Chapter 2

Mercenaries

The subject of the arrival of the Normans in Italy is a difficult one, precisely because the relatively large number of pertinent sources rarely agree with each other. Unsurprisingly, therefore, the topic has been the subject of considerable debate. While it is unnecessary to examine the various theories in detail, a general overview is certainly required. It is possible that a group, or groups, of Normans returning home from pilgrimages to either Jerusalem or the shrine of the archangel Michael at Monte Gargano in southern Italy, were hired as mercenaries by Lombard leaders between the years c. 999–1009. Amatus of Montecassino noted that in c. 999 some Normans assisted Prince Guaimar III of Salerno (r. 999–1027) against Muslim forces then besieging his city. It is also possible that Norman mercenaries may have supported the Lombard leader Meles of Bari in his first insurrection against Byzantine rule in Apulia (1009–11). However, all sources agree that when Meles launched his second foray into Apulia in 1017, Normans were present in his ranks. Three chroniclers, including the two earliest ones, French writers Raoul Glaber (Burgundy) and Adémar of Chabannes (Aquitaine) indicated that the mercenaries were under the command of Raoul of Töeni (Tosny), then in exile having incurred the wrath of Duke Richard II (r. 996–1026). Both Glaber and Adémar stated that the rebellion was supported by Pope Benedict VIII (r. 1012–24) and, while the Italian sources are silent on the matter, it is certainly possible given the papacy’s ongoing desire to restore the ecclesiastical jurisdiction revoked by Byzantine emperor Leo III in the eighth century.¹

The number of men potentially led by Raoul of Töeni cannot be determined. Amatus of Montecassino’s garbled eleventh-century account mentioned that there were initially 250 men, but that 200 of them were killed in the early battles. However, quite unrealistically, he maintained that the depleted ranks had swelled to 3,000 by the second year of the rebellion (1018). In the Montecassino Chronicle, however, Leo Marsicanus stated that when Meles ‘was staying with the prince of Capua [Pandulf IV]’ in 1016/7, present with him were ‘some Normans, about 40 in number’. What does seem clear – if such a word can be used when trying to reconcile the imprecise, conflicting numbers provided by the chroniclers – is that the contingent had increased significantly by 1018, for Leo later noted that double this number survived the final battle at Canne. Adémar of Chabannes was much less specific, merely observing that ‘a large number’ arrived in 1017. Whatever the case, it seems likely that initial numbers were relatively small, but that they increased throughout the course of the rebellion. On the latter point, for once the sources are in general agreement: when word reached Normandy of victory over the Byzantines, others were inspired to make the journey to Italy.²

As for the Lombards, who must have formed the majority of Meles’ forces, only Leo Marsicanus and Raoul Glaber explicitly mentioned them. Writing in reference to Salerno and Benevento, Leo mentioned numerous recruits from among those united in their ‘hatred of the Greeks’, and it is likely that some hailed from Capua given that Meles initially used the city as his base. Unlike the battles fought from the 1040s onwards, for which the primary chroniclers evidently had more detailed sources on which to draw, there is little indication as to what sort of troops were fielded by the Lombards and Normans. While Amatus did mention cavalrymen once, Adémar merely mentioned armati in reference to the Normans, a term normally pertaining to heavy infantrymen. Here, however, he seems to have meant it in the literal sense of ‘armed/armoured men’ – that is, a body of troops protected by haubercs. If later Norman contingents are anything to go by – e.g. Robert Guiscard was said to have arrived in Italy in the 1040s with five cavalrymen and thirty infantrymen – it is possible that these armati consisted of both cavalry and infantry. When it comes to the Lombards, while they are often assumed to have been an infantry-based people by the eleventh century – an assumption shown to be false in the previous chapter – there is no reason to doubt that they too fielded cavalry forces.³

How many battles the Lombardo-Norman forces fought against the Byzantines is, predictably, another tricky matter. Amatus’ account of the engagements fought in 1017–18, for example, is fraught with problems. It seems to be a compressed recollection of the two rebellions (i.e. 1009–11 and 1017–18), and makes for confusing reading. This would explain why he wrote of seven battles, when the other sources mention four, excepting William of Apulia and Raoul Glaber, who suggested three. However, when compared with the other accounts, Amatus’ sixth battle appears to correspond with William’s first, and Glaber’s second. Indeed, the three sources agreed that the outcome of this battle, identified by William of Apulia to have been fought near the River Fortore in northern Apulia in May 1017, was inconclusive. Despite noting significant casualties on both sides, Glaber indicated that ‘the army of the Normans nevertheless emerged triumphant’, whereas both William and Amatus made it clear that the ‘victory’ was in fact a stalemate. If Amatus is to be trusted, the losses among the Normans were substantial – only fifty out of 250 survived. Moreover, he indicated that the figures were learnt from participants, although given that he was writing around sixty years after the battle, these ‘eyewitnesses’ must surely have been sons or grandsons of those who were actually present.⁴

The only battle that received relatively ‘extended’ coverage by two of the chroniclers was the final one, fought at Canne in northern Apulia in October 1018. William of Apulia described it in four lines of verse:

Nearby Canne, where the River Ofanto flows,

It is fought not far from the Kalends of October.

With only a few, the people of Meles are unable to resist;

Fleeing, he is despoiled of his many acquisitions.⁵

William’s summary presents a recurring theme employed by others such as Amatus: at Canne, much like the earlier battles, Meles’ forces were greatly outnumbered. That this was the case prior to Canne seems disingenuous, for imperial garrison troops were rarely substantial; a fact implied by William of Apulia, who had Meles promising the Norman pilgrims an easy victory when attempting to recruit them at Monte Gargano. Moreover, it has already been noted that both sides, according to Amatus and Glaber, sustained heavy losses in the previous engagement, so Byzantine forces cannot have been considerable prior to the battle. However, William of Apulia did note that the newly appointed katepano, Basil Voioannes (r. 1018–28), arrived from Constantinople ‘with a unit of many Greeks’. William, who was better informed on the Byzantines than anyone else writing in Latin in the eleventh century, was regularly specific when it came to ethnic contingents of imperial forces, and accordingly Adémar’s indication that these men were ‘Russians’ (Varangian Guard) – assumed to have been the case by various scholars – should be treated with caution.⁶ Indubitably, Leo Marsicanus did mention the use of ‘Danes, Russians and Varangians’, but he was referring to the first revolt not the second. Since Adémar was writing in faraway Aquitaine – and hence was without the benefit of access to local oral and written sources – he may have confused their decisive involvement in the first revolt of Meles with the second one crushed at Canne some seven years later. However, as this unit of Byzantines was dispatched from the capital, where the élite regiments were stationed, it is likely that these men were kataphraktoi. While their numbers cannot have been high, their quality definitely was. Presumably Meles had previously engaged a mixture of part-time, locally recruited thematic troops, and perhaps some local militia. At Canne, however, his forces would be tested by battle-hardened professionals. Leo’s reference to the battle is particularly revealing:

In the fourth [battle] finally fought near Canne, formerly famous for the defeat of the Romans [216 BC], he [Meles] is overcome by the ambushes and stratagems of the katepano Voioannes, losing easily everything he had taken without difficulty.⁷

The fact that this stratagem was employed potentially suggests that, contrary to the opinion expressed in other accounts, the Byzantines were actually outnumbered. Tactical manuals for almost five centuries had urged imperial commanders to adopt stratagems when outnumbered by the enemy. The famous Strategikon written by, or attributed to, the Emperor Maurice (r. 582–602), devoted an entire book to strategema, and its opening section was entitled: ‘Ambushes and Stratagems Against Superior Enemy Forces’. That generals generally adhered to such exhortations is supported by the fact that imperial forces were regularly noted by historians as adopting the tactic when engaging more numerous forces. It was a tactical reality summarised by Emperor Alexios I Komnenos’ daughter, Anna:

By my reckoning, it has always been best to devise cunning strategy in war and combat, when one’s army is unequal to the strength of the enemy.⁸

At Canne, Voioannes may well have executed a tactical manoeuvre that only the most disciplined troops could employ with success. Perhaps Meles’ forces were lured into a false sense of victory – probably a feigned retreat – and were massacred when the imperial cavalry smashed into the disordered flanks of the pursuers. Judging by the assessment of the battles of 1041 that will be discussed below, it was a military lesson that the Normans in Italy were never to forget.

Following the defeat at Canne, Meles found refuge in Germany, where he gathered support for a third attempt. His brother-in-law and co-conspirator Datto, along with some Normans, escaped to a tower on the border of Capuan and papal territory. Figuratively accepting Byzantine overlordship, Pandulf IV sent the keys of Capua to Voioannes, in addition to allowing Byzantine troops access to his territory so they could apprehend the fugitive. After his capture, Datto was granted the polar opposite of the Romano-Byzantine military triumph: he ‘entered the city of Bari riding an ass’. Based on Lynda Garland’s assessment of other examples of this form of public humiliation, it seems that Voioannes punished Datto with a diapompeusis, ‘a parade of infamy’. But the shaming of Datto did not end with the parade. He was afterwards subjected to a grisly Roman execution known as poena cullei (‘the punishment of the sack’): ‘after a few days he was sewn up in a leather sack in the manner of a parricide, and thrown headlong into the middle of the sea’. As for Datto’s Norman followers, we are told by Leo Marsicanus that Pandulf’s brother, Abbot Atenulf of Montecassino, pleaded successfully for their release.⁹

Capitalising on his victory at Canne, the katepano Voioannes had been busying himself with the fortification of the region corresponding to the northern province of modern-day Puglia, Foggia. The province became known in the vernacular as Capitinata, which Leo Marsicanus noted was a corruption of Catapanata (‘katepanate’ or ‘land of the katepano’). Five fortified towns were established: Civitate, Troia, Fiorentina, Montecorvino and Dragonara. Situated on the extremities of the northern Apulian plain, these fortified settlements were clearly established to prevent further attacks deep into Longobardia. With the north now well defended, Voioannes turned his attention to the southern and western frontiers of Longobardia, although his hand was forced when it came to the former region owing to Muslim attacks on Bari and Palagiano (22km north-west of Taranto). To guard against further incursions into the Tarentine hinterland, in 1023 the katepano erected a fortress at Mottola (11km north of Palagiano), a site of great strategic value given its location on the via Appia, the imperial road connecting Taranto to Rome. Moreover, despite friendly relations existing with the prince of Salerno, Voioannes evidently wanted to defend the western approach into Longobardia: accordingly the town of Melfi, which two decades later would become the base for the Lombard-Norman invasion, was fortified in the same year.¹⁰

Meanwhile, the Normans who survived the defeat at Canne had effectively lost their raison d’être, but they would not wait long for gainful employment. As was presumably the case with the Lombard troops, casualties among the Normans appear to have been high: Leo mentioned eighty survivors, and Amatus noted that out of the 3,000, only 500 remained. Other than the presumably few who followed Datto and Meles into exile, they soon found employment with Lombard princes, the abbots of Montecassino, Meles’ nephews at Comino (Abruzzi) and, interestingly, with the garrison at Troia, one of Voioannes’ newly established towns in the Capitinata. The Normans, then, were clearly ‘soldiers of fortune’; they were happy to offer their services to whoever was willing to pay them.

No doubt prompted by the petitioning of both Meles and Benedict VIII, the western emperor Henry II (r. 1014–24) decided, among other things, to test the new fortifications in the Capitinata. Departing from Germany in 1022, Henry targeted those who supported his eastern counterpart: Pandulf IV of Capua and Guaimar III of Salerno. The campaign was, however, largely a failure. The only short-term gain was the nomination of a new prince in place of Pandulf, but he managed regardless to depose his replacement four years later. Both Salerno and Troia, whose garrisons included Normans, were unsuccessfully besieged. Perhaps Henry’s campaign might have achieved greater support from the Lombards had Meles been with him, but the eminent Bariot, who had challenged Byzantine authority twice, had passed away over a year earlier at Bamberg.

With Meles’ death and the failure of Henry II’s attempt to assert western imperial dominance in southern Italy, the areas under Byzantine rule experienced a period of peace for almost twenty years. Had this been also true of the Lombard regions, a book such as this might not exist. Naturally, mercenaries were better suited to regions where internecine warfare was the rule, not the exception. Without conflict the Normans would have presumably returned home or found employment elsewhere in Europe or the Mediterranean world. As it turned out, there was no need for them to move a figurative inch: the princes of Capua and Salerno now began to wage war against each other, each apparently seeking to establish hegemony over Campania, including the independent, wealthy maritime cities on the west coast.

While both William of Apulia and Geoffrey Malaterra noted that the Normans benefitted financially from the squabbles between the Capuan and Salernitan princes over hegemony in the area of southern Italy not subject to Byzantine rule, only the former suggested that another intentional benefit was gained: that is, from the outset the idea was to weaken the Lombard powers so the Normans could take the reins of power for themselves. Hindsight can be a useful thing, but in this case William was misled by it. The reality was that the incremental increase of Norman fortunes in the region was, by and large, the result of unique and unforeseen circumstances rather than astute, cunning political foresight.

According to Leo Marsicanus in the Montecassino Chronicle, a small group of Normans were hired as mercenaries by Abbot Atenulf (r. 1011–22), the brother of Pandulf IV of Capua, probably soon after the Byzantine victory at Canne in 1018. The soldiers were sent to garrison Pignataro (28km north-west), with the aim of countering incursions into Cassinese territory by the Lombard counts of Aquino (11km north-west of Monte.). As Abbot Desiderius (r. 1058–87) was to write in the 1070s, the Normans initially behaved ‘in the habit of vigorous cavalrymen’, but within a few years had taken the majority of the monastery’s estates. Such behaviour prompted Desiderius to describe the Normans as a people ‘eager for plunder’ and ‘led by impious greed’. In response to this continuing problem – Norman troops were later settled on Cassinese lands by Pandulf IV – Abbot Richer (r. 1038–55) began the construction of fortresses in order to protect the villages throughout the abbey’s territory, known collectively as the ‘land of Saint Benedict’. Clearly the Cassinese Normans were formidable fighters despite their small numbers, for Abbot Richer was not able to drive them from Montecassino and its environs until 1045.¹¹

Pandulf IV of Capua, dispossessed of his principality by the western emperor Henry II in 1022, was able to recover it in 1026 with both Byzantine and Norman assistance. Since his young nephew Guaimar IV (r. 1027–52) ruled Salerno, Pandulf exercised a considerable degree of influence over that principality. Pursuing his evident plan of suzerainty over the non-Byzantine regions of southern Italy, with Norman support Pandulf captured the wealthy city of Naples in 1027. By 1036 he had unsuccessfully attempted to seize control of Benevento from his nephew and namesake Pandulf III, but nonetheless managed to secure control of Gaeta (71km north-west of Naples) shortly afterwards. While most of these acquisitions were held in the short term – Naples, for example, was probably ruled for three years – the Capuan prince’s influence in the region was considerable. The ‘mighty wolf’, as Amatus called him, was nonetheless unpopular for a variety of reasons, most notably his extensive meddling in the internal affairs of the influential abbey of Montecassino. In addition, his previously amenable, even malleable, nephew at Salerno had, to use the words of William of Apulia, begun to display the same ‘great lust for power’ as his uncle.¹²

Duke Sergio IV’s recapture of Naples from the hands of Pandulf in 1030 resulted in the first formal settlement of Norman mercenaries in Italy. No doubt to counter the prince’s likely attempts at retaking the city, Sergio settled a group of mercenaries under the command of Rainulf Drengot at Aversa, a recently fortified town close to the Capuan border. Importantly, Rainulf secured the hand of Sergio’s sister in marriage, establishing a practice that would become common with subsequent Norman leaders in Italy. In addition to the newfound prestige of the Aversan Normans, they were able to gather taxes from the surrounding region. Abbot Desiderius’ words regarding the initial loyalty of Norman mercenaries also apply to those favoured by Sergio for, like their Cassinese counterparts before them, the Aversan Normans transformed from allies into enemies. Upon the death of Sergio’s sister Rainulf evidently felt that his debt to the Neapolitan duke was now at an end, so he forged an alliance with the latter’s enemy: Pandulf IV of Capua who, according to Amatus, cemented the pact by arranging Rainulf’s marriage to his niece.¹³

The Normans who participated in Pandulf IV’s recapture of Capua had been loaned to him by his brother-in-law, Guaimar III of Salerno. Presumably included among these mercenaries were those who had entered Guaimar’s employment following the battle of Canne. However, by c. 1036 amity between the princes had changed to enmity. Deciding that more Norman troops would be required to defeat his uncle, Guaimar IV secured additional mercenaries, luring them in with money, horses and silks. In the process, Amatus related that the Salernitan Normans formally accepted Guaimar’s suzerainty over the other Lombard rulers. As Malaterra observed, with their ‘fidelity to him [Guaimar] inspired by many gifts’, the Norman horsemen ‘attacked the Capuans with diverse and frequent raids, terrorizing every corner of the entire province’. But while the chroniclers of the period – like some of their modern counterparts – tend to imply that only the Normans were involved in the hostilities, interestingly the same Norman/French chronicler noted that ‘the Salernitans … followed suit’.¹⁴

By 1038, Guaimar had effectively ‘won’. The western emperor Conrad II marched south, formally investing Guaimar as prince of Capua, in addition to recognizing Rainulf’s lordship in Aversa. The Norman was now officially a count. Disenfranchised for the second time in sixteen years by a western emperor, Pandulf departed for Constantinople in hope of obtaining military succour. Unfortunately for the ‘mighty wolf’, such help would not be forthcoming. Meanwhile, in the years 1038–40 Guaimar added the prosperous maritime duchies of Amalfi, Sorrento and Gaeta to his dominions, becoming what his uncle had long tried to be: effective lord of the non-Byzantine regions of southern Italy. Among those Normans who helped him to achieve this feat were the first progeny of Tancred of Hauteville to serve as mercenaries in Italy: William, Drogo, and Humphrey.¹⁵

The Byzantines, who previously held Sicily for centuries until its towns were incrementally, but inexorably, captured by the Aghlabid Arabs during the years 827–902 (although Rometta resisted until 965), resolved in 1038 to reconquer the island. While Sicilian attacks on Lombard Campania appear to have ceased in 1016, Byzantine towns in Apulia and Calabria had been subject to various raids, with the most recent occurring in June 1031, when the Calabrian town of Cassano was captured. In the following month, imperial forces led by the katepano Pothos Argyros (r. 1029–31) were defeated in battle by the raiders. While in 1038 there was no immediate threat of raids from Sicily, the decision to undertake a renewed campaign of reconquest in Sicily was presumably influenced by the political instability resulting from the recent assassination of the Kalbid amīr, Ahmad al-Akḥal. Importantly, al-Akḥal was an imperial ally, having earlier sought assistance against the expansionist designs of the Zirid ruler of Ifrīqiya (Tunisia), al-Mu‘izz, who, according to eleventh-century Byzantine chronicler John Skylitzes, had agreed to provide military support to al-Akḥal’s disgruntled brother in return for land in Sicily. While the sources do not always agree on the various facts and events, one thing is clear: with conflict between the Kalbid and Zirid dynasties, in addition to Byzantine involvement in the dispute, as Alex Metcalfe puts it ‘the political coherence of Sicily was now beginning to disintegrate’.¹⁶

In addition to sending experienced troops from Constantinople under the command of the seasoned general George Maniakes, military assistance was secured from the empire’s subordinates and allies in Italy. Depending on which source is favoured, Guaimar IV responded with a cavalry force that was either 300 (Amatus, Leo) or 500 (Skylitzes) strong, consisting entirely of Normans/Franks (Amatus, Leo and Skylitzes), or a ‘mixture of Gauls and Lombards’ (William of Apulia). The latter’s use of ‘Lombards’ is interesting, for it was used in the sense of northerners (Lombardi) rather than southerners (Longobardi). That Lombards from northern Italy were present among the more numerous Normans/Franks is certainly possible, for the man who led the cavalry contingent in question, Arduin, was a mercenary from Milan. To complicate matters, an anonymous Sicilian source of the thirteenth century, which generally relied on Malaterra’s work but nevertheless added original information, emphasized that Maniakes’ army included Lombards from both Salerno and Apulia. Whatever the case, Arduin, whom Malaterra noted as ‘having expertise in the language of the Greeks’, was the logical choice to lead the contingent. Despite Arduin’s overall leadership, if Amatus is to be trusted the Normans acknowledged the leadership of William of Hauteville, who had recently arrived in Italy with his brothers Drogo and Humphrey.¹⁷

Other than the usual Byzantine infantry and cavalry forces, the famous Varangian Guard was also present, under the command of the future King of Norway, Haraldr Sigurðarson (Hardrada), whose last act in life was to fall in battle against the Anglo-Saxons at Stamford Bridge in 1066. Landing at Messina in the summer of 1038, Maniakes’ forces secured this important port, then captured Rometta (12km south-west of Messina). The chronology of the various pertinent sources is almost nonexistent, but it seems likely that the total of fourteen towns mentioned by Skylitzes were captured in 1038–9, and were probably all located in the Val Dèmone (north-eastern province). All the captured towns, according to Skylitzes, were fortified and garrisoned. In 1040, penetrating inland close to the western extremity of the Val Dèmone, Maniakes gained a great victory at Troina. Securing the Val Dèmone, coupled with the capture of the strategic southern port of Syracuse in the Val di Noto (south-eastern province), meant that Maniakes had effectively reconquered eastern Sicily.¹⁸

As Ferdinand Chalandon observed, Amatus and Malaterra ‘exaggerated the role played by the Normans in this campaign’. Military prowess was the exclusive preserve of their protagonists, who thundered through hordes of unbelievers with lethal precision. Clearly a unit of 300–500 experienced cavalrymen, while undoubtedly of great value, could not have been so indispensable. Unfortunately, few details are provided of the battles, other than to say that they were particularly bloody affairs. There is, however, an interesting and relatively detailed account of the battle of Troina contained in the eleventh-century hagiography of the Calabrian saint Agios Philaretos. Evidently the Muslims feared the onslaught of imperial cavalry, for the field around their camp was littered with caltrops. In order to protect against them, Maniakes had the horses’ legs equipped with ‘iron plates’. While The Life of Saint Philaretos notes that the mounts of the Byzantine cavalrymen were protected in this manner, since Malaterra recorded that Arduin’s contingent were deployed in this battle it is likely that his men’s horses were similarly fitted. Malaterra’s insistence that William of Hauteville – called ‘Iron Arm’ after reportedly killing the governor of Syracuse with a single blow – was almost entirely responsible for the routing of the enemy forces can naturally be discounted. The same doubt should be applied to the assertion that William and his fellow cavalrymen not only engaged, but also defeated, the enemy before the rest of the imperial army had the chance to charge. The Life of Saint Philaretos is clear: the imperial cavalry engaged the enemy en masse, albeit deployed ‘in three battalions’ (probably arrayed in wedges, the characteristic formation of the kataphraktoi).¹⁹

Bernard Bachrach has suggested that the art of cavalry transport, so evident when the Conqueror’s forces crossed the English Channel in 1066, may have been an adaptation of what the Normans experienced under Arduin when they were shipped across to Sicily in 1038. While Charles Stanton has since found this view to be ‘improbable’, what is of relevance here is Bachrach’s further suggestion that the bataille, and its subdivision the conroi, were adaptations of Byzantine cavalry organisation (respectively, the tagma/arithmos and vandon). Noting the various mentions of 300–strong Norman cavalry divisions in south Italian sources, presumably subdivided into conrois of 10–20 men, Bachrach, observing that since there ‘is no doubt that the Byzantines used 300–man units’, suggested the former learnt the practice from the latter (and, as was the case with naval matters, transmitted the practice back to Normandy). However, much as the numerical strengths of Roman divisions were never static, the same was true of their continuators. As John Haldon relates, the ‘sizes of individual units on the battlefield varied according to tactical need and was left to the discretion of the commander in the field’. Hence, Byzantine units (tagma, vandon, arithmos) tended to oscillate between 50–400 (vandon) and 400–500 (tagma, arithmos) men. Moreover, although vanda (pl.) usually denoted a subdivision of the tagma or arithmos, all three terms were synonymous on occasion. Hence, since the numerical strength of Byzantine cavalry divisions was never static, it is difficult to support the idea that the Norman system of cavalry organization was modelled on them. That being said, it should be acknowledged that Arduin’s contingent no doubt learnt much from the Byzantines during the Sicilian campaign, a reality underscored by the fact that the Normans would employ similar methods two decades later when undertaking their own invasion of Sicily.²⁰

With eastern Sicily under imperial control, the time was now ripe for Maniakes to turn his attention to the unconquered western half of the island. However, following an altercation with his naval commander Stephan, who also happened to be Emperor Michael IV’s brother-in-law, Maniakes was recalled to the capital. Yet prior to the general’s recall, after the battle of Troina in 1040, he ordered Arduin to be stripped and beaten for insubordination. While Skylitzes attributed the cause of the heated argument between the Byzantine commander and Arduin to the fact that payment of the monthly mercenary wage was in arrears, Amatus, William of Apulia and Malaterra emphasized a disagreement centred on the distribution of spoils after the battle. The latter two noted that Arduin and the Normans immediately abandoned Byzantine employ as a result – clandestinely leaving the Byzantine camp during the night – whereas Amatus suggested that the Lombard bore the punishment, choosing to wait for an opportunity for revenge. Whatever the case, after the arrival of the new katepano Michael Dokeianos (r. 1040–41) in November of 1040, Arduin was appointed garrison/cavalry commander of Melfi in Apulia, one of the towns founded by the katepano Voioannes following his victory at Canne in 1018. In March of 1041, Arduin, with the assistance of Normans from Aversa, overthrew the Byzantine garrison, thereafter using Melfi as a base for an invasion of Longobardia.²¹

Prior to Arduin’s invasion, Byzantine Apulia had been subject to a series of disturbances. Shortly before his death in January 1040, the katepano Nikephoros Dokeianos (r. 1039–40) had discharged the enigmatic conterati, who soon launched a revolt in response. It has been generally assumed that they were lightly-armed troops locally recruited by the Byzantines, and Theotokis has recently suggested they were ‘poorly trained, undisciplined’ and fought ‘mostly on foot’. Furthermore, basing his assertion on emperor Leo VI’s late-ninth to early tenth-century Taktika, Theotokis posits that the weapons wielded by the conterati were ‘short spears’ (i.e. for use by infantrymen). However, in a later military manual attributed to the Byzantine emperor Nikephoros II Phokas (r. 963–69), a kontaratos is unequivocally a lance wielded by both heavy and light cavalrymen, who were placed behind the first four ranks of mace- and sword-wielders in a cavalry wedge. The latter are also mentioned as forming separate, 500–strong divisions, with the suggestion that aside from featuring 100–120 archers, the rest should consist of ‘lancers’ (kontaratoi): a term also used in the same sense by Nikephoros Ouranos in his Taktika, an early eleventh-century embellished version of Phokas’ treatise. Light cavalrymen were used for skirmishing and scouting, and hence were protected only by waist-length corslets of lamellar, scale or mail. However, the kontaratoi among the regular cavalrymen were also equipped with iron helmets and shields (Fig. 3).²²

Conceivably, then, it seems likely the Lombard conterati were so named because they were either based on, or already resembled, their Byzantine counterparts. While most were probably equipped in the manner of light cavalrymen, some were undoubtedly more heavily armed. While there is no explicit evidence for this assertion, in the wider context it needs to be emphasized that, on account of their marked Norman chauvinism, the majority of the sources pertinent to eleventh-century southern Italy are unsurprisingly bereft of insight into Lombard military practices. The two shortcomings in this regard are: firstly, the Lombards seem to have only fielded infantry forces – a tradition solidified by Raoul of Caen’s ‘cavalry of Normandy, infantry of Longobardia’ description in regard to Guiscard’s forces of 1081 – and, secondly, that they were, to use the words of Raoul Glaber, ‘deficient men … insufficiently suited to warfare’. Yet various sections from Amatus’ history negate both assumptions: Pandulf of Aquino, reportedly with only a few mounted troops, defeated some Normans in the 1060s, and Guido of Salerno was said to be a highly skilled cavalryman, capable of unhorsing experienced Norman horsemen in battle. It should also be stressed that, since antiquity, Apulia and Calabria were regions well known for horse-breeding. Indeed, Malaterra noted that Count Roger went to both Apulia and Calabria to procure replacements for ‘the horses which his men had lost’ in Sicily. Lastly, as Amatus, William and Malaterra made clear, the Normans increasingly recruited local soldiers into their ranks. That some of these troops were proficient in mounted combat is suggested by the fact that they were given horses.²³

Having been discharged by the katepano, the conterati, led by a certain Musondo, first captured the fortress at Mottola (55km south of Bari) in May, executing the imperial officials Michael Choirosphaktes and Romanos of Matera in the process. Moving north, in the same month they managed to enter the capital of Longobardia, Bari, and were assisted in this operation by Argyros, whose father Meles had led the Apulian revolt of 1017–18. Argyros, for reasons unexplained, fell out with his allies, imprisoning Musondo and driving the conterati out of Bari. Despite the dire nature of the situation in Apulia, the Byzantines did little to resolve it until November, when a replacement for the deceased Nikephoros Dokeianos arrived, who was none other than his son, Michael. Those (unnamed) members of the conterati who were no doubt directly responsible for the murders of the imperial officials were captured and hanged (or, according to one source, blinded). Since there is no mention of Argyros being punished for his role in the capture of Bari, coupled with the puzzling ten-month delay between the katepano appointments, it seems likely that he had treated with the emperor after dispersing the conterati in May, perhaps being recognized as a caretaker until the arrival of Dokeianos six months later.²⁴

‘At that time there was no imperial army in Italy’, observed William of Apulia, noting that it was still engaged in the ‘war against the Sicilians’. Arduin’s revolt, then, was perfectly timed. According to Amatus, Rainulf of Aversa pledged 300 Norman cavalrymen to Arduin, with whom the Lombard marched on Melfi in early March 1041. Despite being walled and protected by catapults, by means of clandestine entry at night Melfi was taken without violence, as were the nearby towns of Venosa, Lavello and Ascoli. While the fortified towns were spared the horrors of a sack, they were nonetheless despoiled, and their citizens promptly appealed to Michael Dokeianos for succour. The katepano, who had arrived in Bari via Sicily to take up his post, presumably brought some troops with him. Having received an answer in the negative to his demand that the invaders surrender their recently acquired possessions, Dokeianos marched west towards Melfi in mid-March. According to William of Apulia, Arduin’s army by this stage consisted of 700 cavalrymen and 500 foot soldiers. If Amatus’ initial figure of 300 from Aversa is subtracted from William’s overall total of 700, the question remains as to who these additional 400 cavalrymen were. Since Malaterra mentioned two Hauteville brothers present at the ensuing battle, William ‘Iron Arm’ and Drogo, it is tempting to suggest that these extra cavalrymen were the same troops who had served under Arduin in Sicily.²⁵

On 17 March battle was met abutting the banks of the River Olivento, situated between Lavello and Melfi. Despite Amatus’ reference to a multitude, the imperial force cannot have been substantial; it is clear that numerous reinforcements did not arrive from Sicily until after this battle. However, as is evident from William of Apulia’s nuanced, more balanced, better-informed description of the engagement, it is certainly possible that Dokeianos’ men outnumbered Arduin’s, and that he had with him a contingent of kataphraktoi. The sole Byzantine source for the battles of 1041, John Skylitzes, noted that the katepano only possessed ‘a unit of the Opsikion and a division of the Thrakesion’. One of the annals of Bari also suggested that the katepano had just two units, noting ‘many Russi [Varangian Guard] and Opsikions’.²⁶ When reconciled with the varying unit sizes discussed above, even if the Varangian unit is added to those from the Opsikion and Thrakesion themes, the imperial army probably did not number more than 2,000 men.

William of Apulia related that Arduin’s ‘heavy infantrymen’ were grouped into two columns, with some cavalry posted on the flanks to stiffen their resolve. A unit of cavalry advanced towards the Byzantine line, ‘against which a wedge of Greeks was thrown’. William’s language clearly refers to a kataphraktoi charge, presumably actioned by the men from the Opsikion regiment. Both William and Amatus indicated that Dokeianos sent other detachments of two or three units against the rebels and, to judge from Malaterra’s sparser account, the battle hung in the balance for some time. Gradually, however, Arduin’s troops gained the upper hand, managing to drive part of the imperial army towards the river. Encumbered by their arms and panoply, William noted that those Byzantine soldiers not cut down by ‘Gallic’ swords and javelins drowned.²⁷

While the casualties sustained by the Byzantines seems to have been relatively high, since the same contingents were said to have taken part in the next battle a significant number must have managed to escape the slaughter. Judging by the names of the forces enumerated in one of the Bariot annals, the survivors were reinforced by Armenian troops previously serving in Sicily, in addition to those recruited from the mainland, including the Capitinata. Now ready for battle, the Byzantines engaged the rebel army near the River Ofanto on 4 May, not far from where they had crushed Meles’ rebellion at Canne in 1018. According to Amatus, Dokeianos attempted to overwhelm the enemy with a series of cavalry assaults, but his attempts were in vain. The Byzantines were eventually beaten back, and Dokeianos himself was almost killed during the rout. Much plunder was secured from the victory, including arms and horses.²⁸

Since William noted that only ‘a few’ of the Normans possessed chainmail haubercs at the first battle, it can be imagined that some began to resemble Byzantine cavalry in appearance, no doubt donning lamellar and scale corselets stripped from the fallen. Convenience presumably later gave way to common sense, for the campaign season in southern Italy was noted for searing heat, to which the cooler, less weighty panoply of the Byzantines was better suited. Indeed, the increasing preference for splint armour is evident in the doorway frieze at San Nicola di Bari (Fig. 1), which features a mixture of scale, lamellar and chainmail coats. The choice of non-traditional armour is also specifically attested in a literary source: Anna Komnene observed that an arrow ‘passed through’ Count Richard of the Principate’s ‘corslet clad in scales’ in 1096. Moreover, she elsewhere observed that Normans wore ‘breastplates and coats of iron’: the former term being the same one used to denote lamellar corslets worn by kataphraktoi, the latter probably referring to chainmail coats. Although the Byzantine princess wrote her account in the middle of the following century, she did see the southern Italian crusaders as a young girl (aged around thirteen), in addition to drawing on the testimony of her father and his generals.²⁹

Depending on which source is followed, Arduin was either commander at the first two battles or only the first. At any rate, by the time of the third clash the Lombardo-Norman forces had come to acknowledge the leadership of Atenulf, brother of Prince Pandulf III of Benevento. William, despite not explicitly stating the reason for Arduin’s demotion, suggested that the Normans were persuaded to abandon allegiance to their former commander by means of bribery. While numbers were not provided, Amatus noted that Atenulf’s army was substantially larger – a ‘multitude’, no less – for the third and final battle. Those Lombards without horses were granted them, and all new recruits were offered a share in the spoils.³⁰

Michael Dokeianos, scathingly referred to by Skylitzes as a ‘useless man’, was replaced by the son of Basil Voioannes, called Exaugoustos in some of the sources. While Skylitzes insisted that the younger Voioannes engaged Atenulf’s army without reinforcements, William of Apulia noted that Dokeianos, prior to the arrival of his replacement, requested them from Sicily. Voioannes assembled his forces at Montepeloso (Irsina), a fortified hilltop town overlooking the plain watered by the River Bradano. According to Amatus, when it became known that the katepano intended to besiege Melfi, the rebels resolved to force a pitched engagement. Atenulf’s troops marched first to Monte Serico, later assembling in the valley below Montepeloso on 3 September. All relatively detailed accounts of the battle agree that the fighting was particularly fierce. Interestingly, in common with William of Apulia’s description of the previous battle, Amatus observed that the Normans ‘hurled their spears against the Greeks’. Unfortunately, none of the sources mentioned the tactics employed by the Byzantines, but William of Apulia and Malaterra stressed that the imperial army began to gain the upper hand. Indeed, both observed that the Byzantines had all but won the battle, only to have victory snatched from their hands on account of an act of individual valour. Malaterra had William of Hauteville performing the heroic deed, whereas William mentioned Walter, son of Ami I. William/Walter charged impetuously into the fray, exhorting ‘the fleeing Normans to return to battle’. This they did, with the result that the victors soon became the vanquished. Byzantine casualties were said to have been substantial, and the victory was crowned by the capture of the katepano, whom Atenulf later triumphantly paraded before the citizens of Benevento.³¹

Once the news had spread of the Byzantine defeat, most of the coastal cities of Longobardia, including the capital Bari, switched allegiance to the victors. Atenulf’s choice to return to Benevento and ransom the Byzantine captives proved to be short-sighted, for the Normans soon ceased to acknowledge his leadership. While Rainulf Drengot’s contingent from Aversa again observed the suzerainty of Guaimar IV of Salerno, the others chose Argyros: he had reasserted his authority in Bari after the capture of Voioannes, and in February 1042 was formally recognized by the Bariots and Normans as ‘Prince of Bari and Duke of Italy’. In the same month, the Byzantines sent out a replacement for the captured katepano. Synodianos arrived at Otranto, primarily because, aside from Trani in the north, only the southern towns of Apulia continued to acknowledge imperial rule. Aside from unsuccessfully attempting to entice the northern coastal towns to return to the imperial fold, Synodianos tried in vain to form an army of workable size. The emperor Michael V Kalaphates, clearly displeased with the new katepano’s performance, replaced him with a controversial man all Byzantines no doubt felt could solve the problem: George Maniakes, who arrived at Taranto in April of the same year.³²

Judging by the fact that Maniakes assiduously avoided Argyros’ attempts to coax him into battle, the number of soldiers at his disposal cannot have been considerable. The Bariot prince’s army, conversely, seems to have been quite substantial. Hearing of the katepano’s arrival in Taranto, ‘Argyros wrote to the Normans themselves in Aversa and Melfi and all, about 7,000, came to Mottola’. While it is highly unlikely that all of the troops summoned were Normans, the figure of 7,000 is certainly acceptable, but only if taken to reflect the total number of troops available to Argyros, which the same source later records as including Lombards from Bari. Argyros’ decision to muster at Mottola was based on sound strategy. Situated 29km north-west of Taranto and located on a hill 387m above sea level, Mottola is aptly known in Italian as la spia dello Ionio (‘the spy of the Ionian [sea]’). On a clear day, from Mottola’s citadel Argyros could have monitored the Gulf of Taranto and therefore kept abreast of Maniakes’ movements. With Argyros in possession of Mottola, Maniakes, if choosing to march northwest on Melfi, could not traverse the via Appia without the risk of being attacked. Moreover, Taranto was also connected to another major Roman road – the via Appia Traiana – which from Brindisi afforded coastal access to Bari. Argyros clearly needed to prevent his city’s recapture, accordingly choosing to march on Taranto with the hope of bringing the Byzantines to battle. Maniakes, however, refused to give the Bariot what he wanted.³³

While one of the Bariot annals implies that Argyros’ bellicose Norman contingent, frustrated by Maniakes’ refusal to give battle, ‘plundered the entire region of Oria’ as a sort of consolation prize, it is more likely that Argyros shrewdly ordered the attack for a precise reason: to reach Bari from Taranto, Maniakes needed to travel eastwards on the via Appia to Brindisi on the Adriatic coast, then follow the via Appia Traiana northwards to the capital of Longobardia. Since Oria is located around halfway between Taranto and Brindisi, the town was clearly of strategic importance. Indeed, with Mottola and Oria in Argyros’ hands, Maniakes would be effectively ‘boxed in’ at Taranto. While the Normans seemed to have been successful in plundering the hinterland of Oria, the subsequent choice to ‘return to their own [i.e. to Mottola]’, can be explained by their failure to capture the town itself. With access to the via Appia Traiana under threat, Maniakes had little choice but to transport his troops to the Adriatic coast of Longobardia by sea.³⁴

Disembarking on the east coast in June of the same year (1042), Maniakes recaptured Monopoli. As the coastal town lay 40km south-east of Bari, evidently Maniakes aimed to use it as an advance post from which to recapture the capital. However, rather than advancing immediately on Bari, he instead marched inland in the same month, capturing Matera. The town’s location is interesting and says much about the Byzantine commander’s strategy. Matera is centrally located between Mottola (36km east) and Montepeloso (32km north-west), and the latter town was still in Byzantine hands. Possession of Matera not only guarded the mountainous approach to Montepeloso, from where imperial forces could venture further in a north-westerly direction to attack the Norman stronghold of Melfi, but also applied pressure on Argyros’ garrison at Mottola to the south-east. Moreover, as the crow flies, Matera is 55km south-west of Bari, and accordingly Maniakes could threaten the capital from this direction should the coastal advance prove unsuccessful (or Montepeloso, from where Guiscard advanced on Bari in 1068).³⁵

No doubt hearing of Maniakes’ retributive actions against the citizens in the recaptured towns – around 200 were executed at Matera – the citizens of Giovinazzo (18km north-west of Bari), repledged their allegiance to the empire. The coastal strategy appeared to be working, for with the return of Giovinazzo and Monopoli to Byzantine control, Bari was effectively caught in a vice. Accordingly, in the following month (July), Argyros, with a force of Normans and Lombards from Bari, besieged Giovinazzo, capturing it on the third day of the siege. While the town was stripped of all moveable goods, its undoubtedly relieved people ‘were liberated from the hands of the Normans’ by Argyros. Much less fortunate were the members of the Byzantine administration (or garrison?), sixteen of whom were killed. Buoyed by the success of this siege, the Bariot prince resolved to try another fortified coastal town in the vicinity: Trani (42km north-west of Bari). Arriving before its walls in the last week of July, Argyros began a siege that was to last thirty-six days. Interestingly, various siege engines were deployed, although only one of them was described in two annals of Bari: one referred to ‘a tower … like none seen by human eyes in modern times’, and the other a ‘lofty wooden tower’. While these descriptions are hardly precise, it seems likely that Argyros’ engineers built a Byzantine elepolis (‘city-taker’), a mobile siege tower which, on various occasions in the eleventh century, was armed with stone- and flame-throwing devices. Roman armies, like their Hellenistic and Greek predecessors, had for many centuries employed various types of siege engines – also collectively known as ‘city-takers’ – maintained by a corps of engineers present in field or mobile armies. For example, the future emperor Nikephoros Phokas had ‘city-takers built by picked engineers’ before the walls of Chandax (Iraklion, Crete) in 961. These machines, however, were other types of siege engines (e.g. catapults), but in specific regard to siege towers, the emperor Romanos IV had large ones constructed at Mantzikert in 1071, which an eyewitness chronicler noted were large enough to provide ‘a thousand’ with protected access to the city’s walls. Not only did Argyros have a Greek name, but he had also spent much of his life at Constantinople (1011–29). This background, coupled with his control of imperial Bari, means that Argyros’ familiarity with, and access to, Byzantine siege technology is beyond doubt.³⁶

Just how Byzantine in outlook Argyros really was is evident when considering his next action. Upon receipt of a letter in late August 1042 from the newly-installed emperor Constantine IX Monomachos (r. 1042–55), which granted him high court titles such as patrician and magister, Argyros promptly ordered the siege engines to be destroyed. As Constantine ‘ordered Argyros to apply himself to converting the Gauls’, it is likely his letter mentioned the decision – apparently brought about by the machinations of Romanos from the powerful Skleros dynasty – to dismiss Maniakes from office; a guaranteed cause for war given the general’s bellicose disposition. The emperor may have had an additional goal in mind: the suppression of an increasingly dangerous Serbian revolt in the province of Illyria (Albania), which had raged unchecked since 1040. Indeed, at around the same time Argyros received his epistle, another was sent to the commander of the army of Dyrrachion (Dürres, Albania), Michael, ordering him to muster not only his troops but also ‘those armies of the neighbouring themes subject to him’. Although Longobardia was administratively separate from Dyrrachion, Constantine may have intended troops from this province to help Dyrrachion’s general suppress the revolt. While the Normans were clearly upset at Argyros’ decision to abandon the siege – in a fit of rage, one of Ami I’s sons tried to kill him – the Bariot prince nonetheless appears to have persuaded some of them to serve Byzantine interests (at least until 1043 when ‘he allowed the Gauls to depart’). Whether Normans or Bariots joined the large army in Illyria is unknown, but if any forces were transported across the Adriatic, surely only a few returned: according to Skylitzes, the army was disastrously ambushed in a defile, resulting in the deaths of around 14,000 men and seven commanders.³⁷

Maniakes’ replacement Pardos arrived in September 1042, and unsurprisingly the general’s response was rebellion. Surely Constantine had expected this to happen, but he was nonetheless reported by Skylitzes to have been somewhat troubled by the news. The unfortunate Pardos must have been terrified; he was seized, tortured and subsequently executed. Having himself acclaimed emperor by his troops, whom Skylitzes noted were eager to return home, Maniakes marched to Bari in October, hoping to persuade Argyros and the Normans to support his revolt. His attempt to secure military assistance, however, was in vain. In anger, Maniakes split his forces into four divisions, ordering them to pillage the region and bring the spoils back to his base at Taranto, where he now prepared an invasion of the imperial heartland.³⁸

A few months later, Prince Guaimar IV of Salerno, now styling himself as ‘Duke of Apulia’, led an attack on Bari in January 1043 with the assistance of the Normans from Aversa. Since he seems to have had neither siege engines nor a fleet – critically important given that the city was protected on three sides by the Adriatic sea – the Salernitan prince was forced to abandon the siege after only five days. The Normans of Melfi/Apulia had been under the leadership of William ‘Iron Arm’ of Hauteville since September 1042. He was subsequently acknowledged as count by Guaimar, who granted him his niece’s hand in marriage. The towns subject to Norman lordship up to this point were formally divided among twelve self-styled ‘counts’, all of whom acknowledged Prince Guaimar as their lord. Tellingly, coastal towns such as Trani and Siponto that were still subject to imperial rule were also granted, suggesting that Guaimar wanted to be ‘Duke of Apulia’ in more than just name. Moreover, the fact that Guaimar and Count William of Hauteville established a fortress at Scribla in 1044 indicates that the Salernitan prince was also intending to add Calabria to his title.³⁹

In February 1043, the new katepano Basil Theodorokanos arrived in Bari. Basil sailed with the imperial fleet to Otranto, while Argyros marched there with the land forces. Their goal of capturing Maniakes was thwarted, for earlier that month he had crossed the Adriatic with his army, disembarking at Dyrrachion. Maniakes traversed the via Egnatia, the Roman road extending from the Adriatic coast to Constantinople, fighting a pitched battle at Arnissa (95km north-west of Thessaloniki) against the emperor’s troops commanded by Stephan Pergamenos. Despite routing the opposing army, Maniakes was mortally wounded towards the end of the engagement. Stephen later celebrated a somewhat undeserved triumph at Constantinople, parading through the city on a white charger with Maniakes’ head affixed to his lance, and the latter’s supporters formed part of the triumphal procession, mounted backwards on asses with their heads shaven. While Shepard rightly notes that there is ‘no explicit evidence’ for Normans being present with Maniakes at Arnissa, it is certainly possible that the cavalry division clearly named after the general (Maniakatoi), a contingent still in existence during the reign of Alexios I Komnenos (r. 1081–1108), was formed from those who served with the ill-fated general at his final battle.⁴⁰

Basil Theodorokanos, who would win great renown in a sea battle against the Ros (‘Russians’) in July 1043, was recalled to Constantinople not long after Maniakes’ death. Since his replacement Eustathios Palatinos did not arrive in Italy until late 1045, Argyros evidently reigned as katepano in all but name for about two years. His tenure seems to have been relatively quiet when it came to the southern coastal towns, although those to the north-west of Bari came under increasing threat at the hands of Peter I, son of Ami I. Granted lordship of Trani by Guaimar and the Normans, he first captured the principal towns to its north-west (Barletta) and south-east (Bisceglie), respectively 54km and 34km northwest of Bari. Peter also established two settlements in the hinterland of Barletta (Andria) and Bisceglie (Corato), no doubt to guard the inland approaches to these coastal towns. Since, like Guaimar before him, Peter seems not to have had access to Byzantine siege technology, he evidently intended to bring about Trani’s submission by means of starvation. The fact that he was unable to succeed in this aim can be explained by the fact that Trani could easily be provisioned by sea.⁴¹

This period of relative peace ended abruptly in 1046. Towards the end of the previous year, Argyros had been summoned to Constantinople, where he became one of Emperor Constantine IX’s most trusted military advisers. He even led a successful attack on the forces of the rebel Leon Tornikios in 1047, and the army under his command included ‘a considerable number of Franks and Greeks’. The former were probably the aforementioned Maniakatoi, since William of Apulia noted that Argyros disbanded his cadre of Norman troops prior to leaving for Constantinople. Meanwhile, the new katepano Palatinos busied himself with various tasks in Longobardia. According to a charter which he personally witnessed in December 1045, Palatinos decided to reward an official called Vyzantios with the services and taxes heretofore owed to the katepano by the inhabitants of Foliano, a town near Bari. That Foliano was a fortified settlement surrounded by villages is made plain in the Greek: ‘all the services’ were now to be rendered to Vyzantios by those inhabitants variously residing in the village within the fortress, by ‘the people who happen to be outside of this village’, or ‘elsewhere’. Interestingly, among the four ‘services’ mentioned, were military service and payment of the naval tax. The reason given for Palatinos’ gesture was that he appreciated the resistance shown during Maniakes’ revolt, and against the subsequent incursions ‘of the Frankoi’ (e.g. siege of Bari, 1043). In the following year, presumably on account of Norman presence in the region, the katepano marched from Bari to Taranto. A pitched battle was fought on 8 May; the Normans were victorious.⁴²

In the period leading up to this battle and after it, all was not well among the Normans of Apulia and Aversa. Peter I, son of Ami I, was at loggerheads with the Hauteville family; indeed, the Apulian Normans had split into factions. The tension between the two parties came to a head when Count William died during the winter of 1045/6. Peter, preeminent among the leaders owing to his more numerous territorial possessions, intended to succeed the ‘Iron Arm’. Unsurprisingly, William’s younger brothers Drogo and Humphrey objected, managing to solve the problem by defeating Peter in battle. Drogo was then formally invested with the title of count by Guaimar, in addition to receiving the hand of the Salernitan prince’s (potentially illegitimate) daughter, Gaitelgrima, in marriage. A succession crisis also afflicted the Normans of Aversa: Count Rainulf Drengot died in the middle of 1045, and his nephew and successor Asclettin followed him soon afterwards. Asclettin’s cousin, another Rainulf, now contested the succession with a certain Rodulf who, unlike his rival, was not related to the Drengot family. Pandulf IV of Capua, who had returned from exile in Constantinople, supported Rainulf, while his arch enemy Guaimar, in possession of Pandulf’s Capua since 1038, supported Rodulf. Guaimar appointed Rodulf as count of Aversa, but did so without the approval of the Aversans, who desired a successor from the Drengot family. Additionally, Rainulf was imprisoned. He nonetheless managed to escape soon afterwards and, with Pandulf’s backing, replaced Rodulf as count of Aversa. Rainulf II’s position, however, was now threatened by the recent arrival from Normandy of another nephew of Rainulf I, Richard Quarrel. While initially on good terms with the Hauteville family, Richard was subsequently captured and imprisoned by Drogo, who only chose to release him at the behest of Guaimar when news reached the prince of Rainulf II’s untimely death in 1048. Rainulf II was succeeded by Herman (either the son of Rainulf II or Asclettin), but effective power was wielded by the regent William Bellebouche. With the approval of the Aversans, Richard supplanted William, and by the end of the year 1050 had become the undisputed ruler of Aversa.⁴³

Guaimar’s growing prestige and list of titles did not go unnoticed by the western emperor Henry III (r. 1046–56), who decided to reduce the prince’s power when visiting southern Italy in January of 1047. The emperor’s first act was to remove Capua from Guaimar’s authority, returning it to the ‘mighty wolf’, Pandulf IV. Given that the Capuan prince, with use of Norman mercenaries, had striven since returning from Constantinopolitan exile to undermine his Salernitan rival’s authority whenever possible, Guaimar was understandably upset. Moreover, as Guaimar ceased to use ‘Duke of Apulia’ in his charters after the emperor’s visit, evidently he was told not to use a title which, from the perspective of the western emperor, the prince was not entitled to hold. Henry’s other act was to acknowledge Rainulf II and Drogo officially as counts. In theory at least, Henry III’s investiture of the Norman counts could be taken to mean that both were now free of their subordination to Guaimar – indeed, subsequent Aversan charters recognised the emperor rather than prince – but in reality, for the time being at least, the Normans of Aversa and Apulia remained loyal to him. Demoted but not deflated, Guaimar, with ‘three squadrons of Normans’, recaptured Capua after Henry had left Italy. Pandulf, while planning his next move, hired some Norman mercenaries. One of them, a new arrival from Normandy, would forever change the political landscape of southern Italy: Robert of Hauteville, better known to posterity by his nickname Guiscard (‘the Cunning’). As William of Apulia explained in classicizing verse: ‘His nickname was Guiscard, because neither Cicero was more shrewd nor Ulysses so cunning’.⁴⁴

Guiscard’s arrival and early career in Italy are somewhat sketchy, especially when it comes to chronology, but the sources generally agree in outline. He appears to have arrived in Apulia after his half-brother William’s death in 1045/6, perhaps around the time of Drogo’s accession. Depending on which twelfth-century source is favoured, Robert arrived in Apulia ‘with fifteen cavalrymen’ (William of Malmesbury) or five cavalry and thirty infantry (Anna Komnene). Evidently all land had been parcelled out in Apulia, as Guiscard’s choice to serve Pandulf suggests. But service with the Capuan prince ended soon afterwards on a sour note, and again Robert was forced to petition his half-brother for land. Drogo decided to grant him the hilltop fort at Scribla in the Crati Valley of northern Calabria, which William ‘Iron Arm’ and Guaimar had established in 1044. However, pickings in the area were slim, forcing Robert and his men to live as brigands. Robert later moved his headquarters to San Marco Argentano – Scribla had proven injurious to the health of his followers owing to its marshy environs – building a new fortress on a mountain 428m above sea level. From the summit of this lofty stronghold, Guiscard had a commanding view of the plain below him. It was a more than suitable site from which to make further inroads into Calabria but, in order for this goal to have any chance of success, ‘the Cunning’ needed more men and supplies with which to sustain them.⁴⁵

While Calabria was a predominantly Greek-speaking region, there nonetheless existed small communities where languages such as Slavonic and Arabic were spoken. Slavs of the eastern Adriatic coast had long been economically connected with Apulia, and traders from that region had settled in Bari, while others established a settlement at Devia in the Gargano peninsula. The Slavs whom Guiscard encountered in Calabria may have been originally from Apulia although, as Huguette Taviani-Carozzi has suggested, they probably hailed from the Bulgarian population mentioned in contemporary Calabrian hagiographies. Whatever the case, it is clear that the Slavs Guiscard soon came into contact with were locals and had been for some time, since Malaterra referred to them as ‘having knowledge of every part of Calabria’. Robert was able to attract sixty of them into his service, and his new recruits informed him of a town in the vicinity where great spoils could be gained. Having secured it with their assistance, ‘his [Slavic] infantry were made cavalry’ in reward. Enriched with soldiers, horses and supplies, Malaterra observed that Guiscard was now in a much better position to mount incursions deeper into Calabria.⁴⁶

Robert earned his nickname by capturing the governor of Bisignano, a fortified town 16km south-east of San Marco. Arranging for an ostensibly cordial meeting outside Bisignano’s walls, the unfortunate governor Peter was instead wrestled to the ground by Robert and taken hostage. The captive was kept at San Marco until payment of a large ransom was received, some of which Guiscard distributed among his troops. A series of attacks were subsequently launched against Bisignano, Cosenza, and Martirano – the latter two being respectively 31 and 54km south of San Marco – and while none of these fortified towns were captured, all agreed to pay tribute and provide servitium, which in the contemporary northern French context normally denoted military service. Guiscard later journeyed to Apulia, where he met with the powerful Norman leader Gerard – lord of two towns in the vicinity of the southern border of Benevento, Telese and Buonalbergo – who convinced him to take the hand of his aunt Albérade in marriage. Despite Count Drogo’s initial reluctance, he later consented to the marriage, and Robert, according to Amatus, received 200 cavalrymen (perhaps as a dowry) from Gerard to assist in the conquest of Calabria.⁴⁷

After the Norman victory at Taranto in May 1046, the katepano Palatinos returned to Bari. In his absence, citizens unfavourable to the empire had been treating with Count Drogo’s brother Humphrey. Palatinos barricaded himself in the praitorion, the governor’s fortified palace featuring protected access to the port, either built or strengthened by the katepano Basil Mesardonites in 1011. John Raphael was sent from Constantinople with a division of Varangians, gaining access to the palatial stronghold by sea. While Raphael was able to rescue Palatinos, whom he now replaced as katepano, the Norman-sponsored anti-Byzantine faction remained in firm possession of the rest of the city. Raphael and his Varangians therefore had little choice but to ‘establish peace with Bari … and return to Otranto’. The Normans who had defeated Palatinos at Taranto, presumably led by Drogo, meanwhile seem to have pressed further into the Terra d’Otranto, also known as the Salento peninsula (the ‘heel’ of the Italian ‘boot’). Although none of the available sources make explicit mention of this foray, one of the Bariot annals implied that it did occur: ‘The town of Stira was seized by the Varangians in the month of October [1047], and in the month of December they depopulated Lecce’. Feasibly, then, these towns were captured by the Normans when Raphael was detained at Bari, and the new katepano later recaptured them with a detachment of the imperial guard when the time was ripe. There remains, however, another explanation: there is a brief Greek text which mentions acts of brigandage by the local inhabitants in the Salento peninsula, which André Guillou has dated to February 1045. Raphael’s troops, therefore, may have simply contained what was nothing more than a local problem. However, if Guillou’s date is correct, the fact that it took over two years to deal with the brigands seems unlikely, especially given Raphael’s possession of élite troops.⁴⁸

Until April 1051 Bari remained in the hands of the anti-Byzantine faction, led by the Lombards Adralisto, Romuald and Peter. Unsurprisingly, they went to great lengths to prevent the triumphant return of Argyros from Constantinople, who arrived bearing the title of ‘Duke of Italy [= Longobardia], Calabria, Sicily, and Paphlagonia [= N. Anatolia]’. Despite the rebels’ attempts to prevent Argyros’ entry into Bari, the citizens made their wishes clear: Argyros was in, they were out. While Adralisto managed to seek refuge with Humphrey, Romuald and Peter were captured and deported to Constantinople for punishment.⁴⁹

In the same year (1051), Humphrey of Hauteville succeeded his brother Drogo as leader of the Apulian Normans. Perhaps realising that those who were formerly their mercenaries were now on the road to becoming their masters, some unspecified Lombards throughout Apulia formed a conspiracy, the aim of which was to assassinate ‘all the Normans in a single day’. Naturally, murdering them ‘all’ was hardly an achievable task, but several prominent Normans were nonetheless assassinated, including Count Drogo on 10 August. The Norman chauvinist Malaterra, who prefaced the unfortunate episode with a comment on the innate perfidy of the Lombard people, clearly did not appreciate that the Normans were unanimously despised in the region, primarily on account of their violent, rapacious attacks on Apulian towns and villages to date, most recently in the vicinity of the Beneventan border. There was nothing novel about these attacks near Benevento: after their victory at Montepeloso (1041), the Normans ‘uprooted vineyards and olive groves’, in addition to seizing all the livestock in the surrounding countryside. While the method was clearly detrimental to those afflicted by its application, the end result should not be forgotten: starvation. Indeed, one of the Apulian annals recorded a ‘great famine’ occurring throughout Apulia in 1053 and, while the Normans could not have been solely responsible for it, they certainly did not help matters. Guiscard used the scorched-earth tactic in Calabria, as did Richard Quarrel at Capua and Aquino in the 1050s. The method made sound military sense, but its frequent use by Normans in the 1040s–50s hardly endeared them to the local population. While Drogo and the ‘respectable’ Normans had not been involved in these more recent acts of brigandage near Benevento, as their acknowledged leader the Lombards held him responsible all the same. So too did Pope Leo IX (r. 1048–53), who had earlier summoned Drogo and Guaimar to Benevento in April 1051, rebuking them for the attacks on the principality’s environs. Drogo promised that the raids would cease, even though he clearly had no control over those responsible for them. Unsurprisingly, the raids on Beneventan territory continued; accordingly, anti-Norman sentiment reached fever pitch.⁵⁰

Such events were music to the ears of Argyros, ‘the glorious and most faithful duke and magister’, as Pope Leo later referred to him in a letter to Emperor Constantine IX Monomachos. Argyros successfully petitioned the Pope for military assistance, who subsequently appealed to the rulers of France and Germany. While no suitable explanation was given by Amatus as to why King Henry I of France (r. 1031–60) chose not to provide military support, it is interesting to note that in 1052 he was busy waging a campaign against the Normans in France. Emperor Henry III (r. 1046–56), who was the Pope’s cousin, chose to delay for the time being. His decision to prevaricate may have been influenced by the fact that he had originally sanctioned the Norman attacks on the recalcitrant principality of Benevento. When the Beneventans had refused to submit to Henry and Pope Clement II in 1047, the latter placed the city under an interdict, while the former set fire to its suburbs. As neither tactic worked – indeed, the Beneventans would not acknowledge the suzerainty of the emperor and Pope until April 1051 – the hounds were let loose. The Normans were extremely useful, for without their constant harrying it is unlikely that the citizens of Benevento would have overthrown their prince in favour of the emperor (who later granted the principality to the papacy). Accordingly, committing troops against those who had clearly obeyed his wishes and rendered him valuable service, placed Henry, legally their lord and protector, in a predicament. That being said, given that the attacks continued after Benevento’s eventual submission in 1051, the emperor was now well within his rights to support a military intervention. Hoping to expedite matters, Pope Leo later met with his cousin at Worms in December 1052, where he was promised a sizable force by the emperor. However, for reasons unknown, Henry subsequently reneged on the pledge. Despite this significant setback, Leo nevertheless managed to procure through his familial contacts a contingent of mercenaries from Swabia, whose numbers were significantly bolstered by Lombards from various regions of central and southern Italy.⁵¹

While the Lombard coalition was extensive, it lacked the support of the most powerful among them: the prince of Salerno. Cognisant of the fact that Salernitan support was crucial to the campaign’s success, Pope Leo marched into southern Italy at the head of a force comprising troops from the Abruzzi and Gaeta in the spring of 1052. He pleaded with Guaimar to relinquish his support for the Normans, but the prince defiantly refused, warning Leo that any army sent against them would be annihilated. But trouble was brewing in Guaimar’s dominions. The Amalfitans expelled his subordinate Manso in April of the same year, thereafter recalling their former duke John; a Constantinopolitan exile since Guaimar deposed him in 1039. With the considerable tribute exacted from Amalfi now denied to him, Amatus recorded that Guaimar was unable to hire Norman mercenaries to help defend against the impending invasion, which the Amalfitans soon launched against Salerno by sea. During the process of attempting to prevent the Amalfitan ships from making a successful landing in June 1052, the prince was assassinated by his picked mounted troops – called by Amatus ‘the cavalrymen of Salerno’ – and the first of thirty-six blows the prince received was reported to have been delivered by Landulf, the brother of Guaimar’s wife Gemma. The prince’s corpse was then shamefully ‘befouled’ and dragged along the beach for ‘a considerable time’. The slain prince’s brothers were also targeted, and while one was killed (Pandulf), another managed to escape (Guido). He appealed to the Normans for assistance, which they duly gave. On the second day of the subsequent siege, Salerno capitulated. Spurning the opportunity to make himself prince, Guido installed Guaimar’s son Gisulf as prince. While it would be tempting to link Guaimar’s assassination to his long association with, and patronage of, the Normans, the available evidence does not support such an interpretation. Rather, as Loud puts it, the ‘ramifications of the conspiracy suggest that there was considerable dissatisfaction in Salerno with Guaimar’s rule’. While Guaimar’s death had removed a potential hurdle to the Pope’s impending campaign, Prince Gisulf II’s choice to continue his father’s pro-Norman policy meant that Leo could not rely on troops hailing from Salerno and its dependencies.⁵²

The papal plan was to rendezvous with Argyros’ army at Siponto (57km north-west of Trani), from where they would launch an attack on the Normans. Argyros was forced to transport his army by sea, given that Peter I’s control of both the coast and hinterland around Barletta and Bisceglie prevented access by land. Count Humphrey seems to have had access to reliable intelligence for, as Stanton has recently suggested, it is likely that Peter and the count were waiting for Argyros when he landed. Hence, the victory of Peter and Humphrey’s forces recorded in one of the Bari annals was probably more of an ambush than a ‘pitched battle’. However, as Stanton acknowledges, the entry for this battle is under the year 1052 – that is, in the year before the battle of Civitate. Yet since the latter battle is recorded immediately after the one at Siponto under the same year, it is plausible that the annalist made a mistake. At any rate, on his way to the rendezvous with Argyros, Leo’s forces were intercepted by the Normans near the town of Civitate (54km north-west of Siponto).⁵³

The ensuing battle, fought on 18 June at Civitate (near modern-day San Paolo di Civitate), a town situated close to the River Fortore, began, according to the Pope himself in the abovementioned letter to Constantine IX, when the Normans ‘assailed our army with an unexpected charge’. It is important to note, however, that those favourable to the Normans made no mention of the surprise attack. Yet the likelihood that there was one is supported by the Annals of the Beneventans, which also mentioned an ‘unexpected’ or ‘unforeseen’ assault. The Pope informed the emperor that the charge occurred when the Norman emissaries were disingenuously attempting to secure peace – ‘buying time’, as it were – and although Amatus and William of Apulia mention a meeting between the antagonists prior to the battle, they maintained that the attempts to placate the Pope were genuine. It is difficult to determine whether there was a surprise attack, primarily because those in the affirmative might have chosen to forward this ‘dirty’ explanation in order to make the defeat less embarrassing, whereas those in the negative no doubt wanted to memorialize a ‘clean’ victory.⁵⁴

The majority of the available sources agree that the papal army significantly outnumbered the Norman force. No overall numbers, however, were given: all that can be gleaned is that there were about 300 Lombard mounted troops – presumably from Benevento as the local annals recorded that ‘nearly 300 of our cavalrymen’ were killed – and 300–700 Swabian mercenaries. Conversely, William of Apulia did provide a figure for the Normans: ‘These princes were supported with around 3,000 cavalry and a small number of infantry’. Theotokis, quite rightly observing that the figure of 3,000 heavy cavalry is ‘large by the standard of the period’, has suggested it should be divided in half. However, it is likely that in this specific context William meant an assortment of horsemen, rather than a force composed entirely of heavy cavalry (as was the case in regard to a battle in 1041, when he noted that only ‘a few’ possessed haubercs). Indeed, that this was not a homogenous force of cavalry is suggested by William’s language: while some were equipped with javelins, others, most specifically Guiscard’s picked force kept in reserve, wielded lances and swords. Moreover, the words of Fulcher of Chartres come to mind in reference to a battle between crusader and Fatimid armies at Ramla in 1105: ‘There were five hundred of our knights, excepting those who were not thought to be knights by title but were nonetheless mounted’. Taking this passage into consideration, it is important to note that there were three other sources from which the Normans could have feasibly bolstered their cavalry forces: firstly, they had already recruited cavalrymen from among the Lombard and Slavic populations in Apulia and Calabria; secondly, Malaterra mentioned the presence of mounted squires in combat situations, and from another source it is clear that each Norman cavalryman was attended by one. Therefore, a squadron of, say, 300 cavalry had the potential to become 600 strong if the situation called for it (see Chapter 3). Lastly, while the practice was certainly rare, in times of dire necessity (and Civitate was clearly such a time) the numbers of crusader cavalry forces were variously swelled by rank-and-file infantry and squires mounted on horses (e.g. the former at Antioch [1097]; the latter at Ramla [1101 & 1105]). Guiscard, it will be remembered, had earlier mounted his Slavic infantry in Calabria; accordingly a precedent for such a practice had already been set. Indeed, the fact that William drew attention to the small numbers of infantry could well suggest that many of them had been mounted.⁵⁵

In addition to the imprecision when it comes to the question of numbers, is the substantially different information related by those who devoted more than a sentence to the events of the battle (Amatus, William, and Malaterra). However, the most detailed source is also the most feasible and, aside from his heroic sections focusing on Guiscard (a literary motif known to classicists as aristeia [prowess, excellence]), William of Apulia’s better understanding of warfare once again made all the difference.

The papal army camped on the east bank of the Staina, a tributary of the River Fortore. Obscuring the view between the two armies was what Michele Fuiano, following William of Apulia, called ‘a small hill’, which Stanton has since identified as the Coppa Mengoni. The Normans soon ascended it in order to determine the size and disposition of Leo’s army. Having done so, they prepared themselves for battle. Fuiano wrote that, having ‘taken the hill, they assailed the enemy’, implying that the Normans launched their attack from higher ground. Stanton has subsequently agreed, but with the caveat that only Humphrey’s troops were placed on the hill.⁵⁶ Forming the Norman centre, the count’s men faced the Swabian heavy infantry, and Count Richard Quarrel of Aversa’s troops on the right wing – likely arrayed next to the Coppa Mengoni – opposed the Lombards. As William of Apulia observed:

His [Humphrey’s] brother Robert was ordered to watch over the left wing with the Calabrians, and to be prepared to restore his comrades’ strength by providing quick assistance.⁵⁷

Guiscard’s force was clearly kept in reserve and, as Stanton suggests, perhaps hidden behind the hill. Based on William’s description of Guiscard’s weapons, the reserve force consisted of men armed with lance and sword, and its purpose was to charge into any breach created by Richard or Humphrey’s men, who were armed with javelins.⁵⁸

Count Richard, William noted, opened proceedings by launching his ‘illustrious cohort of cavalrymen’ against the left wing of the papal army. The Lombards, described as being poorly arrayed – or caught unaware if there was a surprise attack – fled and were given no quarter; ‘they were slaughtered by javelins and by the sword’. Meanwhile, Humphrey’s centre rained missiles (tela) from a distance. It has been presumed that these missiles were arrows, yet the multipurpose Latin term (‘missiles’, ‘javelins’, ‘spears’) in this context does not necessarily support such an interpretation (see Appendix). When William used the word to denote arrows in other battle descriptions, he was always specific: e.g. ‘From every side missiles [tela] fly; the entire sky is covered in a hail of arrows’.⁵⁹ Significantly, unlike other battles, William does not mention arrows or archers at Civitate; only javelins, lances, and swords were identified. Humphrey’s men, who like most of the Norman army were cavalrymen, were recorded as loosing tela from ‘a spear’s throw off’ or ‘a distance’ at the Swabian heavy infantry formation facing them. This description very much recalls the ranged tactics employed by cavalrymen at Saint-Sever (962), Olivento (1041) and Montepeloso (1041), not to mention their later usage at Hastings (1066). Count Humphrey presumably intended to create openings with javelins into which lance- and sword-based charges could be made. But again, like Duke William’s later experience at Hastings, the count seems to have been unable to prise open the well-disciplined formation of heavy infantrymen. Creating openings was crucial against such a force for, as Matthew Bennett has put it, since a horse is ‘not a tank’ but ‘a flesh and blood animal’, cavalrymen ‘were not capable of “riding over” a formed body of infantry’. Wielding swords William noted as being ‘particularly long and very sharp’ and capable of cleaving a man from head to toe, the Swabians were a seemingly insurmountable obstacle for the Normans and their mounts. While the number of casualties is not provided by any of the sources, William made it clear that some riders and their mounts were dismembered by the formidable Swabian swordsmen.⁶⁰

Sensing the dilemma faced, and undoubtedly champing at the bit to take part in the battle, Robert Guiscard launched his reserve cavalry, presumably at the right flank of the Swabians given his position on the Norman left. It is here that William, true to the classical model that inspired him, switched from a blow-by-blow account of proceedings to a relatively lengthy section on Robert Guiscard’s military prowess. The Norman horseman transforms into a Homeric hero, smiting with lance and sword those unlucky enough to encounter him. Evidently, however, Guiscard did not take all before him. Close-quarter contact with the Swabians was a dangerous activity, for the Norman hero was apparently unhorsed no fewer than three times. Concluding his epic treatment of Robert Guiscard, William noted an event which changed the course of the clash. Count Richard, having returned from his pursuit of the fleeing Lombards, deployed his cavalry at what must have been the Swabian rear. Despite stout resistance, William opined that ‘the addition of the illustrious forces of the victorious Richard were the main reason for the destruction of the enemy’. Swabian casualties were high; Malaterra observed that ‘almost all lay dead’.⁶¹

So shocking was the defeat for the papacy that some chroniclers chose not to mention it. For example, the entire entry for the year 1053 in a redaction of Salernitan annals stated: ‘Pope Leo has fought with the Normans in Apulia’. Amatus, William and Malaterra insisted that the Pope was placed under benevolent Norman protection at Benevento (79km south-west of Civitate). But the pontiff, while undoubtedly well treated, was kept under virtual house arrest for nine months; it would not be until the following year that he was allowed to return to Rome. While in captivity at Benevento, Leo wrote the abovementioned epistle to Constantine IX in 1054. Eager to strengthen ecclesiastical relations, as well as to reaffirm the alliance against the Normans, the Pope accepted Constantine’s request to send legates to Constantinople. But Leo’s intentions were to be thwarted by the histrionics of his legates, Frederick of Lorraine (present at Civitate) and Cardinal Humbert of Silva Candida – most particularly the latter – and by the intransigence of Constantinopolitan patriarch Michael Keroullarios (r. 1043–58). The fact that Argyros, an adherent of the Roman rather than Constantinopolitan Church, held considerable sway over the emperor effectively guaranteed Keroullarios’ vigorous opposition. Having consequently been issued with a sentence of anathema by Leo’s legates, an action which Keroullarios held Argyros personally responsible for, the patriarch arranged for the imprisonment of the duke’s son and son-in-law. Keroullarios also excommunicated the legates, leading to the so-called ‘Great Schism’ of 1054. Despite this inflamed sequence of events, relations remained cordial with Constantine IX. Indeed, when the legate Frederick became Pope Stephen IX in 1057, he looked to the Byzantines for a military alliance. Unsurprisingly, it was Argyros whom the new Pope approached, sending Abbot-elect Desiderius of Montecassino to speak with him at Bari. However, two events in the year 1058 prevented the alliance reaching fruition: Stephen IX’s death and the conquest of Capua by the Norman hero of Civitate, Count Richard Quarrel of Aversa.⁶²

Although groups of Normans would continue to serve the Byzantine emperors well into the following century, they served not in Italy but Anatolia (modern-day Turkey). Their compatriots in Italy continued to be enemies of the empire and, after the stunning victory at Civitate, the Normans had emerged triumphantly as a wholly independent political force in the region. Cordial relations were maintained with the prince of Salerno for the time being, but the decades-long period of mercenary service in Italy was well and truly over. While the rulers of Aversa and Apulia had been happy enough to receive their titles and land formally from the western emperor, after the victory they would never again acknowledge his essentially nominal authority. If the years c. 1016–58 can be seen as a lengthy period of mercenary service segueing towards the end into a time of emergent self-sufficiency, then those that followed (c. 1059–91) can be labelled as a time of consolidation, sanctioned autonomy and, above all, conquest. But there was still much to conquer: various important coastal towns and cities of Apulia were still in Byzantine hands, most of Calabria remained unconquered, and across the Strait of Messina lay the particularly alluring prize of Sicily. As Malaterra observed of the Normans, ‘if not restrained by the yoke of justice, they are most unruly’. The Normans of France had earlier come under the rule of a determined, ruthless and militarily adept ruler known to history as ‘the Conqueror’; by 1057 their counterparts in Italy did as well. Though Duke William is better known to the Englishspeaking world, his counterpart in Italy outstripped him when it came to international fame and infamy. William might have been ‘the Conqueror’, but Robert Guiscard’s epitaph declared him to be nothing less than ‘the terror of the world’.⁶³