see also El. Nisib. AH 65, p. 72; Cont. Byz. Arab. § 29 (attributes the initiative of the treaty to Marwān rather than to ʿAbd al-Malik; see below, note 249, for the editions). Treaty with Justinian II: Theoph. AM 6178, p. 3636–14; Agap. p. 497; Mich. Syr. xi.15, p. 469; Chron. 1234 § 146. See Beihammer, Nachrichten (quoted n. 66), nos. 293 and 295, for Islamic sources. Despite the wording of the LP, p. 36613–14 (qui clementissimus princeps [sc. Justinian II] domino auxiliante pacem constituit cum nec dicenda gente Saracenorum decennio terra marique), I am tempted to identify the peace treaty known in Rome in early September 685 with the treaty concluded by ʿAbd al-Malik with Constantine IV. 75. He is anonymous in Nikeph. § 342–3, 21. Although in his description of the treaty requested by ʿAbd al-Malik from Constantine IV Theophanes refers to an earlier “peace sought after by Muʿāwiyah” (Theoph. AM 6176, p. 36110–13: Ἀβιμέλεχ τὴν ἐπὶ Μαυΐου ζητηθεῖσαν εἰρήνην αἰτεῖται ἀποστείλας πρέσβεις πρὸς τὸν βασιλέα, τὰς αὐτὰς τξεʹ χιλιάδας τοῦ χρυσοῦ νομισμάτων συνθέμενος τελεῖν, καὶ τοὺς τξεʹ δούλους, καὶ ὁμοίως εὐγενεῖς ἵππους τξεʹ), the absence of this information from the 256 MAREk JANkOWIAk attested peace treaty between Constantine IV and the Arabs, concluded with the new caliph ʿAbd al-Malik on the 7 July 685, three days before the death of the emperor.76 The treaty of 685 sanctioned a state of affairs that prevailed after fighting had died out around 680, when the Empire and the Caliphate concentrated on their internal problems, the contested succession of Muʿāwiyah and the conflict between Constantine IV and his brothers. Not much can be thus said on the first siege of Constantinople based on the history of the “patrician Trajan”. He preserved blurred reminiscences of a protracted Arab threat to the Byzantine capital at the beginning of the reign of Constantine IV, of an Arab naval disaster, and of a peace concluded by Constantine IV with a caliph; but all these otherwise attested events have been compressed into an account that leaves the false impression of a close chronological connection between them. In reality, they probably spanned the entire reign of Constantine IV, from its very beginning—if this is how the Arab attack against Constantinople should be dated—to the peace treaty concluded days before its end. Chronology was not a strength of the “patrician Trajan”: we have to turn to the other source of Theophanes for a more precise account of Byzantine-Arab warfare under the reign of Constantine IV.77 heophilus of Edessa While the history of the “patrician Trajan” was apparently based on personal memory, the “Oriental source” of Theophanes was of a fundamentally different nature: structured around a chronological framework, it offered a coherent sequence of precisely dated events that could be easily inserted into the annalistic skeleton of Theophanes’ Chronicle. Theophanes had access to the same body of material covering at least the years ca 630– ca 750 as three extant Syriac and Arabic chronicles: the History written in the early 940s by Agapios/Maḥbūb, probably Chalcedonian bishop of Hierapolis/Membidj;78 the monumental History of Michael the Syrian, Jacobite patriarch of Antioch (1166–99); and the anonymous Edessan chronicle known as Chronicon ad annum 1234 pertinens. The latter two texts used this material by the intermediary of the lost historical work of Dionysius of Tell Maḥrē,79 another Jacobite patriarch of Antioch (818–45), who listed other chronicles depending on the “Oriental source” suggests an authorial comment by Theophanes, who aimed at harmonizing what his two sources had to say on peace treaties with the Arabs. 76. See above, note 74. The date of the treaty is given by Elias of Nisibis. Constantine IV seems to have died on 10 July 685, see Ph. Grierson, The tombs and obits of the Byzantine emperors (337–1042), DOP 16, 1962, pp. 3–60, at p. 50. The Liber pontificalis places the beginning of the reign of Justinian II in early September 685 (p. 36612–13: regnauit domnus Iustinianus Augustus defuncto patre, in initia mensis septembris, ind. XIIII), perhaps when the new emperor was announced in Rome. ʿAbd al-Malik succeeded his father Marwān at his death on 7 May 685, see El. Nisib. AH 65, pp. 71–2 (correct day of the week). 77. See below, pp. 302–3, for a more detailed assessment of the memories of the siege of Constantinople in the work of the “patrician Trajan”. 78. On Agapios, see R. G. Hoyland, Seeing Islam as others saw it : a survey and evaluation of Christian, Jewish and Zoroastrian writings on early Islam (Studies in late antiquity and early Islam 13), Princeton 1997, pp. 440–2, and Id., Theophilus of Edessa’s chronicle and the circulation of historical knowledge in late antiquity and early Islam (Translated texts for historians 57), Liverpool 2011, pp. 14–5. 79. For a tentative reconstruction see A. Palmer, The seventh century in the West-Syrian chronicles (Translated texts for historians 15), Liverpool 1993, pp. 111–221. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 257 among his sources Theophilus of Edessa, no doubt the court astrologer of the caliph al-Mahdi (775–85) and the Syriac scholar who translated the Iliad into Syriac. Theophilus the Astrologer is also quoted by Agapios of Membidj as a prolific historian and an ocular witness of the Abbasid overthrow of the Umayyads.80 This name is currently accepted as a convenient label for the “Oriental source” of Theophanes and for the material shared by the four extant chronicles, even though this attribution remains hypothetical.81 But many uncertainties remain: the transmission to the four extant sources probably involved multiple intermediaries and a significant amount of editing and rewriting in various languages, which raised doubts about the coherence of the common source;82 linguistic analysis gives contradictory pointers to Greek and Syriac as its original language;83 the sources of Theophilus remain almost entirely unknown;84 and the format of his work— annalistic or “rather sparing with dates”—is debated.85 Dionysius of Tell Maḥrē characterised the work of Theophilus as a “narrative resembling ecclesiastical history”.86 This unusual label indicates that it did not fit into any other category of sources listed by Dionysius in the preface to his chronicle, that is “chronographies”, “ecclesiastical histories” and “series of years”. Unfortunately, the historical work of neither Theophilus nor of any of the three other authors named by Dionysius in the same category is extant. It is thus unclear whether, for Dionysius, they resembled ecclesiastical history by their subject and differed from it by their format, or the opposite. The former is suggested by Dionysius’ critique of “the hatred of the Orthodox” exhibited by Theophilus of Edessa, whose work, therefore, touched upon ecclesiastical 80. Syriac scholar and astrologer: Hoyland, Theophilus (quoted n. 78), pp. 6–7; Theophilus’ translation of the Iliad, or perhaps rather of the Iliou persis: A. Hilkens, Syriac ilioupersides : the fall of Troy in Syriac historiography (forthcoming); historian: Agap. p. 525. 81. On the “Oriental source” and its attribution to Theophilus of Edessa, see M. Breydy, Das Chronikon des Maroniten Theophilos ibn Tuma, Journal of Oriental and African studies 2, 1990, pp. 34–46; L. I. Conrad, The conquest of Arwad : a source-critical study in the historiography of the early medieval Near East, in The Byzantine and early Islamic Near East. 1, Problems in the literary source material, ed. by Av. Cameron and L. I. Conrad, Princeton 1992, pp. 317–401; Palmer, The seventh century (quoted n. 79), pp. 95–8 (partly by R. Hoyland); Mango – Scott, Theophanes (quoted n. 22), pp. lxxxii–lxxxiv; W. Brandes, Der frühe Islam in der byzantinischen Historiographie : Anmerkungen zur Quellenproblematik der Chronographia des Theophanes, in Jenseits der Grenzen : Beiträge zur spätantiken und frühmittelalterlichen Geschichtsschreibung, hrsg. von A. Goltz, H. Leppin und H. Schlange-Schöningen, Berlin 2009, pp. 313–43; Howard-Johnston, Witnesses (quoted n. 3), pp. 192–236; Hoyland, Theophilus (quoted n. 78), pp. 4–6, which replaces his Seeing Islam (quoted n. 78), pp. 400–9, 631–71; M. Conterno, Palestina, Siria, Costantinopoli : la « Cronografia » di Teofane Confessore e la mezzaluna fertile della storiografia nei « secoli bui » di Bisanzio, PhD diss. SUM Firenze 2011; papers presented by M. Conterno, M. Debié and R. Hoyland at the Theophanes conference in Paris in September 2012 (publication forthcoming). 82. Hoyland, Theophilus (quoted n. 78), p. 7: Theophilus’ dependants “felt free to creatively revise and reshape it, to abbreviate and reword it, and to supplement it with material from other sources”. See Conterno, La « Cronografia » (quoted n. 81), on the problems raised by the hypothesis of a single common source. 83. Conterno, La « Cronografia » (quoted n. 81), pp. 131–86. 84. See the speculations of Hoyland, Theophilus (quoted n. 78), pp. 23–9. 85. Brandes, Der frühe Islam (quoted n. 81), p. 326; Hoyland, Theophilus (quoted n. 78), p. 19 (quoted). 86. Mich. Syr. x.20, p. 358. 258 MAREk JANkOWIAk affairs,87 but no doubt left much room also for secular history, as was common for the “ecclesiastical histories” listed by Dionysius. What is therefore likely to have set Theophilus’ work apart from Socrates, Sozomen and their successors is its annalistic format. Dionysius’ criticism of the “compartmentalised and discontinuous” character of the works of Theophilus and his three colleagues and their chronological inexactitude would make little sense if these works lacked a chronological backbone. The four extant dependants of Theophilus share sophisticated synchronisms that demonstrate that their common source had an annalistic structure. Discrepancies are not rare, but they are only to be expected in sources that used no fewer than 11 chronological systems.88 However, if we reduce them to the common denominator which, as I argued elsewhere,89 can only be the regnal years of the Roman and Arab rulers, we obtain a significantly more coherent picture. Coherent, but not always factually correct: establishing an absolute chronology on the basis of a series of regnal years carried the risk of adding or omitting a year due to consecutive roundings. This is probably what happened to Jacob of Edessa (d. 709/10), who established a “series of years” later continued by Michael the Syrian: he rounded up the lengths of the reigns of Phocas, Heraclius, and his sons, which pushed the first year of Constans II one year too late, to 642/3.90 Since Jacob’s canon was, according to Michael the Syrian, the only available continuation of the canon of Eusebius of Caesarea,91 we would expect it to have been used as the chronological framework of his work by Theophilus of Edessa,92 by whose intermediary Jacob’s mistake was inherited by Theophanes. But Theophanes started to rely on the “Oriental source” only around the middle of the reign of Heraclius; he thus assigned to Phocas seven, and not eight years as in Jacob’s canon, but kept unchanged Jacob’s lengths of the reigns of Heraclius 87. At least to a larger extent than suggested by the material common to the four dependants and generally assumed in the scholarship, e.g. Hoyland, Theophilus (quoted n. 78), p. 21: “Theophilus [wrote] mostly on secular events.” It is not surprising that the four extant sources share no notices on ecclesiastical events, none of them being of Theophilus’ Maronite creed, on which see The Chronography of Gregory Abû’l Faraj, the son of Aaron, the Hebrew physician, commonly known as Bar Hebraeus, transl. E. A. Wallis Budge, London 1932, vol. 1, p. 116: “he was a skilled astronomer who cleaved to the heresy of the Maronites.” It is, however, tempting to identify Theophilus as the source of the numerous passages on the Maronites in the chronicle of Michael the Syrian (pp. 443–4, 446–8, 451–2, 492–6, 511, perhaps also pp. 520–1), possibly including the passage on the life of Maximus the Confessor (pp. 433–7). 88. Theophanes: years from the Creation and from the Incarnation, regnal years of the Roman emperors and Persian shahs (replaced in the 7th century by the Arab caliphs), and the years of the five patriarchs; Agapios, Michael the Syrian and Chron. 1234: Seleucid and Hijri eras, Roman and Arab rulers. 89. See my paper on the chronology of Theophanes quoted above, note 37. 90. Mich. Syr. xi.17, pp. 482–3. Jacob’s canon is preserved in a single mutilated manuscript (Brit. Libr. Add. 14685, ed. and transl. by E. W. Brooks in Chronica minora, pars tertia [CSCO 6, Scriptores Syri. Ser. 3, 4], Paris 1907, pp. 199–255) which, however, breaks up in AD 629/30. Michael attributes to Phocas 8 years (he reigned for 7 years and 10 months), to Heraclius 31 years (30 years and 3 months) and an entire year to his sons (9 months). The cumulative result is an additional year, as can be seen in the Chronicle of Michael the Syrian, who counts the years of Constans II from AG 954 = AD 642/3 (ii.440, 450, 451 and 536), and those of Constantine IV from AG 981 = AD 669/70 (ii.454). On Jacob’s historical work, see W. Witakowski, The Chronicle of Jacob of Edessa, in Jacob of Edessa and the Syriac culture of his day, ed. by R. B. ter Haar Romeny, Leiden 2008, pp. 25–47. 91. Mich. Syr. xi.17, pp. 482–3. Michael was not aware of the 11th-century continuation of the canon by Elias of Nisibis. 92. See my paper quoted above, note 37. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 259 and his sons. As a result, Theophanes’ chronology found itself one year ahead of that of Jacob and Theophilus. Thus, for instance, events dated in the Syriac chronicles to 642/3, the first year of Constans II, were placed by Theophanes under a year equivalent to 641/2—which is the correct year for the beginning of Constans’ reign, but one year too early for the information borrowed from the “Oriental source”. This is, in my view, the mechanism of the notorious mistake of Theophanes, whose year entries for the reigns of Constans II and Constantine IV contain indications consistently pointing one year earlier than their chronological headings.93 All the events of these two reigns known to Theophanes from the “Oriental source” should consequently be dated one year later than suggested by Theophanes’ annus mundi. Theophilus’ account of the 670s can be reconstructed from the chronicles of Theophanes and of Agapios; the coverage of Dionysius of Tell Maḥrē was more selective, but his dependants still preserve several useful notices, as illustrated in the table 1. With the exception of Michael the Syrian, who confused certain dates in the process of distributing his material into three columns dedicated to political events, ecclesiastical history and natural disasters, the chronology of the remaining three sources is very consistent. The vast majority of the notices appear in more than one source; the rare events attested by only one of the dependants of Theophilus have been included because of their Syrian background or similarity to the other notices. The chronology of Theophilus appears to be sound: the three natural phenomena that can be independently dated—the solar eclipse of 671, the comet of 676, and the earthquake of 679—have all been placed under the correct years.94 We can therefore attempt to reconstruct the Byzantine-Arab warfare in the 670s on the basis of his narrative. The reign of Constantine IV started with a devastating Arab raid in Africa and the wintering of Phadalas in kyzikos, in the vicinity of the Byzantine capital. In the following years, the raids continued with unabated intensity: in 671/2 Busr raided the Empire, in the next year a mighty fleet under the command of a Muḥammad and of ʿAbdallāh b. Qays wintered at Smyrna and in Cilicia and Lycia,95 and in 673/4 Sufyān b. ʿAwf led 93. See the description of the problem in Mango – Scott, Theophanes (quoted n. 22), pp. lxv–lxvii. 94. Solar eclipse: dated by Mich. Syr. to a Sunday in December AG 983, which corresponds to a known eclipse of Sunday 7 December 671, see D. J. Schove, Chronology of eclipses and comets, AD 1–1000, Woodbridge 1984, pp. 132–3. The comet, known also from El. Nisib. AH 56 (who quotes as his source Jacob of Edessa), Liber pontificalis, p. 348, and Chinese and Japanese sources, was visible in August-October 676, see Schove, Chronology, p. 293. The one-year discrepancy between Theophanes and Michael the Syrian may result from the difficulty of dating an event extending over two Greek years. Earthquake: according to Chron. 1234 and Mich. Syr. the earthquake took place on Easter Sunday, which is confirmed by the Chronicle of 846 which provides the date of 3 April AG 990 (AD 679), indeed an Easter Sunday: Chronica minora, vol. 2, Latin transl. I.-B. Chabot (CSCO 4, Scriptores Syri 4), Louvain 1955, p. 175. 95. This information is provided only by Theophanes, but it finds a parallel in Mich. Syr. xi.13, p. 455: “l’année suivante, les Ṭaiyayê pillèrent de nouveau la Lycie et la Cilicie. Étant revenus encore en Lycie, les Ṭaiyayê assiégèrent une ville située sur le bord de la mer. Alors trois patrices des Romains vinrent contre eux, et les vainquirent. En ce jour-là, environ trente mille Ṭaiyayê succombèrent; ceux qui restèrent gagnèrent les navires et eurent à souffrir des tempêtes”. On the other hand, the sequence of items in Theophanes—nocturnal rainbow, attack by a mighty Arab fleet, and plague in Egypt—is exactly mirrored by Agapios, which proves that they all come from a common source. On the names of the generals, see below, pp. 271–2. 260 MAREk JANkOWIAk a big land army. The two latter expeditions targeted Lycia96 and may have in fact been one; it ended in an Arab disaster, when Sufyān was defeated by the patricians Florus, Petronas and Cyprian, losing 30,000 men, while the Arab fleet was annihilated with the help of the recently invented Greek fire. This signal Byzantine victory97 seems to have changed the course of the war: only one Arab raid is mentioned—the wintering of ʿAbdallāh b. Qays and Faḍālah b. ʿUbayd in Crete in 674/5—before the Byzantines launched a counter-offensive.98 In 677/8 a substantial corps of Byzantine troops known as the “Mardaites”99 landed on the Lebanese coast between Tyr and Sidon and occupied the Lebanon mountain range from Amanus to Galilee,100 launching a guerrilla warfare that, according to Agapios, was meant “to deter the Arabs from their incursions”.101 This was the last event of the war reported by Theophilus who apparently made no mention of the peace between Muʿāwiyah and Constantine IV. There is no place for a many-year siege of Constantinople in this narrative.102 On the contrary, the siege would disrupt an otherwise coherent story of a progressive Byzantine recovery, from a nadir in the early 670s when the region of Constantinople was directly threatened by the Arabs, through an important victory in 673/4, to a victorious counterattack in 677/8. By inserting the account of the siege by “patrician Trajan” in the year of a major Byzantine victory, Theophanes distorted Theophilus’ narrative beyond recognition. But the accounts of Theophilus and of the “patrician Trajan” are not irreconcilable. Both speak of a major disaster of the Arab fleet, and both place it in southern Anatolia—at Syllaion or next to a coastal city in Lycia—even though they differ in identifying the cause of the wreckage of the Arab navy, a storm according to “Trajan”, Greek fire in the version of Theophilus.103 We have also seen that the account of peace negotiations led by John Pitzigaudes probably echoes the peace concluded in 685 by Constantine IV and ʿAbd al-Malik. The only element of the account of “Trajan” that is not clearly paralleled in Theophilus is the description of the siege of Constantinople. But before attempting to corroborate it with other early sources, it is necessary to survey other texts that throw light on the history of the Byzantine Empire in the 670s. 96. See below, p. 276–8, on the place of Sufyān’s defeat. 97. Agap. p. 492: “the Romans achieved triumph and victory in this year.” 98. See below, note 151, for an Islamic tradition on a raid against Crete. 99. On the Mardaites, a Byzantine expeditionary corps rather than an obscure ethnic group, see M. Canard, Djarādjima, EI2; now also Howard-Johnston, The Mardaites (quoted n. 3). The speculations of H. M. Bartikian, Η λύση του αινίγματος των Μαρδαϊτών, in Βυζάντιον : αφιέρωμα στον Ανδρέα Ν. Στράτο, vol. 1, Αθήναι 1986, pp. 17–39, carry no weight. 100. Theoph. AM 6169, p. 3557, speaks of the “Holy City”, presumably Jerusalem, where the remaining three dependants of Theophilus have “the mountain of Galilee”. This is probably the result of the mistranslation of the Syriac ṭūrā d-galīlā, which can also be understood as “glorious mountain”; the replacement of “mountain” by “city” was probably the initiative of the translator of the Syriac source into Greek. 101. Agap. p. 493. 102. This crucial fact went almost unnoticed, the only exception known to me being HowardJohnston, Witnesses (quoted n. 3), p. 303. 103. Mich. Syr. xi.13, p. 455, mentions both a storm and a naval battle, see above, note 95. 261 THE FIRST ARAB SIEGE OF CONSTANTINOPLE Event First year of Constantine IV heophanes Constantine 1 = AM 6161 Arab attack on Africa Constantine 1 Agapios Constantine 1 = Muʿāwiyah 10105 = AG 981 “in this year” Chron. 1234 AD Constantine 1 669/70 = Muʿāwiyah 10 = AG 981 Constantine 1 669/70 “in this year” – Mich. Syr. Constantine 1 = Muʿāwiyah 10 = AG 981 “beginning of the reign” AG 980106 – Harsh winter Phadalas winters in kyzikos Raid by Busr Solar eclipse Nocturnal rainbow Great Arab leet Wintering of Arabs in Smyrna, Cilicia and Lycia Plague in Egypt Byzantine victory over Sufyān in Lycia and irst use of Greek ire Abdelas and Phadalas winter in Crete Sign in the sky Rats in Syria Locusts in Syria Invasion of the Mardaites Revolt of John son of Mezezios Earthquake in Batnae and Edessa Death of Muʿāwiyah Constantine 2 Constantine 2 – – 670/1 670/1 Constantine 3 – Constantine 4 Constantine 4 Constantine 4 Muʿāwiyah 12 – Muʿāwiyah 13 Muʿāwiyah 13 – – AG 983 AG 989 – “next year”107 – – Constantine 4 – – 671/2 671/2 672/3 672/3 672/3 Constantine 4 Constantine 5 Muʿāwiyah 13 Muʿāwiyah 14 – no date – – 672/3 673/4 Constantine 6 – – – 674/5 Constantine 7 – Constantine 8 Constantine 9 – – “in this year” – Muʿāwiyah 17 – AG 988 AG 989? AG 990? Constantine 9 “at the same time” AG 990 – – – Constantine 9 – 675/6 675/6? 676/7 677/8 AG 990108 678/9 Constantine 10 “in this year” Constantine 11 Muʿāwiyah 20 Muʿāwiyah 20109 Constantine 11 679/80 = Muʿāwiyah 20 Table 1 – Notices of the dependants of heophilus of Edessa on the reign of Constantine IV and his brothers.104105106107108109 104. Regnal years are provided only when they are explicitly given by the source or, in the case of Agapios, when a notice is placed between two explicitly dated entries. “AG” stands for anno Graecorum, or Seleucid era; a dash indicates the lack of a relevant notice. See above, pp. 7–9, for references to the individual events. Translation in Hoyland, Theophilus (quoted n. 78), pp. 164–72, but his presentation is obscured by the omission from the main text of the dates provided by the four sources. 105. The synchronism is not explicit, but Agapios, p. 490, dates the death of Constans II to the 9th year of Muʿāwiyah. 106. The date is mistaken, as this event is included in the chapter on the reign of Constantine IV who reigned from AG 981. 107. Mich. Syr. xi.13, p. 455, seems to speak of two encounters in Lycia, see above, note 95. 108. Reading AG 990 for AG 950, see Palmer, The seventh century (quoted n. 79), p. 195 n. 480. 109. Mich. Syr. xi.15, p. 468, has mistakenly AG 992, but his canon has the correct date of AG 991 (p. 536). 262 MAREk JANkOWIAk 2. The Empire strikes back: Byzantium in the 670s Sources for the history of Byzantium in the later part of the 7th century are not as scarce as is often believed. Islamic sources provide significant information on the Byzantine-Arab warfare in the reign of Muʿāwiyah, although the mechanisms by which historical information on the early Umayyad period was transmitted to the classical chroniclers remain understudied. Other texts allude to Byzantine activity outside Constantinople in the 670s, during the period traditionally assigned to the siege: dispersed mentions refer to an increased Byzantine activity ca 672–4 which probably culminated with the important defeat of Sufyān b. ʿAwf, while the Miracles of Saint Demetrius and the ecclesiastical correspondence between Rome and Constantinople provide background to the Byzantine victory in the second half of the 670s. Islamic sources on Arab raids against Byzantium Although Muslim historians lost much of their interest in the war with Byzantium after the initial wave of conquests, they still preserve abundant information on the struggle between the Caliphate and the Empire in the 660s and 670s. However, as opposed to the earlier rich accounts of the futūḥ, or Islamic conquests, the Arab raids against the Empire during the caliphate of Muʿāwiyah are known from austere one-line notices. There was no practical purpose in preserving the details of raids that did not result in permanent conquests, since such records no longer served to determine the legal standing of the local population. Besides, the war of attrition waged by the Arabs against the Romans was less spectacular than the first conquests. These reasons, however, only partially explain why almost no tales of two decades of Arab raiding have been transmitted by the classical Muslim historians. Arab warfare against the Empire was waged by Syrian armies, commanded by generals closely connected to the Umayyad court. They would have been best remembered in the Syrian historical tradition, which has been marginalised by the Iraqi and Medinan collectors of oral traditions at work soon after the Abbasid revolution; such texts as the meagre notices on raids against Byzantium collected by Baqī b. Makhlad in Damascus around 850 allow us to appraise the extent of this loss.110 As a result, we have little more than a list of the generals who led the annual raids under the reign of Muʿāwiyah, occasionally supplemented with details on their targets, the season of raiding, and the participation of the fleet. The table below collects such information as is relevant to Anatolia, gleaned from the following early Muslim historians:111 • Muḥammad b. ʿUmar al-Wāqidī (d. 207/822–3) from Medina, a prolific systematizer of information on early Islamic history. Al-Ṭabarī used his lost historical treatise 110. See below, p. 272, and more generally A. Borrut, Entre mémoire et pouvoir : l’espace syrien sous les derniers Omeyyades et les premiers Abbassides (v. 72–193/692–809), Leiden 2011. 111. On all these authors, see the relevant entries in the EI 2, and F. Sezgin, Geschichte des arabischen Schrifttums. 1, Qurʾānwissenschaften, Ḥadīṭ, Geschichte, Fiqh, Dogmatik, Mystik bis ca. 430 H, Leiden 1967. The present survey does not pretend to be exhaustive; other sources contain unique information on Arab raids against Byzantium, in the first place the Tārīkh madīnat Dimashq by Ibn ʿAsākir, on which see below, p. 273. I thank Robert Hoyland, Marie Legendre and Harry Munt for their help with Islamic sources. THE FIRST ARAB SIEGE OF CONSTANTINOPLE • • • • • 263 extensively, but only quoted it explicitly when he found its chronology in disagreement with his other sources; Muḥammad b. Mūsā al-khwārizmī (floruit in the first half of the 9th century), famous mathematician and astronomer who also authored a lost historical work which was systematically used in the early 11th century by Elias of Nisibis for his notices on the annual Arab campaigns against Byzantium;112 khalīfah b. khayyāṭ (d. 240/854–5), a Basran historian whose Tārīkh is the oldest complete extant Islamic history.113 His work survives in the redaction of Baqī b. Makhlad (d. 276/889), an Andalusian scholar who was his disciple and who supplemented it with traditions gathered during his stay in Damascus;114 al-Yaʿqūbī (d. 897 or after 905), author of a history of the world;115 al-Ṭabarī (d. 923),116 author of the monumental History of the prophets and kings; for the sake of completeness, information provided by the dependants of Theophilus of Edessa is also provided. His work, composed probably soon after 750, precedes by two generations the work of al-Wāqidī and reflects a very early stage of the Islamic historical tradition.117 112. Al-khwārizmī’s historical work is mentioned by Ibn al-Nadīm, Fihrist, transl. B. Dodge, New York 1970, vol. 2, p. 652. See A. Borrut, La circulation de l’information historique entre les sources arabo-musulmanes et syriaques : Élie de Nisibe et ses sources, in L’historiographie syriaque, éd. par M. Debié, Paris 2009, pp. 137–59, at pp. 148–55, and more generally W. Witakowski, Elias Barshnenaya’s Chronicle, in Syriac polemics : studies in honour of Gerrit Jan Reinink, ed. by W. van Bekkum, J. Drijvers and A. klugkist, Leuven 2007, pp. 219–37. 113. See C. Wurtzel, The Umayyads in the history of Khalīfa b. Khayyāṭ, PhD diss. Yale 1987, with a useful introduction on khalīfa and a translation of the section on the Umayyad period. Two editions of the single extant manuscript are available: Tārīkh Khalīfah ibn Khayyāṭ, ed. Akram Ḍiyāʾ al-ʿUmarī, 2 vols., Najaf 1967; Tārīkh Khalīfah ibn Khayyāṭ, ed. Suhayl Zakkār, 2 vols., Dimashq 1967–68. 114. Wurtzel, Khalīfa (quoted n. 113), pp. 54–6. 115. Ibn-Wādhih qui dicitur al-Jaʿqubī Historiae, ed. M. Th. Houtsma, Lugduni Batavorum 1883, vol. 2, pp. 285–6 and 302; the relevant fragments are translated in E. W. Brooks, The Arabs in Asia Minor (641–750), from Arabic Sources, JHS 18, 1898, pp. 182–208; see also Id., Additions and corrections to J.H.S. T. XVIII. Pp. 182–208”, JHS 19, 1899, pp. 31–3. 116. Annales quos scripsit Abu Djafar Mohammed ibn Djarir at-Ṭabarī, ed. M. J. de Goeje et al., Lugduni Batavorum 1879–1901; the section on the reign Muʿāwiyah is translated in The History of al-Ṭabarī. 18, Between civil wars : the caliphate of Muʿāwiyah, transl. M. G. Morony, Albany 1987. I quote al-Ṭabarī by the volume and page of the edition, followed by the volume and page of the English translation. Relevant passages are also translated in Brooks, Arabs in Asia Minor (quoted n. 115). See, more generally, The History of al-Ṭabarī. 1, General introduction and From the Creation to the Flood, transl. F. Rosenthal, Albany 1989; Al-Ṭabarī : a medieval Muslim historian and his work, ed. H. kennedy, Princeton 2008 (although R.-J. Lilie, Theophanes and al-Ṭabarī on the Arab invasions of Byzantium, pp. 219–36, brings little new); and recently Borrut, Entre mémoire et pouvoir (quoted n. 110), pp. 103–7. 117. On Theophilus, see above, pp. 256 ff.; on his date: Hoyland, Theophilus (quoted n. 78), p. 21; on his use of early Islamic sources: L. I. Conrad, Theophanes and the Arabic historical tradition : some indications of intercultural transmission, Byz. Forsch. 15, 1990, pp. 1–44, and Hoyland, Theophilus, pp. 29–34. See below, note 143, on the newly discovered text of Agapios on the years 663–8. 264 Ah (AD) MAREk JANkOWIAk al-Khwārizmī (d. ca 850) from Elias of nisibis al-Wāqidī (d. 822/3) from al-Ṭabarī heophilus of Edessa heophilus of Edessa (d. 785) (d. 785) from Agapios from heophanes 41 – (661/2) – – 42 – (662/3) – M2: Busr b. Arṭāt (great Roman defeat, many patricians killed, many captives); ʿAbd al-Malik b. Marwān (W, sea campaign), together with ʿAbd al-Raḥmān b. khālid b. al-Walīd or Busr b. Arṭāt – Muʿāwiyah makes truce with the Byzantines; Ḥabīb b. Maslamah (killed in Armenia IV) 43 k21: Arab expedi- M3: Busr b. Arṭāt Busr b. Abī Arṭāt (663/4) tion (many captives) (Greeks lee to Cple) (W, Cple) Busr b. Arṭāt (W) 44 k22: “part of Sicily” M4: ʿAbd al-Raḥ­ – (664/5) captured mān b. khālid b. alWalīd (Aqlūniyah) ʿAbd al-Raḥmān b. khālid (W) 45 k23: Abderakhman M5: ʿAbd al-Raḥ­mān – (665/6) b. khaled (W, Slavs b. khālid b. al-Walīd join him) (Slavs join him) Suwaid b. kulthūm (W) 46 k24: Bousour (666/7) M6: Busr b. Arṭāt (many captives) – ʿAbd al-Raḥmān b. khālid b. al-Walīd (W) k25: Bousour M7: Busr b. Arṭāt 47 (667/8) (Hexapolis); Phadalas (many captives) (W, Hexapolis) – Mālik b. Hubayrah (W) – Ibn Mukarrir (W) 48 k26: Phadalas (668/9) (Hexapolis), joined by Izid b. Mauias (Chalcedon, Amorion) M8: Arab reinforcements for Shābūr (Malaṭyah), joined by Yazīd b. Muʿāwiyah (Chalcedon) Table 2 – Arab raids against the Empire during the caliphate of Muʿāwiyah.118 118. Abbreviations: Cple = Constantinople; k = regnal year of Constans II; M = regnal year of Muʿāwiyah; W = winter campaign; question marks before the names of generals indicate statements introduced by qīla (“it is said”) or the like. Because of the moving beginning of Hijri years, for AH 41–51 the regnal years of Theophilus correspond to one Julian year earlier than indicated in the first column (e.g. k21 = M3 = AD 662/3). THE FIRST ARAB SIEGE OF CONSTANTINOPLE Khalīfah b. Khayyāṭ (d. 854/5) al-Ṭabarī (d. 923)—sources other than al-Wāqidī al-Yaʿqūbī (d. ca 900) 265 Ah (AD) Muʿāwiyah makes truce with the Byzantines Ḥabīb b. Maslamah (makes peace with the Roman commander) – 41 (661/2) – – great Roman defeat, several patricians killed 42 (662/3) Busr b. Arṭāt (W) Busr b. Arṭāt (W) Busr b. Abī Arṭāt (neither wintered nor reached Cple) 43 (663/4) ʿAbd al-Raḥmān b. khālid (W) ʿAbd al-Raḥmān b. khālid ʿAbd al-Raḥmān [b. khālid] b. al-Walīd (Aqlūniyah) b. al-Walīd (W); Busr b. Abī Arṭāt (sea raid) 44 (664/5) ʿAbd al-Raḥmān b. khālid (W) ʿAbd al-Raḥmān b. khālid ʿAbd al-Raḥmān b. b. al-Walīd (W, Anṭākiyah) khālid b. al-Walīd (W) 45 (665/6) Mālik b. ʿAbdallāh Abū Ḥakīm or Mālik b. ʿAbdallāh alMālik b. Hubayrah al-Fazārī (W) khathʿamī or Mālik b. Hubayrah al-Sakūnī (W) Mālik b. Hubayrah (W); Abū ʿAbd al-Raḥmān al-Qaynī (W, Anṭākiyah) Mālik b. Hubayrah alSakūnī (W) Abū ʿAbd al-Raḥmān al-Qaynī ʿAbd al-Raḥmān al-ʿAtbī (Anṭākiyah al-Sawdā’ or Ibn Mukraz of the Banū ʿĀmir b. Lu’ayy (W, Anṭākiyah) [the Black]) Mālik b. ʿUbaydallāh or ʿAbd alRaḥmān b. khālid b. al-Walīd or Mālik b. Hubayrah al-Sakūnī (W) 46 (666/7) Mālik b. Hubayrah (W); Abū ʿAbd al-Raḥmān al-Qaynī (W, Anṭākiyah) 47 (667/8) ʿAbd al-Raḥmān al-Qaynī (W, Anṭākiyah); ʿAbdallāh b. Qays al-Fazārī; Mālik b. Hubayrah al-Sakūnī (sea raid); ʿUqbah b. ʿĀmir al-Juhanī, al-Mundhir b. al-Zuhayr, khālid b. ʿAbd alRaḥmān b. khālid b. al-Walīd (joint sea raid) 48 (668/9) 266 MAREk JANkOWIAk Ah (AD) al-Wāqidī (d. 822/3) from al-Ṭabarī heophilus of Edessa heophilus of Edessa (d. 785) (d. 785) from Agapios from heophanes al-Khwārizmī (d. ca 850) from Elias of nisibis 49 – (669/70) – – Faḍālah b. ʿUbayd (W) 50 – (670/1) – – Ḥuṣayn b. Numayr (defeated the Roman army) 51 k2: Phadalas (W, (671/2) kyzikos) – – Busr b. Arṭāt (W), Yazīd b. Muʿāwiyah (Cple) 52 (672) k3: Bousour (many M12: Busr b. Arṭāt captives) (many captives) Muḥammad b. Sufyān b. ʿAwf alAzdī (W, died, suc- ʿAbd al-Raḥmān (W) ceeded by ʿAbdallāh b. Masʿadah al-Fazārī) M13: many ships 53 k4: big leet raid the Rūm (672/3) commanded by Mouamed b. Abdelas (W, Smyrna) and kaisos (W, Cilicia and Lycia) Junādah b. Abī Umayyah (Rūdus) Sufyān b. ʿAwf (W) 54 k5: Souphian b. Awf M14: great Arab (673/4) (defeated by Romans defeat on land and on land and sea[?]) sea in Lūkiyah Junādah b. Abī Umayyah (Arwād) ʿAbd al-Raḥmān b. Masʿūd (W) 55 k6: Abdelas b. (674/5) kaisos, Phadalas (both W, Crete) – Sufyān b. ʿAwf alAzdī (W) ʿAbdallāh b. Qays (W) 56 – (675/6) – – Mālik b. ʿAbdallāh (W) 57 – (676/7) M17: Roman – invasion of Lebanon Yazīd (W) Table 2 – Arab raids against the Empire during the caliphate of Muʿāwiyah (continued). THE FIRST ARAB SIEGE OF CONSTANTINOPLE Khalīfah b. Khayyāṭ (d. 854/5) al-Ṭabarī (d. 923)—sources other than al-Wāqidī al-Yaʿqūbī (d. ca 900) Mālik b. Hubayrah or Faḍālah Faḍālah b. ʿUbayd b. ʿUbayd al-Anṣārī (W); (took many captives) ʿAbdallāh b. Masʿadah (W, land raid) 267 Ah (AD) 49 Mālik b. Hubayrah al-Sakūnī (669/70) (W); Faḍālah b. ʿUbayd (W, kharrah[?]); ʿAbdallāh b. kurz al-Bajalī; Yazīd b. Shajarah al-Rahāwī and ʿUqbah b. Nāiʿ (W, sea raid); Yazīd b. Muʿāwiyah and four Companions (Cple) Yazīd b. Muʿāwiyah and Abū Ayyūb al-Anṣārī; ʿAbdallāh b. ʿĀmir (W) Busr b. Arṭāt; Sufyān b. ʿAwf (W) Busr b. Abī Arṭāt and Sufyān b. ʿAwf al-Azdī; Faḍālah b. ʿUbayd al-Anṣārī (sea raid) 50 (670/1) Faḍālah b. ʿUbayd al-Anṣārī (W, naval campaign) Muḥammad b. ʿAbd al-Raḥmān; Faḍālah b. ʿUbayd al-Anṣārī (W) Faḍālah b. ʿUbayd (W); Busr b. Abī Arṭāt 51 (671/2) Busr b. Arṭāt and Sufyān b. ʿAwf Sufyān b. ʿAwf (died, al-Zuhrī (W) succeeded by ʿAbdallāh b. Masʿadah al-Fazārī) Busr b. Abī Arṭāt and Sufyān b. ʿAwf al-Azdī (W); Muḥammad b. ʿAbdallāh al-haqafī 52 (672) ʿAbd al-Raḥmān b. Umm alḤakam (W) Muḥammad b. Mālik; ʿAbd al-Raḥmān b. Umm al?Junādah b. Abī Umayyah Ḥakam al-haqafī (W) al-Azdī (Ṭarsūs) 53 (672/3) Muḥammad b. Mālik (W) – Muḥammad b. Mālik (W); Maʿn b. Yazīd al-Sulamī 54 (673/4) Sufyān b. ʿAwf (W), Yazīd b. Shajarah al-Rahāwī (killed?) Mālik b. ʿAbdallāh alkhathʿamī (W) ʿAmr b. Muḥriz or ʿAbdallāh b. Qays al-Fazārī or Mālik b. ʿAbdallāh (W) 55 (674/5) Masʿūd b. Abī Masʿūd or Junādah b. Abī Umayyah (W) Yazīd b. Muʿāwiyah (Cple); Masʿūd b. Abī Masʿūd (W); Yazīd b. Shajarah (land raid); ʿIyād b. alḤārith (sea raid) (“all this is uncertain”119) Junādah b. Abī Umayyah or ʿAbd al-Raḥmān b. Masʿūd (W); Yazīd b. Shajarah (sea raid); ʿIyād b. al-Ḥārith (land raid) 56 (675/6) ʿAbdallāh b. Qays (W) ʿAbdallāh b. Qays ʿAbdallāh b. Qays (W) 57 (676/7) 119 119. Brooks, Arabs in Asia Minor (quoted n. 115), p. 188, connects the last words of the entry for AH 56, “all this is said” (i.e., uncertain, kullu hadha yuqālu), with the beginning of the following entry, “in the year 57”, and translates: “all these things are also said to have happened in the year 57.” 268 MAREk JANkOWIAk Ah (AD) al-Khwārizmī (d. ca 850) from Elias of nisibis al-Wāqidī (d. 822/3) from al-Ṭabarī heophilus of Edessa heophilus of Edessa (d. 785) (d. 785) from Agapios from heophanes – 58 k9: invasion (677/8) of Mardaites in Lebanon, Muʿāwiyah sues for peace Yazīd b. Shajarah (killed at sea) ʿAmr b. Murrah (W) 59 – (678/9) – no sea raid Junādah b. Abī Umayyah (W, Rūdūs) 60 – (679/80) – Mālik b. ʿAbdallāh – (Sawriyah); Junādah b. Abī Umayyah (Rūdus) 61 – (680/1) – – – Table 2 – Arab raids against the Empire during the caliphate of Muʿāwiyah (continued). THE FIRST ARAB SIEGE OF CONSTANTINOPLE 269 Khalīfah b. Khayyāṭ (d. 854/5) al-Yaʿqūbī (d. ca 900) al-Ṭabarī (d. 923)—sources other than al-Wāqidī Mālik b. ʿAbdallāh (W), Yazīd b. Shajarah al-Rahāwī (killed with his companions); added by Baqī120: Akdar and Saʿīd b. Yazīd (Rūdus); ʿAmr b. Murrah (W, Badhandūn); al-Ḥuṣayn b. Numayr Mālik b. ʿAbdallāh alkhathʿamī; ?ʿAmr b. Yazīd al-Juhanī; Yazīd b. Shajarah (sea raid, killed121) Mālik b. ʿAbdallāh al-khathʿamī; Yazīd b. Shajarah (W); ?Junādah b. Abī Umayyah (sea raid) 58 (677/8) ʿAmr b. Murrah al-Juhanī ʿAmr b. Murrah al-Juhanī (W, land campaign); no naval (land raid), no naval campaign; added by Baqī: campaign Junādah b. Abī Umayyah, ʿAlqamah b. Junādah al-Ḥajarī, ʿAlqamah b. al-Akhtham (Rūdus) ʿAmr b. Murrah al-Juhanī (W, land campaign), ?Junādah b. Abī Umayyah (sea raid) 59 (678/9) – added by Baqī: Egyptians carried food to Rūdus; ʿAmr b. Muʿāwiyah al-ʿUqaylī; ʿAwwām al-Yaḥṣubī, ʿAbdallāh b. Qays al-Fazārī, ʿAbd al-ʿAzīz b. Marwān (m.n.q.l.bah122); Mālik (Sawriyah) – 60 (679/80) – 61 (680/1) Mālik b. ʿAbd al-Raḥmān al-khathʿamī or, according to Baqī, Mālik b. ʿAbdallāh (Qūniyah) Mālik b. ʿAbdallāh alkhathʿamī (Sawriyah) Ah (AD) 120121122 120. See below, p. 272, on the notices inserted by Baqī b. Makhlad into the Tārīkh of khalīfah b. khayyāṭ. 121. Ibn-Wādhih qui dicitur al-Jaʿqubī Historiae, vol. 2, p. 286: wa qīla Yazīd ibn Shajarah fī al-baḥr (“and it is said that Yazīd b. Shajarah [raided] by sea”). Brooks, Arabs in Asia Minor (quoted n. 115), p. 188 n. 3 (note misplaced, see Brooks, Additions [quoted n. 115], p. 31), reads ‫“( قتل‬was killed”) for ‫“( قيل‬it is said”) in order to “get the same as in [al-Wāqidī quoted in] al-Ṭabarī”, and translates: “and Yazīd, the son of Shagara, was killed at sea.” Ibn khayyāṭ, who uses a different verb (‫“ أصيب‬to be killed”), supports Brooks’ reading. 122. The word ‫ منقلبة‬is not intelligible; Wurtzel, Khalīfa (quoted n. 113), p. 438 n. 263, proposes to read ‫“ صقلية‬Sicily”. An expedition led by ʿAbdallāh b. Qays b. Makhlad al-Dizaqī against Sicily in the reign of Muʿāwiyah is mentioned by al-Balādhurī (who quotes al-Wāqidī), Liber expugnationis regionum, ed. M. J. de Goeje, Lugduni Batavorum 1866, p. 235; English translation: al-Balādhurī, The origins of the Islamic state (Kitāb futūḥ al-buldān), transl. P. k. Ḥitti, vol. 1, New York 1916, p. 375. 270 MAREk JANkOWIAk The lists increase in complexity over time: al-Ṭabarī, for instance, rather than selecting one of the multiple traditions he gathered, includes all of them in his History, even if he occasionally marks his disagreement.123 All sources tabulated above appear to share a substantially identical body of material: in all of them, for instance, the same generals led the first raids against Byzantium—Busr b. Abī Arṭāt, ʿAbd al-Raḥmān b. khālid b. al-Walīd, Mālik b. Hubayrah and Faḍālah b. ʿUbayd, in this sequence. They also agree on the rare additional details on the expeditions, such as their targets or the names of the generals who died on campaign. Where the Islamic chronicles are at variance with each other is in the absolute dating of the events. Oral and written traditions, which may have taken the shape of lists of generals at an early stage, seem to have initially lacked exact dates.124 They were fitted into a chronological framework only when the annalistic format of the Islamic chronicles was developed over the course of the 8th and 9th centuries.125 The widely divergent dates of individual raids illustrate this point. The high-profile expedition of Yazīd, son of the caliph, is dated to 49/669–70 by al-Ṭabarī, a year later by Ibn khayyāṭ, two years later by al-khwārizmī, to 55/674–5 by Abū Zurʿah al-Dimashqī (d. 894),126 and finally to 56/675–6 by al-Yaʿqūbī. Elsewhere, al-Ṭabarī gives the date of 52/672.127 Theophanes and Agapios place it in a year that corresponds to 47/667–8 or 48/668–9. If al-Yaʿqūbī’s date is the result of confusion with another Yazīd, son of Shajarah, who campaigned around 56/675–6,128 no year between AH 47 and 55 can be preferred on internal evidence. Other raids are listed in two consecutive years, reflecting 123. See, on this approach, Ch. F. Robinson, Islamic historiography, Cambridge 2003, p. 79: “Medieval Muslim historians held an inclusive and catholic view of the past, which accommodated multiple reconstructions.” 124. Similar, perhaps, to the lists of the leaders of the hajj (see below, note 247, for an example of an uncertain absolute date) or of caliphs; on the latter see now Borrut, Entre mémoire et pouvoir (quoted n. 110), pp. 152–4. 125. On the unlikely existence of an early Islamic annalistic tradition, despite the introduction of the Hijri era not later than in the early 20s/640s, see A. Noth, in collab. with L. I. Conrad, The early Arabic historical tradition : a source-critical study, 2nd ed., Princeton 1994, pp. 40–5; F. M. Donner, Narratives of Islamic origins : the beginnings of Islamic historical writing, Princeton 1998, pp. 230–48 (see p. 231: “for the early period, at least, chronology constitutes one of the weakest points of the Islamic historical tradition”). Hoyland, Theophilus (quoted n. 78), pp. 31–3, proposes a West-Syriac inspiration for Islamic annalistic writing. Absolute dating of reigns of the caliphs still posed problems in the 9th century, as demonstrated by the hesitations over the death of Muʿāwiyah, placed by Ibn khayyāṭ in Rajab 59, but by most later authors in Rajab 60. 126. Abū Zurʿah al-Dimashqī, Tārīkh, ed. Shukrallāh Niʿmatallāh al-Qāwjānī, Damascus 1980, vol. 1, p. 188. 127. Al-Ṭabarī ii.2324 = xxxix.40. 128. Mentions of Yazīd b. Shajarah: al-Wāqidī AH 58, Ibn khayyāṭ AH 55 and 58, al-Yaʿqūbī AH 56 and 58, al-Ṭabarī AH 49, 56 and 58. Al-khwārizmī has a Yazīd without a patronymic under AH 57: such notices gave rise to confusion whenever there were two generals with the same first name (ism). If al-Yaʿqūbī replaced Yazīd b. Muʿāwiyah by Yazīd b. Shajarah in AH 56, al-Ṭabarī did the opposite, listing Yazīd b. Shajarah next to Yazīd b. Muʿāwiyah in AH 49. Similarly, the identity of the Mālik who led the campaign of AH 46 is uncertain, see Yaʿqūbī AH 46: “Mālik b. ʿAbdallāh al-khathʿamī made a raid, but it is (also) said that it was Mālik b. Hubayrah al-Sakūnī, and he wintered in the land of the Romans.” THE FIRST ARAB SIEGE OF CONSTANTINOPLE 271 hesitation as to their correct date rather than multiple campaigns.129 Finally, some dates are internally contradictory: al-Wāqidī has Sufyān b. ʿAwf lead the expedition of AH 55, even though he reported his death three years earlier, during the raid of AH 52. If Muslim historians were unable to establish the precise dates of the raids against the Byzantine Empire under Muʿāwiyah, their relative chronology is more solid. As already remarked, the first several years of raiding were relatively uncontroversial, with the same generals appearing in the same order in all the lists. After AH 47 this consensus breaks down: although the same names still recur in all the sources, their sequence can no longer be reconciled; as a result their absolute dates are disordered, as we have seen with the example of Yazīd b. Muʿāwiyah. The lists start again to look similar towards the end of the caliphate of Muʿāwiyah: in most of them Yazīd b. Shajarah, ʿAbdallāh b. Qays, Mālik b. ʿAbdallāh and ʿAmr b. Murrah led the last Arab raids against Byzantium between 55/674–5 and 59/678–9.130 This is suggestive of a relatively stable list of generals that raided Byzantium during the reign of Muʿāwiyah, which was pinned to the beginning of Muʿāwiyah’s reign, with divergences accumulating over time. The similarities between the lists result from a shared tradition, not from direct connections between the earliest extant Islamic chronicles, which are few: if al-Wāqidī was used by al-Ṭabarī and perhaps also by al-Yaʿqūbī,131 the works of al-khwārizmī, Ibn khayyāṭ and al-Yaʿqūbī show no signs of mutual dependence.132 As for al-Ṭabarī, even though most notices of Ibn khayyāṭ and al-Yaʿqūbī are included in his Tārīkh, it is unclear if he had direct access to their works.133 Thus, if the absolute dates of the Muslim chronographers cannot be trusted, the broadly concordant sequence of generals they transmit points towards a coherent body of traditions. An early witness to these traditions is the chronicle of Theophilus of Edessa, as transmitted by Theophanes and Agapios. All the generals it mentions can be recognized in Islamic sources, including Mouamed son of Abdelas and kaisos who led the big Arab fleet that wintered at Smyrna and in Lycia and Cilicia in AM 6164. The first name of the latter has probably been omitted by haplography, in which case he can be identified with ʿAbdallāh b. Qays, known from the Islamic sources and mentioned by 129. E.g. the raid of ʿAbd al-Raḥmān al-Qaynī against Anṭākiyah, reported by al-Ṭabarī under AH 47 and 48. 130. Only al-khwārizmī has a different sequence of names, with Yazīd between Mālik and ʿAmr. He also dates all these campaigns 1–2 years earlier than the other sources. 131. Al-Ṭabarī quotes al-Wāqidī extensively; al-Yaʿqūbī repeats in similar words al-Wāqidī’s short notices on the successor of Sufyān b. ʿAwf al-Azdī, who died on campaign in 52/672, and on the lack of naval campaign in 59/678–9. Al-Wāqidī does not appear among the sources of Ibn khayyāṭ listed by Wurtzel, Khalīfa (quoted n. 113), pp. 45–54. 132. Al-khwārizmī, in particular, was not used by later historians: he lists two generals unknown otherwise (Suwaid b. kulthūm in AH 45 and Ibn Mukarrir in AH 48, perhaps identical with Ibn Mukraz of the Banū ʿĀmir b. Lu’ayy mentioned by Ibn khayyāṭ in the same year), and two known only from one of the above lists (in 50/670–1 Ḥuṣayn b. Numayr, on whom see below, and in 54/673–4 ʿAbd al-Raḥmān b. Masʿūd, mentioned two years later by al-Ṭabarī). 133. Almost all the notices of Ibn khayyāṭ (with the exception of AH 50, 55 and 56 and single names in AH 48 and 49) and most of those of al-Yaʿqūbī (with the exception of AH 61 and single names in AH 51, 53, 56, 58 and possibly 48) are included in al-Ṭabarī’s work. He does not seem, however, to have directly used the Tārīkh of Ibn khayyāṭ: Wurtzel, Khalīfa (quoted n. 113), pp. 59–60. 272 MAREk JANkOWIAk Theophanes under AM 6166; the former is probably Muḥammad b. ʿAbdallāh al-Thaqafī known from al-Ṭabarī. Theophilus’ sequence of the generals also broadly corresponds to that of the Islamic chronicles. The list of generals in the “Oriental source” is thus based on early Islamic traditions, the only possible source of information on the Arab commanders.134 The chronicle of Theophilus of Edessa, composed probably in the 750s, thus demonstrates that three generations after the events, and three generations before any other extant Islamic tradition, the available information on the raids against the Empire had already essentially the same shape as in the later Islamic chronicles. Where Theophilus diverges from them is either, again, in chronology—he reports, for instance, an expedition of Faḍālah b. ʿUbayd in AD 674/5, although most Islamic sources have him die in Damascus in 53/672–3135—or in reporting expeditions led by pairs of generals where Arab annals record single names.136 Prudence is therefore in order when using Theophilus’ chronology of the Arab raids on Anatolia, but his list of Arab generals confirms the early circulation of the relevant Islamic traditions.137 Their earliest stages cannot, however, be otherwise accessed, because the notices of the Muslim chroniclers on raids against Byzantium often lack isnāds, or chains of transmitters.138 We know that al-Ṭabarī relied on al-Wāqidī (d. 822), and Ibn khayyāṭ on Hishām b. Muḥammad al-kalbī (d. 819), a genealogist and historian from Basra whose father (d. 763) was already an eminent scholar interested in genealogy, but this does not lead us beyond the times of Theophilus. Both al-Ṭabarī and Ibn khayyāṭ, as well as other extant representatives of the Abbasid “historiographical vulgate”,139 relied on traditions gathered by Hijazi and Iraqi traditionists, at the expense of the Syrian tradition, closer to the Umayyad court and to the theatre of the war against Byzantium, which was certainly better informed on events. Glimpses of this rich tradition can be found in the notices on the years AH 58–61 (AD 677/8–680/1) added by Baqī b. Makhlad to the History of his teacher Ibn khayyāṭ on the basis of information he gathered in Damascus from such lost codifications of Syrian traditions as the works of al-Walīd b. Muslim al-Umawī (d. 810).140 Baqī’s additions mention new place names, probably an unknown raid against Sicily in 60/679–80,141 and several generals who do not appear in the parallel lists, such as ʿAbd al-ʿAzīz b. Marwān, no doubt the brother of ʿAbd al-Malik and the future governor of Egypt (685–704), or al-Ḥuṣayn b. Numayr, the bête noire of later Islamic historiography because of his role in the siege of Mecca in 683 when the kaʿbah 134. On Theophilus’ use of Islamic sources, see above, note 117. 135. L. Caetani, Chronographia Islamica, Paris 1912, AH 53, pp. 590–1. 136. Neither Muḥammad and ʿAbdallāh b. Qays (Theoph. AM 6164) nor ʿAbdallāh b. Qays and Faḍālah (Theoph. AM 6166) are mentioned together in one entry by any Islamic source. 137. On the “noyau commun” of early Islamic traditions shared by local historical “schools”, see also Borrut, Entre mémoire et pouvoir (quoted n. 110), pp. 35–6. 138. Numerous isnāds are preserved in the Tārīkh madīnat Dimashq of Ibn ʿAsākir, but their study goes beyond the scope of the present paper. 139. Borrut, Entre mémoire et pouvoir (quoted n. 110), pp. 97–103. 140. See on him Sezgin, Geschichte (quoted n. 111), vol. 1, p. 293; Borrut, Entre mémoire et pouvoir (quoted n. 110), p. 54. 141. New place names: Badhandūn, probably Podandos on the northern approaches to the Cilician Gates in 58/677–8; Qūniyah (Ikonion) or perhaps Qūtiyah (kotyaion?) in 61/680–1. Sicily: see above, note 122. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 273 was burned down.142 The role of the Marwanids—Marwān b. al-Ḥakam and his sons ʿAbd al-Malik and ʿAbd al-ʿAzīz—in the military campaigns of Muʿāwiyah was expunged from the Abbasid vulgate; it is, however, attested by Islamic traditions transmitted by Agapios of Membidj.143 Similarly, al-Ḥuṣayn b. Numayr is credited by al-khwārizmī with a brilliant victory over the entire Roman army with only 3,000 soldiers in 50/670–1,144 an entry absent from the classical chronographies. But the richest mine of the Syrian historical tradition is the still largely untapped monumental Tārīkh madīnat Dimashq of Ibn ʿAsākir (d. 571/1176),145 a biographical dictionary in 80 volumes, containing, among the over 10,000 entries, traditions relative to generals active in the warfare against Byzantium in the reign of Muʿāwiyah.146 With these exceptions, the extant Islamic traditions on the war against the Byzantines have been passed through the filter of classical historiography, hostile to the Umayyads, and provide only the barest and selective outline of events. Is it still possible to build a narrative from their dry notices? As we have seen, the precise chronology of events cannot be recovered, but occasional mentions of geographical names and other details allow us to sketch the progress of the Arab-Byzantine war in the 660s and 670s. Muʿāwiyah broke the peace with the Empire as soon as the first Arab civil war was over. Agapios and al-Ṭabarī report a disastrous defeat of the Roman army and the death of several patricians during the first Arab raid in 42/662–3,147 followed in the next year by an incursion said to have reached Constantinople under the command of Busr b. Abī Arṭāt. Little is known of the next campaigns: in 44/664–5 ʿAbd al-Raḥmān, son of the famous general khālid b. al-Walīd, targeted Aqlūniyah, no doubt koloneia in north-eastern Anatolia, and in the late 40s/660s an Antioch was attacked, but it is difficult to decide whether Antioch of Pisidia, Antioch of Isauria, or perhaps Antioch of Syria is meant.148 According to al-Ṭabarī, raiding peaked in 48/668–9 and 49/669–70, with six generals invading the Empire in each of these years: two big sea campaigns are mentioned, of the Egyptians and the people of Medina in 48/668–9, and of the Syrian and Egyptian fleets in the 142. Other otherwise unattested generals are Akdar, Saʿīd b. Yazīd (both AH 58), ʿAmr b. Muʿāwiyah al-ʿUqaylī and ʿAwwām al-Yaḥṣubī (both AH 60). ʿAlqamah b. Junādah al-Ḥajarī and ʿAlqamah b. al-Akhtham (both AH 59) are mentioned by Ibn ʿAsākir, see below, note 151. See the notes of Wurtzel, Khalīfa (quoted n. 113), on all these individuals, and the entry in EI 2 on al-Ḥuṣayn b. Numayr. 143. In the passage on years 663–8 recently discovered by Robert Hoyland in the cod. Laurentianus Or. 323, fol. 97v–98r, to be published in the proceedings of the Theophanes conference held in Paris in September 2012. I am grateful to Robert Hoyland for making this text available to me. 144. El. Nisib. AH 50, p. 69: “Ḥuṣayn b. Numayr met the entire Roman army when he had 3,000 men from the Arabs, and defeated the Romans.” 145. ʿAlī ibn al-Ḥasan Ibn ʿAsākir, Tārīkh madīnat Dimashq, ed. ʿUmar ibn Gharāmah al-ʿAmrawī, Bayrūt 1995–2000; see also a recent collection of studies: Ibn ʿAsākir and early Islamic history, ed. by J. E. Lindsay, Princeton 2001; Borrut, Entre mémoire et pouvoir (quoted n. 110), pp. 120–4. 146. Information deriving from the Tārīkh madīnat Dimashq will be used only occasionally; this source awaits a systematic survey. 147. Both passages are textually close, compare al-Ṭabarī ii.16, lines 15–6, and Agap. p. 487, lines 4–5. 148. Anṭākiyah: Ibn khayyāṭ AH 47, 48; al-Yaʿqūbī AH 45 and 48; al-Ṭabarī AH 47, 48. Brooks, Arabs in Asia Minor (quoted n. 115), pp. 184–5 thought of Antioch in Pisidia, and tentatively proposed to identify “Antioch the Black” (mentioned by al-Yaʿqūbī in AH 48) with Antioch in Isauria. But Ibn khayyāṭ, AH 47, opposes arḍ al-Rūm and Anṭākiyah, as if the latter did not belong to the former. 274 MAREk JANkOWIAk following year. In the same year 49/669–70, al-Ṭabarī briefly reported the expedition against Constantinople commanded by Yazīd b. Muʿāwiyah, who was accompanied by four Companions of the Prophet, and the raid of Faḍālah b. ʿUbayd against a place often identified, without sufficient support, with the Tunisian island of Jerba.149 After this climax notices again become more sketchy. Judging from the place-names mentioned by the chroniclers—Tarsus, Rhodes, Podandos (Badhandūn) at the Cilician Gates, and perhaps Isauria150—warfare moved away from Constantinople. It probably consisted in tit-for-tat naval campaigns: Syrian traditions report an Arab expedition against Iṣṭādh.nah (on the African coast?) and Crete,151 and a Byzantine attack on the Syrian coast just south of Bāniyās, where the governor of Ḥimṣ ʿAbdallāh b. Qurṭ al-Thumālī al-Azdī was killed; both events are dated to 56/675–6.152 The impression that the Byzantines were on the counterattack is confirmed by the death of two leaders of Arab raiding parties in the 670s: Sufyān b. ʿAwf al-Azdī probably in 52/672 and Yazīd b. Shajarah, explicitly said to have been killed together with his companions, around 58/677–8. No Arab general is known to have died in the 660s, the reports of their deaths sound therefore like reluctant acknowledgements of major Arab defeats. The raids continued until the year of Muʿāwiyah’s death, albeit with a significantly decreased momentum: sources notice that there was no sea raid against Byzantium in 59/678–9,153 as if this was the first year 149. Constantinople: see below, pp. 290. Jerba: Arabic Jarabbah (‫ ) َج َربَة‬is a conjecture proposed by the modern editors (al-Ṭabarī ii.86, note e) for kharrah (‫)خ ّره‬, or perhaps Jarah or Ḥarah, of the manuscripts, on the basis of Yāqūt, Geographisches Wörterbuch, ed. F. Wüstenfeld, vol. 2, Leipzig 1867, p. 47, where a raid led by Faḍālah b. ʿUbayd against Jarabbah is mentioned; Yāqūt seems, however, to place Jarabbah in Yemen. The connection with Jerba, although universally assumed (e.g. W. E. kaegi, The interrelationship of seventh-century Muslim raids into Anatolia with the struggle for North Africa, Byz. Forsch. 28, 2004, pp. 21–43, at pp. 26–7; Id., Muslim expansion and Byzantine collapse in North Africa, Cambridge 2010, p. 180; Howard-Johnston, Witnesses [quoted n. 3], pp. 389 and 491–2), remains to be proved. 150. Another placename, al-M.ndrūn (‫ )المندرون‬or al-M.ndūn (‫)المندون‬, attested as the winter quarters of the army led in 58/677–8 by ʿAmr b. Murrah (Ibn ʿAsākir, Tārīkh madīnat Dimashq, vol. 14, p. 385), may be a corruption of otherwise attested al-Badhandūn (‫البذندون‬, Podandos). 151. Ibn ʿAsākir, Tārīkh madīnat Dimashq, vol. 11, p. 298: “in the year 56 there was the raid of ʿĀbis b. Saʿīd and Mālik b. ʿAbdallāh al-khathʿamī against Iṣṭādh.nah. ʿĀbis led the people of Egypt, Junādah b. Abī Umayyah the people of Syria, and Mālik b. ʿAbdallāh was in command of them all. They wintered in Iqrīṭiyyah in the year of hunger after their return from Iṣṭādh.nah, and in the year 59 there was the raid of Junādah b. Abī Umayyah, ʿAlqamah b. Junādah al-Ḥajarī and ʿAlqamah b. al-Ajtham [read al-Akhtham, see above, note 142] against Rūdus.” Yāqūt, Geographisches Wörterbuch, vol. 1, p. 299, adds that ʿĀbis was sent by Maslamah b. Mukhallad, the governor of Egypt, and accordingly places Iṣṭādh.nah in the Maghrib. 152. Ibn ʿAsākir, Tārīkh madīnat Dimashq, vol. 32, pp. 5–14, esp. pp. 9–10: “ʿAbdallāh b. Qurṭ was the governor of Ḥimṣ; he came out by night to watch over the sea coast and encountered †māthūr of the Romans and was killed by him in the place called the Tower of Ibn Qurṭ, between Bāniyās and Maraqiyyah”. He was also in charge of guarding the coast ( ʿalā shāṭiʾ al-sāḥil), which indicates an external threat to the Syrian littoral. On Maraqiyyah, a small castle mid-way between Bāniyās and Arwād, see G. Le Strange, Palestine under the Moslems : a description of Syria and the Holy Land from AD 650 to 1500, London 1890, p. 400. On ʿAbdallāh b. Qurṭ, see also A. Mazor, The Kitāb futūḥ al-Shām of al-Qudāmi as a case study for the transmission of traditions about the conquest of Syria, Der Islam 84, 2007, pp. 17–45, at pp. 23–4. 153. ʿAmr’s campaign on land: Ibn khayyāṭ, al-Yaʿqūbī and al-Ṭabarī; no sea raid in 59/678–9: al-Wāqidī, Ibn khayyāṭ and al-Yaʿqūbī. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 275 without a naval campaign since the beginning of the war, while the last recorded raid took place in 60/679–80 or 61/680–1 against Sawriyah (Isauria?).154 An episode of the last phase of Byzantine-Arab warfare under Muʿāwiyah has received much attention from Lawrence I. Conrad in his important study on the “conquest of Arwad”:155 the garrisoning of a Byzantine island by an Arab force led by Junādah b. Abī Umayyah al-Azdī. All the early Islamic chronographies studied in this chapter list at least one raid by Junādah between 53/672–3 and 60/679–80; al-Ṭabarī mentions him in nearly every entry between these dates, although he frequently notes that his sources provide conflicting information. Junādah’s target is usually named as Rhodes (Rūdus), although al-Yaʿqūbī has Ṭarsūs, and al-Wāqidī both Rūdus and Arwād.156 Al-Ṭabarī uncharacteristically copied the circumstantial story of the Arab garrison from al-Wāqidī under 53/672–3, and again, in different terms but also from al-Wāqidī, in the following year.157 Junādah is said to have conquered Rūdus or Arwād, “an island in the sea near Constantinople”, where he established a permanent garrison. Rather than pose a danger to the Byzantines, however, the Muslims were constantly harassed by the enemy who “blockaded them by sea and cut off their ships”. They nonetheless held for seven years, until Yazīd b. Muʿāwiyah brought them back after the death of his father. This obscure episode has been interpreted by Conrad as a distorted echo of the Arab conquest of Arwad 20 years earlier and construed as an archetypal example of the unreliability of the early Islamic traditions. This is to go too far. Although the stories of Junādah’s occupation of Rhodes and of the conquest of Arwad are confused time and again in Islamic sources,158 nothing indicates that they are two versions of the same episode: the two narratives have little in common, and the expedition of Junādah is firmly placed in the last years of Muʿāwiyah’s caliphate. The judgment passed by Conrad on the story of Junādah as an “artificial composite elaborated from earlier undifferentiated fragments in a completely arbitrary fashion”, and thus “for the most part ahistorical”,159 is therefore peremptory. There are no grounds to reject Junādah’s occupation of Rhodes as an outright fabrication,160 even if it is unclear why precisely this episode of at best secondary importance is the only circumstantial narrative of the Byzantine-Arab war under Muʿāwiyah that al-Ṭabarī deemed important to include in his History. 154. See below, note 278, for a mention of ship construction continuing into the reign of Yazīd. 155. Conrad, The conquest of Arwad (quoted n. 81). 156. These names are corrupt in the manuscripts of al-Ṭabarī: D.rūs in AH 53, ii.157, and Arwādah or Awrād in AH 54, ii.163. 157. Al-Ṭabarī AH 53, ii.157 = xviii.166 and AH 54, ii.163 = xviii.172. 158. Already by Ibn Aʿtham al-kūfī (fl. ca 819?, see Borrut, Entre mémoire et pouvoir [quoted n. 110], pp. 91–3) in a passage quoted by Conrad, The conquest of Arwad (quoted n. 81), pp. 350–1, where Junādah appears as the conqueror of Arwad, and by al-Wāqidī, the common source of al-Ṭabarī and al-Balādhurī, Kitāb futūḥ al-buldān, p. 236 de Goeje (pp. 375–6 Ḥitti) [quoted n. 122], who also has two reports, first on Junādah’s conquest of Rūdus in AH 52, and then his occupation of Arwād in AH 54. 159. Conrad, The conquest of Arwad (quoted n. 81), pp. 374–5 and 365. 160. Nor does an earlier attack on Rhodes in 653/4 (see e.g. Theoph. AM 6145, p. 3458–11) preclude the reality of an Arab occupation of this island in the 670s, pace Conrad, The conquest of Arwad (quoted n. 81), pp. 376–7. 276 MAREk JANkOWIAk Other historians associated the seven-year occupation of an island near Constantinople around 673–680 with the Arab winter base in kyzikos during the seven-year siege of Constantinople described by the “patrician Trajan”. 161 However, apart from the number of seven years, neither the place names, nor the names of the generals, nor the circumstances—Junādah’s garrison was blockaded by the Byzantines rather than blockading the Byzantine capital—coincide; and, of course, Constantinople was not besieged by the Arabs in the 670s when Junādah was starving at Rhodes. Theophilus and the Islamic traditions recorded in the 9th century thus present a parallel narrative of Byzantine-Arab warfare in the 660s and 670s: a period of devastating raids culminating at the end of the 660s in the campaign led by the son of the caliph, followed in the 670s by a period more favourable to the Byzantines, who were able to regain initiative and confine Arab raids to the border area. Although Theophilus is not independent from the Islamic traditions, he also used other sources: details on Byzantine politics, on the patricians who defeated Sufyān b. ʿAwf, on the discovery of Greek fire, and on natural phenomena, all absent from Islamic historiography, derive from Byzantine and Syriac sources. His success in combining them in a seamless fashion indicates that they offered complementary narratives. None of them featured a siege of Constantinople in the 670s, when the Empire was occupied on other fronts, as confirmed by a number of sources to which we now turn. he turning of the tide (ca 672–4) Theophilus of Edessa devoted unusual attention to the Byzantine victory over Sufyān b. ʿAwf in 673/4, implying that this encounter put an end to the Arab inroads deep into Anatolia and thus changed the course of the war. Theophilus’ notice is preserved by three of his dependants—Theophanes, Agapios and Michael the Syrian162—whose narratives largely overlap: they all speak of a battle, in which the Byzantines, led by three patricians, killed 30,000 Arabs before destroying the Arab fleet with Greek fire; and they all date it to 673/4. They differ only in details that they chose to include: while Theophanes gives the names of the generals of both armies, Michael the Syrian specifies the location of the encounter as a coastal city in Lycia, besieged by the Arabs and relieved by the patricians. No other source refers explicitly to this battle, but it is tempting to associate it with the wreckage of the Arab fleet at Syllaion mentioned by the “patrician Trajan”, as both events are presented as major Arab disasters, both constitute a turning point in the respective narratives of the Byzantine-Arab war, and both took place in the same part of Anatolia. It remains, however, an open question whether the coastal city in Lycia mentioned by 161. See, for instance, J. Wellhausen, Die kämpfe der Araber mit den Romäern in der Zeit der Umaijiden, Nachrichten von der königlichen Gesellschaft der Wissenschaften zu Göttingen, philologischhistorische Klasse 4, 1901, pp. 414–47, at p. 425; English translation by M. Bonner: Arab wars with the Byzantines in the Umayyad period, in Arab-Byzantine relations in early Islamic times, ed. by M. Bonner, Aldershot 2004, pp. 31–64, at p. 42. 162. Theoph. AM 6165, p. 35411–17, quoted above, p. 8; Agap. p. 492; Mich. Syr. xi.13, p. 455. All these texts are translated in Hoyland, Theophilus (quoted n. 78), pp. 167–8. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 277 Michael the Syrian can be identified with Syllaion, technically a city in Pamphylia, although situated at small distance from both the Mediterranean coast and Lycia.163 The “patrician Trajan” is not the only source to mention the city of Syllaion in connection with the Byzantine-Arab warfare in the 670s. A curious reference to this fortress can be found in the Life of Saint Andrew the Fool,164 a baroque roman hagiographique that contains an apocalypse offering an original account of the last decades before the end of the times. The date of the Life gave rise to a controversy between its editor, Lennart Rydén, who placed its redaction in the 10th century, and Cyril Mango who detected a number of hints pointing to the 670s.165 The two positions need not be mutually exclusive; but Mango’s arguments cannot be disregarded and point to the oldest layer of the text authored in the reign of Constantine IV. Mango was surprised by the unusual role played by Syllaion in the apocalypse of Andrew the Fool.166 In the description of the civil war that three young emperors, “shameless, foolish and good-for-nothing”, will wage on each other in the last days, the hagiographer lists the armies that will be recruited by each of them: the first of the brothers will enroll the people of Thessalonica and Rome, the second will “march out to Mesopotamia and the Cyclades of the islands”, while the third will also march out, and he will enroll Phrygia, Karia, Galatia, and Asia, and Armenia, and Arabia. And coming to Syllaion he will say to her: “You have been called Syllaion, but you will not be pillaged nor taken by any of your enemies.” And having said this he too will go to an unallied people, that is one which is not under his or his colleagues’ sway.167 The three brothers will then kill each other in a terrible battle and be succeeded by a “base woman” whose depravity will cause God to destroy Constantinople. After 40 days 163. On the site of Syllaion and its Middle Byzantine kastron, see C. Foss, The cities of Pamphylia, in Id., Cities, fortresses and villages of Byzantine Asia Minor, Variorum 1996, IV, pp. 19–23; and also G. E. Bean, Turkey’s southern shore : an archaeological guide, New York 1968, pp. 59–66; V. Ruggieri – F. Nethercott, The metropolitan city of Syllion and its churches, JÖB 36, 1986, pp. 133–56; TIB 8.1, pp. 395–402. 164. The Life of St. Andrew the Fool, ed. and transl. L. Rydén, Uppsala 1995. 165. L. Rydén, The date of the Life of Andreas Salos, DOP 32, 1978, pp. 127–55; C. Mango, The Life of Saint Andrew the Fool reconsidered, Rivista di studi bizantini e slavi 2, 1982, pp. 297–313, esp. p. 309: “reign of Constantine IV […] or in the first reign of Justinian II, i.e. approximately between the years 680 and 695”; The Life of St. Andrew the Fool, ed. Rydén, vol. 1, pp. 41–56. 166. Mango, The Life of Saint Andrew the Fool (quoted n. 165), p. 307. 167. The Life of St. Andrew the Fool, ed. Rydén, lines 3939–44: ὁ δὲ τρίτος ἐξελεύσεται καὶ στρατοπεδεύσει καὶ αὐτὸς Φρυγίαν, Καρίαν, Γαλατίαν καὶ τὴν Ἀσίαν καὶ Ἀρμενίαν καὶ τὴν Ἀραβίαν. Καὶ ἐν Συλαίῳ γενόμενος τάδε ἐρεῖ πρὸς αὐτήν· “Σύλαιον ἐπικέκλησαι, ἀλλ’ οὐ συληθήσῃ οὐδὲ παραληφθήσῃ ὑπ’ οὐδενὸς τῶν πολεμούντων σε.” Καὶ ταῦτα εἰρηκὼς ἐλεύσεται καὶ αὐτὸς ἐν λαῷ ἀσυνθέτῳ, τουτέστι μὴ ὄντι ὑποκάτω αὐτοῦ ἢ τῶν ἑταίρων αὐτοῦ. In an earlier paper, Rydén established a slightly different text: L. Rydén, The Andreas Salos Apocalypse : Greek text, translation, and commentary, DOP 28, 1974, pp. 197-261, at p. 207141–146. 278 MAREk JANkOWIAk of mourning, “the imperial power will be given to Rome, Syllaion and Thessalonica, the end already approaching”.168 The role of Syllaion in the apocalypse of Andrew the Fool is as puzzling as it is prominent. It would be rash to try to identify historical events behind individual episodes related in the apocalypse,169 but the prophecy that it should never be captured, clearly more than just a pun on its name (Σύλαιον—συληθήσεται “to strip off, pillage, plunder”), and the predicted transfer of the Empire to Syllaion, placed on a par with Rome and Thessalonica after the destruction of Constantinople, may evoke memories of the reign of Constantine IV and his two brothers, all teenagers when they succeeded their father in 668, which suggests their identification with the three young and foolish emperors. The only 7th-century event that can account for these associations is the disaster of an Arab army, perhaps that of Sufyān b. ʿAwf, placed by the “patrician Trajan” at Syllaion. The integration of the defeat of Sufyān into apocalyptical literature may have been facilitated by strong eschatological expectations, enhanced by the sighting of an aurora at an unusually low latitude in March 673.170 The expedition of Sufyān b. ʿAwf against Byzantium is also recorded in Islamic sources. They date it between 52/672 and 55/674–5171 and cast some light on its leader. In the earliest extant sources, such as al-Wāqidī (d. 822/3), Sufyān died when leading a winter raid against the Empire: Al-Wāqidī claimed that the raid of Sufyān b. ʿAwf al-Azdī occurred during this year, as well as his winter campaign in Byzantine territory. He also claimed that Sufyān died there and appointed ʿAbdallāh b. Masʿadah al-Fazārī as his successor.172 At an early stage of transmission the disastrous campaign of Sufyān b. ʿAwf was amalgamated with an earlier, equally unfortunate Arab expedition, decimated by famine and smallpox during a difficult wintering in the Byzantine territory and purportedly succoured by Yazīd, son of Muʿāwiyah.173 But the earliest Muslim chronographers and Theophilus of Edessa make it clear that these were two distinct events, both remembered as disasters that ended in the loss of a significant part of an Arab army. 168. The Life of St. Andrew the Fool, ed. Rydén, lines 4048–50: ἀπὸ γοῦν τῶν ἡμερῶν ἐκείνων δοθήσεται τὸ βασίλειον Ῥώμῃ καὶ τῷ Συλαίῳ καὶ τῇ Θεσσαλονίκῃ, τοῦ τέλους ἤδη ἐγγίσαντος. 169. But the recent assessment of the apocalypse of Andrew the Fool by A. kraft, The last Roman emperor topos in the Byzantine apocalyptic tradition, Byz. 82, 2012, pp. 213–57, at pp. 240–4, as “remarkably uninterested in its contemporary political environment” (p. 241) goes too far in the opposite direction. The arguments to the contrary of Mango, The Life of Saint Andrew the Fool (quoted n. 165), pp. 305–8, remain valid. 170. Mich. Syr. xi.13, p. 456: “at the third watch of the night, a complete bow was seen. It is something most unnatural for a bow to be seen when the sun is underneath the earth. All who saw it thought that the end of the world would come that year” (transl. R. Hoyland). 171. See the table above, pp. 266–7. 172. Al-Ṭabarī AH 52, ii.157 = xviii.165. 173. Masʿūdī, Les prairies d’or, § 1818–9; al-Balādhurī, Ansāb al-ashrāf, vol. IV A, ed. by M. Schloessinger, Jerusalem 1971, p. 70, translated into Italian in Aḥmad ibn Yaḥyā al-Balādhurī, Il califfo Muʿāwiya I secondo il « Kitāb ansāb al-ašrāf » (Le genealogie dei nobili), trad. O. Pinto e G. Levi della Vida, Roma 1938, § 237, p. 91. See below, pp. 290 ff., on Yazīd’s campaign. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 279 The 12th-century biographical dictionary Tārīkh madīnat Dimashq (“History of the city of Damascus”) of Ibn ʿAsākir adds several anecdotes that demonstrate Sufyān’s anti-Byzantine zeal.174 Famous for “humiliating the cities of Caesar”,175 he was appointed by Muʿāwiyah as the leader of a raid against the Byzantines for his readiness to trespass the caliph’s orders should they be too conservative. Sufyān is said to have ridden at the head of 3,000 cavalrymen to Constantinople, where he engaged in the following dialogue: [Sufyān] raided with us as far as the Golden Gate, so that the people of Constantinople were alarmed and rang the bells. They came to meet us and said: — What do you want, o troop of Arabs? Why did you come? — We have come to destroy this city of unbelief, and God will destroy it by our hands. — We do not know whether your plan is wrong, or the book [i.e., Qur’ān] lied, or you have arrived before the preordained time. By God, we know that the city will be conquered one day, but we do not think that the time is now.176 According to another anecdote ascribed to al-Wāqidī, Sufyān died at a place called al-Randāq or al-Rindāq (‫ )الرنداق‬in the land of the Rum, “which in the Roman language means ‘piercing’ (‫”)خازق‬. No similar Greek word comes to my mind, but the placename makes one think of the river Rhyndakos at the border of Hellespontos and Bithynia, or of the Bithynian city of Apollonia ad Rhyndacum located on the same river, 70 km south-east from kyzikos. The Arab army was already in trouble when Sufyān died; it seems that his successor, ʿAbd al-Raḥmān b. Masʿūd al-Fazārī, failed to extricate it and bring it safely back home, even though an Arab defeat is only implied.177 Islamic sources thus preserve an echo of Sufyān’s death during a disastrous campaign against Byzantium, but their lack of a consistent narrative, their uncertain chronology, and their unwillingness to acknowledge the Arab defeat—or at least their attempts to exonerate Sufyān from the responsibility of the defeat, projecting it instead on his successor, ʿAbd al-Raḥmān b. Masʿūd al-Fazārī178—make them all but unusable. As a result, they have little to add to the account of Theophilus of Edessa, except, perhaps, for the information on Sufyān’s reconnaissance to Constantinople and his advance to the district of the Rhyndakos. If the significance of a major Byzantine victory in 673/4 is implied rather than explicitly stated by the sources, there exists evidence for a sudden surge in the activity of the Byzantine fleet around 672/3, when naval warfare was waged by the Byzantines at a considerable distance from Constantinople. In his Book of the governors of Egypt, al-kindī (d. 961) describes an event of the governorship of Maslama b. Mukhallad: 174. Ibn ʿAsākir, Tārīkh madīnat Dimashq, vol. 21, pp. 347–52. 175. Ibid., pp. 349–50; same anecdote earlier in al-Balādhurī, Ansāb al-ashrāf, pp. 86–7 (transl. in al-Balādhurī, Il califfo Muʿāwiya I, § 269, pp. 111–2). 176. Ibn ʿAsākir, Tārīkh madīnat Dimashq, vol. 21, p. 350. 177. Ibid., p. 351. 178. Some Islamic sources hesitatingly report the tradition on Sufyān’s death in action, see al-Wāqidī quoted above, note 172, and Ibn ʿAsākir, Tārīkh madīnat Dimashq, vol. 21, p. 352: “Abū Saʿīd ʿAbd al-Raḥmān b. Aḥmad b. Yūnus said: Sufyān b. ʿAwf al-Azdī was killed in the land of the Romans in the year 55, according to what said Ibn Yūnus. The opinion of those who say: ‘He died [i.e., was not killed]’ is more correct, but God only knows.” 280 MAREk JANkOWIAk when he was amīr, the Romans landed at al-Burullus in the year 53/672–3, and there were martyred at that time Wardān, a client of ʿAmr b. al-ʿĀṣ, and ʿĀidh b. Thaʿlabah al-Balawī and Abū Ruqayyah ʿAmr b. Qays al-Lakhmī with a large number of men.179 The number of victims and the death of Wardān, a mawlā and a close associate of ʿAmr b. al-ʿĀṣ, indicate the importance of this encounter. A Byzantine attack against the district of Paralos at the northernmost tip of the Nile delta is not implausible: after the Arab conquest of Egypt, the Empire attempted at least twice to re-establish its control over this province.180 A puzzling appeal for divine help for the emperors (in plural) in a paschal letter issued by a patriarch of Alexandria in early 672 suggests perhaps that the attack took place in winter 671/2 and that the Byzantines may have even temporarily occupied Alexandria.181 On the other hand, Al-kindī mentions an Egyptian raid against Constantinople led by ʿĀbis b. Saʿīd al-Murādī, head of the Egyptian police and judiciary under Maslama b. Mukhallad, but neither the date nor any further details are known.182 Thus not only was the naval war not limited to the region of Constantinople in the early 670s, but the Byzantines had the capacity to counter-attack as far as Egypt. Byzantine activity well outside Constantinople around 673 is also suggested by the seals of the general kommerkiarioi (γενικοὶ κομμερκιάριοι).183 Particularly frequent between ca 660 and ca 730, their seals attest to their very high rank—that of general, honorary consul or patrician—and to their rare prerogative to place the imperial effigy on the obverses of their seals. But despite their apparent importance, the office of genikos 179. Al-kindī, The governors and judges of Egypt or Kitâb el ʾumarâʾ (el wulâh) wa Kitâb el quḍâh, ed. R. Guest, Leyden – London 1912, p. 38; transl. by E. W. Brooks, The relations between the Empire and Egypt from a new Arabic source, BZ 22, 1913, pp. 381–91, at p. 389. Same information in Ibn ʿAbd al-Ḥakam, Futūḥ Miṣr, ed. C. C. Torrey, New Haven 1922, p. 124; Russian translation: ибн ʿАбд ал-Хакам, Завоевание Египта, ал-Магриба и ал-Андалуса, перевод С. Б. Певзнера, Москва 1985, p. 124. 180. Probably in 645, see below, p. 287, and around 660 when troops paying allegiance to Byzantium are attested in Egypt: The Armenian history attributed to Sebeos, transl. R. W. Thomson, Liverpool 1999, p. 154, with the commentary of J. Howard-Johnston, pp. 284–7; Chron. 1234 § 137. 181. P.Grenf. II 112, lines 16–9: οὕτω γὰρ ἡμῶν ἑορταζόντων τῶν ἐχθρῶν ἡ ὑποταγὴ τοῖς βασιλεῦσ<ε>ι διὰ τῆς ὑμῶν πρὸς θεὸν ἱκετείας εὖ οἶδα ὅτι γενήσεται καὶ τῇ καθ’ ἡμᾶς ἐκκλήσιᾳ τὴν εἰρήνην αἰτουμένοις ἐλπίζομεν μηδαμῶς ὑμᾶς ἀποτεύξεσθαι. The date of Easter (25 April) limits the choice to the years 577 or 672, but the earlier date is unlikely in view of the reference to the emperors in plural. The letter was issued before the beginning of the Lent i.e., before 15 March. See, more in detail, my forthcoming study of the festal letters preserved in the papyri. 182. Al-kindī, The governors and judges of Egypt, p. 39; Brooks, New Arabic source (quoted n. 179), p. 389. 183. See most recently F. Montinaro, Études sur l’évergétisme impérial à Byzance (IV e-IX e siècles), PhD diss. EPHE Paris 2013, vol. 1, pp. 115–350, and his contribution to the present volume; as well as W. Brandes, Finanzverwaltung in Krisenzeiten : Untersuchungen zur byzantinischen Administration im 6.-9. Jahrhundert, Frankfurt am Main 2002, in particular pp. 239–426 and 511–610 (catalogue and summary lists of seals); L. Brubaker and J. Haldon, Byzantium in the iconoclast era c. 680–850. A history, Cambridge 2011, pp. 682–95. The seals are quoted according to the catalogues of Montinaro (M), Brandes (B), G. Zacos, A. Veglery, Byzantine lead seals, vol. 1, Basel 1972 (ZV), and J.-C. Cheynet, C. Morrisson, W. Seibt, Les sceaux byzantins de la collection Henri Seyrig, Paris 1991 (Seyrig). THE FIRST ARAB SIEGE OF CONSTANTINOPLE 281 kommerkiarios is not attested in this period in written sources.184 Their seals nearly always carry the mention of an apotheke (ἀποθήκη), i.e. storehouse, and—from 673/4—an indictional date. I follow Federico Montinaro in seeing the general kommerkiarioi as central officers of the highest rank, not local managers of the provincial apothekai.185 This, and the fact that one of the first known general kommerkiarioi was an official responsible for the finances of the army,186 strengthens John Haldon’s conclusion that the irregular pattern of apothekai appearing on seals in a given year, often clustered by two or three, “emphasises the […] ad hoc nature of the activities with which the kommerkiarioi seem to have dealt”, which in all likelihood were military campaigns.187 One would only expect such activities to be coordinated centrally by “financial crisis-managers”188 who were responsible for mobilising resources for the army through a network of provincial apothekai, rather than for farming tax revenues or controlling silk trade—at least during the most dramatic phase of the Empire’s struggle for survival in the 660s and 670s. The exact nature of the link between the seals of the general kommerkiarioi and military activity remains, however, hazy, and it is impossible to use the seals of the general kommerkiarioi as direct evidence for military campaigns. But they still can provide an indication of where resources were being mobilized. Between 659 and 672/3, the apothekai of most eastern Anatolian provinces appear on the seals of the general kommerkiarioi in a chaotic pattern consistent with the picture of generalized raiding in most of the Anatolian provinces conveyed by written sources. Starting from 673/4, the seals bear an indictional date and their distribution can be traced more precisely. An important group of six seals issued in that year by the apotheke of Africa in the name of the general kommerkiarioi Mikkinas and Gregorios has been found in Carthage.189 These are among the very few known seals of the apotheke of Africa,190 the only apotheke attested in the West; they may bear witness to Byzantine military activity in Africa in 673/4. The names of the general kommerkiarioi Mikkinas and Gregorios appear also on the seals of the apotheke of Honorias in 673/4 and 674/5,191 and on a uniface seal that bears neither the date nor 184. See the catalogue of earlier and later mentions of the kommerkiarioi in the written sources in Montinaro, Études sur l’évergétisme (quoted n. 183), vol. 1, pp. 169–89. 185. Montinaro, Études sur l’évergétisme (quoted n. 183), vol. 1, pp. 147–8; the idea has already been prudently voiced by Brandes, Finanzverwaltung (quoted n. 183), pp. 310–1. 186. Stephanos, genikos kommerkiarios between 659 and 668, was a military logothete (στρατιωτικὸς λογοθέτης): M 8 = B 52 = ZV 144; see also PmbZ #6909, and Brandes, Finanzverwaltung (quoted n. 183), p. 576. The title of “military commander” (στρατηλάτης) on the seals issued in the name of the genikos kommerkiarios kosmas (PmbZ #4066) in 679/80–681/2 is a dignity, not an office, see Brandes, Finanzverwaltung, pp. 85 and 571. 187. Brubaker – Haldon, Byzantium in the iconoclast era (quoted n. 183), pp. 692–3. 188. Brubaker – Haldon, Byzantium in the iconoclast era (quoted n. 183), p. 695, in line with the conclusions of Brandes, Finanzverwaltung (quoted n. 183), pp. 425–6. 189. C. Morrisson, W. Seibt, Sceaux de commerciaires byzantins du viie siècle trouvés à Carthage, RN 6e série, 24, 1982, pp. 222–41. The inscription reads: Ἰνδ(ικτιὼν) βʹ. Μικκί<ν>ου (καὶ) Γρηγορίου γενικῶν κομμερκιαρίων. Ἀποθήκης Ἀφρικῆς (M 18). 190. A seal of the apotheke of Africa issued by an anonymous predecessor of Mikkinas and Gregorios during the reign of Constantine IV and his brothers is mentioned by Morrisson – Seibt, Sceaux (quoted n. 189), p. 236 n. 15. 191. The name of the apotheke is hypothetical on the later seal: Morrisson – Seibt, Sceaux (quoted n. 189), p. 235; Brandes, Finanzverwaltung (quoted n. 183), no. 64, p. 523. 282 MAREk JANkOWIAk the name of the apotheke, but was interestingly found on a beach in Jebleh on the Syrian coast, which suggests a link with the Byzantine raid of 675/6.192 In 675/6 Mikkinas and Gregorios were succeeded by the honorary consul Peter who remained in the position until 679/80. The apothekai known to have operated during his tenure were located further east, along the Byzantine-Arab border: First or Fourth Armenia in 675/6 and on a seal with a missing date,193 Isauria in 676/7, and the two Cilicias in 679/80. They indicate Byzantine activity in southeastern Anatolia in a period when Constantinople is supposed to have been besieged by the Arabs. Peter’s successor, the stratelates kosmas, was first active in northern Anatolia—the apothekai of Helenopontos and of “Pylai and Sangaros” on the southern shore of the Gulf of Nicomedia are attested in 679/80—before shifting his attention back to the southeastern border of the Empire, where the apothekai of Isauria and of the Cappadocia I were active in 681/2. This brief shift of focus from the Arab border to the region of Constantinople and northern Anatolia corresponds to the decrease in the intensity of the Arab raids, the Bulgar danger on the Danube, and the convocation of a potentially perilous ecumenical council to Constantinople. The seals of the apothekai can thus be plausibly linked with military operations and testify to Byzantine activity in Anatolia throughout the 670s. On the other hand, the introduction of the indictional date in 673/4 is thought to reflect a reform of the office of the general kommerkiarioi, endowing a makeshift institution with a more regular character. This reform has been connected with the beginning of a new indictional cycle and with the imminent siege of Constantinople,194 but no seals dated to indiction 1 (672/3) are currently known, and, as we have seen, the siege of 674–678 has no historical substance. The reform can rather be interpreted as an attempt at regularising, in the aftermath of the first important Byzantine victory over the Arabs, an institutional solution that proved efficient in providing logistical support to the Byzantine armies.195 Taken together, sources studied in this chapter imply a turning point in the Byzantine-Arab war around 672/3, when the Byzantine fleet, absent from earlier operations, suddenly appeared on the stage and allowed the Byzantines to regain the strategic initiative and to destroy a major Arab army in 673/4. This success was followed by a Byzantine counter-offensive, glimpses of which can be gained from sources composed in Italy and Thessalonica. 192. M P28 = B 56 = ZV 145 = Seyrig 149. Pace earlier editors, Morrisson – Seibt, Sceaux (quoted n. 189), p. 235, and Montinaro, Études sur l’évergétisme (quoted n. 183), vol. 1, p. 346, showed that it dates from 673/4 at the earliest. Find place: H. Seyrig, Bulle de Magnus le Syrien, in G. Tchalenko, Villages antiques de la Syrie du Nord : le massif du Bélus à l’époque romaine, vol. 3, Paris 1958, pp. 40–1 n. 3. On the Byzantine raid of 675/6, placed by an Islamic source just south of Jebleh, see above, p. 274. 193. M 21 (not in Brandes) and M 19 = B 66 = ZV 155. On the latter seal, the numeral of the province of Armenia can be read either as αʹ or as δʹ; the latter is preferred by Brandes, Finanzverwaltung (quoted n. 183), no. 66, p. 523. 194. Brandes, Finanzverwaltung (quoted n. 183), p. 312; Brubaker – Haldon, Byzantium in the iconoclast era (quoted n. 183), p. 689. 195. Another reform, of Byzantine copper coinage in the Western provinces, has been conjectured around 674, but the chronology of provincial issues is too tentative to allow firm conclusions, see V. Prigent, Nouvelle hypothèse à propos des monnaies de bronze à double marque de valeur de l’empereur Constantin IV, in Puer Apuliae : mélanges offerts à Jean-Marie Martin, éd. par E. Cuozzo et al., Paris 2008, vol. 2, pp. 567–79 at pp. 571–2. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 283 brighter horizons (ca 676–8) If the patriarch of Constantinople Thomas, during the two years in the office (667–9), was unable to send his synodical letter to Pope Vitalian because of the “presence of the godless Saracens”, his three successors, John V (669–75), Constantine (675–7) and Theodore (677–9),196 faced no such obstacle. Their relations with Rome are indirectly referred to by the emperor Constantine IV in his sacra to Pope Donus dated 12 August 678.197 After several attempts of the patriarchs of Constantinople to establish communication with Rome had failed because of a schism between the two churches,198 the emperor in person made overtures to the pope: we inform your fatherly beatitude that Theodore, [recently] ordained as the most holy and most blessed patriarch of this our God-protected and imperial city, related to our serenity that he hesitated before sending the usual synodical letter to your fatherly beatitude, lest it be refused, as happened to the patriarchs before him, and had decided instead to address a letter of exhortation to your fatherly beatitude, which indeed he sent, and because your fatherly beatitude is entirely cognizant of this, we do not intend in our present pious sacra to discourse at length about this letter.199 The emperor refers to synodical letters, in plural, sent by Theodore’s predecessors and refused by the popes. To avoid a similar misadventure, the new patriarch, ordained in August or September 677, decided to enter into communication with the pope by means of a less formal letter exhorting to reconciliation. The patriarchs mentioned by Constantine IV can only be the two direct predecessors of Theodore, John V and Constantine,200 given that the synodical letter of John V’s predecessor, Thomas, had never reached Rome, and we know that Thomas’ predecessor, Peter (654–66), had 196. See, on them, van Dieten, Patriarchen (quoted n. 10), pp. 121–9, and PmbZ #2704, 3708 (with a mistaken quote from Zon. xiv.21, where the initiative of calling the ecumenical council is attributed to the emperor Constantine, not to the homonymous patriarch) and 7317. Their dates have been established by Brooks, On the lists of the patriarchs (quoted n. 11), pp. 47–9; no documents issued by them are extant: V. Grumel, Les regestes des actes du patriarcat de Constantinople. 1, Les actes des patriarches. 1, Les regestes de 381 à 715, 2e éd. rev. et corr., Paris 1972, pp. 235–9. 197. ACO, ser. sec. II, pp. 2–10. The date is preserved in the Latin translation, p. 310–11: data pridie Idus Augustas Constantinopoli indictione sexta. 198. The emperor used the word “schism” at ACO, ser. sec. II, p. 219. In his answer, Pope Agatho spoke of σκάνδαλον διχονοίας (p. 11819), see also p. 1381–5. 199. ACO, ser. sec. II, p. 42–10 : ἐν εἰδήσει ποιούμεθα τὴν ὑμετέραν πατρικὴν μακαριότητα, ὡς ὁ προχειρισθεὶς Θεόδωρος, ὁ ἁγιώτατος καὶ μακαριώτατος πατριάρχης τῆς θεοφυλάκτου ἡμῶν ταύτης καὶ βασιλίδος πόλεως, ἀνήγαγε τῇ ἡμετέρᾳ γαλήνῃ ὑφορᾶσθαι αὐτὸν στεῖλαι τὰ πρὸς συνήθειαν συνοδικὰ πρὸς τὴν ὑμετέραν πατρικὴν μακαριότητα, μήπως ἀπρόσδεκτα γένωνται, καθὼς καὶ τὰ ἐπὶ τῶν πρὸ αὐτοῦ πατριαρχῶν γεγόνασιν· ἀλλ’ ὅτι μᾶλλον συνεῖδε προτρεπτικῇ ἐπιστολῇ χρήσασθαι πρὸς τὴν ὑμετέραν πατρικὴν μακαριότητα, ἥντινα καὶ ἔστειλε, καὶ ταύτης ἐν εἰδήσει πάντως γέγονεν ἡ ὑμετέρα πατρικὴ μακαριότης καὶ οὐ συνείδομεν διὰ τῆς παρούσης ἡμῶν εὐσεβοῦς σάκρας περὶ τῆς τοιαύτης ἐπιστολῆς πολυλογίᾳ χρήσασθαι. I thank Richard Price for discussing this passage with me. 200. Same opinion in the works quoted above, note 196. 284 MAREk JANkOWIAk re-established the communion between the two churches in 658, after over a decade of violent controversy over the wills of Christ.201 The schism referred to by Constantine IV must therefore have begun around 670, when Pope Vitalian (657–72) refused to accept the synodical letter of John V, patriarch since November 669. No source explains Vitalian’s motivations.202 Although in his sacra the emperor presented the schism as another episode of the controversy over the wills of Christ, it is unlikely that theological subtleties motivated the pope who reconciled with the imperial court immediately after his ordination, much to the displeasure of the Greekspeaking monks settled in Rome who had entangled the papacy in the confrontation with Constantinople in the 640s and 650s. A list of patriarchs of Constantinople copied in Rome around 676 also suggests that theological controversy was not the prime reason of the schism: the patriarchs John and Constantine, whose synodical letters had been rejected by the popes, are not qualified as heretici, as opposed to their “Monothelete” predecessors, Sergius, Pyrrhus, Paul and Peter.203 The matter of the wills of Christ appears rather to have been exhumed by the pope to counter the grant of autocephaly to the church of Ravenna by Constans II in 666, which was perceived in Rome as a challenge to papal supremacy over the churches of the West. As long as an emperor resided in the West, Vitalian hesitated to retaliate and even lent support to the sons of Constans II against Mezezios in Sicily in 668–9. But he took his decision soon afterwards: the insistence of the patriarchs of Constantinople and Antioch on removing Vitalian’s name from the diptychs of the church of Constantinople shows that they held him responsible for the schism.204 Nothing is known of the relations between the churches of Rome and Constantinople until the next patriarch, Constantine, ordained on 2 September 675, sent his synodical letter to Rome, probably in 676. He was no more successful than his predecessor. His successor Theodore, patriarch since August or September 677, despatched his letter of exhortation before August 678, the date of the sacra of the emperor addressed to Pope Donus. Thus at least three letters carried by official embassies205 were despatched from Constantinople to Rome between 676 and 678, when Constantinople is thought to have been besieged by the Arabs. News also circulated in the opposite direction, from Rome to Constantinople: the name of Pope Donus, who was consecrated on 2 November 676, was known in Constantinople, even though he probably did not send his synodical letter to the patriarch. 201. The restoration of communion between Rome and Constantinople is mentioned by Maximus the Confessor in his letter to Anastasius the Monk dated 18 April 658: Maximus, epistula ad Anastasium Monachum, lines 3–6, in CCSG 39, p. 161. 202. See, on what follows, my forthcoming monograph on the Monothelete controversy. 203. L. Cuppo, I Pontifices di Costantinopoli nel Liber pontificalis del settimo secolo : note sul codice BAV, Vat. Lat. 3764, Rivista di storia e letteratura religiosa 43, 2007, pp. 359–72, at p. 362: Sergius hereticus / Pyrrus hereticus / Paulus hereticus / Petrus hereticus / Thomas / Iohannes / Constantinus. Cuppo’s interpretation of this text does not take into account the schism between the churches of Rome and Constantinople in the 670s. That the synodical letters of John V and Constantine were accepted by the Monothelete patriarch of Antioch Makarios (ACO, ser. sec. II, p. 616) does not need to mean that they were explicit about the question of the wills of Christ. 204. ACO, ser. sec. II, p. 811–13. 205. Such as the one headed by Epiphanius gloriosus a secretis which brought the imperial sacra of 12 August 678 to Rome: LP, p. 3505. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 285 There are also hints that, even before the imperial sacra announced the imperial intention of restoring the unity of the church to the bishop of Rome, the papacy had been anticipating such an initiative. This means that around 676–8 Rome judged the military prospects of the Empire to be sufficiently good to allow the emperor to devote more attention to church affairs. According to the Liber pontificalis, Pope Donus (676–8) “discovered that the Syriac monks” of one of monasteries in Rome “were Nestorians; he divided them among different monasteries, and established Roman monks in that monastery”.206 Such a “discovery” is rather unlikely in the heart of western Christendom; but it can be interpreted as the elimination of the party of the Oriental monks, which 30 years later had pushed the popes towards a violent confrontation with the imperial court. The choice of Agatho to succeed Donus after the latter’s death on 11 April 678 also suggests that Rome was preparing for difficult negotiations with Constantinople: as opposed to his predecessors, Agatho was a native of Sicily, therefore in all likelihood a native Greek speaker, the first on the papal throne since the 640s. Finally, soon after the death of Adeodatus on 17 June 676 the Roman curia updated the Liber pontificalis207 and possibly prefixed it with an up-to-date list of bishops of Constantinople, ending with the patriarch Constantine (675–7).208 The picture of information flowing freely from and to Constantinople and the impression of a positive outlook for the Empire are consistent with the presentation of the crisis by the emperor in his sacra to Donus. On the one hand, he assumed that the pope was well informed of court politics in Constantinople.209 On the other, he depicted an Empire emerging from a lengthy crisis which prevented him from devoting sufficient attention to church affairs since the time he succeeded to the throne.210 Although the crisis was not yet over and conditions were not favourable for a general council to be convened, the emperor was at least at leisure to hold a debate on the controversial theological issues, with the hope that the reconciliation of the churches would propitiate the Divinity to lend further support to the Romans.211 The imperial sacra of 678 was less the result of a recent military success, such as the end of the alleged siege of Constantinople, than of a steady improvement of Byzantine fortunes over the preceding years. 206. LP, p. 3484–6: hic repperit in urbe Roma in monasterio qui appellatur Boetiana Nestorianitas monachos Syros, quos per diuersa monasteria diuisit; in quo predicto monasterio monachos Romanos instituit. This is the only reference to the Boetiana monastery, see J.-M. Sansterre, Les moines grecs et orientaux à Rome aux époques byzantine et carolingienne (milieu du VI e s.-fin du IX e s.). 1, Texte, Bruxelles 1983, p. 37. 207. The existence of such an update is demonstrated by: 1) the sharp difference in the level of detail of the life of Vitalian before and after the description of the visit of Constans II to Rome in 663; 2) errors in the chronology of the death of Constans II and the revolt of Mezezios, unlikely to have been committed by a strictly contemporary author (see below, p. 310); 3) a colophon following the life of Adeodatus in the manuscripts of the E family: a tempore ordinationis sancti Gregorii papae usque hunc sunt anni XCV m. V dies XIIII (LP, p. 347, app. crit. ad l. 5), even if this date translates to 17 February 686, ten years after the death of Adeodatus (17 June 676). On the textual history of the Liber pontificalis, see most recently the relevant chapter of Montinaro, Études sur l’évergétisme (quoted n. 183), vol. 1, pp. 33–113. 208. See above, note 203. 209. ACO, ser. sec. II, p. 210–15. 210. See above, note 69, for the most relevant passages of the sacra. 211. ACO, ser. sec. II, p. 827–29. 286 MAREk JANkOWIAk A similar picture of an emperor able to devote increasing means and attention to Western affairs in the late 670s emerges from the Miracles of Saint Demetrius. The fourth miracle of the second collection tells the story of a siege of Thessalonica by the Slavs and of its miraculous salvation through the protection of the saint.212 The narrative begins with the arrest of Perboundos, king of the Slavic tribe of Rynchinoi settled in the vicinity of Thessalonica. After Perboundos had been sent in chains to Constantinople, a delegation of Slavs and of inhabitants of Thessalonica followed him to ask the emperor for mercy, but, “since they found His Piety busy arming an expedition against the apostate Agarenes, it was agreed with all the envoys that Perboundos would be set free after the war”.213 Perboundos was meanwhile granted freedom of movement in Constantinople, which he used to escape. Alarmed by his flight, the emperor, who feared that Perboundos might exact vengeance on Thessalonica, ordered a thorough search. Constantinople was sealed off, navigation to and from the city forbidden, Perboundos’ guards executed, and the prefect of Constantinople apparently exiled to Thessalonica. The emperor also dispatched a dromon to that city to warn it of the impending danger. After 40 days Perboundos was found and soon executed after another attempt to escape. At the news of his death the Slavs blockaded Thessalonica. They harassed the city from land and sea during the next two full years. 214 The city, reduced to an extreme famine, received little help from the emperor who sent “ten armed ships also carrying provisions, but was unable to send more troops, because he happened to be busy with another war”.215 This was too little to relieve the city, but the ships could be used to bring additional grain from Thessaly. Taking advantage of their absence and of the small numbers of defenders left in the city, the Slavs launched a general assault on the 25 July of the 5th indiction.216 Helped by Saint Demetrius, the defenders withstood three days of intense fighting; a little later the fleet returned with fresh provisions. Despite this setback, the Slavs not only maintained the blockade, but also extended their naval operations towards Constantinople and plundered ships “transporting crops to the imperial city from the islands, the straits, Parion and Prokonnesos, and even captured the staff of the custom house with their ships”.217 This bold inroad into the sea of Marmara, which 212. Lemerle, Les plus anciens recueils (quoted n. 7), vol. 1, pp. 198–221 (Greek text and French paraphrase), and vol. 2, pp. 111–36 (commentary). 213. Mirac. Dem. ii.4.232, p. 20916–18: εὑρημένης τῆς αὐτοῦ εὐσεβείας πρὸς παράταξιν τῶν θεοπτώτων Ἀγαρηνῶν ἀνθοπλιζομένης, συντάξασθαι τοῖς αὐτοῖς ἅπασιν ἀποκρισιαρίοις τὸν αὐτὸν Περβοῦνδον μετὰ τὸν πόλεμον ἀπολύειν. 214. Mirac. Dem. ii.4.243, p. 21120: καὶ τοῦτο [sc. the blockade] παρ’ αὐτοῖς διηνεκῶς ἐπράττετο μέχρι χρόνον δύο πληρεστάτων. 215. Mirac. Dem. ii.4.251, p. 21322–24: δέκα ἐνόπλους καράβους μετὰ καὶ δαπανῶν ὁ τῶν πραγμάτων κύριος τῇ πόλει κατέπεμψε, διὰ τὸ μὴ δύνασθαι στρατὸν πλείονα στεῖλαι· καὶ γὰρ καὶ αὐτὸς συνέβη ἐν ἑτέρῳ ἀσχολεῖσθαι πολέμῳ. 216. Mirac. Dem. ii.4.255, p. 21428–29: τῇ εἰκάδι πέμπτῇ τοῦ ἰουλίου μηνὸς ἰνδικτιῶνος πέμπτης. 217. Mirac. Dem. ii.4.277, p. 2206–11: τῶν γὰρ ἁπάντων Σκλαβίνων τῶν ἀπὸ τοῦ Στρυμῶνος καὶ Ῥυγχίνου λοιπὸν ἐκ τῶν ἔνθεν μερικῶς καταπαυσάντων, καὶ διὰ ζευκτῶν πλοίων τοὺς θαλαττίους πλωτῆρας, τοὺς ἐπὶ παρακομιδῇ καρπῶν ἐν τῇ βασιλευούσῃ ἀνιόντας πόλει, παμπόλλους ἐκπορθήσαντες, ἀπό τε τῶν νήσων καὶ τῆς στενῆς θαλάττης καὶ τῶν ἐπὶ τὸ Πάριον καὶ Προκόννησον τόπους, καὶ αὐτοὺς τοὺς εἰς τὸ τελωνεῖον ἅμα τῶν πλωΐμων αἰχμαλωτίσαντες μετὰ πλείστων νηῶν οἴκοι ἐπὶ θυλάκους ἀπίασι. The custom house (τὸ τελωνεῖον) can hardly be other than that in Abydos, although Lemerle, Les plus anciens recueils (quoted n. 7), vol. 2, pp. 125–6, thought rather of the teloneion of Constantinople. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 287 threatened the provisioning of Constantinople itself, finally prompted the emperor to action. A Byzantine land army defeated the Strymon Slavs; at this news the blockade of Thessalonica was at last lifted. The Slavs retreated in disorder leaving behind provisions, immediately plundered by the starved population of the city, half-clothed because of the heat of the summer. At the same time, the emperor despatched a grain fleet to resupply the city. Learning of this, the Slavs sued for peace. This account has a complex internal chronology, but only one absolute date: the 25 July of the 5th indiction. The blockade that is said to have lasted for entire two years probably started in the summer of the 4th indiction.218 The affair of Perboundos should then be dated to the spring of the 4th indiction, when the emperor was preparing an expedition against the Arabs. He was still unable to relieve the city in the 5th indiction when he was “busy with another war”, but in the summer of the next 6th indiction he could marshal enough troops to pacify the Strymon Slavs.219 The interpretation of these events is thus contingent upon the identification of the 5th indiction when the Slavic assault took place. The possible years are 647, 662 or 677, after the beginning of the Arab conquests and before Justinian II visited Thessalonica in 688/9, an event which, according to Paul Lemerle, would have left traces in the Miracles had they been produced after it. We can dismiss the year 662 on account of the peace between Constans II and Muʿāwiyah: it is unlikely that in the previous year the emperor was “busy arming an expedition against the apostate Agarenes”. Paul Lemerle quickly brushed off also the year 647, a date defended by James Howard-Johnston, who recently proposed to identify the expedition against the Arabs prepared by the emperor when envoys from Thessalonica arrived to Constantinople with the campaign of the eunuch Manuel that resulted in a short-lived Byzantine reoccupation of Alexandria.220 But its date is uncertain: Islamic sources suggest that the Byzantine attack took place in the sailing season of 645,221 while Agapios seems to place Manuel’s expedition in 647/8, in the year of the Arab invasion of the province of Africa.222 Neither of these dates coincides with the preparations for a major 218. Assuming, additionally, that the Slavic forays into the sea of Marmara took place in the summer following the general assault of the 5th indiction, see Lemerle, Les plus anciens recueils (quoted n. 7), vol. 2, p. 130. I follow Lemerle’s chronology, although the two years of the blockade could also alternatively be counted back from the summer of the 5th indiction. 219. Lemerle, Les plus anciens recueils (quoted n. 7), vol. 2, pp. 128–33; Howard-Johnston, Witnesses (quoted n. 3), p. 154 n. 55. 220. Howard-Johnston, Witnesses (quoted n. 3), pp. 153–4, esp. n. 55, but his argument for dating the Byzantine attack on Alexandria to 646/7 rather than 645/6 is circular: the new date depends on the chronology of our miracle, which in turn is taken to mention the reoccupation of Alexandria because of its changed date. 221. Ibn khayyāṭ AH 25 (p. 114 Dhakkār) places the expulsion of the Byzantines from Alexandria in Rabīʿ I AH 25 (26 December 645–24 January 646); other Islamic sources, mostly concurring with the date of 25/645–6, are listed in L. Caetani, Annali dell’Islam. 7, Dall’anno 24 al 32 H, Milano 1914, pp. 103–19 (AH 25 §§ 72–104). See also al-Balādhurī, Kitāb futūḥ al-buldān, p. 221 de Goeje (p. 348 Ḥitti), who hesitates between AH 23 and 25; and A. J. Butler, The Arab conquest of Egypt and the last thirty years of the Roman dominion, 2nd ed., Oxford 1978, p. 469 (also 25/645–6). 222. Agap. p. 479. Agapios seems to have the same date of the Arab invasion of Africa as Theoph. AM 6139, p. 34324–28, i.e., the 6th year of Constans II and the 2nd year of ʿUthmān. The solar eclipse of 5 November 644 placed by both of them in the 11th year of the 12 years of ʿUmar makes it clear that AD 647/8 is meant, which may be the actual date of the Arab invasion of Africa, see R. Guéry, 288 MAREk JANkOWIAk campaign in the spring of the year preceding the Slavic general assault on Thessalonica i.e., in 646. If, however, we opt for the year 676, it is easy to imagine Constantine IV preparing for another season of fighting with the Arabs; perhaps he was equipping the fleet that attacked the coast of the jund Ḥimṣ that year.223 Paul Lemerle recognized that the decisive argument in favour of the later date is provided by the structure of the second collection of the Miracles of Saint Demetrius. At the beginning of the fourth miracle, the author of the collection strongly emphasized the transition between the ancient miracles and those that happened in his own times.224 The fourth and the fifth miracle of the second collection225 belong to the latter group; they are therefore chronologically close to each other. Lemerle convincingly dated the fifth miracle to the 680s:226 its main characters, kouber and Mauros, can be plausibly identified respectively with the fourth son of the Bulgar khan kubrat, who was active in the 680s, and with a “strategos of the Sermesianoi and of the Bulgars” known from a late 7th-century seal;227 furthermore, the mention of the Karabisianoi, the Byzantine war fleet created by Constans II during his stay in Sicily, points to the period after the 660s.228 It follows that the fourth miracle should also be dated to the reign of Constantine IV, and the Slavic general assault on Thessalonica—to 25 July 677. Having determined this date, Lemerle was surprised by the incompatibility of the apparent freedom of movement to and from Constantinople with the traditional vision of a long siege or blockade of the Byzantine capital in the 670s. This chronology, he wrote, “conduit à réviser l’image traditionnelle d’un siège de Constantinople par les Arabes qui aurait duré sept ans. Il n’y eut pas de siège à proprement parler, et d’ailleurs les chroniques grecques n’emploient pas ce mot. Il y eut, dans les eaux de Constantinople, une série d’opérations navales menées pendant la belle saison […] par des bateaux arabes qui, l’autre moitié de l’année, se retiraient à Cyzique”.229 As a matter of fact, the fourth C. Morrisson, H. Slim, Recherches archéologiques franco-tunisiennes à Rougga. 3, Le trésor de monnaies d’or byzantines (Collection de l’École française de Rome 60), Rome 1982, p. 57. 223. See above, p. 274. 224. Mirac. Dem. ii.4.230, p. 208: τὴν ἤδη προτεθεῖσαν μερικῶς τῶν θαυμάτων πληθὺν τοῦ ἀοιδίμου καὶ συμμάχου καὶ κηδεμόνος ἡμῶν Δημητρίου […] ἐν βραχεῖ λόγῳ ἐκ τῶν πλείστων διεξελθών, μετελεύσομαι εἰς τὴν νῦν καθ’ ἡμᾶς προελθοῦσαν παρ’ αὐτοῦ τοῦ ὑπὲρ ἡμῶν τὴν ψυχὴν προϊεμένου μάρτυρος Δημητρίου προμήθειαν. 225. According to Lemerle, Les plus anciens recueils (quoted n. 7), vol. 2, pp. 163–4, the last, sixth miracle was written by a different author. 226. Lemerle, Les plus anciens recueils (quoted n. 7), vol. 2, pp. 161–2. 227. On the identification of kouber with the son of kubrat, see the summary of the discussion in PmbZ #4165; Lemerle, Les plus anciens recueils (quoted n. 7), vol. 2, pp. 143–5, was favourable to this hypothesis. The short description of the fourth son of kubrat in Nikeph. § 3517–19—“the forth went over the river Istros and settled in Pannonia, which is now under the Avars, becoming an ally of the local nation”—intriguingly hints at the events described in the fifth miracle. On Mauros, see PmbZ #4911; his seal (ZV 934) bearing the inscription Μαύρῳ πατρικίῳ καὶ ἄρχοντι τῶν Σερμησιανῶν καὶ Βουλγάρων confirms the narrative of the fifth miracle. 228. Mirac. Dem. ii.5.295, pp. 23030–2312: Σισιννίῳ στρατηγῷ τότε τῶν καράβων ὑπάρχοντι […] ὅπως μετὰ τῶν ὑπ’ αὐτὸν ὄντων καραβισιάνων στρατιωτῶν. On the Karabisianoi, see Lemerle, Les plus anciens recueils (quoted n. 7), vol. 2, pp. 154–7; C. Zuckerman, Learning from the enemy and more : studies in « Dark centuries » Byzantium, Millennium 2, 2005, pp. 79–135, at pp. 107–8 and 117–25. 229. Lemerle, Les plus anciens recueils (quoted n. 7), vol. 2, p. 133. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 289 miracle mentions no less than six sea journeys (the land route being permanently blocked by the Slavs) between Thessalonica and Constantinople in a single year, probably 676.230 In 677, ten dromones sailed out from Constantinople, apparently without any hindrance. In the next year, grain transports heading for Constantinople were threatened by Slavic pirates, not the supposed Arabic garrison in kyzikos. The Slavs also sacked a τελωνεῖον, no doubt that of Abydos, which suggests that it operated as usual, controlling the ships sailing through the Hellespont. There is also no trace of a hostile land army at the walls of Constantinople: on his escape, Perboundos simply walked out of the Blachernai Gate and hid in one of the suburban estates.231 As a result of his disappearance, the emperor ordered that the gates of Constantinople be closed and forbade navigation to and from the city, an order that would make no sense if the city was blockaded by the Arabs. Lemerle’s conclusion should therefore be brought one step further: the fourth miracle positively rules out the Arab presence in the vicinity of Constantinople in 675 or 676. There was no siege of Constantinople at that time. There was, however, an exacting war with the Caliphate, which put such a strain on the Empire that it was not able to relieve its second biggest town. Warships, the dromones, were in particularly short supply: only ten of them could be sent to Thessalonica in the critical year 677. In the following year, however, an entire fleet carrying 60,000 measures of grain was despatched by the emperor. To conclude this section, the many-year siege of Constantinople in the 670s can safely be removed from history. It came into being as a side effect of Theophanes’ trouble with his sources, and has been too readily accepted by modern historians. Other sources, Byzantine and Islamic, tell a different story: after a deep crisis in the first years of Constantine IV’s reign, the Empire struck back, mainly thanks to its newly developed naval capabilities. The Byzantines attacked Egypt in 672 or 673 and possibly operated in Africa in 673/4; in the same year they won a major victory in southern Anatolia and destroyed an Arab land and naval force. Probably in spring 676 the emperor Constantine IV was preparing an expedition against the “Hagarenes” which it is tempting to identify with the Byzantine raid against the Syrian coast that cost the Arabic governor of Ḥimṣ his life.232 By that time, the end of the war was already looming, as results from the ecclesiastical correspondence between Constantinople and Rome. In 677 the Byzantine fleet was still too busy to relieve Thessalonica, no doubt because of the Byzantine invasion of Lebanon, the preparation of which is probably reflected in the miracle of Saint Demetrius. But the Byzantine victory over the Arabs was complete by 678, when Thessalonica was finally relieved and when Constantine IV took the initiative of healing the rift with the church of Rome. It now remains to be seen if it is possible to find a place for a siege of Constantinople at another date in the 660s or 670s. 230. These are: 1) the report of the prefect of Thessalonica to the emperor about the treacherous behaviour of Perboundos; 2) imperial order of arrest; 3) despatching of the arrested Perboundos to Constantinople; 4) the joint Slavic-Thessalonian embassy to Constantinople; 5) its return to Thessalonica; 6) a dromon is sent by the emperor after the flight of Perboundos to Thessalonica in order to warn the city of the impending Slavic attack. To these we can add 7) the exile of the prefect of Constantinople to Thessalonica, plausibly suggested by Lemerle (Les plus anciens recueils [quoted n. 7], vol. 1, p. 199 n. 3). 231. Mirac. Dem. ii.4.235, p. 20929–31: ὁ ῥὴξ Περβοῦνδος […] ἔξεισι τῆς ἐν Βλαχέρναις πύλης. 232. See above, p. 274. 290 MAREk JANkOWIAk II. The blockade of Constantinople in 667–9 and the siege of 668 1. The expedition of Yazīd b. Muʿāwiyah against Constantinople In an article published in 1926, Marius Canard presented Islamic sources on Arab attacks against Constantinople in the 7th and 8th centuries.233 This rare study of the expedition of Yazīd b. Muʿāwiyah has not been given sufficient consideration by Byzantinists, perhaps because Canard unsuccessfully sought to find in Islamic sources a confirmation of the siege placed by Theophanes in the 670s. An opposite approach will be proposed below: after presenting the non-Byzantine traditions on Yazīd’s campaign against Constantinople, I will gather the fragmentary Byzantine evidence attesting a major crisis of the Empire around 668. Yazīd at Constantinople: Islamic sources Islamic sources associate the most important effort at conquering Constantinople with Yazīd b. Muʿāwiyah. They have, however, surprisingly little to report on the expedition he led. Its importance is emphasized by the participation of four Companions of the Prophet and leaders of the Medinan aristocracy: ʿAbdallāh b. ʿAbbās, paternal cousin of the Prophet and ancestor of the Abbasids; ʿAbdallāh b. ʿUmar, son of the second caliph and one of the most important transmitters of hadith; ʿAbdallāh b. al-Zubayr, son of a sister of the Prophet’s wife ʿĀʾisha, grandson of Abū Bakr and future caliph during the second civil war (683–92); and Abū Ayyūb al-Anṣārī, who hosted the Prophet during his stay in Medina.234 The caliph’s son and soon heir apparent, Yazīd, led these most respected figures of the Islamic community to take possession of the last substantial bastion of resistance against the Caliphate. But Islamic chronicles are discreet as to the outcome of the expedition and contradictory as to its date: as we have seen, they place Yazīd’s campaign hesitatingly between 49/669–70 and 55/674–5, whereas Theophilus of Edessa suggests rather 47/667–8 or 48/668–9.235 Alongside Yazīd’s expedition, al-Ṭabarī listed under the years 48/668–9 and 49/669–70 an exceptional number of campaigns, nine, but he did not make it clear whether they were part of a coordinated action aimed at conquering Constantinople. That this was indeed the case is confirmed by other sources, such as the Futūḥ Miṣr of Ibn ʿAbd al-Ḥakam, which shows some of the generals mentioned by al-Ṭabarī at Constantinople.236 The classical Muslim historians downplay the expedition of Yazīd b. Muʿāwiyah, no doubt on account of the opprobrium later attached to his person. Still, several episodes 233. M. Canard, Les expéditions des Arabes contre Constantinople dans l’histoire et dans la légende, JournAs 208, 1926, pp. 61–121. 234. See, for instance, al-Ṭabarī AH 49, ii.86 = xviii.94: “the raid of Yazīd b. Muʿāwiyah against the Byzantines occurred during this year. He reached Qusṭanṭīniyyah accompanied by Ibn ʿAbbās, Ibn ʿUmar, Ibn al-Zubayr, and Abū Ayyūb al-Anṣārī”, and the relevant articles in EI 2, and, on ʿAbdallāh b. ʿAbbās, in EI 3 (online edition accessed on 19 Dec. 2012). 235. See above, p. 270. 236. Ibn ʿAbd al-Ḥakam, Futūḥ Miṣr, p. 269 (Russian translation: ибн ʿАбд ал-Хакам, Завоевание Египта [quoted n. 179], p. 289): “we were at Constantinople and at the head of the Egyptians was ʿUqbah b. ʿĀmir, companion of the Prophet, and at the head of the Syrians Faḍālah b. ʿUbayd.” Both names are listed by al-Ṭabarī in AH 48–9. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 291 of his attack on Constantinople survived in traditions recorded in the 10th century in such compilations as the Kitāb al-aghānī or al-ʿIqd al-farīd.237 In accordance with his portrayal as an idle prince devoted to courtly pleasures, Yazīd is said to have composed the following poem upon hearing about the sufferings of an Arab army wintering in Byzantine territory: When I lie on carpets drinking a morning draught in Dayr Murrān with Umm Kulthūm What do I care about troops suffering from fever and smallpox at Chalcedon? 238 Informed of this poem, Muʿāwiyah despatched his son, aged at that time between 20 and 25,239 at the head of a corps destined to succour the Arab troops—hence the name of the “succouring expedition” (ghazwah al-rādifah)240—and to attack Constantinople. Another story, featuring a rare heroic deed by Yazīd, is placed during the siege itself.241 Two tents had been put up in Constantinople next to the city wall: one of them resounded with tambourines, drums and flutes whenever the Arabs were attacking, while the other responded in a similar way to every Roman charge; the former belonged to the daughter of the emperor, the latter to the daughter of Jabalah b. al-Ayham, the last Ghassanid ruler exiled to Byzantium.242 Yazīd, determined to please the Ghassanid princess, defeated the Byzantines at the walls, forced them to flee to the city and hammered a city gate with an iron club until it broke in two. A golden plate was later affixed to it to commemorate this feat. The story ends here, with the Arabs victorious in the encounter, but unable to penetrate into the city. But the most popular traditions related to the Arab siege of Constantinople replaced the controversial Yazīd with Abū Ayyūb, a respected Companion of the Prophet.243 They report his death in the land of the Rūm—on the way to Constantinople or during the siege, because of his age or by the epidemic of smallpox that decimated the army succoured by Yazīd—and burial next to the walls of the Byzantine capital. His tomb was either concealed to prevent its profanation by the Byzantines or protected by their guarantees of immunity. It was “rediscovered” by the Turks after the fall of Constantinople in 1453 and is still venerated in the mosque of Eyüp not far from Blachernai. Yazīd plays a central role only in traditions that survived in a non-Arab environment. A French scholar recorded in 1936 a kurdish panegyric recited by a Yezidi sheikh on 237. Canard, Les expéditions (quoted n. 233), pp. 67–77; and also H. Lammens, Études sur le règne du calife omaiyade Moʿawia Ier, MUSJ 3, 1, 1908, pp. 145–312, at pp. 306–11. 238. Abū al-Faraj al-Iṣbahānī, Kitāb al-aghānī, vol. 17, ed. ʿAbd al-Sattār Aḥmad Farrāj, Beirut 1959, p. 141. The poem is transmitted, with slight variations, by numerous sources, see Canard, Les expéditions (quoted n. 233), p. 69 n. 3. Umm kulthūm was a wife of Yazīd. 239. Yazīd was born between 22/642–3 and 27/647–8, see Lammens, Études (quoted n. 237), p. 189. 240. Masʿūdī, Les prairies d’or, § 1819. 241. Al-Iṣbahānī, Kitāb al-aghānī, vol. 17, p. 141, transl. by Canard, Les expéditions (quoted n. 233), pp. 69–70. 242. On whom see EI 2, s.v. Djabala b. al-Ayham, and J. Bray, Christian king, Muslim apostate : depictions of Jabala ibn al-Ayham in early Arabic sources, in Writing « true stories » : historians and hagiographers in the late antique and medieval Near East, ed. by A. Papaconstantinou, Turnhout 2010, pp. 175–203. 243. Sources in Canard, Les expéditions (quoted n. 233), pp. 70–6; see also EI 2, s.v. Abū Ayyūb al-Anṣārī. 292 MAREk JANkOWIAk the exploits of the Sultan Ezi during the siege of Constantinople.244 Having attached 18,000 tents to the fish on the sea surrounding Constantinople, Sultan Ezi converted the qāḍī of Constantinople to Yezidism and changed the stones hurled against him by the people of Constantinople into onions. Impressed by the powers of Ezi, Hejīya Sofīya, the daughter of the emperor Qostantīn, persuaded her father to give his throne up for Sultan Ezi who then ruled for 72 years. Even if the historical details of the siege of Constantinople, considered to be a Muslim city, vanished from the Yezidi traditions, it is remarkable that, at the remove of over twelve centuries, the kurdish panegyric preserves memories of the encounter between Yazīd b. Muʿāwiyah and Constantine IV. The siege of Constantinople is the main event associated by the Yezidis with their eponym.245 Overall, information on Yazīd’s expedition against Constantinople has been reduced in Islamic sources to the barest outline and a handful of minor anecdotes, in a striking confirmation of the efficiency of the “historiographical filter” applied by Abbasid historians to the memories of the Umayyad era.246 The political significance of the Arab attempt to conquer the Byzantine capital is, however, undeniable: after his return from Constantinople, Yazīd led the pilgrimage to Mecca, probably in Dhū al-Ḥijjah 50/ December 670, while in the next year the hajj was led by Muʿāwiyah himself, who used this occasion to declare Yazīd as heir apparent and to secure the support of the key figures of the ummah for this controversial decision.247 The three Companions who survived the campaign against Constantinople—Ibn ʿAbbās, Ibn ʿUmar and Ibn al-Zubayr—opposed Muʿāwiyah’s choice; but otherwise Yazīd’s campaign must have been considered successful enough to lend legitimacy to Muʿāwiyah’s attempt to establish the dynastic principle of succession in the Caliphate. The only sources traceable to early Muslim informants to have avoided the Abbasid rewriting of the Umayyad history are two Spanish eighth-century chronicles.248 They preserved a matter-of-fact description of the expedition of Yazīd b. Muʿāwiyah against Constantinople which allows us to identify it with the two-year blockade of the Byzantine capital during the patriarchate of Thomas. They are both remarkably well-informed 244. R. Lescot, Enquête sur les Yezidis de Syrie et du Djebel Sindjār (Mémoires de l’Institut français de Damas 5), Beyrouth 1938, pp. 61–4 (French summary) and 236–8 (text in kurdish). 245. On the role and picture of Yazīd in Yezidism, see J. S. Guest, Survival among the Kurds : a history of the Yezidis, London 1993, pp. 11–2 and 32. 246. Even Ibn ʿAsākir, overall sympathetic to Yazīd (see J. Lindsay, Caliphal and moral exemplar? ʿAlī Ibn ʿAsākir’s portrait of Yazīd b. Muʿāwiya, Der Islam 74, 1997, pp. 250–78), barely mentions the siege of Constantinople: Ibn ʿAsākir, Tārīkh madīnat Dimashq, vol. 65, p. 404, transl. in J. Lindsay, Professors, prophets, and politicians : ʿAlī ibn ʿAsākir’s « Tārīkh madīnat Dimashq », PhD diss. University of Wisconsin, Madison 1994, p. 254. See also Borrut, Entre mémoire et pouvoir (quoted n. 110), pp. 231–59, on the transformations of the siege of 717–8 and its leader Maslamah b. ʿAbd al-Malik. 247. The leaders of the hajj in AH 50–1 are uncertain (see Lammens, Études [quoted n. 237], p. 265 n. 6): Ibn khayyāṭ has Yazīd or Muʿāwiyah in AH 50 and Muʿāwiyah in AH 51, al-Yaʿqūbī— Muʿāwiyah in AH 50 and Yazīd in AH 51, and al-Ṭabarī—Yazīd or Muʿāwiyah in AH 50 and Yazīd in AH 51. Ibn khayyāṭ, who may be the closest to the truth, explicitly connects the hajj led by Yazīd in AH 50 with the expedition against Constantinople: “Yazīd b. Muʿāwiyah led the pilgrimage after he had returned from the land of the Romans.” 248. Brooks, The Sicilian expedition (quoted n. 14), pp. 457–8, is one of the few historians to have brought up the Spanish chronicles in connection with the siege of Constantinople, but he discarded their testimony as conflicting with the traditional date of 674–8. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 293 about Islamic history, which suggests an early Arabic informant, if not an early written source. The earlier of the chronicles, baptised Continuatio Byzantia Arabica by Theodor Mommsen, is also known as the Spanish-Arab Chronicle of 741.249 The name Chronicle of 724 would be better suited, as the text breaks off at the death of caliph Yazīd II in 724 and correctly reports, in the future tense, the arrangements for his succession— Yazīd was to be succeeded by his brother Hishām who would then pass the throne to Yazīd’s son Walīd.250 That these arrangements were respected when Hishām died almost 20 years later, in 743, is no argument for a later date of the chronicle, no more than the correct length of Leo III’s reign (717–41), no doubt supplemented at an early stage of transmission.251 The Continuatio Byzantia Arabica was thus composed only a decade after the Arab conquest of Spain; conspicuously uninterested in Spanish affairs, it reads like an attempt to explain who the new masters of Spain were. The later chronicle, Mommsen’s Continuatio Hispana, more commonly known as the Mozarabic Chronicle of 754,252 took a different perspective, centred on Spain and less sympathetic to the Arabs, but drew its 249. Henceforth abbreviated Cont. Byz. Arab., ed. J. Gil, Corpus scriptorum Muzarabicorum, vol. 1, Madrid 1973, pp. 7–14. I use the numbering of paragraphs of the old edition of Th. Mommsen, MGH aa 11, pp. 334–59. English translation: Hoyland, Seeing Islam (quoted n. 78), pp. 612–27; Spanish translation: J. C. Martín, Los « Chronica Byzantia-Arabica » : contribución a la discusión sobre su autoría y datación, y traducción anotada, e-Spania 1, 2006, http://e-spania.revues.org/329 (accessed 5 January 2013), §§ 21–63. See also C. E. Dubler, Sobre la crónica arábigo-bizantina de 741 y la influencia bizantina en la Península Ibérica, al-Andalus 11, 1946, pp. 283–349; Hoyland, Seeing Islam, pp. 423–6; C. Cardelle de Hartmann, The textual transmission of the Mozarabic Chronicle of 754, Early medieval Europe 8, 1999, pp. 13–29, at pp. 15–7; and works quoted below, note 252. 250. Cont. Byz. Arab. § 41: igitur Yzit rex Sarracenorum quarto explicato anno ab hac luce migrauit fratri regnum relinquens Hesciam nomine et post fratrem natum proprii seminis regnaturum adsciscit nomine Hulit. The existence of such an arrangement at the time of Yazīd’s death is confirmed by Islamic sources, e.g. al-Ṭabarī ii.1740–1 = xxvi.87–8. 251. A new argument for the date of 743/4 has been recently adduced by Martín, Los « Chronica Byzantia-Arabica » (quoted n. 249), § 17: Cont. Byz. Arab. § 29 speaks of Maroan, cuius ex filio nepos hactenus nostris temporibus illorum obtinet principatum, which Martín translates: “Maroán—precisamente el nieto del hijo de éste ejerce el principado entre los sarracenos en nuestros días”, and understands that a great-grandson (“el nieto del hijo”) of Marwan was ruling at the time of the composition of the chronicle. This would provide the terminus post quem of 743, when Walīd II became the first great-grandson of Marwan to succeed to the caliphal throne. But in classical usage ex filio nepos means “grandson born of a son” i.e., “son of the son”, see e.g. Laterculus Regum Vvandalorum et Alanorum, MGH aa 13, p. 458: Guntamundus nepos ex filio Geiserici Gentune reg. Cartagine (Guntamund was son of Gento, son of Geiseric, see PLRE 2, stemma 41), or Scriptores Historiae Augustae, Maximus et Balbinus iii.4, on Gordian III: hic nepos erat Gordiani ex filio, in the context of a discussion whether Gordian III was a grandson of Gordian I through the latter’s son or daughter. The year 743 thus becomes a terminus ante quem for the Cont. Byz. Arab. I thank krzysztof Rzepkowski for discussions on this passage. 252. Henceforth abbreviated Cont. Hisp. Editions: Gil, Corpus scriptorum Muzarabicorum (quoted n. 249), vol. 1, pp. 15–54; J. E. López Pereira, Crónica mozárabe de 754 : edición critica y traducción (Textos medievales 58), Zaragoza 1980; Id., Continuatio isidoriana hispana : crónica mozárabe de 754, León 2009, pp. 175–289. I use the numbering system of Th. Mommsen, MGH aa 11, pp. 334–68, whose edition is widely accessible. English translation: k. B. Wolf, Conquerors and chroniclers of early medieval Spain (Translated texts for historians 9), 2nd ed., Liverpool 1999, pp. 111–60. See also J. E. López Pereira, Estudio crítico sobre la crónica mozárabe de 754, Zaragoza 1980; A. M. Moure Casas, En torno a las fuentes de la Crónica Mozárabe, in Humanitas in honorem Antonio Fontán, Madrid 1992, pp. 351–63; Cardelle de Hartmann, The textual transmission (quoted n. 249). 294 MAREk JANkOWIAk knowledge of the Arabic and Byzantine history from the same source as the Continuatio Byzantia Arabica. The very close textual parallels between the two chronicles point towards a written source,253 translated into Latin independently by the two chronicles. The original language of this source is unknown: Greek or Syriac have been proposed,254 but I am not aware of any linguistic analysis to support this view; on the other hand, the hypothesis of a translation directly from Arabic is the most economical. In that case, the common source of the two Spanish chronicles, composed before 724, would be the earliest witness to Arab oral traditions on the early Caliphate, the only one set in writing before the end of the Umayyad period and free from Abbasid interference.255 This explains, to my mind, why the Spanish chronicles are the only sources to offer a comprehensive narrative of the expedition of Yazīd b. Muʿāwiyah against Constantinople: Muʿāwiyah, king of the Saracens, despatched a hundred thousand men to subdue Constantinople; they were to serve under his son Yazīd to whom he had also decreed his kingdom. They surrounded it with a siege through all the spring time until they could no longer bear the pain of hunger and pestilence, and abandoning the city they captured many towns and, after two years, safely returned laden with plunder to Damascus and the king, by whom they had been despatched.256 Although the Spanish chronicles do not provide an absolute date, they situate the siege of Constantinople by Yazīd around the time of the accession of Constantine IV, placed just before (Cont. Byz. Arab.) or just after Yazīd’s campaign (Cont. Hisp.). This corresponds to spring 668 or 669. We will see that the former date is more likely. The importance of this early Islamic tradition cannot be overestimated. It allows us to put together the pieces of the puzzle presented above: for the author of the Continuatio 253. Rather than two separate sources for Byzantine and Arabic material proposed by R. Collins, The Arab conquest of Spain : 710–797, Oxford 1989, pp. 54–6. Pace Dubler, Sobre la crónica arábigobizantina de 741 (quoted n. 249), I cannot see any connection between the Spanish chronicles and Byzantine or Syriac sources, such as Theophilus of Edessa or John of Nikiu, other than reporting the same events. 254. Th. Nöldeke, Epimetrum, MGH aa 11, p. 369; Hoyland, Seeing Islam (quoted n. 78), p. 425. 255. I will elaborate elsewhere on this important point. The debate on the place of composition of this source, usually thought not to be Spain, is not relevant to my argument; see Th. Nöldeke, Epimetrum (quoted n. 254), pp. 368–9 (Syria); L. A. García Moreno, Elementos de tradición bizantina en dos Vidas de Mahoma mozárabes, in Bizancio y la Península Ibérica : de la Antigüedad tardía a la Edad Moderna, ed. I. Pérez Martín and P. Bádenas de la Peña, Madrid 2004, pp. 247–71, at pp. 255–6 (Egypt); Hoyland, Seeing Islam (quoted n. 78), p. 424 (Syria); Martín, Los « Chronica Byzantia-Arabica » (quoted n. 249), § 10 (“origen oriental”). 256. Cont. Byz. Arab. § 26: Moabia Sarracenorum rex centum milia uirorum, quae Yzit filio suo, cui et regnum decreuerat, famularentur obsequio, direxit ad Constantinopolim debellandum. quam dum per omne uernum tempus obsidione cingerent et famis ac pestilentiae laborem non tolerarent, relicta urbe plurima oppida capientes onusti praeda Damascum et regem, a quo directi fuerant, post biennium salutifere reuiserunt (transl. Hoyland, Seeing Islam [quoted n. 78], p. 619, slightly modified). Parallel account in Cont. Hisp. § 39, where Mommsen has Yazīd return to Damascus post triennium rather than post biennium, but this must be a mistake, as no such lesson is to be found in the manuscripts, see Gil, Corpus scriptorum Muzarabicorum (quoted n. 249), vol. 1, p. 25, § 237 (salutifere post uiennium reuiserunt), and López Pereira, Crónica mozárabe de 754 (quoted n. 252), p. 48, § 295 (same lesson). THE FIRST ARAB SIEGE OF CONSTANTINOPLE 295 Byzantia Arabica, who wrote a generation earlier than Theophilus of Edessa, the siege of Constantinople by Yazīd, sent by his father in order to conquer the Byzantine capital— ad Constantinopolim debellandum—was the highlight of Muʿāwiyah’s caliphate. Far from being just one of the twenty annual Arab invasions against the Empire, as Abbasid chroniclers would have us believe, it was the culmination of Muʿāwiyah’s efforts to capture Constantinople; there is also no trace of Yazīd’s negative picture, so prominent in the later Abbasid sources.257 The Spanish chronicles thus allow us to evaluate the transformation of the Islamic historical memory that took place over the course of its successive rewritings.258 Their description of the two-year expedition of Yazīd corresponds remarkably closely to that of “a long-lasting incursion by the godless Saracens and their presence throughout the two years when [Thomas] was bishop”, recalled by the chartophylax George at the Sixth Council. Both events coincide chronologically. The Spanish chronicles also make clear why George spoke of Arab “presence” (παράστασις) rather than of a “siege”: the Byzantine capital was subjected to a close siege only during one spring, as the Arab army, too big to sustain a longer siege, was quickly confronted with famine and diseases, not unlike during the second attempt at conquering Constantinople in 717–8.259 The Arabs remained, however, for two years in Byzantine territory, which they plundered thoroughly before returning to Damascus. Yazīd at Constantinople: other sources Even though the expedition of Yazīd b. Muʿāwiyah against the Empire was known to Theophilus of Edessa, he curiously does not seem to have mentioned his attack against the Byzantine capital. Among his four dependants, the Chronicle of 1234 reports: Muʿāwiyah sent his son Yazīd, who caught up with Faḍālah and then together they invaded the territory of the Romans. They took captives and plundered, and generally did as they wanted. Under the pressure of this aggression the Romans offered them gifts; the Arabs made peace and went back to their country.260 Theophanes, Agapios and Michael the Syrian add that the Arab army advanced as far as Chalcedon;261 in addition, Theophanes tells the story of the garrison left by the Arabs in Amorion after the return of the main army to Syria, and of the Byzantine reoccupation of the city after a daring attack led by a snowy night by “the same cubicularius Andrew”. This 257. See the characterisation of Yazīd in Cont. Byz. Arab. § 27: iucundissimus et cunctis nationibus regni eius subditis uir gratissime habitus, qui nullam umquam, ut omnibus moris est, sibi regalis fastigii causa gloriam appetiuit, sed communis cum omnibus ciuiliter uixit. 258. Borrut, Entre mémoire et pouvoir (quoted n. 110), pp. 61–108. 259. As opposed to the narrative of the “patrician Trajan”, the descriptions of the two Arab sieges of Constantinople in the Spanish chronicles do not share common features, compare Cont. Byz. Arab. § 26 and 36, and Cont. Hisp. § 39 and 78. Roman resistance, for instance, is mentioned among the factors that forced the Arabs to retreat in 718, but not in 668. 260. Chron. 1234, § 138 (transl. Hoyland, Theophilus [quoted n. 78], p. 161). 261. Theoph. AM 6159, pp. 35027–3519; Agap. p. 489; Mich. Syr. xi.12, p. 454. 296 MAREk JANkOWIAk cross-reference proves that this otherwise unknown episode also comes from Theophilus of Edessa.262 This omission of Yazīd’s attack on Constantinople by otherwise well-informed Theophilus is all the more remarkable as an episode of the siege is extensively described in another early Syriac chronicle. On the last folio of the mutilated Maronite Chronicle, a long passage tells the story of an unsuccessful Roman sortie from the besieged Constantinople. It is worth quoting in extenso: Yazīd b. Muʿāwiyah went up again with a large army. While they were encamped in Thrace, the Arabs scattered for the purpose of plunder, leaving their hirelings and their sons to pasture the cattle and to snatch anything that should come their way. When those who were standing on the wall saw this, they went out and fell upon them and killed a great many young men and hirelings and some of the Arabs too. Then they snatched up the booty and went in to the City. The next day, all the young men of the City grouped together, along with some of those who had come in to take refuge there and a few of the Romans and said, “Let us make a sortie against them.” But Constantine told them, “Do not make a sortie. It is not as if you had engaged in a battle and won. All you have done is a bit of common thieving.” But they refused to listen to him. Instead, a large number of people went out armed, carrying banners and streamers on high as is the Roman custom. As soon as they had gone out, all the gates were closed. The King had a tent erected on the wall, where he sat watching. The Saracens drew them after them, retreating a good long way away from the wall, so that they would not be able to escape quickly when put to flight. So they went out and squatted in tribal formation. When the others reached them, they leapt to their feet and cried out in the way of their language, “God is great!”. Immediately the others turned tail in flight, chased by the Saracens, who fell on them, killing and making captives right up to the point where they came within range of the catapults on the wall. In his fury with them Constantine was barely willing to open the gates for them. Many of them fell and others were wounded by arrows.263 This is an account of a regular siege of the Byzantine capital conducted by an Arab army encamped in Thrace and led by Yazīd b. Muʿāwiyah. It cannot refer but to the only known campaign of Yazīd against Byzantium, which culminated in the siege of Constantinople in spring 668. The Maronite Chronicle, however, placed this event in the 21st year of Constans, i.e. 662/3.264 The reason for this discrepancy is not known; perhaps there was some confusion with the raid of Busr b. Abī Arṭāt which reached Constantinople in that year.265 It is also puzzling that the description of Yazīd’s campaign consists of a single anecdotal episode and lacks any context; this suggests that it was separated from 262. The cubicularius Andrew is known exclusively from the dependants of Theophilus of Edessa, see PmbZ #353. 263. Chronicon Maroniticum, ed. E. W. Brooks in Chronica minora (quoted n. 90), pp. 72–3 (textus), p. 56 (versio); English transl.: Palmer, The seventh century (quoted n. 79), pp. 32–3. 264. The date of Yazīd’s attack on Constantinople is lost with the missing previous folio of the Maronite Chronicle, but it can be recovered from the date of the next event, placed in 663/4. 265. As suggested by Robert Hoyland in Palmer, The seventh century (quoted n. 79), p. 32 n. 143, and Brooks, The Sicilian expedition (quoted n. 14), at p. 457. Raid of Busr: Agap. p. 488; al-Ṭabarī AH 43 from al-Wāqidī. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 297 the main narrative of Yazīd’s expedition which perhaps remained at the proper date, in the lost section of the Maronite Chronicle. Assuming that the Maronite Chronicle belongs to the same nexus of sources as the chronicle of Theophilus of Edessa,266 this confusion might explain the disappearance of the siege of Constantinople from the notice on Yazīd’s campaign in the dependants of Theophilus. Some disorder in the coverage of the years 663–8 in the chronicle of Theophilus of Edessa, or perhaps in his Byzantine or Syriac source(s), is also hinted at by the exceptionally extensive use of Islamic sources by Agapios of Membidj in the last years of Constans’ reign.267 At any rate, the anecdote reported by the Maronite Chronicle indicates that Yazīd’s army camped at the land walls of Constantinople, and that both armies came to close quarters. The same is suggested by an inscription from the Rhesion gate (Mevlevihane kapı), located in the central section of the Theodosian Wall of Constantinople, which probably attests to the renovation of this section of the wall in the year of Yazīd’s siege of the city. Two tabulae ansatae engraved on the northern wall of its southern tower have been inscribed with the following text: + Νικᾷ ἡ τύχη Κωνσταντίνου τοῦ θεοφυλάκτου ἡμῶν δεσπότου + + + + Ἀνενεώθη ἐπὶ [--]ο[-- τοῦ ἐνδοξο]τάτου ἀπὸ [ὑπάτω]ν πατρ[ικίου καὶ κουρά]τορος τοῦ βασιλικ[ο]ῦ οἴκου [τῶν] Μαρίν[ης] ἐν ἰνδ(ικτιῶνι) ιαʹ +268 The two inscriptions are no doubt contemporary and should be read together. The last line of the first inscription has been hammered out and replaced with two crosses, added to the one that since the beginning had its place at the end of the text. Attempts to reconstruct the original text have so far not been successful: they have inexplicably focused on circus factions, although a damnatio memoriae of an emperor would be more readily expected.269 The epigraphy of the text together with the name of the emperor, Constantine, point to the 7th century, a date consistent also with the role of the anonymous curator domus Marinae.270 As a result, the 11th indiction corresponds to the years 652/3, 266. The relationship between the Maronite Chronicle and Theophilus of Edessa has not been studied; I will return to it in a future paper. 267. See the forthcoming publication of this section by Robert Hoyland referred to above, note 143. 268. B. Meyer-Plath, A. M. Schneider, Die Landmauer von Konstantinopel. 2, Aufnahme, Beschreibung und Geschichte, Berlin 1943, no. 36, pp. 133–4. 269. H. Grégoire, Une inscription au nom de Constantin III, ou La liquidation des partis à Byzance, Byz. 13, 1938, pp. 165–75, thought of καὶ τῶν Βενέτων; see also the review of F. Dölger in BZ 38, 1938, pp. 582–3 (καὶ τῶν Πρασίνων?); A. Van Millingen, Byzantine Constantinople : the walls of the city and adjoining historical sites, London 1899, pp. 79 and 102 (καὶ (τῶν?) Ῥουσίων); D. Missiou, Who was the Constantine in the inscription no 8788 CIG IV? : a contribution to the study of the demes in the period of Herakleids, Byzantina 13, 1985, pp. 1477–86 (proposes to identify the emperor as Constantine IV). The Whites have not yet been proposed. 270. It is unclear why a curator domus diuinae took care of the restoration of the walls of Constantinople, but the importance of this office under Constantine IV is confirmed by the high position of a curator domus Hormisdianae among the dignitaries present at the Sixth Ecumenical Council, see e.g. ACO, ser. sec. II, p. 1425–26. On the increasing role of the curatores domus diuinae in the late 6th century, see D. Feissel, Magnus, Mégas et les curateurs des « maisons divines » de Justin II à Maurice, TM 9, 1985, pp. 465–76. 298 MAREk JANkOWIAk 667/8 or 682/3, but only in 667/8 there were co-emperors who could suffer a damnatio memoriae, a misadventure that indeed happened to the brothers of Constantine IV, Heraclius and Tiberius, whose names were removed from some of the documents of the Acts of the Sixth Council after their dethronement in 681.271 The inscription of the Rhesion gate thus in all likelihood bears witness to Byzantine preparations to the siege of Constantinople of early 668, or perhaps to the restoration of the city wall after Yazīd lifted it. The siege was also commemorated in liturgy. On 25 June, the church of Constantinople remembered the help provided contrary to reason and beyond any hope by the great God and our Saviour Jesus Christ, through the prayers of our Lady the all-holy Theotokos and ever-virgin Mary who bore him without seed, against the godless Saracens who were besieging our imperial city by land and sea.272 Although neither the synaxarion nor the typikon preserve any chronological details of the siege, this notice can only refer to the first of the two regular sieges of Constantinople by the Arabs, that of 717–8 being commemorated on 16 August.273 Thus, not only does the notice confirm that the city was besieged both by a land army and a naval force, but—assuming that the liturgical celebration commemorated the end of the siege, as in the case of the second Arab siege—it also tallies perfectly with the report of the Spanish chronicles that the siege took place in spring time. We can conclude that Yazīd lifted the siege of Constantinople in late June 668. Other contemporary documents can be tentatively associated with Yazīd’s campaign. A papyrus from the dossier of Flavius Papas274 mentions sailors sent from as far as Apollonos Ano (Edfu)—located a thousand kilometres up the Nile—for a kourson, or a naval raid, in an 11th indiction, which can only correspond to the year 667/8; it consequently refers to 271. The most salient example is the sacra to Pope Donus, whose heading bears the name of Constantine IV alone (ACO, ser. sec. II, p. 27–8), although it was issued by the three emperors, as showed by the answer of Pope Agatho (ACO, ser. sec. II, pp. 5215–17 and 12210–12) and a passage of the Liber pontificalis (p. 3504–5). 272. J. Mateos, Le Typicon de la Grande Église. 1, Le cycle des douze mois, Rome 1962, p. 3201–6: τῇ αὐτῇ ἡμέρᾳ, ἀνάμνησις τελεῖται τῆς ὑπὲρ λόγον καὶ πᾶσαν ἐλπίδα δωρηθείσης βοηθείας παρὰ τοῦ μεγάλου Θεοῦ καὶ Σωτῆρος ἡμῶν Ἰησοῦ Χριστοῦ διὰ τῶν πρεσβειῶν τῆς ἀσπόρως αὐτὸν τεκούσης παναγίας Δεσποίνης ἡμῶν Θεοτόκου καὶ ἀειπαρθένου Μαρίας κατὰ τῶν διά τε γῆς καὶ θαλάσσης κυκλωσάντων τὴν καθ’ ἡμᾶς βασιλίδα πόλιν ἀθέων Σαρακηνῶν. Almost identical text in Syn. CP, col. 7728–16. 273. Commemoration of the siege of 717–8: Mateos, Le Typicon (quoted n. 272), p. 3727–12; Syn. CP, coll. 90130–90427. The siege ended on 15 August 718 according to Theoph. AM 6210, p. 3996–7, and Nikeph. § 56. 274. Published by R. Rémondon, Papyrus grecs d’Apollônos Anô, Le Caire 1953, and J. Gascou, Papyrus grecs inédits d’Apollônos Anô, in Hommages à la mémoire de Serge Sauneron. 2, Égypte post-pharaonique, Paris 1979, pp. 25–34, see pp. 25–9 for P.Apoll.106 describing preparations for a kourson. On the date of the dossier, probably the 660s and 670s, see J. Gascou – k. A. Worp, Problèmes de documentation apollinopolite, ZPE 49, 1982, pp. 83–95, at pp. 83–9; C. Foss, Egypt under Muʿāwiya. 1, Flavius Papas and Upper Egypt, BSOAS 72, 2009, pp. 1–24, at pp. 4–5; and below, note 275. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 299 the Egyptian fleet which participated in the expedition of Yazīd against Constantinople.275 Although much less detailed than the Aphrodito papyri describing requisitions devoted to the naval raids, or koursa, in the first two decades of the 8th century,276 the dossier of Papas illustrates the strain put on the Egyptian administration and population by the requirements of the Arab fleet. The papyri mention drafting men for the fleet and for the shipyards of Babylon;277 another source shows that even the patriarch of Alexandria had to contribute to the effort.278 Although no similar documentation survives from Syria, the participation of the Syrian fleet in an important expedition against Constantinople is alluded to in one of the “Tales useful for soul” of Anastasius of Sinai. Sartabias, an inhabitant of Damascus possessed by a demon, gained some respite from his affliction when the demons joined the Arabs in their attack against the Byzantine capital: when the Saracens were about to launch a naval attack against the straits of Abydos by Constantinople, the demon appeared to Sartabias and told him: “Our commander sends us as soldiers to help our companions the Saracens in their expedition (κοῦρσον) against Constantinople, and I was also ordered to go. You will have respite, and no one will cause you annoyance until we return from the expedition.” And this is what happened.279 275. P.Apoll.83 l. 8–10: Βίκτωρι διακ(όνῳ) (ὑπὲρ) [ἀ]γορασία(ς) ψωμ(ίων) λόγ(ῳ) δαπά(νης) ναυτ(ο)καρ(άβων) πεμφθ(έντων) δ(ιὰ) Γρηγορίου πιστικ(οῦ) ἐπὶ τ(ῆς) ια ι(ν)δ(ικτίονος). No other 11th indiction within the brackets determined by Gascou – Worp, Problèmes (quoted n. 274), i.e. 652/3 and 682/3, saw any Muslim naval activity; this overlooked information allows therefore to establish the only firm date in the archive of Fl. Papas. See above, note 236, on the Egyptian fleet at Constantinople. 276. Greek papyri in the British Museum : catalogue with texts. 4, The Aphrodito papyri, ed. by H. I. Bell, London 1910, pp. xxxii–xxxv. On the koursa, see also A. M. Fahmy, Muslim sea-power in the eastern Mediterranean, London 1950, pp. 87–92, and F. Trombley, Sawirus ibn al-Muqaffaʿ and the Christians of Umayyad Egypt, in Papyrology and the history of early Islamic Egypt, ed. by P. Sijpesteijn and L. Sundelin, Leiden 2004, pp. 199–226. The Byzantines appear to have imitated this system: Zuckerman, Learning from the enemy (quoted n. 228), pp. 83–4 and 108. 277. For instance P.Apoll.28, 29, 38 and 106. See Foss, Flavius Papas and Upper Egypt (quoted n. 274), pp. 15 and 18–22, for the analysis of these documents. 278. History of the patriarchs of the Coptic Church of Alexandria. 3, Agatho to Michael I (766), ed. by B. Evetts (PO 5, 1), Paris 1910, p. 5 [259]: “in those days Alexandria was governed by a man whose name was Theodore, who […] tyrannised over the father, Abba Agathon, and troubled him; not only demanding of him the money which he was bound to pay, and taking from him thirty-six denarii as poll-tax every year, on account of his disciples, but that which he spent upon the sailors in the fleet he also exacted from him. […] The patriarch needed seven thousand denarii to satisfy the demands of Theodore the Chalcedonian, besides the taxes upon his property”; see also PO 10, 5, pp. 372–3 [486–7]: the patriarch Agathon “was compelled to build ships for the fleet. So Theodore the Chalcedonian, the governor of Alexandria, treated Agathon in the reign of the caliph Yazīd, son of Muʿāwiyah”. 279. A. Binggeli, Anastase le Sinaïte, « Récits sur le Sinaï » et « Récits utiles à l’âme » : édition, traduction, commentaire, PhD diss. université Paris IV-Sorbonne 2001, récit II 3, pp. 221 (text) and 533 (transl.): ὅτε οὖν ἤμελλον διὰ θαλάσσης ἀνελθεῖν οἱ Σαρακηνοὶ ἐπὶ τὰ στενὰ Ἀβύδου Κωνσταντινουπόλεως, φαίνεται ὁ δαίμων τῷ Σαρταβίᾳ καὶ λέγει αὐτῷ· ταξάτους ἔπεμψεν ὁ ἄρχων ἡμῶν, ἵνα βοηθήσωμεν τοῖς ἑταίροις ἡμῶν τοῖς Σαρακηνοῖς εἰς τὸ κοῦρσον κατὰ Κωνσταντινουπόλεως, καὶ ἐτάγην κἀγὼ ἀπελθεῖν. καὶ ἰδοὺ ἔχεις ἄδειαν, καὶ οῦδείς σοι παρενοχλεῖ 300 MAREk JANkOWIAk The date of the composition of this collection of stories, the early 690s,280 makes it likely that the demons took part in the expedition of Yazīd, no doubt in the Syrian fleet led by the qāḍī of Damascus, Faḍālah b. ʿUbayd. If the connection of the anecdote on Sartabias and of P.Apoll.83 with the siege of Constantinople is correct, both these texts confirm the participation of a substantial naval force in the army of Yazīd. Other sources attest to the devastation of the region of Constantinople.281 A late th 7 -century kontakion, composed probably in Syracuse, beseeches Saint Euphemia to “guard your city of Chalcedon and that of the Syracusans and every city” and to “drive the barbarous nations and the impious tribe of Hagar from them”.282 The implication is that it was composed when both cities were simultaneously under Arab occupation, which happened only around 668.283 The kontakion reads as a desperate plea for Euphemia’s intercession against enemies who had captured her home city. Another liturgical hymn indicates that the relics of Euphemia were evacuated from Chalcedon to Constantinople by Constantine IV,284 perhaps before Yazīd and Faḍālah b. ʿUbayd established their winter quarters there in 667/8.285 On the other hand, large-scale transfers of the Slavs and the inhabitants of Cyprus to Bithynia and the region of kyzikos under Justinian II286 hint at the extent of the desolation wrought by the Arabs, no doubt during the campaign of Yazīd and such raids as those of Faḍālah b. ʿUbayd, whose wintering in kyzikos in 670/1 was perhaps echoed by the “patrician Trajan”, and of Sufyān b. ʿAwf who probably reached the nearby Rhyndakos region around 672. Sources shed little light on Yazīd’s activities during a year or so separating the end of the close siege from his return to Damascus. He no doubt remained in the region of the Byzantine capital, impeding communication between Constantinople and the Western provinces: in autumn 668, the news of the death of Constans II in Syracuse probably took ἕως οὗ ὑποστρέψωμεν ἐκ τοῦ κούρσου. ὅπερ καὶ γέγονεν. I thank André Binggeli for putting his dissertation at my disposal. 280. Binggeli, Anastase le Sinaïte (quoted n. 279), p. 332. 281. On the impact of the Arab invasions of Anatolia in the 660s and 670s, see, more generally, my Notitia 1 and the impact of the Arab invasions on Asia Minor, Millennium 10, 2013. 282. A. Acconcia Longo, Il concilio Calcedonense in un antico contacio per S. Eufemia, AnBoll 96, 1978, pp. 305–37, at p. 337 (lines 5–8 of the οἶκος ιβʹ): αὕτη ταῖς σαῖς πρεσβείαις πόλιν σου Καλχηδόνα καὶ τῶν Συρακουσαίων τε καὶ πᾶσαν πόλιν φύλαξον, τὰ ἔθνη ἐξ αὐτῶν τὰ βάρβαρα ἐκδίωξον καὶ τῆς Ἄγαρ τὴν ἄνομον φυλήν. See pp. 320–2 for date (Constans II’s stay in Sicily, probably too early). The kontakion also mentions the emperors in plural (p. 331, prooemium l. 4). I cannot detect any allusions to the Monothelete controversy, which does not necessarily indicate a rigorous application of the Typos (pace Acconcia Longo, pp. 324–5), but the extinction of the theological debate after the council of 662. 283. See below, pp. 313–4, for the Arab attack on Syracuse after the death of Constans II. 284. P. Canart, Le palimpseste Vaticanus gr. 1876 et la date de la translation de sainte Euphémie, AnBoll 87, 1969, pp. 91–104, pp. 97–8 for the date, and p. 104 for the reference to Constantine IV: ἐπέδωκας ἑαυτὴν πρὸ καιροῦ τῇ πίστει τοῦ ἀναπεπαυμένου πατρὸς αὐτοῦ [sc. of Justinian, under whose reign the text was probably composed] Κωνσταντίνου τῆς ἀπὸ Χαλκηδόνος μετατεθεῖσα ταφῆς. 285. See below, p. 304. 286. Cyprus: B. Engelzakis, Cyprus, Nea Justinianoupolis, in Studies on the history of the Church of Cyprus, 4th–20th centuries, Variorum 1995, pp. 63–82 (I owe this reference to C. Zuckerman); the main sources for the transfer of the Cypriotes are: canon 39 of the Council in Trullo; Theoph. AM 6183, p. 3659–13; Constantine Porphyrogenitus, De administrando imperio, ed. Gy. Moravcsik, transl. R. J. H. Jenkins, Washington DC 1967, § 47. Slavs: Theoph. AM 6180, p. 36411–15. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 301 unusually long, more than three months, to arrive to Constantinople,287 and the patriarch Thomas was still unable to dispatch his synodical letter to Rome. It is also unclear how Yazīd’s invasion ended. The Chronicle of 1234 reports that the Arabs “made peace” in exchange for Byzantine gifts. Negotiations that led to the withdrawal of Yazīd’s army from the Byzantine territory are perhaps recorded in the panegyric of Juansher, prince of Caucasian Albania, produced and recited in the 33rd year of his rule, i.e. in 669/70,288 and preserved in the History of the Caucasian Albanians composed in the 10th century by Movses Dasxuranc’i. The panegyric culminates in the description of Juansher’s recent visit to the court of Muʿāwiyah: The king of the South again summoned the praiseworthy prince of the East, Juansher, with an invitation promising many gifts to add to his previous honours and to increase his glory. For leading citizens of the town of Byzantium had come to him to accept the yoke of tribute to the sons of Hagar, and since eunuchs had undertaken to kill Constantine in the thirtieth year of his reign, the king had ordered them to remain where they were until the arrival of the prince of the East. The celebrated lord arrived with his previous retinue and was welcomed with more reverence and esteem than on the previous occasion, for [the caliph] commanded that the palace of his own brother be prepared for his repose and that he be always seated with him at table. He then ordered him to enter into peaceable negotiations with the honourable gentlemen who had come from the imperial city. The king was greatly amazed at his judicious and profound knowledge, and the ambassadors from the kingdom of Greece were no less grateful to him, for he endeavoured in many ways to advise the king to their advantage.289 The 30th regnal year of Constans II, for he is the emperor who had been killed by the eunuchs, corresponds to AD 670/1, an impossible year, since not only does it fall three years after Constans II was assassinated in Syracuse, but it also postdates the declamation of the panegyric, which was presented at the winter “national assembly” in 669/70, little after the triumphal return of Juansher from Damascus. It results that the Albanian prince went to the caliphal court in 669, and that “the honourable gentlemen who had come from the imperial city” were sent to Damascus by Constantine IV at a time when Yazīd was still stationed in the region of Constantinople.290 The Byzantine embassy offered effectively a capitulation: it was ready to “accept the yoke of tribute to the sons of Hagar” 287. See below, pp. 308–9. 288. Date of the panegyric: C. J. F. Dowsett, The History of the Caucasian Albanians by Movsēs Dasxuranci (London Oriental series 8), London 1961 (henceforth Movs. Dasx.), II 28, p. 130: “in the course of these thrice ten and three years the Lord afforded great honour unto the victorious prince of Albania.” Juanšer counted his regnal years since his nomination as the sparapet of Albania shortly before the battle of Qadisiyya (6 January 638), a date confirmed by a coherent synchronism in Movs. Dasx. II 19, where the 15th year of Juanšer is equated with the 20th year of Yazdgird III and AH 31, all equivalent to AD 651/2. On Movses’ sources, in particular the “Eulogy of Juansher”, see C. Zuckerman, The khazars and Byzantium : the first encounter, in The world of the Khazars : new perspectives, Leiden – Boston 2007, pp. 399–432, at pp. 407–9. 289. Movs. Dasx. II 28 (p. 127 Dowsett). 290. Which excludes, pace Dowsett, The History (quoted n. 288), p. 127 n. 1, the identification of the Byzantine embassy with the mission of the cubicularius Andrew, on which see below, p. 303. 302 MAREk JANkOWIAk and relied on the intermediation of a peripheral ruler to mitigate the Arab demands. These harsh conditions—no different in substance from those offered by Muʿāwiyah two years earlier to the cubicularius Andrew i.e., the payment of a heavy tribute and the acknowledgment of Muslim suzerainty—are confirmed by the Chronicle of 1234 (tunc oppressi sunt Romani, et eis dederunt munera). They were accepted by the Byzantines no doubt because of the turmoil into which the Empire had been thrown by the assassination of Constans II in July 668: Constantine IV urgently needed free hands to deal with the rival emperor who succeeded to his father in Sicily.291 Finally, a poem celebrating the destruction of an Arab fleet should be removed from the dossier of sources related to the siege of Constantinople during the reign of Constantine IV.292 Its title—“iambic verses by Theodosius the Grammarian on the Arab ships, when the Christians destroyed them at Constantinople during the reign of Heraclius the pious”293—is misleading, since there was no Arab attack on Constantinople during the reign of Heraclius. The text celebrates in vague terms the destruction of an Arab fleet, perhaps in a storm294 or in a naval battle. It alludes to the Arabic war-cry Allahu akbar, speaks of ships built of Lebanese cedars, and, puzzlingly, refers to “fire-throwing ships” that were part of the Arab fleet.295 David Olster proposed to connect it with the siege of 674–678, but this hypothesis is based on an incorrect translation of line 13, where τὰς ἐκείνων [sc. Ἀράβων] ἀνθυπέστρεψας κάρας, addressed to Christ, does not mean “you took away the fear of their returning shadows”—which Olster takes to be an allusion to the Arab fleet returning every spring to Constantinople—but rather “you turned them back”. The poem is also unlikely to refer to Yazīd’s attack on Constantinople in 668, which ended in a voluntary retreat of the Arabs, not in a Byzantine victory; it probably commemorated the second siege of Constantinople under Leo III, as proposed by Spyridon Lampros and recently supported with new arguments by Marc Lauxtermann.296 Even though sources are fragmentary, they converge to demonstrate the reality of an Arab blockade of Constantinople between 667 and 669 and of a siege in 668. The vague memories of the “patrician Trajan” were not entirely misguided: the Arabs besieged Constantinople at the beginning of Constantine IV’s reign, they wintered in kyzikos in 670/1 and suffered a major setback at Syllaion in 673/4, if this event is to be identified with the defeat of Sufyān b. ʿAwf. The title of “caliph” perhaps reverberates in the name of the Arab leader Chaleph, echoing the connection between the expedition led by Yazīd b. Muʿāwiyah against Constantinople and his designation as the future caliph. Even the 291. See the next chapter. 292. To which it was added by D. Olster, Theodosius Grammaticus and the Arab Siege of 674–78, BSl. 56, 1995, pp. 23–8. See Σπ. Π. Λαμπρος, Ἱστορικὰ μελετήματα, Ἐν Ἀθήναις 1884, pp. 129–32 (edition) and 132–44 (commentary). 293. Λαμπρος, Ἱστορικὰ μελετήματα (quoted n. 292), pp. 132–3. Theodosius the Grammarian is not otherwise known, see PmbZ #7817. 294. See S. O’Sullivan, Sebeos’ account of an Arab attack on Constantinople in 654, BMGS 28, 2004, pp. 67–88, at pp. 80–1, where she associates the poem with the raid of the Arab fleet against Constantinople in 654, but see below. 295. “Allahu akbar”: lines 6–8; Lebanon: line 38; fire-throwing ships: line 39 (διήρεις νῆες αἱ πυρεκβόλοι). 296. M. Lauxtermann, Byzantine poetry from Pisides to Geometres : texts and contexts, vol. 2, chapter 10 (forthcoming). I thank Marc Lauxtermann for making his text available to me. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 303 length of the “siege”, seven years, can be accounted for: it may refer to the permanent Arab threat looming over the region of Constantinople between the expedition of Yazīd and Faḍālah in 667/8 and the last major action of the Arab fleet in 673/4. But memory failed the “patrician Trajan” on two critical points: he amalgamated a series of campaigns into a single “siege” of Constantinople, and focused on naval, rather than land, warfare, no doubt under the influence of the recent siege of 717–8.297 The presence of the Arab army at the land walls of Constantinople implies that the Byzantines were unable to oppose its crossing of the Bosphorus, no doubt because their fleet was stationed in Sicily at the time of the siege; it is thus unlikely that there was any naval warfare at all. But even if the depiction of the siege by the “patrician Trajan” is imprecise and at times erroneous, he was right to emphasize an event that shook the very foundations of the Byzantine Empire. he revolt of Saborios The chronicle of Theophilus of Edessa is the only source to report the events that led to the expedition of Yazīd b. Muʿāwiyah.298 The caliph sent his son in order to assist the Arab general Faḍālah, who had in turn been entrusted with the task of supporting the Byzantine general Saborios in his revolt against Constans II and his sons. None of the four dependants of Theophilus explains the motives that pushed Saborios, patrician and strategos of the Armeniakon, to renounce allegiance to the emperor and to promise to “subject the land of the Romans to Muʿāwiyah’s rule if he would send him an army and help him to kill Constans”;299 they take more pleasure in telling the story of the quarrel that opposed, in the presence of the caliph, Saborios’ envoy Sergius and the eunuch Andrew, one of the top officials at the court of Constantinople. Although Andrew, offended by Sergius, eventually exacted vengeance by castrating and hanging him, he failed to outbid Sergius’ offer of the entire tax revenue of the Empire and returned empty-handed to Constantinople to warn of the pact between Muʿāwiyah and Saborios. The imperial government tried to prevent Saborios from joining with the Arab army despatched by Muʿāwiyah under the command of Faḍālah b. ʿUbayd al-Anṣārī. But even before the imperial commander Nicephorus was able to offer battle to the rebel, an unusual accident changed the course of the events: when Saborios had learnt that Nikephoros was marching against him, he trained himself for battle. It happened that one day he was going out of the town on horseback as was his custom. When he came near the town gate, he struck his horse with his whip. The horse became restive and dashed his head against the gate, thus causing him to die miserably. In this way God granted victory to the emperor.300 Theophanes locates Saborios’ death in Hadrianoupolis; since among the many Anatolian cities named after Hadrian none was located in Eastern Anatolia, Saborios 297. See above, p. 253. 298. Theoph. AM 6159, pp. 34829–3519; Agap. pp. 488–9 and the folios newly published in Hoyland, Theophilus (quoted n. 78), pp. 320–1 (text) and 156–8 (translation); Mich. Syr. xi.12, pp. 451–4; Chron.1234 § 138. 299. Chron. 1234 § 138 (transl. Hoyland, Theophilus [quoted n. 78], p. 158). 300. Theoph. AM 6159, p. 35022–27. 304 MAREk JANkOWIAk must have started his advance towards Constantinople without awaiting the Arab army.301 The news of his death surprised Faḍālah in Melitene at the Byzantine-Arab border; deprived of the support of his Byzantine ally, he consulted his strategy with Muʿāwiyah, who decided to mount a major offensive against the Empire and to put his son at the head of the reinforcements. After both armies had made junction, the Arabs invaded the Empire and advanced as far as Chalcedon. Theophilus dated these events to the 26th year of Constans i.e., to 667/8.302 Assuming that Yazīd and Faḍālah spent the winter 667/8 in Chalcedon—a distant echo of which sounds in the poem on the hardships of the Arab army attributed to Yazīd—their invasion of the Byzantine territory started in summer 667, and the revolt of Saborios took place around spring 667. As for the Arab fleet, it must have arrived to the Sea of Marmara no later than in autumn 667, in time to prevent the despatch of the synodical letter of Patriarch Thomas to Pope Vitalian. Contrary to what might be expected, newly consecrated patriarchs needed, as a rule, several months to compose their synodical letters;303 the letter of Thomas, consecrated patriarch on 17 April 667, was therefore not ready before late summer or early autumn 667. By that time Arab forces were converging on Constantinople; after a winter in Chalcedon, they launched an assault against the Byzantine capital in spring 668. Although the siege of Constantinople was quickly lifted, perhaps on 25 June 668, because of the famine and disease that struck the Arab army, Yazīd stayed in the Byzantine territory for another year, and returned to Damascus post biennium, probably in late 669. During that time, the patriarch Thomas was still unable to send his letter to Pope Vitalian; but he may have been additionally deterred from communicating with the West by the disturbance in Sicily after the assassination of Constans II in July 668. The Arab invasion was over in winter 669/70, the probable date of the Byzantine reoccupation of Amorion, where Yazīd had left a strong garrison perhaps to serve as a foothold for a future attack on Constantinople. 301. Theoph. AM 6159, p. 35021: ἦν δὲ Σαβώριος εἰς Ἀδριανούπολιν. Agap. p. 489 has Awdīnā (‫)أودينا‬, possibly a corruption of Hadriana or a similar name (‫)?أدرينا‬. Mich. Syr. and Chron. 1234 do not mention the name of the city. The likely candidates are Hadrianoupolis in Honorias (300 km E from Constantinople), Hadrianoupolis in Pisidia (380 km SE from Constantinople), Hadrianoupolis/ Adriane in Pamphylia, and Hadrianeia in Hellespontos. 302. Theoph., Mich. Syr. and Chron. 1234 provide this date explicitly; Agapios places the events in the 8th year of Muʿāwiyah, whose first year coincides in his chronicle with the 19th year of Constans II (Agap. p. 486). 303. I know no synodical letter despatched earlier than 5 months after ordination: Gregory the Great, ordained on 3 September 590, sent his synodical letter in February 591 (Reg. I 24), which caused perplexity of the modern editor of his letters: quomodo autem factum sit, ut synodica quam ep. I 4 (a. 590, Sept.) “de subsequenti sub festinatione” promittit, nunc demum post quinque menses (a. 591, Febr.) detur, non intellegimus, nisi posteriori tempore, quam missa est, in registrum eam receptam esse dicamus (MGH epp. I.1, p. 29), an unnecessary hypothesis; Cyriacus, patriarch of Constantinople since late 595 or early 596, sent his synodical letter to Gregory around June 596, which Gregory answered in October 596 (Reg. VII 5); Epiphanius, patriarch since 25 Feb. 520, sent his synodical letter on 9 Sept. 520 (Epistolae imperatorum pontificum aliorum inde ab a. CCCLXVII usque DLIII datae Avellana quae dicitur collectio, ed. O. Günther [Corpus scriptorum ecclesiasticorum Latinorum 35], vol. 2, Vindobonae 1898, no. 233); the synodical letter of Vitalian, ordained on 30 July 657, arrived to Constantinople little before 18 April 658 (Maximus, epistula ad Anastasium Monachum, lines 3–4, in CCSG 39, p. 161). THE FIRST ARAB SIEGE OF CONSTANTINOPLE 305 2. The Byzantines in panic The fall of Constantinople in 668 would not have amounted to the end of the Roman Empire: the senior emperor, his court and the central administration all had been based since 663 in Syracuse. Constans II’s decision to leave Constantinople and to transfer the central administration of the Empire and a significant part of the troops to Italy is best seen as a bold strategic retreat aimed at buying time and building a naval force capable of conducting a counter-offensive against the Arabs.304 Although Constans’ wife and his three sons and co-emperors still resided in Constantinople, assisted by a regency council headed by the patrician Theodore of koloneia and the cubicularius Andrew,305 Syracuse was the main decision centre of the Byzantine world at the time of Yazīd’s expedition. It witnessed momentous events at the time of the siege of Constantinople, and although no source connects them with Yazīd’s campaign, it is worthwhile to speculate on their link with the Arab assault on the Byzantine capital. he death of Constans II Very little is known of the Sicilian period of Constans II’s reign.306 A small dossier of papal letters related to a trivial affair of church discipline, addressed on 27 January 668 to the metropolitan of Gortyna, to George bishop of Syracuse, and to the imperial cubicularius and chartularius Vaanes,307 shows business as usual: they confirm that relations between the pope and the imperial court were correct and display no hint of the impending Arab assault against Constantinople. The emperor also does not seem to have been preoccupied with the gathering storm. According to Agapios of Membidj, he campaigned against the Slavs before his death,308 perhaps in relation to the complex warfare in northeastern Italy that opposed the Avars, the Slavs, the Lombard dukes of Cividale, and Grimoald, the Lombard king of Pavia (662–71).309 Paul the Deacon, our main source, is too vague to allow a reconstruction of Constans II’s Italian policy, but we can conclude that on the eve of the Arab siege of Constantinople the emperor was more preoccupied by his Italian possessions than by the Byzantine capital. The campaign against the Slavs is the last known action of Constans. He was assassinated after his return to Syracuse, probably on 15 July 668.310 Theophilus of Edessa presents the crime as an unpremeditated act of the cubicularius Andrew, son of Troilos. When assisting the emperor at the bath, he covered “his head with shampoo and soap, 304. Zuckerman, Learning from the enemy (quoted n. 228), in particular pp. 80–4 and 108. Contra: S. Cosentino, Constans II and the Byzantine navy, BZ 100, 2007, pp. 577–603, at pp. 597–602. 305. Theoph. AM 6160, p. 35125–28. 306. The main, though not unproblematic, study remains P. Corsi, La spedizione italiana di Costante II, Bologna 1983, summarized in Id., La politica italiana di Constante II, Settimane 34, 1986, pp. 751–96. 307. R. Schieffer, kreta, Rom und Laon : vier Briefe des Papstes Vitalian vom Jahre 668, in Papsttum, Kirche und Recht im Mittelalter : Festschrift für Horst Fuhrmann zum 65. Geburtstag, hrsg. von H. Mordek, Tübingen 1991, pp. 15–30. 308. Campaign against the Slavs: Agap. p. 490; the date is given as the 9th year of Muʿāwiyah i.e., 668/9. The other three dependants of Theophilus of Edessa did not preserve this notice. 309. Paulus Diaconus, Historia gentis Langobardorum v. 17–28. 310. See below for the date. 306 MAREk JANkOWIAk so that he was unable to open his eyes, and took a silver bucket, which he had placed in front of the king, and brought it down on his head and battered him”.311 Andrew escaped unhindered, while “the servants remained outside waiting for the king to come out, but when they had been sitting for a long time and it was getting late and he still had not come out, they entered the bath and found him unconscious. They brought him out and he lived for that day, but then died having reigned for twenty-seven years”.312 To succeed him, the Romans “made king by constraint a certain Mezezios, an Armenian, for he was very comely and handsome”.313 This is a rather peculiar account. Theophilus insists heavily that Andrew acted singlehandedly, that other courtiers were unaware of the agony of the emperor, and that Mezezios became emperor against his will. His account reads like a vigorous denial of the existence of a conspiracy and an attempt to reject the entire responsibility for Constans’ death on Andrew son of Troilos.314 But other sources disagree both on the instrument of crime—the Greek Life of Pope Martin and the Sermo adversus Monotheletas of Anastasius of Sinai sensibly replace the silver bucket with a sword315—and on its sponsors: the Spanish Continuatio Byzantia Arabica attributes the death of Constans II to a “conspiracy of ministers”, Movses Dasxuranc’i speaks of “eunuchs [who] had undertaken to kill Constantine”, and even Theophilus himself mentions, in the description of the Sicilian expedition of Constantine IV, the murderers, in plural, of Constans II.316 Moreover, several manuscripts of the Chronicle of Theophanes include, in addition to the narrative derived from Theophilus of Edessa, a parallel story of Constans’ death: when the emperor Constans was in Sicily, Mezezios rebelled against him with others in the bath, and they killed him. The emperor [sc. Constantine IV] punished Mezezios and those who dared to do that with him and beheaded them, including the patrician Justinian, father of the future patriarch Germanos, and he castrated Germanos, although he had already reached puberty.317 311. Chron. 1234 § 139 (transl. Hoyland, Theophilus [quoted n. 78], p. 163). See also Theoph. AM 6160, pp. 35114–3529; Agap. pp. 490–1; Mich. Syr. xi.12, pp. 450–1, for parallel accounts of the death of Constans and revolt of Mezezios. 312. Agap. p. 490 (transl. Hoyland, Theophilus [quoted n. 78], p. 163). 313. Theoph. AM 6160, p. 3523–4. 314. A similar reasoning in P. Peeters, Une vie grecque du pape S. Martin I, AnBoll 51, 1933, pp. 225–62, at pp. 230–1, and D. Motta, Politica dinastica e tensioni sociali nella Sicilia bizantina : da Costante II a Costantino IV, Mediterraneo antico 1, 1998, pp. 659–83, at p. 660. 315. Peeters, Une vie grecque (quoted n. 314), p. 253 § 1: ὅστις Κωνσταντῖνος ἐν Σικελίᾳ, ἐν τῷ Φάτνης λουτρῷ μαχαίρᾳ ἀνῃρέθη; see also the useful discussion at pp. 228–31. Anastasius Sinaita, Sermo aduersus Monotheletas qui dicitur homilia tertia de creatione hominis, line 101, in Anastasii Sinaitae Sermones duo in constitutionem hominis secundum imaginem Dei necnon opuscula adversus monotheletas, ed. k.-H. Uthemann (CCSG 12), Turnhout 1985, p. 61: ὁ Μαρτίνου διώκτης ἐν Σικελίᾳ ὑπὸ μαχαίρας ἀπέθανεν. The dependants of Theophilus of Edessa (including the Greek lives of Maximus the Confessor dependant on Theophanes) are the only sources to mention the silver bucket. 316. Cont. Byz. Arab. § 41: Constans Augustus […] ministrorum coniuratione peremptus est (similar words in Cont. Hisp. § 34); Movs. Dasx. ii.28 (p. 127 Dowsett); Theoph. AM 6160, p. 3526–7; Agap. p. 491; Chron. 1234 § 139; a slightly different text in Mich. Syr. xi.12, p. 451. 317. Theoph. AM 6160, p. 352, app. crit. ad l. 9: τοῦ δὲ βασιλέως Κώνσταντος ἐν Σικελίᾳ ὄντος ἐπανέστη αὐτῷ Μιζίζιος μετὰ ἑτέρων ἐν τῷ λουτρῷ καὶ ἐθανάτωσαν αὐτόν. ἠμύνατο δὲ ὁ βασιλεὺς τὸν Μιζίζιον καὶ τοὺς σὺν αὐτῷ τοὺς τοῦτο τετολμηκότας καὶ ἀπεκεφάλισεν αὐτούς. ὡσαύτως δὲ THE FIRST ARAB SIEGE OF CONSTANTINOPLE 307 Names associated with the murder of Constans II point to a coup d’état in which the Byzantine government decided to remove the emperor:318 the father of the cubicularius Andrew was probably the Troilos who presided over the trials of the Dyothelete leaders Maximus the Confessor and Pope Martin between 653 and 656; Mezezios may have been the count of the Opsikion; and Justinian and Germanos were thought by Ernst Stein to be descendants of the Justinianic dynasty.319 The interpretation of these events depends on their date. The only source to give the full date of Constans’ death is the Liber pontificalis, but its date of 15 July 669320 is contradicted by documents from the imperial and papal chancelleries. They demonstrate that Constantine IV, Constans’ son and successor, reckoned the years of his “post-consulate” i.e., of his rule as the senior emperor, from a day between 1 October and 7 November 668, probably the date of the arrival of the news of Constans II’s death to Constantinople. Given that the date of 669 has recently found much currency,321 these documents need to be recalled: • all the sessions of the Sixth Ecumenical Council, held in Constantinople from 7 November 680 to 16 September 681, are dated to the 13th year of Constantine IV’s “post-consulate”.322 It results that Constantine counted the years of his “post-consulate” from a day between 17 September and 7 November 668; • letter of Pope Adeodatus to Abbot Hadrian of 23 December 673 (6th year of the post-consulate, counted from a day between 23 December 667 and 22 December 668);323 • acts of a synod held in Rome on an unknown day of October 678 (10th year of the post-consulate); they allow to narrow the interval down to 1 October–7 November 668;324 καὶ Ἰουστινιανὸν τὸν πατρίκιον, τὸν πατέρα Γερμανοῦ τοῦ γενομένου πατριάρχου‧ τὸν δὲ Γερμανὸν τραχύτερον ὄντα εὐνούχισεν. 318. I remain doubtful of hypotheses seeking to identify the organisers of the plot against Constans II at the court of Muʿāwiyah (Howard-Johnston, Witnesses [quoted n. 3], pp. 126, 225 and 490–1) or of Constantine IV (V. Prigent, La Sicile de Constant II : l’apport des sources sigillographiques, in La Sicile de Byzance à l’Islam, études réunies par A. Nef et V. Prigent, Paris 2010, pp. 157–87, at pp. 176–7). I am grateful to James Howard-Johnston for discussions on this matter. 319. Troilos: PmbZ #8524. Mezezios: J. Gouillard, Aux origines de l’iconoclasme : le témoignage de Grégoire II, TM 3, 1968, pp. 243–307, at p. 295 l. 251; see also PmbZ #5163; his affiliation with the Gnuni family is based exclusively on his name, see C. Toumanoff, Caucasia and Byzantium, Traditio 27, 1971, pp. 111–58, at pp. 135 and 149. Justinian and Germanos: E. Stein, Die Abstammung des ökumenischen Patriarchen Germanus I, Klio 16, 1920, p. 20; see also PmbZ #3557. 320. LP, p. 3446–7: XV die mensis iulii per XII indictionem praedictus imperator in balneo occisus est. 321. See, for instance, kaegi, Muslim expansion (quoted n. 149), p. 180; Howard-Johnston, Witnesses (quoted n. 3), p. 491; T. Greenwood, A corpus of early medieval Armenian inscriptions, DOP 58, 2004, pp. 27–91, at p. 49; Sarris, Empires of faith (quoted n. 36), p. 292. Most scholars consider that the indiction number in the Liber pontificalis is a slip, see e.g. Duchesne in LP, pp. 344–5 n. 7; Grierson, Tombs and obits (quoted n. 76), p. 49. 322. ACO, ser. sec. II, passim. The years in the dating formulae of the council are spelled out, and the lessons of both Greek and Latin manuscripts concord. 323. Thomas of Elmham, Historia Monasterii S. Augustini Cantuariensis, ed. Ch. Hardwick, London 1858, p. 246; Councils and ecclesiastical documents relating to Great Britain and Ireland, ed. by A. W. Haddan and W. Stubbs, vol. 3, Oxford 1871, p. 124. 324. W. Levison, Die Akten der römischen Synode von 679, Zeitschrift der Savigny-Stiftung für Rechtsgeschichte. Kanonistische Abteilung 2, 1912, pp. 249–82, at p. 277. Levison dated the synod of 308 MAREk JANkOWIAk • sacra of the emperor to the patriarch of Constantinople George issued on 10 September 680 (12th year of the post-consulate, counted from a day between 10 September 668 and 9 September 669);325 • the only document to suggest a different date is the sacra of Constantine IV to the synod of Rome, dated 23 December 681, in the 13th—instead of the expected 14th— year of Constantine’s post-consulate. Its heading, however, can otherwise be shown to be corrupt.326 Constantine IV thus counted his reign as the senior emperor from October or early November 668, perhaps from 5 November 668, the date given in the imperial obituary as that of the death of Constans II, which, however, falls within the interval determined above and may represent the day of Constantine’s accession.327 Not a single source as much as hints at Constantine IV having declared himself senior emperor before the death of his father; on the contrary, the beginning of his rule is always placed after the murder of Constans II.328 Furthermore, sources attribute 27 years of reign to Constans II, making it likely that he died around September 668;329 the only two texts dated by higher years of Constans’ rule come from the Caucasus and illustrate the slow propagation of the news of Constans’ murder and succession outside the borders of the Empire.330 We can conclude that the Liber pontificalis is mistaken331 and that Constans II died in Syracuse on 15 July 668; the fact that the news of his assassination reached Constantinople only Rome to October 679; see my forthcoming monograph on the Monothelete controversy for a detailed argument for a date one year earlier. 325. ACO, ser. sec. II, p. 1119–20, date preserved only in the Latin translation. 326. ACO, ser. sec. II, p. 86710–12. In addition to the mistaken post-consulate, the dating formula, preserved only in the Latin translation, contains a corrupt regnal year (29th instead of 28th) and an unintelligible mention of the place where the document was issued (in opostulorum: in the church of the Holy Apostles?). 327. Grierson, Tombs and obits (quoted n. 76), pp. 49–50. Grierson initially hesitated to draw this conclusion, but he later changed his mind (DOC II, 1, p. 403 n. 8). 328. See, for instance, the Spanish chronicles quoted below, note 338. 329. 27 years: Theoph. AM 6160, p. 35113; Agap. p. 490; Mich. Syr. xi.12, pp. 450–1; Chron. 1234 § 139; Cont. Byz. Arab. § 25; Cont. Hisp. § 34. Only Nikeph., Chronographikon syntomon, p. 9912, has 28 years, probably because he does not have a separate year for Heraclius’ sons. Constans was probably crowned in September 641, see Mango, Nikephoros (quoted n. 3), p. 192. 330. These are an Armenian inscription from Aruč of 24 March 670 dated to the 29th year of Constans (Greenwood, Early medieval Armenian inscriptions [quoted n. 321], p. 86, no. 11, see also pp. 39, 48–9 on the date), and a reference of Movses Dasxuranc’i to the 30th regnal year (Movs. Dasx. II 28; see above, p. 301). On the limited circulation of information across the political borders in the periods of conflict in the 7th century, see M. Jankowiak, Travelling across borders : a church historian’s perspective on contacts between Byzantium and Syria in the second half of the 7th century, in Arab-Byzantine coins and history (quoted n. 3), pp. 13–25. 331. This is not an isolated mistake in the section of the Liber pontificalis covering the second part of the pontificate of Vitalian and that of Adeodatus: Pope Vitalian is said to have died “not long after” the death of Constans II (LP, p. 3448), which is not exact for an interval of three and a half years; and the dating of the revolt of Mezezios to the pontificate of Adeodatus is contradicted by the reference to “the love shown to us by Vitalian in his lifetime during the revolt of our usurpers” in the letter of Constantine IV to Pope Donus (ACO, ser. sec. II, p. 820–22, transl. R. Price). See above, note 207, on the date of these biographies. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 309 after more than 3 months illustrates the difficulty of communicating with the Byzantine capital, still blockaded by Yazīd’s army. If Constans was assassinated on 15 July 668, twenty days after Yazīd had lifted the siege of Constantinople, but before this news reached Sicily, it is not too far-fetched to suppose a link between his death and the Arab assault against the Byzantine capital. A parallel with the events of 717–8 is instructive. According to the description of the second siege of Constantinople by the “patrician Trajan”, “the patrician Sergios, who was strategos of Sicily, along with the inhabitants of the West, gave up hope for Byzantium and the emperor himself on account of the enemy attack against them, and crowned their own emperor, one of the adjutants of Sergios, called Basil, the son of Gregory surnamed Onomagoulos, and renamed him Tiberios”.332 The revolt collapsed as soon as an envoy of Leo III brought news from the besieged Constantinople and reassured the people of Syracuse “that the Empire stood firm and that the City was confident as regards the enemy”.333 Sergios, who was later forgiven by Leo III, did not rebel against the regime in Constantinople. On the contrary, his revolt is best understood as a manifestation of Byzantine patriotism triggered by the prospect of the imminent fall of Constantinople. The goal of the strategos Sergios, the highest-ranking Byzantine official in the West, was not to challenge the emperor, and even less to express a separatist sentiment, but to ensure the survival of the Empire should its capital fall to the Arabs.334 Similarly, there is no reason to see the coup against Constans II in terms of social tensions between the nobility and an allegedly autocratic emperor, centrifugal tendencies, or religious antagonism.335 What seems to have prompted the Byzantine elites to withdraw their support from Constans II was his inactivity in the face of the supreme danger threatening the Byzantine capital. Constans’ withdrawal from Eastern affairs after 663, his campaign against the Slavs probably in early 668, and his intention to transfer his entire family from Constantinople to the West suggest that he did not favour the strategy of defending Constantinople at all costs. This approach was manifestly not shared by his entourage which removed him in order to succour the Byzantine capital. he revolt of Mezezios and the Sicilian expedition of Constantine IV Nothing is known of Mezezios’ reign, except that he issued coinage.336 There exist two conflicting accounts of his fall. According to Theophilus of Edessa, “when Constantine had heard of his father’s demise, he arrived in Sicily with a great fleet and, having captured 332. Nikeph. § 551–7 (p. 124 Mango); parallel account in Theoph. AM 6210, p. 3987–11. 333. Theoph. AM 6210, p. 39824–25. 334. On the Sicilian revolt of 717–8, see S. Caruso, Sulla rivolta in Sicilia dello stratego Sergio, in Byzantina Mediolanensia : 5 Congresso nazionale di studi bizantini, Milano, 19-22 ottobre 1994 : atti, a cura di F. Conca, Messina 1996, pp. 87–95; M. Nichanian – V. Prigent, Les stratèges de Sicile : de la naissance du thème au règne de Léon V, REB 61, 2003, pp. 97–141, at pp. 103–5, with older bibliography. The parallel with the revolt of Mezezios has already been made by Prigent, La Sicile de Constant II (quoted n. 318), p. 181. 335. See the catalogue of hypotheses in Motta, Politica dinastica (quoted n. 314), pp. 660–1, and the scepticism of Prigent, La Sicile de Constant II (quoted n. 318), pp. 175–7, esp. notes 96 and 107. 336. Mezezios’ solidi and semisses appear to be genuine, pace W. Hahn, Mezezius in peccato suo interiit : kritische Betrachtungen zu einem Neuling in der Münzreihe der byzantinischen kaiser, JÖB 29, 1980, pp. 61–70: see Ph. Grierson, A semissis of Mezezius (668–9), NC 146, 1986, pp. 231–2; 310 MAREk JANkOWIAk Mezezios, put him to death together with his father’s murderers”.337 The reality of the Sicilian expedition of Constantine IV is confirmed by the Spanish chronicles, even if they confusingly place Constantine IV, and not Constans II, in Syracuse at the latter’s death: “hearing at Syracuse that his father had been killed in a revolt of his own men, he [Constantine IV] headed for the palace with as large a fleet as he could muster and ascended the throne in glorious triumph”.338 Despite these explicit texts, E. W. Brooks, followed by many scholars, concluded to the “internal improbability of the Sicilian expedition” of Constantine IV.339 He based his verdict on the second tradition on the fall of Mezezios, transmitted by the Liber pontificalis: In his [Pope Adeodatus’ (672–676)] time Mezezios, who was in Sicily with the eastern army, rebelled and seized the kingship. The army of Italy made their way, some through the districts of Histria, some through the districts of Campania, and yet others through the districts of Sardinia and of Africa. In this way they all came to Sicily and to Syracuse, and with God’s help the unspeakable Mezezios was killed. Many of his judges were taken mutilated to Constantinople along with the rebel’s head.340 The revolt of Mezezios is here said to have been put down by Byzantine troops stationed in the West that spontaneously converged upon Syracuse. There is no mention of Constantine IV or of the Byzantine fleet; it is not specified who punished Mezezios’ supporters and sent them to Constantinople; finally, the role of the pope is left in the dark. This last point is surprising, not only on account of the Roman origin of the text, but also because in his sacra to Donus Constantine IV himself referred to “the love shown to us by Vitalian in his lifetime during the revolt of our usurpers”,341 no doubt alluding to Mezezios and his son John.342 It is therefore certain not only that the pope opposed the usurpation of Mezezios, but also that the notice on the revolt is misplaced in the Liber pontificalis, since it at least started during the pontificate of Vitalian. Even more curiously, the author of this section of the Liber pontificalis ignored Constantine IV and also C. Morrisson, Note de numismatique byzantine à propos de quelques ouvrages récents, RN 6e série, 25, 1983, pp. 213–26, at pp. 215–6 n. 8. 337. Theoph. AM 6160, p. 3524–7 (transl. Mango – Scott, Theophanes [quoted n. 22]). Parallel accounts: Agap. p. 491; Mich. Syr. xi.12, p. 451; Chron. 1234 § 139. 338. Cont. Hisp. § 40: hic aput Siracusam audiens seditione suorum occisum patrem cum classe qua potuit palatium petiit et tronum gloriose triumphando conscendit (transl. Hoyland, Seeing Islam [quoted n. 78], p. 619 n. 50); Cont. Byz. Arab. § 25 “has abbreviated this notice to the point of incomprehensibility” (Hoyland, Seeing Islam, p. 619 n. 50): Constantinus apud Syracusam audiens seditione suorum occisum patrem coronatur imperium. Despite an earlier notice on the death of Constans II at Syracuse, the chroniclers misunderstood that it was Constantine IV who was at Syracuse when his father died. 339. Brooks, The Sicilian expedition (quoted n. 14). 340. LP, p. 3465–9: Mezezius, qui erat in Sicilia cum exercitu Orientali, intartizavit et arripuit regnum. Et perrexit exercitus Italiae per partes Histriae, alii per partes Campaniae, necnon et alii per partes Sardiniae Africae; pari modo venerunt Sicilia in civitate Syracusana, et Deo auxiliante interemptus est nec dicendus Mezezius; et multi ex iudicibus eius truncati perducti sunt Constantinopolim, simul et caput eiusdem intartae (transl. Davis, The Book of Pontiffs [quoted n. 19], pp. 70–1, modified). 341. See above, note 331. 342. See below, note 344. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 311 altogether: his name does not appear in the papal biographies covering the first decade of his reign, not even on the occasion of his accession. In the context of the schism between the churches of Rome and Constantinople in the 670s,343 this silence is meaningful: even if the “unspeakable Mezezios” is pictured in a negative light, the biographer also marked his distance from the sons of Constans II. His silence about the “Sicilian expedition” of Constantine IV is motivated by Roman hostility to Constantinople and does not call into question its historicity; on the contrary, the intervention of troops from Constantinople is hinted at by the mention of the deportation of the mutilated supporters of Mezezios to the Byzantine capital, which finds confirmation in the notice on the castration of the future patriarch Germanos in some of the manuscripts of Theophanes. The sequence of events and their chronology are difficult to recover. The oft-quoted date of the fall of Mezezios, February 669, is based on the following obscure notice of Michael the Syrian: “at this time John, son of Mezezios, rose up against Constantine. His rebellion lasted seven months, until he was killed by the army which the emperor had brought to Sicily.”344 E. W. Brooks considered this notice to be a garbled duplicate of the account of the revolt of Mezezios; he therefore took the seven months of John’s rebellion to refer to that of his father, whose fall he consequently placed in February 669.345 Even if Michael’s notice does not inspire much confidence—John’s revolt is not reported by other dependants of Theophilus of Edessa—the existence of John son of Mezezios has recently been confirmed by a seal published by Vivien Prigent;346 there is thus no reason to reject it, although it is difficult to find a chronological space for his revolt.347 More importantly, it is impossible to date with any precision the fall of Mezezios. His revolt does not necessarily need to have been short-lived: the narrative of the life of Adeodatus, although biased, can perhaps be trusted for the date of the fall of Mezezios, around 672,348 a hypothesis supported by the sudden appearance of the Byzantine fleet in the war against the Arabs around that time, which suggests that Constantine IV did not control it before that date. The movements of the Byzantine fleet in the aftermath of the death of Constans II are referred to by several sources. According to a tradition recorded in the Patria Konstantinoupoleos, a tenth-century collection of Constantinopolitan folklore, it was brought from Sicily to Constantinople after the coup of Mezezios by a dignitary loyal to the dynasty: Severus, patrician and adoptive brother of the emperor Constans, the grandson of Heraclius, who was killed in a bath in Sicily, built himself a gerokomeion because it was his house; 343. On which see above, pp. 283–4. 344. Mich. Syr. xi.13, p. 455 (transl. Palmer, The seventh century [quoted n. 79], p. 195 n. 479). 345. Brooks, The Sicilian expedition (quoted n. 14), p. 459. 346. Prigent, La Sicile de Constant II (quoted n. 318), pp. 178–80; another seal had already been published in ZV 388. 347. Prigent, La Sicile de Constant II (quoted n. 318), p. 183, proposes to date both revolts, of Mezezios and of his son, before the Vitalian’s death in January 672. 348. Thus also Howard-Johnston, Witnesses (quoted n. 3), p. 492. 312 MAREk JANkOWIAk and his wife built a temple, because after the killing of the emperor in Sicily he brought the Roman fleet and went as far as Phoinix.349 On face value, this account does not appear trustworthy: neither Severus nor the gerokomeion or the church are otherwise known,350 and the name of Phoinix points to the Battle of the Masts fought in 654, fourteen years before Constans’ death. But adoptive kinship was a political strategy used by the Heraclians,351 and the transfer of the Byzantine fleet from Sicily to Constantinople soon after Constans II’s death must correspond to reality: this was the only fleet that Constantine IV could use to quell the revolt of Mezezios. Another trace of the movements of the fleet is perhaps preserved in the commentary on the Epigrams of Gregory of Nazianzus composed by kosmas of Jerusalem in the eighth century. Commenting on transporting warships over “the land of Thrace”, kosmas associated it not with the expedition of Xerxes, but with recent history: As for carrying the ships over the land of Thrace, many did it before and also Constantine the Younger did it, having driven the ships overland. For there is a place next to Thrace where dry land extends over six miles and which stands between (two) seas.352 Constantin Zuckerman proposed to identify “Constantine the Younger” as Constantine IV, the “dry land six miles wide” as the Thracian Chersonese, and the context of this manoeuvre as the siege of Constantinople.353 The reason why Constantine IV decided to “drive” his fleet overland was, still according to Zuckerman, the blockade of the Hellespont, no doubt by the Arab navy. Given that the presence of the Arabs in the region of the Sea of Marmara is attested only between winter 667/8 (wintering of Faḍālah in Chalcedon) and 670/1 (wintering of Faḍālah in kyzikos), Constantine’s stratagem is likely to be connected with the Sicilian expedition, but neither its exact date nor the direction of the crossing can be determined. 349. Patria Cpoleos iii.108, pp. 251–2 Preger: Σευῆρος, πατρίκιος καὶ ἀδελφοποιητὸς Κώνστα βασιλέως τοῦ ἔγγονος Ἡρακλείου τοῦ ἀναιρεθέντος ἐν τῷ βαλανείῳ εἰς Σικελίαν, ἔκτισεν αὐτὸς τὸ γηροκομεῖον διότι οἶκος αὐτοῦ ἦν· καὶ ναὸν ἤγειρεν ἡ γυνὴ αὐτοῦ, διότι μετὰ τὸ σφαγῆναι ἐν Σικελίᾳ τὸν βασιλέα ἐκεῖνος παρέλαβεν τὸν στόλον τὸν Ῥωμαϊκὸν καὶ ἀνῆλθεν ἕως τὸν Φοίνικα. 350. PmbZ #6696; R. Janin, La géographie ecclésiastique de l’Empire byzantin. 1, Le siège de Constantinople et le patriarcat œcuménique. 3, Les églises et les monastères, 2e éd., Paris 1969, p. 556. 351. Niketas, Heraclius’ cousin, was the adoptive brother of John the Almsgiver (C. Rapp, All in the family : John the Almsgiver, Nicetas and Heraclius, Nea Rhome 1, 2004, pp. 121–34); Constans II proposed Maximus the Confessor to be his adoptive father (Disputatio Bizyae, lines 630–45, in CCSG 39, pp. 131–3—I owe this interpretation to Gilbert Dagron). 352. PG 38, col. 534 (commenting PG 38, col. 82C): τὸ δὲ διὰ ξηρᾶς ναῦς ἀγαγεῖν τῆς Θρᾴκης, πολλοὶ μὲν πρότερον πεποιήκασιν, ἀλλὰ καὶ Κωνσταντῖνος πεποίηκεν ὁ νέος διὰ ξηρᾶς ἐλάσας τὰς ναῦς. Ἔστιν γὰρ περὶ τὴν Θρᾴκην τόπος ἓξ μιλίων διάστημα ξηρᾶς ἔχων καὶ τῶν θαλασσῶν μεταξὺ διιστῶν (transl. C. Zuckerman, A Gothia in the Hellespont in the early eighth century, BMGS 19, 1995, pp. 234–41, at p. 234, slightly modified). On kosmas of Jerusalem (d. ca 752), see ODB s.v. kosmas the Hymnographer. 353. Zuckerman, A Gothia (quoted n. 352). Zuckerman’s conjecture is reinforced by the allusion to Hexamilion (“where dry land extends over six miles”), a locality situated at narrowest place of the Thracian Chersonese. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 313 The reconstruction of the events is further complicated by an Arab attack against Syracuse. In the passage immediately following the description of the fall of Mezezios, the Liber pontificalis reports: Afterwards the Saracens came to Sicily, occupied Syracuse and caused much slaughter among the people who had fled to the walled towns and the hills. They returned to Alexandria taking with them enormous booty and the bronze which had been brought there by sea from Rome.354 Several other sources probably refer to the same event. A late Medieval catalogue of bishops of Syracuse attributes a violent death to the bishop George, one of the addressees of Vitalian’s letters in January 668, but the connection with the Arab invasion is not explicit.355 The History of the Patriarchs of Alexandria describes inhabitants of Sicily led into captivity to Egypt during the patriarchate of Agatho (661–77): the Muslims were fighting against the Romans furiously. And the Romans had a prince whose name was Tiberius, whom they had made their ruler, and who possessed many islands. So the Muslims took the Romans captive, and carried them away from their own country to a strange land. Thus with regard to Sicily and all its provinces, they took possession of that island, and ravaged it, and brought the people captives to Egypt. And this holy patriarch Agathon was sad at heart when he saw his fellow-Christians in the hands of the Gentiles; and as the conquerors had offered many souls of them for sale, he bought them and set them free.356 Tiberius “who possessed many islands” and, as we learn a little further, was “slain” at the beginning of the term of the next Coptic patriarch, John III (677–86), can only be the young brother of Constantine IV, dethroned in 681;357 that he is named as the ruler of Sicily indicates perhaps that the fall of Mezezios preceded the Egyptian raid against Syracuse. 354. LP, p. 3468–11: postmodum uenientes Sarraceni Siciliam, obtinuerunt praedictam ciuitatem et multa occisione in populo qui in castris seu montanis confugerant fecerunt, et praeda nimia uel aere qui ibidem a ciuitate Romana nauigatum fuerat secum abstollentes Alexandriam reuersi sunt (transl. Davis, The Book of Pontiffs [quoted n. 19], p. 71). 355. The catalogue, known as Episcoporum Syracusanorum numerus, was first edited by C. Eschobar, De rebus praeclaris syracusanis, Venetiis 1520 (non uidi), and used by R. Pirro, Sicilia sacra, 3rd ed., Panormi 1733, vol. 1, p. 609: Georgius, qui Constantinopoli didicit, hic troparia, qua in natiuitate Christi recitantur, et in Epiphania, fecit; occisus est autem in loco, qui uocatur […] Tritamari. On George of Syracuse as a hymnographer, see C. Émerau, Hymnographi Byzantini, Échos d’Orient 22, 1923, pp. 11–25 and 419–39, at p. 427. See also PmbZ #1966. 356. History of the patriarchs (quoted n. 278), p. 4 [258]. 357. History of the patriarchs (quoted n. 278), pp. 10–1 [264–5]. Other sources also hint at the division of the fronts of the war against the Arabs between the three sons of Constans II, see The Life of St. Andrew the Fool, ed. Rydén, lines 3921–44, and perhaps Mich. Syr. xi.13, p. 454 (at the beginning of the reign of Constantine IV and his brothers, “they”—referring probably to Heraclius and Tiberius—“went off to Gaul and to Italy and subjected all the peoples of the western region”, see Palmer, The seventh century [quoted n. 79], p. 194 n. 476). 314 MAREk JANkOWIAk On the Islamic side, al-Balādhurī reports that “the first to invade Sicily was Muʿāwiyah b. Ḥudayj al-kindī in the days of Muʿāwiyah b. Abī Sufyān”.358 Muʿāwiyah b. Ḥudayj—a comrade-in-arms of Maslama b. Mukhallad, governor of Egypt since 47/667–8, which confirms the Egyptian origin of the fleet—is known to have conducted several raids against the Byzantine province of Africa, the last of which, dated by Islamic sources to 50/670–1, should perhaps be identified with the major Arab raid against Africa recorded by Theophilus of Edessa under the year 669/70.359 Perhaps Sicily was attacked on that occasion, as part of a coordinated action aiming at undoing Constans’ efforts to consolidate the position of the Empire in the West. Arab raids may have targeted Mezezios, a hypothesis suggested by the description of a naval Arab raid against Sicily in the Futūḥ al-Shām of pseudo-al-Wāqidī, to which Vivien Prigent has recently drawn attention.360 Uncharacteristically for an Islamic source, it presents in some detail the political situation of the Empire: at the time of the Arab attack, the Romans were ruled by two hostile kings, a king of Sicily and a king of the Rūm called Qusṭantīn b. Hiraql. Despite his initial self-confidence, the king of Sicily, unable to withstand the Arab invasion, solicited the king of the Rūm for help; the Arabs preferred to avoid an encounter with the 600 ships sent by Qusṭantīn and returned to Syria. Although the Futūḥ al-Shām places these events in the caliphate of ʿUthmān, it is tempting to follow Prigent in identifying the two Byzantine rulers as Mezezios and Constantine IV; this suggests that the fall of Mezezios was accelerated by the Arab incursion against Syracuse. As we see, sources give contradictory indications as to the chronology of events. There is, however, no reason to contest the reality of the Sicilian expedition of Constantine IV. It is unlikely that the new emperor left Constantinople before the army of Yazīd returned to Damascus; but as soon as he had free hands, he restored the unity of the Empire. The rift between the two Byzantine courts of Constantinople and Syracuse explains the apparent inaction of the Byzantines in the first years of Constantine IV’s reign: only after the Byzantine army had been transferred from Sicily to the East could the Byzantines launch a counter-offensive that within several years allowed them to regain initiative and decisively defeat the armies of the Caliphate. A inancial collapse? Finally, two developments in the realm of Byzantine monetary economy datable to around 668 deserve a mention: a reform of the copper coinage, and the disappearance of copper coins of Constantine IV and his successors from the archaeological material. To the debated matter of the relationship between these two phenomena, we can now add the question of their connection with the siege of Constantinople in 668. 358. Al-Balādhurī, Kitāb futūḥ al-buldān, p. 235 de Goeje (p. 375 Ḥitti). 359. Islamic sources: Ibn ʿAbd al-Ḥakam, Futūḥ Miṣr, p. 194 (Russian transl.: ибн ʿАбд ал-Хакам, Завоевание Египта [quoted n. 179], p. 213); EI2 s.v. Muʿāwiya b. Ḥudaydj. Theophilus of Edessa: Theoph. AM 6161, p. 35212–14; Agap. p. 491; Mich. Syr. xi.13, p. 454; Chron. 1234 § 140. 360. Prigent, La Sicile de Constant II (quoted n. 318), pp. 181–2. The relevant passage is translated into Italian in M. Amari, Biblioteca arabo-sicula, Torino 1880, vol. 1, pp. 331–8. THE FIRST ARAB SIEGE OF CONSTANTINOPLE 315 The radical reform of the copper coinage continues to puzzle numismatists.361 At the very beginning of his reign, Constantine IV multiplied the weight of the copper follis by four, restoring it to the weight it had a century earlier, during the reign of Justinian.362 This drastic move cannot be explained merely by imitatio Iustiniani, the main motif of Constantine IV’s early propaganda, evidenced also in the change of the iconography of the solidi and in the name of his son, the future Justinian II, born in 669. In economic terms, the fourfold increase of the weight of the standard copper coin suggests either a radical change in the value of copper in respect to gold, or a deep restructuring of the Byzantine monetary system. In either case, an overnight change of the value of the basic monetary unit by 400 % could only destabilise the monetary system. It may, therefore, not be too far-fetched to attribute to Constantine’s “reform” the sudden disappearance of copper coins from the archaeological material in the cities of Anatolia and Greece after Constans II’s reign.363 With the exception of a limited number of sites, copper returned to circulation only in the 9th century. What rationale can we conjecture behind this unparalleled reform, that seems to have brought the Late Antique monetary economy to an end? Wolfram Brandes interpreted it as an adjustment of the central government to the changed economic and fiscal realities of the devastated Empire.364 The spectacular increase in the value of the follis suggests that the new copper coins were meant to substitute monetary units of a higher order i.e., silver and gold coins, which, in turn, implies a shortage of precious metals, plundered by the Arabs or paid as tribute. This interpretation fits well the picture of the crisis experienced by the Empire at the accession of Constantine IV. Confronted with an empty treasury, devastated territory, revolting Western provinces and a besieged capital, the new emperor had no choice but to take radical measures to finance the war with the Arabs. In these circumstances, payments to the troops in a high-value, good quality copper coin, rather than in gold or silver, are not inconceivable.365 In an Empire robbed of its riches, copper became the noblest metal. The monetary reform of 668 appears as a desperate move to deal with the crisis: the deregulation of the monetary system inherited from late antiquity was the price paid by the Empire for its survival. 361. See L. Schindler, Die Reform des kupfergeldes unter konstantinos IV, Numismatische Zeitschrift 86, 1955, pp. 33–5; D. M. Metcalf, The currency of Byzantine coins in Smyrna and Slavonia, Hamburger Beiträge zur Numismatik 4, 1960, pp. 429–44; A. R. Bellinger, The copper issues of Constantine IV, American numismatic society. Museum notes 12, 1966, pp. 120–1; DOC II, 1, pp. 514 and 517; BNC 1, p. 376; MIB III, pp. 17 and 157–9; Brandes, Finanzverwaltung (quoted n. 183), pp. 323–9; Prigent, Nouvelle hypothèse (quoted n. 195). 362. I omit other puzzling aspects of Constantine IV’s copper coinage: the introduction of the reform only in Constantinople and the continuation of the old coinage in Western mints, in particular in Sicily; double indications of value (on which see Prigent, Nouvelle hypothèse [quoted n. 195], pp. 568–70); and the countermarks on coins from Cyprus (MIB III, pp. 159–60). 363. The connection has been proposed by Brandes, Finanzverwaltung (quoted n. 183), pp. 325–8. On the end of the monetary circulation around 668, see also C. Morrisson, Survivance de l’économie monétaire à Byzance (viie-ixe s.), in The dark centuries of Byzantium (7 th–9 th c.), ed. E. kountoura-Galake, Athens 2001, pp. 377–97. 364. Brandes, Finanzverwaltung (quoted n. 183), pp. 325–6. 365. See Brandes, Finanzverwaltung (quoted n. 183), pp. 326–7, on the donativa in copper. See also BNC 1, p. 376: “la série ‘réformée’, surtout les folles et les pièces de trente noummia, semble avoir été une émission de prestige”. 316 MAREk JANkOWIAk Conclusion We can now conclude that the chartophylax George was right: the patriarch Thomas could not have possibly sent his synodical letter to Rome. During the two years of his patriarchate, from 667 to 669, Constantinople was threatened by a powerful Arab army led by the son of the caliph. This was the most determined effort undertaken so far by the Islamic state to conquer the Byzantine capital. It failed because of its logistic infeasibility, but it had a deep impact on the Empire, both in the short term—it probably caused the death of Constans II—and in a longer perspective, causing financial pressure that possibly ended a millennium of monetary circulation in Anatolia. On the other hand, Theophanes the Confessor was wrong: despite his ingenuity in connecting the disparate sources he had at his disposal, the long siege of Constantinople in the 670s has no historical substance and should now be removed from the handbooks. In the light of these findings, the history of the Byzantine Empire in the obscure decades of the 660s and the 670s can be rewritten. Arab forays into Anatolia started probably in 662/3, after Muʿāwiyah had emerged victorious from the civil war in the Caliphate. The first Arab raid reached Constantinople, and the following campaigns devastated vast stretches of Anatolia. In 667, the rebellion of Saborios, the general of the Armeniakon, offered the Arabs an opportunity to bring the Empire to heel. Although Saborios died accidentally before making junction with an army sent by Muʿāwiyah, the caliph decided to continue the offensive. The task of capturing Constantinople was entrusted to the caliph’s son, Yazīd, accompanied by the most prominent members of the Islamic community. The blockade of the Byzantine capital started probably in autumn 667, preventing the patriarch Thomas from despatching his synodical letter to Rome. After a winter in Chalcedon, the Arabs gave assault in spring 668. Although Yazīd was soon forced to lift the siege of Constantinople (on 25 June 668?), the prospect of the fall of the city sent shockwaves throughout the Empire and probably cost Constans II his life. His successor Mezezios, however, apparently took no action to relieve the besieged city. Meanwhile the Arab army roamed around Constantinople for another year or so, impeding communication with Rome until the death of Thomas in November 669. The events of the following 3–4 years are obscure, but by 672 or 673 Mezezios had been eliminated by Constantine IV and the Byzantine fleet launched a counter-offensive against the Arabs. In 672 or 673 Egypt was raided; in 673/4 the Byzantines won a major victory on land and see on the southern coast of Anatolia, perhaps not far from Syllaion; in 675/6 the Arab governor of Ḥimṣ fell victim to a raid on the Syrian coast; and finally, in 677/8 the Byzantines invaded the Lebanese coast and paralysed the Arab offensive capability. Even if no formal peace was concluded, the Arab raids ceased around 680. This reversal of fortunes in the 670s was enabled less by the discovery of Greek fire than by the return of Byzantine troops from the West after the fall of Mezezios, confirmed by the high proportion of Western names among the Byzantine elites during the reign of Constantine IV.366 The decisive role was played by the fleet built by Constans II during 366. Among the three patricians who defeated Sufyān b. ʿAwf, two have Latin names more frequent in the West than in the East in the 7th century, see PmbZ under Florus/Phloros and Cyprianus/ kyprianos. The rare name of Mikkinas, genikos kommerkiarios in the early 670s, is attested in Africa, Sardinia and Rome, see Morrisson – Seibt, Sceaux (quoted n. 189), p. 240, and PLRE 3, p. 889 THE FIRST ARAB SIEGE OF CONSTANTINOPLE 317 his Western stay, which enabled the Byzantines to transfer the warfare to the coasts of the Caliphate around 672/3 and to disembark in 677/8 troops that tied the Arab armies to the defence of Syria and effectively ended 15 years of annual devastation of Anatolia. This reconstruction builds on the advances in the understanding of the chronicle of Theophanes the Confessor and on recent work on such 7th-century sources as the Acts of the Sixth Ecumenical Council. But no study of the first Arab siege of Constantinople would be complete without Islamic sources. If not pressed too hard for absolute chronology—which they are unable to provide for the first several decades of the Hijri era—Muslim historians transmit a generally reliable, although highly selective, account of events. The narrative of the early Islamic history in the classical historians such as al-Ṭabarī results from a long process of selection and rewriting; only when these “historiographical filters” are removed, with the help of such early sources as the Spanish chronicles or the chronicle of Theophilus of Edessa, can we reach to the oldest layers of the Arab historical memory. The siege of Constantinople in 668 illustrates these mechanisms: the suppression of positive references to Yazīd resulted in the amplification of the anecdotal episode of the death and burial of Abū Ayyūb at the walls of Constantinople. This shift in emphasis did not make Yazīd’s expedition less historical, but it explains why it was slighted over by the Byzantinists. The reconstruction of the first decades of Islamic history requires thus a more sophisticated methodology than the traditional approach consisting in harmonising at all costs the Muslim historians with the framework provided by Byzantine sources.367 In the case of the first Arab siege of Constantinople, where difficulties arising from the transformations of the Islamic historical memory are compounded by the incoherences intrinsic to the account of Theophanes, this approach led to the overlooking of the expedition of Yazīd. This undervaluing of the Islamic historiography is probably the main reason why the problem of the first Arab siege of Constantinople, sometimes thought to be “l’un des plus difficultueux de toute l’historiographie byzantine”,368 waited so long for a solution. (Micinius). A patrician Innokentios who had served in the Opsikion is attested in Constantinople in 681, see W. Brandes, Philippos ὁ στρατηλάτης τοῦ βασιλικοῦ Ὀψικίου : Anmerkungen zur Frühgeschichte des Thema Opsikion, in Novum millennium : studies in Byzantine history and culture dedicated to Paul Speck, ed. by C. Sode and S. Takács, Aldershot 2001, pp. 21–39. See also Prigent, La Sicile de Constant II (quoted n. 318), p. 179 n. 123, where further bibliography. 367. See, for such an approach, Wellhausen, Die kämpfe der Araber mit den Romäern (quoted n. 161). 368. Grégoire, Une inscription au nom de Constantin III (quoted n. 269), p. 170 n. 1. 318 MAREk JANkOWIAk Chronology 662/3 667, 17 April 667 beginning of the annual Arab raids against Anatolia homas II becomes patriarch of Constantinople revolt of Saborios, Faḍālah b. ʿUbayd sent to assist him, death of Saborios, Yazīd b. Muʿāwiyah sent to reinforce Faḍālah 667/8 renovation of the Rhesion gate in Constantinople 667/8, winter Yazīd and Faḍālah winter in Chalcedon 668, spring siege of Constantinople by Yazīd 668, 25 June? Yazīd lifts the siege 668, 15 July Constans II assassinated in Syracuse, succeeded by Mezezios 668, 5 November? beginning of the reign of Constantine IV 668, spring—late 669? the army of Yazīd plunders the region of Constantinople 669 visit of Juansher in Damascus, negotiations with Byzantine envoys between 669 and 672 Sicilian expedition of Constantine IV, fall of Mezezios, Arab attack against Syracuse 669, 14 or 15 November death of homas II, succeeded by John V as patriarch of Constantinople 669/70, winter Byzantines reconquer Amorion 669/70? Arab raid against Africa 670? Pope Vitalian rejects the synodical letter of John V 670/1, winter Faḍālah winters in kyzikos 670, November/ Yazīd leads the hajj December 671, November Muʿāwiyah leads the hajj and declares Yazīd his heir 672 (or 673?) Byzantine raid against Egypt 672, March? paschal letter of the patriarch of Alexandria suggesting Byzantine presence in Egypt 672/3, winter big Arab leet under Muḥammad b. ʿAbdallāh and ʿAbdallāh b. Qays winters at Smyrna and in Cilicia and Lycia 673, March aurora borealis seen in Syria 673/4 defeat and death of Sufyān b. ʿAwf (at Syllaion?) 673/4 seals of the general kommerkiarioi issued by the apotheke of Africa THE FIRST ARAB SIEGE OF CONSTANTINOPLE 674/5 675, 2 September 675/6 676? 676?, spring 677, spring 677, 25 July 677, August/September 677/8 678 678, 12 August 680, 6 May 680/1 680, 7 November– 16 September 681 681, 28 March 319 ʿAbdallāh b. Qays and Faḍālah winter in Crete death of John V, succeeded by Constantine as patriarch of Constantinople governor of Ḥimṣ killed in a Byzantine raid against the Syrian coast the pope rejects the synodical letter of the patriarch Constantine afair of Perboundos, Constantine IV prepares a major naval expedition against the Arabs, beginning of the Slavic blockade of hessalonica ten warships are sent to relieve hessalonica, the rest of the Byzantine leet “busy with another war” Slavic general assault on hessalonica heodore becomes patriarch of Constantinople, sends a “letter of exhortation” to the pope Byzantine troops known as the Mardaites invade Lebanon hessalonica relieved by a Byzantine land army sacra of Constantine IV and his brothers to Pope Donus death of Muʿāwiyah last recorded Arab raid Sixth Ecumenical Council in Constantinople declaration of the chartophylax George at the 13th session of the Sixth Ecumenical Council Abbreviations Agap. = Kitab al-ʿUnvan (Histoire universelle), écrite par Agapius (Mahboub) de Menbidj, éd. et trad. en français par A. Vasiliev (PO 8), Paris 1912, pp. 397–550. Chron. 1234 = Chronicon anonymum ad annum Christi 1234 pertinens. 1, Praemissum est Chronicon anonymum ad A.D. 819 pertinens, interpretatus est I.-B. Chabot (CSCO 109. Scriptores Syri 56), Lovanii 1937. El. Nisib. = Eliae Metropolitae Nisibeni Opus chronologicum. Pars prior, ed. E. W. Brooks (CSCO Scriptores Syri. Versio, ser. 3ª, 7), Paris 1910. LP = Le Liber pontiicalis. 1, texte, introd. et commentaire par L. Duchesne, 2e éd., Paris 1955. Mich. Syr. = Chronique de Michel le Syrien, patriarche jacobite d’Antioche 1166-1199. 2, éd. pour la première fois et trad. en français par J.-B. Chabot, Paris 1901. Nikeph. = Nikephoros patriarch of Constantinople, Short history, text, transl. and comment. by C. Mango (CFHB 13), Washington 1990. heoph. = heophanis Chronographia. 1, Textum Graecum continens, rec. C. de Boor, Lipsiae 1883. 320 MAREk JANkOWIAk Main places mentioned in the text. B  /O  : Association des amis du Centre d’histoire et civilisation de Byzance 52, rue du Cardinal Lemoine – 75005 Paris. Tél. 01 44 27 17 79 – fax 01 44 27 18 85 e-mail : achcbyz@college-de-france.fr ou commande en ligne sur www.achcbyz.com Titres TM 15 16 17 BR 1 2 3 4 5 6 7 8 MTM 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 PCBE Prix unit. • Mélanges Jean-Pierre Sodini • Mélanges Cécile Morrisson • Constructing the seventh century, edited by Constantin Zuckerman • J. Lefort, Société rurale et histoire du paysage à Byzance • M.-Fr. Auzépy, L’histoire des iconoclastes • J.-Cl. Cheynet, La société byzantine : l’apport des sceaux, 2 volumes • J. Gascou, Fiscalité et société en Égypte byzantine • G. Dagron et V. Déroche, Juifs et chrétiens en Orient byzantin • J. Beaucamp, Femmes, patrimoines, normes à Byzance • D. Feissel, Documents, droit, diplomatique de l’Empire romain tardif • G. Dagron, Idées byzantines, 2 volumes • A. Laniado, Recherches sur les notables municipaux dans l’Empire protobyzantin • La pétition à Byzance, éd. D. Feissel et J. Gascou • Recherches sur la Chronique de Jean Malalas I, éd. J. Beaucamp et al. • C. Zuckerman, Du village à l’Empire : autour du Registre iscal d’Aphroditô • Byzance et les reliques du Christ, éd. J. Durand et B. Flusin • M. Loukaki, C. Jouanno, Discours en l’honneur du patriarche Georges Xiphilin • Lire et écrire à Byzance, éd. B. Mondrain • D. Feissel, Chroniques d’épigraphie byzantine (1987-2004) • K. Smyrlis, La fortune des grands monastères byzantins • Les nécropoles de Viminacium à l’époque des Grandes Migrations, éd. V. Ivanisevic et al. • Pèlerinages et lieux saints dans l’Antiquité et le Moyen Âge. Mélanges Pierre Maraval • Recherches sur la Chronique de Jean Malalas II, éd. J. Beaucamp et al. • La Crimée entre Byzance et le Khaganat khazar, éd. C. Zuckerman • M.-H. Congourdeau, L’embryon et son âme dans les sources grecques • M. Detoraki, Le Martyre de saint Aréthas et de ses compagnons (BHG 166) • J.-P. Grélois, Pierre Gilles, itinéraires byzantins • Oralité et lien social au Moyen Âge, éd. M.-Fr. Auzépy et G. Saint-Guillain • Puer Apuliae. Mélanges offerts à Jean-Marie Martin, 2 volumes • Guerre et société au Moyen Âge, éd. D. Barthélemy et J.-Cl. Cheynet • Le massacre de Najrân. 2, Juifs et chrétiens en Arabie, éd. J. Beaucamp et al. • J.-Cl. Cheynet et D. Theodoridis, Sceaux byzantins de la collection D. heodoridis • A. Mardirossian, La Collection canonique d’Antioche • Liquid & Multiple : Individuals & identities in the thirteenth-century Aegean (eds.) G. Saint-Guillain & D. Stathakopoulos • V. Ivanišević and M. Kazanski, he Pontic-Danubian Realm in the Period of the Great Migration • Les réseaux familiaux Antiquité tardive et Moyen Âge, (éd.) Béatrice Caseau • Géoponiques, (trad.) J-P. Grélois et J. Lefort • Réduire le schisme ? Ecclésiologies et politiques de l’Union entre Orient et Occident (XIIIe-XVIIIe siècle), édité par Marie-Hélène Blanchet et Frédéric Gabriel 75 € 90 € 90 € 40 € 40 € 60 € 40 € 40 € 40 € 40 € 60 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 30 € 60 € 30 € 30 € 30 € 30 € 30 € 24 € 60 € 24 € 30 € 24 € 30 € 24 € 30 € 24 € 180 € 144 € • Prosopographie chrétienne du Bas-Empire. 4, la Gaule Chrétienne sous la direction de Luce Pietri et Marc Heijmans 360 € 288 € • Participation forfaitaire aux frais de port / Postage (ixed rate for the entire order) Total 60 € 72 € 72 € 32 € 32 € 45 € 32 € 32 € 32 € 32 € 45 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € 24 € • S. Destephen, Prosopographie chrétienne du Bas-Empire. 3, Diocèse d’Asie Occ. Mon. • Kiev-Cherson-Constantinople, Ukrainian Papers • Collectanea Borisoglebica. 1, éd. C. Zuckerman • F. Androshchuk V. Zotsenko / Scandinavian Antiquities of Southern Rus’ • N. Alekseyenko L’Administration Byzantine de Cherson Catalogue • E. Vardanyan, Relets d’Arménie - Relections of Armenia • Jean-Bernard de Vaivre, Monuments médiévaux de Chypre : Photographies de la mission de Camille Enlart en 1896 N° d’ex. 40 € 40 € 50 € 50 € 29 € 32 32 40 40 23 € € € € € 36 € 29 € 6€ BON DE COMMANDE AU VERSO Association non assujettie à la TVA BON DE COMMANDE ET DE LIVRAISON COORDONNEES DU CLIENT nom / raison sociale adresse e-mail n° fax OUVRAGES ACHETES références prix TTC Frais de port Montant à payer 6€ € J’ai bien reçu / Je commande les ouvrages décrits ci-dessus. J’ai bien noté qu’une facture me sera envoyée à l’adresse indiquée. J’autorise ACHCByz à débiter, sur son site www.achcbyz.com (sécurisé par la Banque Postale), ma carte bancaire pour le montant à payer. n° carte bancaire : date de validité : cryptogramme (3 derniers chiffres au dos de la carte) : Date et signature Nota important : - seules les cartes bancaires françaises au logo « CB » et internationales appartenant au réseau VISA ou EUROCARD / MASTERCARD sont acceptées. Règlement par chèque bancaire ou postal français joint, à l’ordre de ACHCByz ou par virement : Code banque Code guichet N° compte Clé RIB Domiciliation IBAN BIC 30002 00420 0000009504J 03 Paris Maubert Mutualité FR76 3000 2004 2000 0000 9504 J03 CRLYFRPP Je demande l’envoi du volume par la poste à l’adresse suivante :