Chapter 4
Robert Guiscard’s entry into the Byzantine nobility in 1074 was unpopular with many of its members, and the worst-case scenario they envisioned became a reality when, in March 1078, Emperor Michael VII Doukas was deposed by the general Nikephoros Votaneiates. The new emperor immediately repudiated the treaty with Guiscard, ensuring that the duke’s daughter’s impending marriage to Michael VII’s son Constantine was cancelled. Naturally the duke was angered by this turn of events, and he no doubt began to contemplate an invasion of the empire upon hearing the news. While Guiscard already had some justification for doing so, Michael VII’s subsequent ‘appearance’ in southern Italy bolstered the legitimacy of his planned invasion: that is, not only was he avenging the insult to his daughter, but he was also working to reinstate the rightful emperor. Guiscard’s nemesis Pope Gregory VII promptly excommunicated the usurper Votaneiates, although this was both a futile and pompous gesture given that the empire did not acknowledge the spiritual, let alone political, authority of the Roman patriarch. This act, however, did not amount to a vote of confidence in the duke, whom the Pope excommunicated for the third time in March 1080. However, the bitter enemies managed to reconcile their differences at Ceprano in June, and in the following month Gregory wrote to the bishops of Apulia and Calabria, implying that supporting the reinstatement of ‘Michael’ as emperor was both important and just. There was, however, a sizeable chink in the moral armour of Gregory and Guiscard: as even the pro-Norman Malaterra admitted, the deposed ‘Michael’ was, to use the words of William of Apulia and Anna Komnene respectively, in fact a ‘trickster’ or ‘false emperor’. The duke probably knew this, or came to know it, but he nonetheless maintained the charade, evidently managing to dupe the Pope himself. Preparations began for an invasion of the empire, the scale of which would make previous campaigns seem small in comparison.1
Malaterra and Anna Komnene, both writing with considerable hindsight, believed that Guiscard’s intention was to depose Votaneiates and become emperor himself; a view reiterated by Chalandon at the beginning of the twentieth century. However, as Komnene herself noted, a ‘barbarian’ could not expect the loyalty ‘of the Roman people and army’, a fact later demonstrated when the dynasty violently founded by Count Baldwin IX of Flanders in 1204 was overthrown by the Byzantine Palaiologos family in 1261. While Guiscard might have intended to become emperor eventually (as Komnene claimed), initially at least he needed to rule by proxy; hence the importance of ‘Michael’ to his campaign. However, others have suggested that Guiscard actually intended to acquire land in the western Balkans for his son Marc Bohemond. Owing to the duke’s need to maintain legitimacy among the Lombards of Campania, Bohemond’s half-brother Roger Borsa, son of Sichelgaita of Salerno, was nominated as the ducal successor. As the duke’s eldest son, Bohemond was therefore disenfranchised by this decision. The idea that Guiscard wanted to create a patrimony for the son he had passed over is certainly supported by the fact that he concentrated on securing the coastal regions of Illyria and northern Greece which faced his possessions in Apulia (although subjugation of the south-western regions of Constantinople was clearly required if there was to be any hope of creating a viable lordship). If he could install a puppet, the possession of such territories could then be ratified and hence legitimately incorporated into his already extensive dominions. That being said, despite the hindsight of Malaterra and Komnene, the fact remains that their interpretation does accord with both the character and the obvious ambition of Robert ‘the Cunning’.2
Guiscard’s chances of success, however, were significantly lessened when the septuagenarian Votaneiates was overthrown in April 1081 by his talented general, Alexios I Komnenos, who at twenty-five years of age was already a military veteran. The new emperor’s family had previously provided the empire with a ruler – Alexios’ uncle Isaac I (r. 1057–59) – whose bellicose image on imperial coins was copied on those struck by William the Conqueror’s mints in the kingdom of England. Not only had Alexios defeated a much larger army of Byzantine rebels in 1078 (c. 6000 vs 12,000 men), but four years earlier he had also been responsible for ending the rebellion led by the Norman commander of the Frankish regiments, Roussel of Bailleul, who had in the previous decade served Count Roger in Sicily. Moreover, since he had long fought both with and against Normans and other French-speakers during the civil wars of 1077–81, Alexios was familiar with their tactics. The army Alexios inherited, while significantly smaller than it had been prior to the reign of Constantine IX Monomachos (r. 1042–55), was still a formidable force: the internecine warfare, both internal and external, of the years leading up to his accession to the throne meant that it was highly experienced and capable of defeating a variety of foes. Nevertheless, Alexios had no desire to wage a war against Guiscard, and since he had solidified his power base by reconciling with the Doukas dynasty which Votaneiates deposed, he was in a position to make amends for the treatment meted out to the duke’s daughter Olympias (now known as Eleni). According to both Byzantine and Norman sources, Alexios took pains to treat Eleni with great deference, ensuring that she was readopted into the imperial family. While never explicitly stated, as the emperor nominated Eleni’s intended husband Constantine Doukas to be his successor, the potential for Guiscard’s daughter to become a future empress as previously promised was implied. The duke and Pope remained unmoved, however. Alexios was simply yet another usurper to a crown belonging to their ally Michael VII Doukas, who, contrary to their belief or official position, was not seeking help to regain his throne in Italy, but was instead a resident of a monastery in Constantinople.3
Pope Gregory’s reconciliation with, and support of, Guiscard was borne out of political necessity. The western emperor Henry IV – technically a king at this stage – was planning to invade Italy and depose the man whom he famously dismissed as ‘not pope but false monk’. The timing was perfect for Alexios; he promptly contacted Henry and organised a mutually beneficial alliance. In return for Henry’s invasion of northern Italy, Alexios agreed to deal with those who, after the battle of Civitate in 1053, no longer recognized the nominal overlordship of the western emperor (and were accordingly viewed by Henry as rebellious subjects). The eastern emperor’s strategy was twofold: repel the impending invasion, and foment rebellion throughout Longobardia via the services of yet another Norman in imperial service – the duke’s frequently rebellious nephew, Abelard (son of Count Humphrey). Guiscard was therefore now in the unfavourable position he had found himself in during 1074; he was opposed by powerful enemies to his north and east. Nonetheless, exhibiting the strenuitas (‘resoluteness’, ‘dynamism’) that Malaterra believed to be an innate characteristic of the Normans, he now began levying troops and organizing a fleet with which to invade the Byzantine empire. Despite the opposition faced – all sources attested to the unpopularity of the expedition – the duke managed to assemble his largest fleet and army recorded to date. Mustering at Otranto, where additional ships were built, the first target was set: Dyrrachion in modern-day Albania, at that time the empire’s fourth-largest city.4
As would be expected, the numerical estimates provided by the Latin sources vary: Orderic Vitalis gave 10,000, whereas the Montecassino Chronicle opted for 15,000. Since medieval authors regularly ignored the contribution of infantrymen, Malaterra, although sometimes forthcoming in this regard, wrote only of 1,300 cavalrymen; it was also stated that the figure was based on eyewitness testimony. However, there may have been more than 1,300 given the attested presence of ‘squires’ or ‘arms-bearers’. While Chalandon felt that this figure was the only reliable one to be found in the Latin sources, Taviani-Carozzi has more recently weighed in favour of the numbers given by Orderic and the Montecassino Chronicle – that is, between 10,000 and 15,000 men in total. Willoughby and Birkenmeier, however, accept the figure of 30,000 given by Alexios’ daughter Anna Komnene, and while Haldon concedes that the total was ‘probably exaggerated’, he nonetheless believes ‘the emperor almost certainly had a smaller army’. Although an exact figure for Guiscard’s army is clearly unobtainable, 10–15,000 would seem to be the most likely figure; indeed, any estimate higher than that fails to consider a particularly salient point: the duke did not take every single individual capable of military service with him to Illyria. After all, unlike Alexios, he was undertaking a campaign outside his borders. Conversely, aside from some garrison forces, the emperor was able to commit much of the Byzantine military to the battle. Malaterra made it clear that Robert did not leave his numerous towns and fortresses ungarrisoned, and the bulk of his brother’s troops did not participate in the expedition: as the same monk put it, ‘Count Roger of Sicily was attending to his brother’s affairs in Calabria and Apulia as if they were his own’. But Malaterra also mentioned the ducal business to which the count was most devoted: the campaign in Sicily. As related in the previous chapter, it was still being actively fought, and it would not be finished until ten years after the launch of Guiscard’s invasion.5
In order to get his troops to Illyria, Guiscard of course needed a sizeable fleet to ferry them across the Adriatic. Alexios Komnenos’ daughter maintained that the duke did so with 150 ships, a figure that jars considerably with William of Apulia’s mention of fifty. Conceivably, however, to this figure can be added the squadron of fifteen commanded by Bohemond, who in March 1081 was sent ahead to both reconnoitre and raid the Illyrian coast. Sixty-five ships, then, might seem more feasible, yet the fact remains that a minimum of ninety-three ships capable of ferrying up to fourteen horses was required to transport 1,300 cavalry, not to mention the mounts of squires. At any rate, it was evidently a large fleet not only comprising transports, but war galleys as well. In regard to the former ship type, Komnene stated the ‘horses and armed cavalrymen embarked on the dromons’. This, coupled with her observation that the landing was expected to be opposed by mounted troops, attests to the use of Byzantine ramps (klimakes) enabling cavalrymen to ride fully equipped onto the beach. While William of Apulia believed these transports derived from Dalmatia – i.e. Croatia and Montenegro, on the northern border of the duchy of Dyrrachion – he seems to have overstated the involvement of Slavs from this region. While Komnene’s husband Nikephoros Vryennios noted that the ‘Croatians and Dukljans [‘Montenegrins’]’ rebelled against the empire in the 1070s, relative order had been established by the reign of Alexios, a fact attested to by the presence of their prince and subsequent king, Constantine Bodin (r. 1081–1108), whom Komnene noted led ‘the Dalmatians’ in the imperial army at the ensuing battle fought before the walls of Dyrrachion. Moreover, given the large number of naval operations launched by Guiscard and Count Roger to date, the dockyards of Apulia and Calabria were more than capable of maintaining existing ships inherited from the Byzantines, in addition to constructing new ones: as Malaterra wrote in regard to the campaign’s preparations, ‘Not just a single ship is produced, but a fleet is fitted’. The matter, however, is complicated by the fact that, as mentioned in one of the Apulian annals, Governor Argyritzes of Bari married his daughter Jacquinta to Constantine Bodin in April 1081 (or Paul Stephenson suggests perhaps as early as 1078). While this pact clearly establishes a link between Dalmatia and Norman Italy, it is interesting to note that if Dalmatians were present in the duke’s naval forces, at least some of them fought against the side featuring their king and his troops. That, however, is not to discount the participation of some Dalmatians for, as will now be related, the strategy employed indicates that Guiscard used locals who knew not only the coastal regions intimately, but also the inland river systems.6
Reconstructing the precise sequence of events is a difficult task, for William, Malaterra, and Komnene not only differ when it comes to the places reconnoitred and captured, but also in regard to the sequence. What seems to be the likely chain of events is the following: in the vicinity of the modern border between Greece and Albania, Vonitsa and Vouthroton (Butrint) were captured, as were two strongholds on the island of Kerkyra (Corfu, vital given its position at the mouth of the Adriatic). The fleet then sailed north to the protected harbour of Jericho (Orikum, Albania), presumably to reconnoitre Avlona (Vlorë) lying 15km to its north. However, Komnene noted that the duke’s ships were subsequently caught in a tempest ‘near the promontory called Glossa’ (Kepi i Gjuhëzës), the base of which forms the western flank of Jericho’s harbour. The princess recorded the destruction of a considerable number of ships, and William observed that one of the affected vessels was Guiscard’s. Following the testimony of Alexios’ daughter, Stanton asserts that the storm was encountered after Avlona’s capitulation, but it is important to note that William of Apulia, writing fifty years earlier, placed it beforehand; indeed, the duke sailed for Dyrrachion after capturing Avlona. Moreover, while Malaterra did not mention the storm, having noted the fleet docking at Jericho, he related that ‘afterwards it crossed over to a calm riverbank, where the River Vjosë flows into the sea’ – that is, 23km north-west of Avlona.7
The River Vjosë was an important route linking northern Greece to Illyria, and the latter destination to Italy. For example, in 1149 a large Byzantine fleet sailed up the river in order to reach Avlona, from where it intended to make the crossing to Ancona, Italy. While not explicitly mentioning the navigation of this river, Komnene nonetheless implied it: she observed that after the storm Guiscard recuperated at Glavinitza (also known as Glavinicia), which Taviani-Carozzi and Stephenson identify as Ballsh, situated 35km south-east of the Adriatic entrance to the River Vjosë (and 8km north of the river’s eastern bank). At first sight the choice might seem to be a strange one. Yet control of a river flowing all the way to northern Greece has often been deemed a necessity by armies campaigning in Illyria/Albania: for example, in an attempt to prevent the Romans from invading northern Greece, Philip V of Macedon engaged them on the west bank of the Vjosë near Tepelenë in 198 BC; and in 1943 Albanian battalions from Avlona/Vlorë secured a stunning victory over the Germans, capturing their strongholds at Drashovicë and Mavrovë (respectively 9 and 11km south-west of Avlona) lying on the western bank of the River Shushicë, a tributary of the Vjosë. From this tributary, the mouth of which is 14km from Glavinitza, Avlona could be accessed. That this route was taken is likely given Malaterra noted, in addition to targeting Avlona, that the forces focused on the hilltop fortress of Kanina (Kaninë). The logical place to disembark and encamp would have been Sherishtë, located centrally between Avlona and Kanina (respectively 5km west and 3km south-west), and situated just 3km from the River Shushicë. The garrisons of Avlona and Kanina surrendered after seeing the size of the Hauteville army, after which the duke set sail for Dyrrachion in June. Guiscard’s strategy up to this point can be likened to that employed previously in Calabria: capture or obtain the submission of smaller towns prior to advancing on the provincial capital. Not only could such towns be exploited for provisions and money, but men as well. Since Malaterra related that supplies were secured locally, and William and Komnene noted that Guiscard’s forces were bolstered by the local inhabitants – called ‘allies’ by the latter – the strategy (initially, at least) was clearly as successful as it had earlier been in Calabria.8
Alexios meanwhile dispatched his trusted general and brother-in-law, George Palaiologos – one of Komnene’s named sources – to prepare Dyrrachion for a siege. Moreover, in return for trading privileges, Alexios secured naval assistance from Venice (a former imperial dependency). The Venetian fleet docked at Pallia (Kep i Bishti i Palles) in July, the northernmost section of the promontory on which Dyrrachion was situated. From this location the fleet’s commanders learnt the Normans had blockaded the harbour, and that the land forces had set up a fortified camp before the city’s walls. Komnene related that the Venetians now lashed their largest vessels together with chains, forming what was ‘called a sea harbour’. Smaller skiffs were hoisted and lashed to the mastheads of the ships, and the marines placed within these interesting siege platforms were equipped with heavy planks fitted with large nails, which Komnene observed were around a cubit in length (i.e. up to 45cm). Bohemond’s squadron was dispatched towards the Venetian fleet, with orders to encourage the Venetians to acclaim ‘Michael’ as emperor. Like Gisulf II of Salerno before him, evidently Bohemond had adopted the Byzantine fashion of wearing a beard, which the Venetian sailors mocked when he addressed them. Ridiculing Normans could only bring about a single response; smarting from the insult to his appearance, Marc led an immediate attack. It was then that the purpose of the aforementioned planks became clear: the Venetian marines hurled them at the hulls of the Norman ships, puncturing them in the process. Bohemond’s ship was one of those affected; as his vessel started to sink, he hastily abandoned it for another. His fellow marines were clearly not so lucky, for Komnene noted that many drowned, and others were killed in either ranged or close-quarters combat. Pressing their success, the Venetians drove Bohemond’s squadron onto the beach, rapidly disembarking and attacking the fleeing Norman troops. Witnessing the spectacle from the city’s walls, George Palaiologos sallied out with the city’s garrison and, in conjunction with the Venetian marines, drove the Hauteville troops back to their fortified camp.9
In the next day or two, another naval engagement, or series of engagements, was fought. While some – most recently, Stanton and Theotokis – believe the second battle was in fact part of the first, this is unlikely to have been the case. Firstly, the engagement just described was hardly on a grand scale and hence cannot have taken an entire day (as Malaterra related in regard to the second one, observing that it resumed the following morning). Secondly, William of Apulia recorded a series of clashes occurring over three days. Thirdly, Komnene made it clear: what was different between the two discernible battles was that, in the second one, the Venetians were joined by a Byzantine squadron commanded by Mavrikas, almost certainly the same Dyrrachion-based admiral who defeated Geoffrey of Otranto’s Illyrian foray in 1066 (see Chapter 3). Since Komnene drew on the testimony of George Palaiologos, the princess’ uncle and Dyrrachion’s commander, her testimony is likely to have been correct, although it is important to note that her subsequent episode – i.e. the refloating of the ducal fleet in the River Acherontas – was prematurely dated (discussed below). Lastly, William, Komnene and Malaterra implied that all battle-ready ships were deployed by Guiscard, not just the detachment led by Bohemond in the previous engagement. The duke’s fleet appears to have been caught in a vice during the second engagement: that is, the Venetian ‘sea harbour’ attacked from the north, while Mavrikas’ fleet assailed from the rear. Indeed, Guiscard’s ships might have been completely captured had it not been for the marines armed with bows, whose rapid salvos prevented the enemy from boarding. While unattested elsewhere, it is possible that one of the large Norman galleys was burnt by the flammable liquid propelled through pressurized ‘siphons’ or ‘tubes’ – ‘Greek fire’, actually called ‘Roman/liquid/sea fire’ by its Byzantine inventors – although one of the best sources for this secret weapon, Komnene, does not mention it. Since Malaterra believed that only the Venetians fought in this battle, he erroneously thought the ships fitted with the technology to be Venetian. This belief, while understandable, is nonetheless unlikely: since its invention in the seventh century, the Byzantines had taken great pains to prevent both its allies and enemies from acquiring it; a tenth-century manual on statecraft made this very clear.10
Equally lacking in success was the duke’s attempt to capture the city, which began before (and continued after) the naval clashes. Much like the siege of Palermo, when Guiscard focused on one section of the fortifications while Count Roger targeted another, Dyrrachion’s walls were assailed from two sides. Theotokis believes that the sections besieged were to the north and east, positing that these areas were the accessible ones. As an anonymous traveller observed in the sixteenth century, the ‘city is situated on the western slope of the hill facing eastwards and stretches down to the seaside’. He also noted that ‘at a shot’s distance’ from the walls ‘there is a well from which the city draws its water’. This language recalls Komnene’s observation that Guiscard established his fortified camp at a bowshot from the fortifications, and the fact that there was a suitable site located at a similar distance from them makes it the obvious choice. Access to drinking water was obviously vital, and importantly its procurement was difficult in a region the anonymous traveller noted as being ‘abundant in brackish water, marsh and saltwater’. Now the reliance on a sixteenth-century source might at first seem questionable, yet it appears that the city and its environs had changed little over the centuries (other than some polygonal towers added in the thirteenth century): firstly, the anonymous source referred to the city’s ‘ancient and weak walls … which have not been kept up’, meaning the Ottomans had neglected to modify the Byzantine fortifications to cope with cannon fire; secondly, William noted the presence of a ‘river’ that Komnene observed was ‘salty’ or ‘brackish’. Given its saline composition, the lagoon’s source was clearly the sea – that is, from the bay to the city’s south where the Hauteville fleet was moored. The salt marsh evidently ran in an easterly direction, forming a natural perimeter around the fortified camp, hence offering protection from both the south and from the plain to the east where the battle was later fought (although access to it was afforded by a bridge). Accordingly, while Guiscard’s men may have focused on the city’s eastern walls as Theotokis suggests, they were encamped to the south, close to the beach. As for the location of the other sieging party led by Bohemond, it was likely to have been on the north side of city, presumably camped near to where the anonymous itinerant noted there was another water source ‘further on’ from the aforementioned well.11
While unsuccessful, the siege was by no means lacking in energy. A Byzantine-style ‘city-taker’, protected by a large stone-throwing catapult encased in leather hides to protect it from incendiary fire, was deployed against the fortifications. Said by Komnene to have exceeded the height of the city’s walls, it was clearly a leviathan; she later related that the city’s many towers extended ‘eleven feet’ above the walls (which must have been at least double that height). The walls were not only wide enough for four horsemen to ride along them side-by-side, but also to house heavy artillery engines – i.e. ‘stone-throwers’ (see Appendix). Naturally, like Bohemond prior to engaging the Venetians at sea, Guiscard preceded hostilities with an attempt to convince the defenders to hail the legitimate emperor ‘Michael’: he received an abusive response, and the impostor was ridiculed when paraded in regal garb before the jeering citizens. In order to counter the threat of Guiscard’s imposing siege tower, Palaiologos had one prepared in front of it, presumably mounted on the targeted wall since it was wide enough for such a structure. George was clearly a resourceful, knowledgeable student of war, for he equipped the top of the tower with a large beam: it was designed to be thrust against the door of the enemy’s tower, which served as a bridge when lowered. The following day, a force of 500 light and heavy infantry entered the duke’s tower once it had reached the city’s walls. Having ascended to the top of it, the troops attempted to burst out, but the beam prevented the drawbridge from being lowered. The upper section of the siege tower was then set alight with incendiaries, and its lower section was demolished by a team of men armed with axes, who were themselves screened by a contingent of heavy infantry. Angered but characteristically resolute nonetheless, the duke ordered the construction of another tower. Progress, however, was soon halted: Alexios and the imperial army arrived in the evening of 17 October, setting up camp on the western bank of the River Charzanes (Erzen, c. 7km east of the beach).12
That night Alexios held a council of war with his generals, all of whom were linked to him by familial ties (very important given that he was effectively the third usurper in ten years, and had himself made enemies when defeating two attempted coups prior to becoming emperor). He asked his men whether Guiscard should be brought to battle or not. Echoing the exhortations of various imperial military manuals, the elders, including George Palaiologos, advised against a hasty pitched battle, suggesting that the enemy could be whittled down by stratagems and skirmishing. Predictably the younger element, including Constantine Doukas (betrothed to Guiscard’s daughter in 1074), was eager for battle, and it appears that while the youthful Alexios acknowledged the sagacity of the elders’ assessment, his younger commanders reaffirmed a decision he may well have already made: with his greater numbers, and backed by highly experienced battalions of kataphraktoi and Franks/Normans, he would literally drive the duke’s army into the sea. Nonetheless, Alexios appears to have accepted that an ambush should be tried first, ordering a nocturnal attack on Guiscard’s camp. However, evidently anticipating Alexios’ plan to launch this sally, the duke ordered the camp to be burnt, moving his forces across the bridge on the lagoon, and onto the plain early the following morning. Having made the crossing, the bridge was promptly destroyed, thus preventing an attack from the left flank and rear.
Alexios, meanwhile, sent a detachment of men from his Varangian bodyguard, whose customary Scandinavian make-up had been significantly lessened, to use the words of Orderic Vitalis, by Anglo-Saxons ‘who had abandoned the kingdom of Albion [Engl.] after King Harold and the nobles were killed’ at Hastings. They were sent around the salt marsh with orders to attack the camp, and the raid was to be coordinated with a sally from the city’s garrison. Having found Guiscard’s camp empty and the bridge destroyed, both the garrison of George Palaiologos and the detachment of Varangians were unable to take part in the battle that was to follow. The invaders were in a particularly precarious position for two reasons: firstly, their forces would now be fighting with their backs to the lagoon (Bohemond) and bay of Dyrrachion (Guiscard and Count Ami). This posed a particular problem given the harbour was being patrolled by imperial and Venetian ships. Secondly, William of Apulia noted that the area in which the duke drew up his battle lines was narrow and restrictive.13
The Norman battle line was as follows: Guiscard commanded the centre; Count Ami II of Giovinazzo the right wing; Bohemond the left. Facing the duke was Alexios, screened by javelineers and ‘many soldiers adept at archery’, who were themselves fronted by heavy infantry and dismounted Varangians commanded by Nabites. To Count Ami’s immediate north was the general, Gregory Pakourianos, and the caesar, Nikephoros Melissenos, positioned his troops opposite Bohemond’s. Komnene also mentions that Pakourianos had brought with him ‘all the heavy infantry’, whom she had earlier defined as ‘spear-bearers’. Although their placement was not indicated, it can be safely assumed that they formed the front ranks of the three battle columns – that is, behind the Varangians (discussed presently). It is evident from Komnene’s account that the imperial army possessed a sizable number of élite regiments, including the same battle-hardened kataphraktoi who had previously fought on both sides during the revolts against Alexios’ predecessor, Nikephoros Votaneiates. Fighting alongside these imperial cavalrymen were ‘the battalions of the Franks’ – Normans and other French-speakers led by ‘Constantine, son of Humbert’. Also present were ‘The Immortals’, who since their reformation in the 1070s, Shepard notes, were trained to charge at full gallop with their lances couched. Some distance behind the Byzantine battle line was the abovementioned Constantine Bodin, who commanded an unspecified number of Dalmatian troops, apparently held in reserve. Lastly, greatly aiding Alexios’ ranged capabilities, was a contingent of Turkic horse archers led by Tatikios, a member of the imperial household whose father was of Arabic extraction. While the location of these horsemen was not given, they were probably interspersed throughout the imperial battle line, as they had been when Alexios fought the would-be emperor Nikephoros Vryennios at Kalavryai in 1078.14
Based on the estimates of Birkenmeier and Haldon, the army probably consisted of 18–25,000 men, and hence outnumbered Guiscard’s estimated 10–15,000 (although his initial forces had since been increased by Illyrian troops). Other than Komnene’s mention of Guiscard’s ‘cavalry and infantry’, none of the sources provided specific information regarding troop types. However, Calabrians were mentioned, and among them were the ‘excellent infantry’ from Cosenza, who were subsequently used by the duke as assault troops for a special mission (javelineers, based on the description). Lombards were also identified, including troops from Salerno who had chosen to accompany Duchess Sichelgaita. As for their troop types, nothing can be determined, but they were probably a mixture of cavalry as well as light and heavy infantry. Lastly, William also mentioned sailors participating in the battle; they were probably archers of Lombard, Byzantine, and Dalmatian extraction.15
Before describing the stages of the battle, it is necessary to consider the manner in which the Varangian guardsmen were arrayed, precisely because their position is crucial to gaining an understanding of both the various stages and outcome of battle. Birkenmeier believes the Varangians and foot archers were placed ‘in front of the imperial battle line’, a tactical decision deemed ‘peculiar’. Komnene’s language is ambiguous in this respect; the ‘axe-bearers’ were placed ‘a small interval from the battle line’. Since their duty was to guard the emperor in battle, it has understandably been assumed by Birkenmeier, Haldon and Theotokis that the Varangians were formed in front of Alexios. However, Komnene does not actually state that, merely suggesting that they formed the first line of the imperial array. Consultation of the tenth-century military manual ascribed to Nikephoras Phokas sheds considerable light: the general advised that 1,200 élite soldiers – Haldon estimates the Varangians to have totalled 1,400 – should be placed ‘before the formation of heavy infantry’ which fronted the entire line of battle. Noted by his daughter to have been an avid reader of military treatises, Alexios seems to have been influenced by the one in question: not only in this respect, but also in regard to his use of missile troops in the manner recommended by the same author (‘engage with opposing cavalry and handle them’ – see below). Lending credence to the idea that the ‘axe-bearers’ were likely to have been spread across the entire frontage is Malaterra’s observation: ‘the Varangians asked the emperor for the first line of battle’. Moreover, it is important to emphasize that Count Ami ‘struck near to the extremity of the battle line of Nabites’ (as Komnene put it). Despite the contrary views of Haldon and Theotokis, it is unlikely that Ami who, since he deployed both ‘infantry and cavalry’ clearly committed the entire wing, would have charged the left flank of the imperial centre from the Norman right: such an assault would have exposed the right flank of Guiscard’s centre, in addition to allowing Pakourianos’ left to envelop Ami’s exposed right flank. Accordingly, while there were undoubtedly Varangians arrayed before the emperor – perhaps double the number posted on the wings – the logical wing for the experienced Norman count to attack was the one facing him led by Gregory Pakourianos.16
Guiscard began the battle by deploying units of skirmishing cavalry (conterati?), with the intention of enticing the imperial front line to break formation and charge. As was earlier the case at Aiello (1065), the duke’s cavalrymen were again resisted by men hurling javelins, who sallied out from the intervals of the infantry columns, as did their comrades armed with bows. According to Komnene, while both armies drew inexorably closer together, the skirmishing continued. William, however, had Alexios sending out ‘many divisions’ simultaneously, observing that the duke’s cavalry were ‘terrified by the first collision’. He also noted that the burdensome charge afflicted the troops positioned in ‘a confined place’ – that is, Bohemond’s left wing with the salt marsh to its left and rear (where it met the sea). Precisely who performed the charge cannot be determined, but the two most likely candidates were the Franks and Immortals. Such regiments were evidently picked cavalry intended to complement each other, for both Vryennios and Komnene recorded them being grouped together by Alexios in the late 1070s. Indeed, since William related the severe discomfort caused by the ‘hailstorm’ of arrows up to this point, it follows that openings had been created into which the cavalry could charge. The charge, or series of cavalry assaults, wrought utter havoc. The soldiers became entangled and, as a result, Bohemond was unable to deploy his cavalry, some of whom seem to have abandoned their mounts in the general panic. It was probably at this stage that ‘Ami’s formation of infantry and cavalry’ (the entire right wing) engaged the abovementioned ‘extremity’ of the Varangians. The onslaught was so forcefully repulsed that his men retreated towards the sea.17
Having routed the left and right wings, some of the emperor’s units believed the battle was won; they now began looting the abandoned horses and baggage. Tellingly Komnene did not mention this fatal lack of discipline, no doubt because it would have reflected badly on her uncle Melissenos, and ultimately on her father. It was now, according to William, that those driven into the sea and the lagoon’s mouth were eventually able to regroup around Guiscard, whose centre, he added, was unaffected by the aforementioned ‘first collision’. This island of relative cohesion in a sea of flight effectively won the battle. Once again the position of the Varangians becomes crucial to understanding what happened next. While others including Theotokis feel that those targeted were in front of Alexios, this is unlikely to have been the case given their ‘exposed flank’ (Malaterra) was attacked with ‘some infantry’ (Komnene). If the Varangians were advancing on his centre, Guiscard cannot have dispatched a force of infantry arrayed in front of him to attack their unprotected side. Rather, given the angle required for a flank assault, the duke’s infantry must have attacked the exposed right side of the Varangians who had helped to drive back Ami’s assault on the left wing, and were therefore the ones who had advanced the furthest forward (and were stated by Komnene to have been exhausted by their exertions up to this point). Moreover, as his daughter made clear, Alexios’ centre was not only still advancing on Guiscard at the time, but was also maintaining its formation. The flank assault brought about the decimation of the Varangians – aside from a few who managed to flee to a nearby church, which those who routed them promptly set fire to, burning their remaining foes alive.18
It was from the duke’s centre that not just the Varangians, but by implication given the plundering reported and casualties sustained, apparently a great many of the imperial troops, were routed by a series of devastating cavalry charges. For Guiscard, a sexagenarian who had fought in just about every type of battle imaginable, the disorganized wings were easy pickings. Alexios’ centre, however, remained a problem.
The Roman battle line continued to fight it out strongly. Yet like a winged-horseman, Robert thrust forward with his remaining forces, drove against the Roman [infantry] phalanx, forced it back, and broke it up into various sections.19
Alexios’ cavalry force behind the infantry column, ‘like an unshaken tower’, continued to hold its ground. No doubt seeking to redeem himself, Ami broke off from the main body, leading what can only have been a couched-lance assault: ‘taking long lances in their arms’, he and a few others charged directly at the emperor. Alexios managed to dodge the thrust of Ami’s lance, thereafter severing the arm of Peter of Alifa, who interestingly would soon enter imperial service and found a dynasty which faithfully served Alexios’ son and grandson (John II and Manuel I). Like William’s description of Guiscard at Civitate, his fellow classicist Komnene at this point in the narrative switched to an aristeia: a Homeric rendering of her father’s military excellence, which to a modern audience reads very much like a ‘heroic retreat’. As an aristeia focused on individual valour, this section of her battle account switches to the singular, but clearly Alexios’ trusted cavalry also engaged their mounted counterparts. Some of them had already fallen, including two men of particularly high birth, but others obviously still rallied behind their emperor. The imperial cavalry continued to resist, but when Alexios realized that both Constantine Bodin’s reserve cavalry and the Turkic horse archers had abandoned the field, he ordered a retreat.20
Robert Guiscard had triumphed yet again, inflicting a crushing defeat on an army that should have carried the day. If two Latin sources are to be trusted, 5–6,000 imperial troops were killed. Probably most of the dead were infantry, but William did relate that many of the Turkic horse archers were killed. As for the Hauteville casualties, only William provided a figure: thirty cavalry, although a substantial number of infantry must have been killed given the carnage reported by himself and Komnene, not to mention Orderic Vitalis’ observation that both sides sustained ‘enormous loss’. Casualties among the kataphraktoi seem to have been relatively slight, for mounted contingents were reported to have been present at subsequent battles, although Komnene did relate that various nobles were killed during the battle’s closing stages. While the indiscipline of some of the imperial divisions undoubtedly turned a defeat into a victory, Guiscard’s abilities as a commander should nonetheless be underscored. Not only did he triumph over the largest and most professional army he had ever faced, but he had also defeated one featuring units of skilled heavy cavalry, meaning that the Hauteville army did not possess the mounted advantage it often held in the past. Indeed, while Guiscard was confident in the prowess of his ‘élite cavalrymen’ prior to the battle, he advised them not to be ‘cocksure’. The reason for this was made plain by William of Apulia: the duke was concerned at the prospect of engaging Alexios’ ‘hostile wedges’. Lastly, Guiscard’s ability to rally routing forces and launch a series of devastating counterattacks attests to the steely élan he inspired among his troops.21
George Palaiologos, who had launched a sortie during the attack on the Norman camp, had been unable to return to the city once Guiscard’s forces had achieved victory. Alexios accordingly entrusted the city’s affairs and its citadel to the leaders of the Venetian colony, while command of the lower city was entrusted to an unnamed ‘Albanian’. The city’s garrison resisted the ongoing siege from autumn into winter. Yet in February 1082, Guiscard, having been granted access by traitors from within, captured the city with his most valued troops: ‘The duke chose some [Byzantine] men from Cosenza, whom he recognized took the lead in speed, adding to them some picked cavalry’, as William put it. According to Komnene, having captured Dyrrachion the duke now assessed his troops, grouping them into contingents based on ethnicity, in addition to organizing the hiring of additional local ones. Alexios meanwhile regrouped with his remaining forces at the empire’s second-largest city, Thessaloniki, where he began to organize replacements for those lost, as well as requesting assistance from allies. The emperor was able to attract numerous recruits, whom he chose to drill personally. A delegation was also sent to Henry IV, requesting that he, in accordance with the agreement reached in the previous year, march into northern Italy and undermine Guiscard’s ally, Gregory VII.22
The logical road for the duke to traverse was the via Egnatia, which extended in an easterly direction from Dyrrachion to Constantinople. But he no doubt expected considerable resistance along that most obvious of routes, choosing instead to march south into northern Greece. The target chosen was the city of Kastoria, from where the duke could march in a north-easterly direction to Thessaloniki, and thence to Constantinople. Malaterra noted that the city was both well fortified, and contained a garrison of 300 Varangian guardsmen. The investment appears to have been largely unsuccessful, but the determined resistance of the garrison, much like their predecessors at Reggio in 1060, appears to have faltered when the duke’s men began preparing siege engines. Following the capitulation of Kastoria, other fortresses within the region surrendered, and the citizens subsequently provided Guiscard with valuable intelligence. However, word now reached the duke of Henry IV’s advance into northern Italy, although of much greater concern was the news that, encouraged by Byzantine agents, various cities and towns in Apulia had risen in rebellion. Entrusting Bohemond and the constable Brienne with command of his army, Guiscard returned to Apulia with two ships, probably in April 1082.23
As Theotokis aptly puts it, Bohemond’s martial ability would now be firmly put to the test by ‘an opponent of about the same age but with much greater experience in warfare’. Soon after the capture of Kastoria, Alexios arrived in the theme of Ellas, encamping near Ioannina (101km south-west) in May. The emperor chose this location, for its garrison had recently surrendered to Bohemond who, taking a leaf out of his uncle Roger’s book, had reinforced its fortifications, improving the ‘citadel’ and adding a second one. Evidently with a force much smaller than the one led by Bohemond, the emperor started the campaign with skirmishing tactics, effectively utilizing the advice of his elders before the battle at Dyrrachion. After various raids and reconnaissance missions had been conducted, Alexios now decided to risk a pitched engagement. Doing so against a larger force was clearly ill-advised, although the emperor seems to have felt confident enough in his troops to risk a pitched clash. Confident to an extent, that is, for he seems to have possessed fewer cavalry with which he could protect the flanks of the infantry formation. Hence, in order to negate the certainty of a frontal charge by Bohemond’s horsemen, the emperor placed spear-armed wagons before his (mostly) light infantry. Heavily armoured men were placed nearby, with orders to roll the wagons forward when the enemy cavalry approached. While Alexios implemented these stratagems under the cover of darkness, judging by what happened when the sun rose, the scouts of Bohemond and Brienne were nonetheless able to discover the nature and location of the traps. Two versions of what happened next exist: Komnene stated that Bohemond’s cavalry simply avoided the traps and attacked the imperial flanks, while William recorded that since it was a foggy morning, Bohemond’s men were able to traverse unseen through the vineyards and bushland flanking Alexios’ position, thereafter launching a surprise attack. Whatever the case, both sources agree that Alexios and his men offered fierce resistance – Komnene recorded substantial casualties on both sides – but they were ultimately unable to withstand the attack on their flanks. Interestingly, while the assault cavalrymen at Dyrrachion seem to have been lancers, the divisions led by Bohemond and Brienne were armed with javelins: Komnene wrote of ‘missiles thrown’ at her father ‘from both sides’.24
The emperor reunited with his scattered forces at Ohrid, a city in the modern Republic of Macedonia, situated 162km north of Ioannina. Bolstering his forces with mercenaries, the emperor decided to take another crack at Bohemond’s army. In another attempt to disadvantage the enemy cavalry, at night Alexios had ‘iron caltrops constructed’, which were then scattered in front of his infantry line. Placed behind the infantry column were missile troops, and cavalry were posted on the wings. As Komnene explained, her father’s intention was to coax Bohemond’s cavalry into the area peppered with caltrops, weaken their ranks by means of projectiles, then envelop them with a cavalry assault. It was a good plan, but Bohemond and Brienne’s scouts did their job. Cognisant of the trap, the Hauteville cavalry again overran the imperial flanks. Although Alexios resisted in the centre, after receiving a wound and witnessing the flight of his wings the emperor abandoned the field. In the wake of the second victory, Bohemond dispatched contingents throughout the region to secure towns and cities; one of them, led by Raoul of Pontoise, managed to capture the modern-day capital of the Republic of Macedonia, Skopje (110km north-east of Ohrid). Bohemond’s strategy was clear: gain control of all important settlements to the north and west of Thessaloniki, for hemming in the second-largest city in Europe was crucial to protecting his rear when advancing eastwards on the imperial capital. He was, however, less successful than Raoul: despite endeavouring to capture Ohrid, Bohemond’s attempt was repulsed by a determined garrison, as was his subsequent attempt to seize Ostrovo – that is, Arnissa (95km north-west of Thessaloniki), where the formidable George Maniakes had met his death in battle forty years earlier (see Chapter 2). The dejected Norman now chose to abandon the via Egnatia, marching into Thessaly, but again he unsuccessfully besieged another well-fortified town (Veria, 44km south-east of Arnissa). Subsequently an old fort at Moglena (Almopia, 59km north of Veria) was rebuilt and garrisoned: it was, however, soon captured by a force led by Gregory Pakourianos, who executed its commander and demolished the fortress.25
Despite such reverses and lack of success, in the winter of 1083 Bohemond was able to capture Pelagonia (Bitola, 50km north-west of Arnissa), a town situated close to the modern border between Greece and the republic of Macedonia. Leaving Macedonia, he returned to Thessaly where his army successfully besieged Trikala (168km south-east of Bitola). While the bulk of Bohemond’s forces invested Trikala, a detachment captured Tziviskos, which Alexios Savvides plausibly identifies as Grizano (26km north-east). The next target was the Thessalian city of Larissa (57km east of Trikala), and the full-scale siege of its walls began on 23 April 1083. Despite the presence of Bohemond’s entire army, in addition to the use of siege engines, the investment dragged on for six months. Alexios, meanwhile, had been organizing reinforcements. The emperor was lucky to receive 7,000 horse archers from his great enemy Süleyman I (r. 1077–86), the Selçuk sultan of Rum (eastern Anatolia). In addition to levying more troops, the emperor dispatched agents, whose brief was to entice Bohemond’s men to desert. While not as successful as the subsequent attempt, Alexios’ operatives managed nevertheless to entice the aforementioned Raoul and two other men, Reynald and William, to betray their commander. Their plot, however, was soon discovered and, while Raoul managed to find safe haven with the emperor, Reynald and William were blinded. Having gathered his forces, Alexios advanced on Larissa, in the vicinity of which he held a council of war with his senior officers. It was decided that the emperor would give command of the army to Caesar Nikephoros Melissenos and Basil Kourtikios, the latter a solider of considerable renown. Alexios advised them to begin hostilities by skirmishing, then feign retreat in order to entice a cavalry charge. The emperor then reconnoitred the environs of Larissa before the ensuing battle, searching to its west for a suitable spot to set an ambush with his picked cavalry. Evidently learning from his previous mistakes in regard to keeping his stratagems secret, a detachment of troops was sent to harry Bohemond’s forces encamped on the plain before the city’s walls. The ruse was successful, for Alexios was able to find a low-lying position in which to conceal his men.26
The following morning, Bohemond awoke to see the imperial battle line arrayed to the east of his position before the city’s walls. The cavalry behind the imperial centre were garbed in finery normally reserved for the emperor and his retinue, with the idea of tricking the enemy into thinking that Alexios was present. The stratagem worked as, dividing their cavalry into two squadrons, Bohemond and Brienne charged the centre, hoping to capture the emperor. Following some skirmishing, the imperial battle line retreated as planned. Alexios now took the opportunity to launch an assault on Bohemond’s camp, killing most of the infantry and seizing the baggage and considerable spoils accumulated to date. The emperor then ordered a contingent of infantry archers to target the Hauteville cavalrymen’s mounts, which were in the process of chasing the retreating army. Having come within firing range of the enemy cavalry, the archers successfully maimed and killed various horses, the panic of which created a vast cloud of dust. Unable to determine the position of the archers, Brienne’s ranks were decimated. Shortly afterwards, some Byzantine infantry unwisely chose to make an assault on Bohemond’s cavalry, resulting in the death of 500 of them. Another detachment, including some Selçuk horsemen, was afterwards sent to deal with Bohemond, but they were also driven off. Hostilities finished when darkness fell, only to be resumed at dawn the following morning. Brienne’s thinned ranks had earlier been able to unite with Bohemond’s, and they now advanced along the Pineios River, finally encamping in a defile on the edge of the plain of Larissa. The emperor sent out some Selçuk and Pecheneg cavalry to entice a cavalry charge, but to no avail. Sensing an all-out assault on his position, Bohemond, as he would later do on crusade at Dorylaion (1097), wisely ordered his cavalry to dismount and form a shield wall. Despite driving back a concerted assault, Bohemond’s standard-bearer was skewered by a lance thrust in the process, and the banner was captured. Seeing their lord’s standard in enemy hands annihilated the defenders’ morale; the determined resistance now turned into a rout.27
Having suffered a great defeat in addition to losing his supplies, food and money, Bohemond’s campaign was now in tatters. His increasingly hungry men were now forced to retreat to Avlona and Kastoria in order to procure food. Not one to bask in the glory of his artful victory, Alexios again sent agents to infiltrate and sow dissension among the Norman ranks. Knowing that his adversary’s cash reserves were at an all-time low, he no doubt felt confident that many would find the prospect of imperial service attractive. If Komnene is to be trusted, the intrigues of imperial agents appear to have provoked a mutiny of sorts, for she related that Bohemond’s men vehemently began to demand immediate payment. Their lord managed to prevaricate successfully, no doubt offering the promise of future rewards, as had his father and uncle before him in similarly straitened circumstances. Nevertheless, all was not well among the men commanded by Brienne and Bohemond, the former now based at Kastoria, the latter at Avlona.28
Reconstructing the timeline of this period is difficult given both William and Komnene consistently failed to mention seasons, let alone dates. Having related the retreat to Avlona and Kastoria, William observed that the Venetian navy managed to recapture Dyrrachion in the autumn of 1083, although the keep’s garrison managed to hold out. Hearing that Bohemond was on his way from Avlona to the city’s south, the Venetians wintered in their ships. Interestingly, William observed that they added roofs to their fleet, in addition to barring a potential land-based offensive by means of a wooden fortress protected by siege engines. Once the winter ended, the Venetians rendezvoused with Mavrikas’ fleet at Kerkyra, whose citizens had rebelled when news reached them of the naval defeat inflicted on Guiscard’s flotilla in 1081. With the coastal regions of northern Greece and Illyria now in a much healthier position, it was determined that the homesick Venetians should return home.29
Probably at the same time, that is in the late autumn or early winter of 1083–84, Alexios arrived before the walls of Kastoria. Luckily Komnene decided at this point to describe the city’s site and fortifications. Based on Lake Orestiada, the city’s fortified area was located on a precipitous promontory; it was therefore only accessible by land from one side. Alexios blockaded the landward walls with palisades lashed together with chains, and linked by towers. Lofty ‘city-takers’ (mobile siege towers) were deployed against the walls, while large ‘rock-throwers’ bombarded, and afterwards destroyed, the battlements. This land-based assault was coordinated with an approach by water: marines under the command of George Palaiologos were placed on skiffs, which approached the tip of the spit jutting into the lake. The goal was to use the vessels as platforms from which the soldiers could scale the cliff face. After the signal was issued, Alexios and George led a two-pronged assault on the walls. Knowing that all was lost, Brienne’s garrison mutinied, demanding the city be surrendered to the emperor. Those who chose not to enter imperial service, including Brienne himself, were granted safe passage to Avlona. With all possessions lost and his troops either demoralized or now serving the emperor, the humbled Bohemond had little choice but to return to Italy.30
Henry IV, while presumably pleased by the largesse received from Alexios, needed little inducement to invade northern Italy. The Salian dynasty (1024–1125), of which Henry was the third ruler, had inherited an intricately interwoven ‘Church and State’ system from its Ottonian predecessors (919–1024). The Salians were also bequeathed an attitude that can be traced back to Charlemagne who, in R.H.C Davis’ classic description, felt the Pope to be little more than the head of ‘a sort of ministry of prayer’. In a 796 letter to Pope Leo III, the same king had famously referred to his expansive realm not as a kingdom, but as a ‘Church’ which he, a figurative David, ruled as God’s anointed. Modern historians call this ideology ‘caesaropapism’, which Charlemagne, as with various other things, modelled on Byzantine practice. Unsurprisingly, therefore, ecclesiastical appointments, including the highest one, had regularly been the prerogative of the western emperors. Gregory VII, whose pontificate was the culmination of earlier attempts to break the power of western emperors over ecclesiastical affairs, asserted that only popes could appoint bishops; he also maintained that supreme pontiffs possessed the power to depose emperors. As the Ottonian dynasty had managed to break the power of semi-independent provinces by replacing hereditary dukes with bishops who were normally drawn from their own family, Gregory’s directive, if followed, meant that Henry IV would effectively relinquish authority of up to half of his kingdom (more territory than this was subject to bishops and abbots in the Ottonian period, but Henry’s predecessors had annexed some of it to the crown). War was therefore inevitable, especially given the Pope’s support for Rudolf of Rheinfelden, who was elected king in 1077 by pro-Gregorian clerics within the empire following Henry’s deposition and excommunication in the previous year. While Gregory’s notion that a pope could depose an emperor was highly controversial, he was able to move in this direction given that prior to being crowned as emperors at Rome by the reigning pope, rulers of the western empire were only kings. Henry’s plan was therefore simple: march on Rome, remove Gregory from office, and replace the intransigent patriarch with someone who would crown him emperor. He was successful in this aim, and while Alexios probably predicted that his western colleague would return to Germany after achieving his goal, he nonetheless knew that Guiscard was bound by oath to protect his theoretical lord, Gregory VII. In addition, since western emperors had personally intervened in south Italian affairs (1022, 1038, and 1047), the duke must surely have feared that Henry would at least provide military assistance to the rebels in Apulia. Moreover, as he discovered in early 1082 that Richard of Capua’s son and successor, Jordan, had not only submitted to Henry, but was also helping him to besiege Rome, Guiscard must have been very worried indeed.31
Despite Henry’s presence in northern Italy, Guiscard had more important matters to attend to: the suppression of various revolts in Apulia. The serious nature of these uprisings is clearly attested by the fact that they took the duke over a year to suppress. Fanning the fires of rebellion on behalf of Alexios was Guiscard’s nephew Abelard who, it will be recalled, had already been involved in several rebellions (he probably expected to succeed his father Humphrey as count in 1057, and hence believed the duke had usurped his inheritance). Guiscard’s son Roger managed to suppress the uprising at Troia in the Capitinata, while his father dealt with Geoffrey of Conversano at Oria in April 1082. The duke then marched on Bari, whose citizens were punished with a particularly onerous tax. Then, in May the following year, Guiscard besieged the stronghold at Canne ruled by Abelard’s half-brother Herman, and its garrison managed to hold out until 10 July. But Robert’s job was not yet done; Prince Jordan of Capua remained a problem. Having raided the principality’s environs, in addition to scorching its crops, the duke was able to bring about his nephew’s submission towards the end of 1083. In an effort to bind its prominent citizens more closely to their Pope, Guiscard sent ‘more than 30,000 [Byzantine] gold coins to the Romans’. Yet in March 1084 Henry IV breached the city’s defences, although Gregory VII managed to resist capture by seeking refuge in the Castel Sant’Angelo (Fig. 14). Undeterred, Henry lay siege to this virtually impregnable fortress. When the king formally deposed the Pope for a second time, in addition to nominating Guibert of Ravenna as the new one (Clement III), the duke chose to march on Rome, arriving there in May.32
While William of Apulia usually provided believable figures in regard to Norman forces, the license associated with the poetic craft came to the fore on this occasion: the duke, he opined, led an army of 6,000 cavalry and 30,000 infantry (the latter figure is also given by Guido of Ferrara). Clearly the total figure is much too high, but Guiscard’s army was nonetheless formidable. Geoffrey Malaterra also emphasized the large numbers of cavalry (4,000) and infantry (unspecified), and one of the Apulian annals attested to ‘a multitude of Normans, Lombards, and other peoples’. As had been the case at Salerno eight years earlier, the ‘other peoples’ must have been Byzantines and Muslims (specifically mentioned by Landulf of Milan). Evidently Henry IV’s scouts gave an accurate report of the duke’s numerical strength, for when informed the king opted for a tactical retreat rather than, as Orderic Vitalis observed, committing to the uncertain outcome of a ‘hasty battle’. The ensuing siege and subsequent sack is difficult to reconstruct given that none of the sources agree on which section of the walls were scaled. While some suggested the city’s northern gates (Porta Flaminia or Pinciana), others either indicated or implied its south-eastern (S. Giovanni) or southern counterparts (Latina). The duke encamped before the Porta S. Giovanni to the city’s south-east, for Malaterra noted that the unnamed gate was located on the via Tuscolana. However, with a detachment of 1,300 cavalry, Guiscard chose to reconnoitre the rest of the city’s walls, hoping to find another gate that was less well manned. The monk then had Guiscard’s men climbing the walls near the Porta S. Lorenzo, adding the confusing clarification that the gate in question was ‘beneath the aqueduct, close to the Tiber’. At first sight Malaterra’s knowledge of Rome seems a little confused: the eastern gate connected to an aqueduct was in fact the Porta Maggiore to S. Lorenzo’s south and, moreover, neither entrance to the city was anywhere near the Tiber, the river skirting the entire length of Rome’s western fortifications. However, there is another way of interpreting Malaterra’s description: he may well have meant the city’s western gate – Porta Aurelia (= S. Pancrazio) – confusing its name with the nearby S. Lorenzo in Damaso (1km north-east), a venerable church constructed in the fourth century and hence one of the city’s oldest. Various other facts support this theory: not only was there was an aqueduct (aqua Traiana) in close proximity, but the Aurelian bridge (= Ponte Sisto) could be used to cross the west bank of the nearby Tiber. Furthermore, it is important to emphasize that the Porta Aurelia afforded much quicker access to the duke’s target, the Castel Sant’Angelo, situated just 2km north-east of the gate. Lastly, Malaterra related that Guiscard’s picked men, having scaled the walls and opened the gate for the rest of their detachment, rendezvoused with the rest of the army at a ‘bridge’ (the latter evidently having overcome the garrison defending the Porta S. Giovanni by this stage, or perhaps the southern Latina gate as Louis Hamilton theorizes). Unfortunately he did not name the bridge (Sisto or Fabricio?), but its mention could indicate that the duke’s cavalry were separated from the greater host by a body of water, which can only have been the Tiber.33
What all sources generally agree on is that once Guiscard’s men had scaled the walls and let the others in – or rendezvoused with those who had burst through another gate – the combined forces advanced on the Castel Sant’Angelo, successfully freeing the Pope. The Castel, originally Emperor Hadrian’s mausoleum but known in this period as the Tower of Crescentius, was close to the Vatican palace, although the papal residence prior to the fourteenth century was the Lateran (near the Porta S. Giovanni). Guiscard accordingly restored this palatial residence to Gregory VII, but the latter would not enjoy it for long. Three days after the duke’s forces had entered the city, groups of citizens formed militias who launched a series of counterattacks. Not only were those inside the walls afflicted, but so too were Roger Borsa’s cavalry outside of them. The counterattack seems to have been particularly threatening, and the duke accordingly opted for the strategic use of arson. As many of the city’s ecclesiastical buildings were both lofty and fortified – the duke himself appears to have been using the fortified Lateran as his base – ‘some churches were devastated by fire’. So began what has been termed by various historians as one of the city’s most terrible sacks, a belief enshrined to this day in the affected area’s street name, via dei Normanni. Yet much like an earlier one – as Peter Heather terms it, Alaric’s ‘highly civilized sack’ in 410 – the damage caused has been greatly exaggerated. Indeed, Hamilton notes that only the firing of SS. Quatro Coronati (connected to the Lateran by an underground tunnel), is firmly supported by the archaeological evidence. Therefore, the notion that much of the city was wantonly burnt should be dispensed with.34
While the sources favourable to the Normans stated that Guiscard was able to gain the upper hand, given his subsequent return south with the Pope in tow, the duke was more likely forced to follow in Henry IV’s footsteps – that is, opt for a tactical retreat. Whatever the case, Gregory would never again see the Lateran: he died the following year and was interred at the Salernitan cathedral begun by Guiscard after the capture of the city (1076). Conversely the duke was presumably glad to see the back of ‘the eternal city’, and he was no doubt concerned by Bohemond’s news that the initially successful campaign in the Balkans had failed. The duke’s resolve, however, was not shaken: he immediately began organizing another invasion force, which departed for Illyria in October 1084. While Alexios’ position on the throne remained tenuous, and the empire’s Danubian frontier was yet again threatened by Pechenegs, the emperor was nonetheless in a better position to repulse another invasion; indeed, as Loud observes, Guiscard ‘had missed his best chance’.35
The size of the duke’s invasion force appears to have been considerable, as William noted the fleet consisted of 120 ships, in addition to an unspecified number of merchant vessels. Similarly, a set of Apulian annals attested to ‘a large provision of ships’. As had been the case with the previous campaign, the duke dispatched an advance fleet to (re)secure bridgeheads on the Illyrian coast (probably in September). Led by his sons Guy, Roger Borsa and Bohemond, the fleet managed to recapture Avlona and Vouthroton. The main flotilla rendezvoused with the advance fleet at the latter location, presumably because of its close proximity to the next target: Kerkyra, whose citizens had managed to expel the garrison at Kassiopi in the north, but not the one holding out in the keep of the island’s capital to the south (which to this day bears the same name as the island). The emperor again successfully appealed to the Venetians for naval assistance, only this time a substantial imperial fleet was mobilized to assist them. Guiscard now advanced with his fleet to Kerkyra, specifically to the northern harbour of Kassiopi, where a series of naval engagements were fought against the allied fleet. The latter navy appears to have bested the duke’s twice, but in January 1085 it was itself defeated at a third engagement in the harbour of Kerkyra (18km south of Kassiopi). As one of the Apulian annals observed of the duke’s formidable victory:
More than one-thousand men were killed in this battle, in addition to the capture of five ships. Two were sunk with all men, thus those whom were able to evade the warrior’s sword, the abyss of the sea swallowed up.36
The Venetians, however, were soon able to exact revenge. With a fleet consisting of heavy, medium, and light ships, Guiscard’s navy was engaged off the coast of Vouthroton, where it was decisively defeated. The emperor was so pleased with his ally’s performance that he granted the Venetian doge a court rank and annual pension, in addition to endowing his people with free trade privileges, thus paving the way for the Venetian mercantile empire. While Alexios’ magnanimous gesture was not only understandable, but also sound diplomacy, the imperial fisc would henceforth be deprived of a significant source of income, later leading to great tension and, ultimately, riots and warfare when Alexios’ son and grandson variously revoked and reinstated the free-trade agreement.37
Following the third naval clash, Guiscard’s troops disembarked and swiftly lifted the siege of Kerkyra’s citadel. The fleet subsequently wintered in the mouth of what Komnene called the ‘Sweet River’ (Acherontas) near Ammoudia in north-western Greece, while the tireless duke advanced north with his best cavalry, choosing to bivouac either in the Ambracian Gulf near Vonitsa (Komnene), or at Vouthroton (William), situated 10km south-east of Kassiopi. However, disease spread throughout the main camp, indiscriminately killing a large number of infantry, in addition to 500 cavalrymen. Although Theotokis believes that the culprit was malaria, it is more likely that Guiscard’s men were afflicted by one of the maladies regularly afflicting sedentary armies until the modern period: flux/dysentery, or cholera, commonly the result of having latrines located too close to sources of drinking water, but also contracted from the transfer of bacteria contained in excreta to food by flies. Indeed, this theory appears to be confirmed by one of the Apulian annals, which referred to death caused ‘by the flowing forth of the bowel’. Bohemond became violently ill, and his concerned father arranged for a ship to take him with all haste to the coast of Campania, hoping that the pinnacle of Lombard civilization could save his son’s life: the Schola Medica Salernitana, the medical school of Salerno founded in the ninth century. He did not hope in vain; Bohemond was soon brought back to full health. Many others, however, were much less fortunate. Both William and Komnene believed that 10,000 died, and while that figure is too large, it seems possible that at least a third of Guiscard’s men perished. Had the duke’s troops been pagans, the location where so many met their end (according to Komnene) was in another sense fortuitous: in Greek mythology, the Acherontas was believed to have been the river over which the dead were ferried to Hades by the boatman Charon. Yet in the Latin Christian tradition, what Dante would later refer to as the ‘sorrowful shore of Acheron’ was believed to be the entrance to Hell.38
At the onset of the summer of 1085, Roger Borsa was dispatched with a detachment to capture Kephallonia, an island to the south-east of Kerkyra. As Theotokis notes, the Grecian isle was chosen for a precise reason: it was the capital of the theme bearing Kephallonia’s name, which comprised the seven islands of the Ionian Sea. If the duke was to advance further to the south and capture mainland towns and cities in the theme of Ellas (which included the Peloponnese), securing bridgeheads in the theme of Kephallonia was crucial, not only for supplies, but also for the procurement of additional ships. The duke joined his son and heir later that season, but was first beset by a significant impediment to the deployment of his fleet. Towards the end of spring, during which time the fleet had been beached near Ammoudia, the Acherontas had reached its lowest ebb; it was now too shallow for the fleet to return to the Ionian Sea. Guiscard, however, was characteristically unfazed; ‘by craft the duke makes a difficult labour easy’, as William of Apulia put it. Ordering his men to erect a network of makeshift dam walls composed of heaped sand strengthened by stakes fashioned from felled willow trees, Guiscard’s men were able to narrow the channel and thus raise the water level sufficiently to allow his ships to depart. By July 1085 the fleet had reached the north-eastern tip of Kephallonia, yet the many toils had evidently taken their toll on the immune system of the seventy-year-old conqueror: soon after arriving in the bay that to this day bears his name in tribute (Fiskardo), the duke fell victim to the ‘flowing forth of the bowel’ malady that had already killed so many of his men.39
Duchess Sichelgaita, who earlier had joined her son in Kephallonia, was made by William to voice the fears of both herself and her son to the dying duke. She reproved Guiscard for antagonizing the Byzantines; surely without the duke’s steely leadership his people would not have the courage to resist an invasion? But there was more:
Behold! The son and wife are allowed to be dragged off by wolves,
And your people will never be secure without you!40
Both fears were very real, but that posed by William’s ‘wolves’ was the most serious: in addition to the threat of serial malcontents such as Abelard and Geoffrey of Conversano, was that of another scion of the Hauteville dynasty: Bohemond, whose various incursions into Roger Borsa’s territory soon after the duke’s death indicated that he was more than a little upset about his disenfranchisement. The civil wars between the half brothers might have been averted if Guiscard had been able to secure permanent possessions in Illyria and Greece for his eldest son to rule, but this possibility had slowly but surely been eroded when the duke was recalled to Italy in 1082.
No doubt advised by his strong-willed and politically astute Lombard mother, Roger Borsa wisely decided to abandon the campaign. He then secured oaths of fidelity from his father’s retinue, emphasizing that they must help him to counter the inevitable challenges to his ducal inheritance. There were, however, two last crises for the young duke to face. Terrified that Alexios’ forces would swoop down and massacre them, the army burned the larger ships, keeping only the smaller ones deemed suitable to ensure a speedy return to Italy. Not only were horses abandoned, but so too was baggage and money. Yet the panic-stricken burning of the larger ships introduced an insurmountable logistical problem: limited space on the smaller vessels meant that many were unable to gain passage and, according to William of Apulia, most of these stranded troops had little choice but to accept Alexios’ offer of gainful employment in the imperial army. Adding insult to injury, a considerable number of the fleeing flotilla were caught in a tempest off the coast of Otranto, resulting in the loss of numerous ships and men. Sichelgaita’s flagship was itself caught in the storm, and Guiscard’s corpse was washed overboard, although her sailors were quite miraculously able to retrieve it in the tempestuous chaos.41