The population of Constantinople at the turn of the eighth century is difficult to calculate and that is even before trying to estimate how many inhabitants listened to the order that any who could not feed themselves for three years should leave.37 The previous 150 years had seen the city repeatedly ravaged by plague, warfare, siege, and disruption to its food supply. Therefore, despite it being a major hub for refugees fleeing the Arab conquests, the population of Constantinople had likely significantly declined from its peak of several hundred thousand. It has even been suggested that it had slumped to 50,000 by the mid-eighth century.38 Such a drastic decline may explain why the Aqueduct of Valens, cut during the Avar siege of the city in 626, was not repaired until 768, by which time the population of the city was starting to recover. The system of 100 internal collectors and cisterns39 was considered adequate to water the remaining population. This might present the unexpected benefits of a reduced population in times of siege. While there were fewer potential militia to defend the walls, it allowed for more concentration of resources for use by the soldiery. More generally, there were fewer mouths to feed and water. The reduced population will also have seen sizeable portions of the land encompassed by the walls left uninhabited, providing free space to plant and grow vegetables.40 Such intra muros allotments are unlikely to have made the city self-sufficient, even with its drastically reduced population, but it will have helped relieve some of the pressure caused by the Arab cutting of supply lines.41 The Great Chain might have also helped in the logistics of the capital. By maintaining control of the Golden Horn, the Romans made it impossible for the Arabs to enforce a complete blockade of Constantinople as a circumvallation of the full length of the Golden Horn and Galata’s defences may have been beyond even the vast Umayyad forces. Some supplies may therefore have entered the city from the Black Sea coast through the Golden Horn estuary.

  In general, logistical support was going to be vital to both the besieged and the besieger. As it was, due to their preparations and the defences of Constantinople, the Romans would prove more adept and ready for the fight, although the Arabs were by no means arrogant in their assessment of the resources needed for the mammoth task. In the early stages, the logistical planning by the Arabs met with success – their armies crossed Anatolia with its own supplies and a fleet carrying provisions, weapons, and men reached the coastlines around the Roman capital with little difficulty; however, upon arrival at Constantinople, the Arabs faced sterner Roman opposition than they initially thought; the winter weather was going to be substantially worse than planned for and their naval power was going to prove insufficient. But make no mistake, while there may have been some limited relief of the pressures imposed by the siege, the inhabitants and defenders of Constantinople suffered deprivation during the Arab blockade. Indeed, such was the importance of Constantinople, the siege of 717–718 was the Roman Empire’s ‘ultimate trial’.42

  That this contest could be viewed as ‘mission impossible’ and the ‘ultimate trial’ for each side demonstrates its importance. And beyond logistics, both sides had cause for confidence. The Arabs were buoyed by a century of almost unbroken success and by the substantial victories and conquests achieved by the Marwanid branch of the Umayyad dynasty since the mid-680s. And yet, the Romans could look back at their ability to survive the seventh century calamities through their infrastructural and military organisation, improvisational skills, and their toughness and sheer bloody-mindedness as a people. Lesser nations would have capitulated during the various nadirs of the 600s, but would the Romans crumble under the test from the Umayyad juggernaut?

  The Siege Begins – Early Umayyad Setbacks

  The scattered information about the siege of 717–718 does not only affect the make-up of the forces involved and their preparations; it also provides issues about its exact date. Again, we face deviations through inclusive counting, the differences between Roman and Muslim calendars, and generally different notions of what constituted the Arab attempt to capture Constantinople – the actual siege itself or the various campaigns that formed its prelude. There may even be some intentional changes to make the dates of the siege seem more symbolic.

  Some Arab sources mention the siege lasting two years, but this is either from counting inclusively – even a single day over a year would be considered ‘two years’ – or counting the ‘siege’ as involving the entire extended campaign from the overland advance of the Arab vanguard and the activities surrounding Amorion. Such a counting of preliminary operations as part of the ‘siege’ campaign as a whole makes it difficult to know how far back we should go. A combination of Tabari and Roman hagiography suggests an attack on a Roman island, possibly Crete, in AH97 (5 September 715 – 24 August 716) by Umar b. Hubayra, the Arab admiral for much of the siege campaign, which could see this attack be taken as part of the Arab advance against Constantinople. Do we take in the attempted naval and marine actions of Anastasius II and the Arab preparations they were meant to disrupt? Or possibly even the Arab capture of Tarsus in 712?

  In terms of the actual beginning of the siege, it is surely to be dated to the arrival of Maslamah’s forces before the walls of Constantinople, and maybe a few days after that as it would have taken some time for the Umayyad army to establish a landward blockade of the Constantinopolitan peninsula. Theophanes gives the date of the Arab arrival as 15 August 717,43 with the Synaxarium Ecclesiae Constantinopolitanae probably following Theophanes, but getting its date wrong by saying 16 August.44 However, when taken with the Arab sources, which have 15 August 718 as the date of the lifting of the siege, this would mean that it had lasted one year to the day in Theophanes’ dating, which seems a little too coincidental, particularly as 15 August was the date of the Feast of Dormition to the Virgin Mary, who would be given the credit for the survival of Constantinople in the years to come. It could even be that the Romans altered their own records to have this Marian feast play an even more prominent role than it did.45 Modern scholars tend to follow Nikephoros over Theophanes on this occasion, with the former stating that the siege lasted 13 months,46 which when coupled with the Arab sources, would suggest that Maslamah’s siege of Constantinople started around 15 July 717 and ended on 15 August 718.47

  The Arab sources are not completely clear on their dating of the siege either. That it lasted more than a calendar year caused some historiographical and chronological issues for Arab historians, with the likes of Tabari and al-Mas’udi placing it in AH98, which equates to 14 August 716 to 13 August 717, which seems a year early, although would incorporate the potential starting date of Nikephoros. On the other hand, Ibn Khayyat and Ya’qubi place it in AH99 (14 August 717 – 2 August 718), which is close to the dates attributed by Theophanes. Such discrepancy in the Arab sources may reflect the murkiness of information regarding the siege in the decades and centuries after it took place, particularly to its starting date. If that was what these four named Arab historians were focusing on, they could demonstrate a similar divide between them as between Nikephoros (July 717) and Theophanes (August 717).

  Whatever date the siege began, the main part of the Arab army arrived outside Constantinople and made Hebdomon, a coastal suburb of Constantinople, its main base of operations. This meant that, as the closest entrance to the main concentration of Arab forces, the Golden Gate became the focus of any Arab attempts to force a way into Constantinople. The Arabs then undertook the surrounding of all Constantinople’s land approaches ‘with a stockade, digging a great ditch and erecting above it a parapet-like wall of unmortared stone.’48 These siege walls extended from Hebdomon on the coast of the Sea of Marmara in a crescent north to the northern tip of the Golden Horn and then on to the shore of the Bosphorus to blockade Galata and the Golden Horn. As already mentioned, the coastline and estuary of the Golden Horn was difficult to surround, and it may be that the Roman garrison targeted this area for raids. However, the source for these raids, the Chronicle to 1234, speaks of the Arabs having to deal with Roman scouts attempting to disrupt supply lines ‘on the opposite coast,’49 without clarity as to which coast it is talking about – the opposite coast of the Constantinopolitan peninsula and therefore the Golden Horn, or that of the Sea of Marmara and therefore in the Asian lands of the Opsikon. But even if the Chronicle to 1234 is unclear, it is likely that the Arab presence on both of these ‘opposite coasts’ came under attack from Roman raids throughout the siege.

  Rather than challenge Maslamah’s crossing or his march east to the capital, Leo resorted to pulling most, if not all of his Thrace-based forces back to defend Constantinople. He also had them undertake a scorched earth policy, stripping the extra muros suburbs of the capital and its immediate vicinity of as much materiel and resources as they could.50 However, this lack of open Roman contesting of the Arab approach to Constantinople does not mean that Maslamah’s forces were one-dimensional in that approach or even that they had it all their own way. Maslamah seems not to have made a mad dash to Constantinople from Abydos through Rhegion to Hebdomon. On their arrival in Europe, Theophanes notes the Arabs ‘punishing the Thracian fortresses,’51 which could mean that part of the Umayyad army overcame some Roman forces stationed there. Tabari suggests that the Arabs conquered a ‘city of the Slavs’52 at around this time; and while he does not specify where this was (there were settlements of Slavs in Roman Asia Minor and even Arab Syria), that he mentions it in relation to Maslamah’s operations before Constantinople makes it likely that he is talking about a settlement in Thrace.

  This Arab ‘punishing of Thracian fortresses’ may have been a consequence or even a catalyst of another major part of the Umayyad advance towards Constantinople – the movements and near-demise of Maslamah himself. The Arab general may have crossed to Thrace ‘six miles below the city’53 – this could suggest that rather than from Abydos to Sestus, which was some 300 km from Constantinople, Maslamah having crossed directly to Hebdomon once his advance force was established there. However, this does not necessarily fit in with the report that he was in direct command of the Arab rear-guard of 4,000 cavalry, which presumably would have been marching east from Sestus to Hebdomon. It could be that rather than a rear-guard, Maslamah took command of a cavalry force to scout the area that was to be the hinterland of his camp for the foreseeable future. He is also recorded sending out foraging parties, which reputedly brought back mountains of grain,54 although it is unlikely that Maslamah would partake in such foraging.

  Whatever the circumstances for his presence in Thrace at the head of 4,000 cavalry, it almost came to fatal grief for Maslamah. During the night, ‘the Bulgar allies of the Romans fell upon him unexpectedly and slaughtered most of the force that was with him. Maslamah [only] escaped by a hair’s breadth.’55 We might enquire why the Bulgars launched such an attack. First and perhaps foremost, the diplomacy of Justinian II and Theodosius III had obtained a treaty and possibly even an alliance with the Bulgar khan. And it is quite possible that Leo III reached out to the khan, no doubt with his agents offering financial inducements, to ensure the continuation of these good relations. The Bulgars and their Slavic subordinates may also have been annoyed by any wide-ranging foraging and capture of the ‘city of the Slavs’ undertaken by Arab forces.

  There is no record of the Umayyads contacting the Bulgars in an attempt to achieve their neutrality or even an alliance against the Roman Empire. Of course, this is an argument from silence, and it does seem like a big enough oversight for it to be difficult to believe. Given the extent of their realm and the prominence of diplomacy in Maslamah’s expedition so far, the likelihood is that some sort of Umayyad embassy was sent to the Bulgars. Even if they were approached, the Bulgars stayed true to their agreements with the empire or decided that preserving a weakened Roman Empire was more to their benefit than helping the vigorous Arab caliphate take Constantinople. The potential benefit of capturing some imperial territory in Thrace, Macedonia, and along the Black Sea coast would surely be outweighed by being almost certainly the next target of the rampant forces of Islam, who would be looking to subdue the pagan Bulgars as they moved into the Balkan peninsula. ‘Better the devil you know… than the devil you do not…’ Any Bulgar rejection of Umayyad overtures might explain some of the more aggressive moves by Maslamah’s forces – ‘punishing the Thracian fortresses’, capturing the ‘city of the Slavs’, the widespread foraging and possibly whatever operation he was undertaking with the 4,000 cavalry. Perhaps the Umayyads arrived in Thrace knowing that the Bulgars were going to be hostile.

  The unclear chronology and gaps in the historical record regarding Bulgaro-Arab relations prior to the siege of Constantinople leaves us uncertain of what was the cause and what was the effect. Were the Bulgar attacks on the Arabs retaliatory or opportunistic first strikes? Were the Arabs attempting to forcefully assert their dominance in eastern Thrace over an already proclaimed Roman ally, reacting to Bulgar raids, or looking to secure the surrounding area without a specific target in mind beyond the Romans? Regardless of its origin, the threat posed by the Bulgars and Slavs to the Arab lines, potentially trapping them against the Theodosian Walls, will certainly have been much more acute after the near-death experience of Maslamah. The recognition of this threat was reflected in the dual aspect of the Arab circumvallation of Constantinople. The combination of stone wall, palisade and trench not only faced towards the immense defences of the imperial capital, but also out into the Thracian hinterland to protect the Arab position from attack.

  This might not have been just out of fear or wariness of the Bulgars. We have already seen that the Arab action involved in the ‘First Siege’ of Constantinople was possibly interrupted by the return of Roman forces from the west in the early years of the reign of Constantine IV. Maslamah may have been wary of a similar episode playing out once more, for given the amount of preparation the Arabs put into this siege, it would be expected that they tried to learn from previous experiences and planned to counter any issues their predecessors had come up against. Even if there was no sizeable Roman force abroad, with the forces of Roman Italy and Hellas either too distracted or not strong enough to inflict any real damage on Maslamah’s forces, a seaborne relief force from the Asian themes would almost certainly approach the Arab position through the Aegean Sea and then attempt to march along the Via Egnatia to threaten the Arab rear. Because of such considerations, Maslamah’s dual aspect camp meant that there was no real ‘rear’ to the Arab lines, which essentially constituted a fortified camp stretching from Hebdomon, around the Golden Horn to the Bosphorus shore north of Galata. Maslamah also seems to have charged a sizeable detachment of 20,000 men under the command of Sharahil b. Abd with patrolling the western approaches.56

  With his army in position, the next piece of Maslamah’s siege puzzle arrived – his fleet. Under the command of Suleiman, the Arab fleet appeared outside Constantinople on 1 September. This seems a little late compared to the arrival of the army in mid-August or even mid-July, but then the fleet had been overseeing, facilitating, and protecting the crossing from Abydos to Sestus and possibly that of Maslamah to Hebdomon. The Arab commander will also have been wary of bringing up their ships in dribs and drabs as that could have allowed the imperial navy to pick them off one at a time. Suleiman may also have advanced carefully through the Sea of Marmara to flush out any Roman squadrons looking to raid the Arab fleet, coastal positions, and supply lines.

  Initially, the Arab fleet anchored ‘between the Magnaura and the Kyklobion,’57 which is essentially along the coast of Hebdomon; however, two days later, with a south wind blowing, the fleet sailed past Constantinople to take up positions along the European and Asian coasts of the Bosphorus and the Sea of Marmara, from Galata to Kleidon in Thrace and Eutropios and Anthemios on the Asian side. Maslamah and Suleiman were either looking to reposition their ships or were forced into it. Might the anchorage off Hebdomon have not provided sufficient protection from either the weather or Roman naval raids? Or might this have been the next move in tightening the blockade of Constantinople by seizing control of the Bosphorus? A tighter blockade would have seen the imperial fleet confined to the Golden Horn – this would have allowed the Arabs to attack the sea walls, which would in turn have forced Leo to detach men from the land walls.

  However, the imperial fleet was still free to operate, and this dispersal of Arab ships presented Leo with an opportunity.58 The large, heavily-laden Arab katenai provided a tempting target. The Romans were further aided by a change in the weather. As the Arab rear-guard of 20 ships, with 2,000 marines on board, sailed north past the city, the wind first slackened and then turned from southerly to northerly. This and the current saw the Arab rear-guard pushed back towards Constantinople. Leo and his admirals sent out a squadron of their Greek-fire wielding dromones. This attack saw these 20 Arab ships (and maybe some supply ships they were escorting) left as ‘blazing wrecks’, with some crashing into the sea walls, others sinking, and others scattered along the coast and islands of the Sea of Marmara – Theophanes mentions the islands of Oxeia and Plateia, some 20 miles south of Constantinople.59

  On the surface, losing 20 ships from a supposed fleet of 1,800 may not seem like much of an issue, but even looking past the exaggeration of Arab ship numbers and the possibility that these 20 ships were a more significant proportion of a core of 200–300 Arab warships, the Roman ability to quickly destroy the Arab rear-guard left lasting psychological effects. It was a reminder to the Romans that their fleet was capable of dealing damage to the Arabs and delivered a general morale boost to the defenders; to the Arabs, it was a lesson in the dangers of the Greek fire-wielding Roman navy and of manoeuvring in the waters around the imperial capital, making them wary of pressing their naval blockade as tightly as they would have liked. Even when the Arab fleet was further reinforced as the siege went on, ‘they dared not confront the [Romans] in open waters.’60

 

1 2 3 4 5 6 7 8 9 10 11 12 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 40 41 42 43 44 45 46 47 48