In the period from the late 1040s until the late 1060s the Seljuk Turks under the Sultan Alp Arslan (1063-1073) had made considerable inroads into formerly Byzantine territory in eastern Anatolia. As ruler of Iran, Iraq and northern Syria – nominally for the Abbasid Caliph in Baghdad – Arslan had faced a problem with his unruly Turkmen nomads, whose desire for further booty and conquests could not be accommodated within the Islamic world. And so he had released them against the Christian powers to the north and west, where they were soon able to bring the kingdom of Armenia under his sway – the capital, Ani, was sacked in 1064 – and mount yearly raids deep into imperial lands. Plundering and destroying as they went, sacking several major cities, they had soon created a situation which the regular armies of the empire were only with extreme difficulty able to contain. In a series of campaigns between 1068 and 1070 the emperor Romanos IV (1068-1071) had led his armies in an attempt to bring the Turks to battle and destroy their power, but had been thwarted by their mobility, small numbers, and the large number of incursions which they mounted. The Turks were able literally to ride circles around the much slower moving imperial forces which, as we have noted already, were no longer as well-led or as effective as in the heyday of the Byzantine reconquests under Basil II.
In the winter of 1070-1071 Romanos prepared another major expedition, directed against the Seljuk garrisons which had been placed in the Byzantine border fortresses at Khliat and Manzikert in the east. His intention was certainly to re-establish the frontier defences as far as he could, although whether he also hoped to meet Alp Arslan himself in battle is a moot point. For in March or April 1071, when his preparations were well advanced, the emperor proposed a treaty with the Turkish Sultan, who was engaged upon the siege of Edessa, by which the latter would abandon his siege and the former would return the city of Hierapolis (Manbidj) to the Sultan’s authority, taken by the emperor during his campaign in 1068. This would then enable the Sultan to resume his war with the Fatimid-controlled cities of Syria, in particular Aleppo, removing him from the scene in Asia Minor and leaving Romanos a free hand to reassert Roman authority over the region, and perhaps set up a defensive chain that would hinder or prevent future Turkmen raids. Romanos’s first offer of a truce was reinforced by a second embassy, which arrived at Aleppo in May, more or less demanding the exchange of the towns named in the first offer, and threatening war if no agreement was reached.
Romanos had already left Constantinople, and had begun mustering his troops in late February and March. By the time the Sultan received the second embassy, therefore, he must also have received news of the imperial advance towards his Armenian territories. Abandoning his negotiations with the governor of Aleppo, he seems precipitately to have headed back east, crossing the Euphrates with his own guards and a small retinue, to take command of the forces in Armenia and Mesopotamia and deal with the threat from Romanos.
The emperor’s advance across Asia Minor was accompanied by several events that did not augur well for the success of the expedition. In the first place, he chose to leave the general Nikephoros Botaneiates behind, a competent officer whom, however, he suspected of potential disloyalty. Yet he chose to take with him Andronikos Doukas, eldest son of John Doukas, one of his rivals for the imperial throne, who certainly was disloyal to him. At the same time, Romanos began to distance himself from his troops and his officers, insisting eventually on establishing a separate baggage train and encampment for the imperial party, refusing to share in the hardships of the campaign but taking with him elaborate materials and equipment for his own accommodation. On the march from the Halys river to Sebasteia his guard of German mercenaries, the Nemitzoi, suffered some casualties at the hands of the local population, whose property and lands they had pillaged en route, and then complained to the emperor about their treatment. Romanos was forced to cow them into submission by threatening them with force from the other units encamped around them, and then dismissed them to a distant posting, away from the immediate area of the campaign; while at Sebasteia the emperor had to deal with possible opposition from the local Armenian population, who had been accused of making common cause with the Seljuk raiders during the previous year.
By late June the imperial forces had reached Erzurum (Theodosioupolis), where a decision had to be made as to which direction the army should take and how exactly the emperor wished to implement his strategy. There appears to have been some dissension. On the one hand, some of the generals suggested he move on, try to outflank the sultan and take the war into Seljuk territory, and bring him to battle. The emperor’s last report of the Sultan was that he had departed in haste, indeed in a panic, from Aleppo, and it was assumed that he would need to go first to Iraq to raise his forces there before he could deal with the Roman attack. Others, including the generals Joseph Tarchaneiotes and Nikephoros Bryennios, argued rather that the emperor should wait, fortify the surrounding towns and strengthen their garrisons, lay waste the countryside to deprive the Turks of necessary supplies when they approached, and await events. The latter course of action seemed inappropriate, the more so since the army was clearly in danger of running out of supplies if it waited in one place for too long, and so the order was given to move on.
An estimate of the forces at the emperor’s disposal at this point is difficult, but it is clear that he had by no means denuded the empire of troops for this campaign. A detachment of Varangians was certainly left in the imperial palace; a detachment of Frankish heavy cavalry under their leader Krispos had been left at Abydos; and since both the Normans and the Hungarians were a threat at different points in the Balkans, the garrisons in these regions will certainly not have been reduced. It may have been to these areas that the unruly German contingent was posted. The field troops in Syria, and in particular those under the doux or military governor of Antioch remained substantial, as later events demonstrate, even though some reinforcements to the emperor’s field army had been sent from Syria. The contemporary sources also make it clear that, after the battle, considerable numbers of troops were still in their garrisons and posts throughout Anatolia.
Of the units which accompanied the emperor, some are mentioned in the sources by name. The Franks under Roussel de Bailleul, who may have numbered 500 or more; the five tagmata of the West, each of perhaps 1,000 men; a number of detachments of Oğuz (Turk) mercenaries, whose exact number is unknown; troops from Bulgaria; indigenous eastern thematic tagmata from Cappadocia, and probably also from Koloneia, Charsianon, Anatolikon (units from Pisidia and Lykaonia are mentioned in earlier campaigns for the 1050s), Chaldia (Trebizond) and Armeniakon, again perhaps in each case as many as 1,000 strong, but of dubious quality in many cases. Units from Cilicia and Bithynia are also mentioned in one of the sources. Tagmata from the field armies of Syria were also present, although it is not certain how many. In addition to these troops, there were also substantial numbers of Armenian infantry units. Where these were drawn from is unknown: possibly from the regions around Sebasteia and Theodosioupolis, as well as from the Syrian forces. In addition, there was a substantial body of Pecheneg mercenaries and allies and some units from allied or vassal states in the Balkans. Of the palace regiments, the soldiers of several other units, the Hetaireai, the Scholai, and the Stratelatai made up a reserve division, and there were detachments of Varangians also present. The total of the forces thus assembled can only be guessed at. The medieval Islamic sources reckon it at anything from 100,000 to 300,000, both preposterous in view of both the demography of the empire at the time and the logistics involved. But a grand total of perhaps 40,000 may be reasonable, and would certainly explain the emperor’s apparent confidence and the fact that the Turkish Sultan was clearly worried by the size of the threat.
The emperor’s plan seems to have been to take both Manzikert and Khliat, which lay somewhat to the south on the western shore of Lake Van. But he was completely misinformed of the movements of Alp Arslan and his troops. The latter, in fact, had not returned to Iraq at all; rather, he had marched directly towards the Armenian border by way of Amida and Mosul, then on to Khoi just to the north of Lake Urmia. There his vizier had proceeded to Azerbaijan to raise further troops, while he himself, having collected some 10,000 cavalry from his allies and vassals en route, had by now assembled a force of some 30,000 horsemen. Thus while Romanos thought that the Turkish leader was some way away, he was in fact just over 100 miles distant, with his scouts covering and reporting every move made by the Roman emperor. Already, therefore, Romanos was at a disadvantage, even if his forces were in a substantial majority.
From Erzurum/Theodosioupolis the emperor advanced eastwards. The troops were ordered to collect enough provisions for two months – a very considerable amount that necessarily entailed the use of large numbers of pack-animals and, possibly carts, slowing the army down somewhat. A substantial body of the Pecheneg allied force, closely followed by the Frankish troops under Roussel, were ordered ahead to the region around Khliat, which Romanos clearly perceived as the more difficult of his two first objectives, with instructions to collect fodder and provisions, prevent enemy damage to the harvest and, presumably, to secure it for the imperial advance. The emperor must have continued his march east along the same route, before turning south to cross the Araxes, and then east, either along the valley of the Murat Su, or a little further south (which is the route the forces under Roussel will have followed) at Muş (Taron), towards Manzikert itself.
Before reaching this first objective, he detached a further substantial force under Tarchaneiotes, with orders to assist Roussel in taking and garrisoning Khliat. According to Attaleiates, this included the elite of the army, most of the better and more battle-hardened units, including the Varangians and some of the Armenian infantry from the field forces under the doux of Theodosioupolis. He also notes that the troops remaining to Romanos were now fewer than those he had sent off to Khliat. We may surmise that, after the separation of these various detachments, it is likely that the forces remaining with the emperor at this point numbered only some 20,000 or so, and were therefore – contrary to Romanos’s expectation and assumptions – barely superior in numbers, if at all, to the main Turkish host.
The detachment of the troops under Roussel and then Tarchaneiotes, based on the false assumption that the enemy would approach from the south or east of Khliat and was still some distance away, proved to be a major blunder. Unaware of the closeness of the Seljuk forces, which were by now approaching both Khliat and Manzikert from the east, the two Roman commanders were suddenly confronted by what seemed to be a substantial enemy force. What happened next has no explanation in the sources, for both forces appear simply to have about turned and moved with great haste away from the Seljuks, whom they seem neither to have reconnoitred, nor to have reported to the emperor, a mere 50km or less to the north. Both divisions simply marched off towards Melitene/Malatya on the Euphrates, and took no further part in the campaign.
There are two possible routes between Manzikert and Khliat, and it is likely that the troops under Tarchaneiotes took the slightly quicker, more easterly road, across the plain stretching south-eastwards from Manzikert, down towards the Süphan Daği. Since they were clearly able eventually to reach Melitene (Malatya), they must certainly have got as far as the junction with the westerly road down from Manzikert, just north of Khliat, where they perhaps also joined up with the troops under Roussel. It can, therefore, only have been at this point that the Seljuk troops appeared in strength, compelling the imperial forces to turn north along the westerly route back towards Manzikert, before turning westwards after a few kilometres, back through Taron, Harput and on to Malatya. It is likely that they may also have found this route back to the north, along which they would march to rejoin the emperor, cut by Seljuk forces, which would explain the decision to turn westwards, although a few fast riders might still have been able to get through to warn the emperor. Whether or not treachery played a role is unclear; more probably their understanding of the overall strategic situation encouraged their response, and the information required to assess this is simply not at our disposal.
Whatever the reason for this loss, the emperor was now deprived of some of his best and most reliable units. Unaware of the events to the south, he proceeded to Manzikert, which capitulated without a struggle, the garrison being released without punishment. Romanos set up his camp outside the fortress and on the banks of a small tributary of the Murat Su which flowed down from the Süphan Daği. The city was located on the northwestern edge of a roughly quadrangular rocky steppe region which stretches for some ten miles along a northwest/south-east axis, before rising gradually towards the foothills of the Süphan Daği, north-east of Khliat. This was an area thoroughly known to the Turks, but less familiar to both Romanos and his commanding officers, a fact which again proved to be a significant disadvantage to the Romans.
On the morning after the occupation of Manzikert, probably Wednesday 24 August, the emperor was informed that some of the detachments sent out to forage for supplies along the route south – towards Khliat, in fact – had been attacked and driven back by Turkish warriors. The commander of the left wing, Nikephoros Bryennios, was ordered to chase these raiders off; but in dealing with what turned out to be a much larger force than expected, soon found his units lured into ambushes and surrounded, so that he was compelled to withdraw to the camp. The emperor, still believing that this could not be the main Seljuk army, now sent out a much stronger force of cavalry under the Armenian commander Nikephoros Basilakes, the doux of Theodosioupolis. But Basilakes, ignoring the tenets of Byzantine tactics in respect of feigned retreats and the tactics of nomadic peoples, allowed his forces to engage in an uncontrolled pursuit of what he took to be retreating Seljuk troops. In fact, of course, the Turks had set an ambush, and not only were Basilakes’s troops cut to pieces and driven back in flight, but he himself was captured.
Realizing that the Turks were present in greater strength than he had thought, but still entirely oblivious of Arslan’s nearness, Romanos then ordered the whole left wing division to advance and drive the Turks off. But the latter had now retreated into the foothills surrounding the plain, and it was only when Bryennios reached the spot of the ambush on Basilakes that he learned, from a wounded survivor, what the true situation was. By this time – probably towards the middle of the afternoon – the Turks had arrived in real strength, and launched a determined attack on Bryennios’s division, attempting to encircle it. The latter sensibly ordered a disciplined withdrawal, covering the movement of his own units by occasional charges, and at one point forcing the Turks into a real retreat. But the latter by now had the initiative, and it was with relief that Bryennios and his troops eventually reached the safety of the imperial camp. Bryennios himself was wounded – reportedly having two Turkish arrows fast in the armour on his back and a spear-thrust in the chest – although he was able to fight again the next day.
The emperor now realized that he was facing the main Turkish force, and that his information regarding Arslan’s movements was clearly inadequate. The army was readied for a general assault but when drawn up for the advance the Turks had withdrawn completely into the hill country away to the south-east. In spite of scouts sent out to locate them, the whole enemy force had moved out of range, and the emperor was forced to withdraw to his camp.