Montreal, from the west. (Courtesy of APAAME)
Siege weapons and the means of defence changed relatively little during the twelfth and thirteenth centuries, yet significant changes to the political environment led to notable developments in the ways that castles were constructed and attacked. The numerical superiority of Muslim armies inspired the construction of larger Frankish castles, made possible by the considerable resources of the military orders and investment by visiting crusaders. Muslim defences also developed, due in large part to the Ayyūbid power struggle and subsequent threat posed by the Mongols to Mamlūk rule in western Syria.
It has long been debated whether the Franks or Muslims possessed a superior building tradition or siege technologies. To support either side, however, arguments have inevitably relied on abstract criteria and the unique nature of each siege has often been neglected. Frankish and Muslim rulers adopted the fortifications, siege engines, fighting styles, etc., best suited to their environmental, political, cultural and social contexts, as well as the skill and resources at their disposal. The considerable variance within these groups of ‘Franks’ and ‘Muslims’ speaks to the importance of context, and is enough to cast doubt on the utility of the binary framework that continues to dominate the way the crusades are studied. For example, the Frankish fortifications built in the kingdom of Jerusalem in the early twelfth century looked different from those constructed around the same time in the county of Edessa, and the Muslim armies assembled by al-Afḑal Shāhinshāh in Egypt looked and fought differently from contemporary forces fighting under Īlghāzī’s Artuqid banner. When considering the spectrum of traditions and degree of interaction among them, it is natural to question the extent to which various parties were influenced by the practices and technologies of their neighbours.
While it is possible to trace the diffusion of certain architectural features from Europe to the Middle East, and vice versa, it is much harder to judge the effectiveness of the defences, or siege equipment, at any particular siege. Very generally, certain patterns can be seen in the ways that sieges were undertaken through the period; however, the success or failure of most seems to have had more to do with local circumstances and contemporary political realities than the siege traditions or technologies employed by the besiegers.
In the nineteenth century, the famed architect and architectural restorer Eugène Emmanuel Viollet-le-Duc remarked, ‘We cannot doubt that the crusades, during which so many memorable sieges were effected, perfected the means of attack, and that consequently important modifications were introduced into the defence of fortified places.’35 This sentiment has been echoed by countless other historians since, some of whom have looked to the crusades as an almost mythical nexus of technological exchange.
As noted above, there is little evidence to support theories that the Franks or Muslims learned about certain siege technologies through their interactions with each other. Most siege weapons were well known before the start of this period, while the development of the counterweight trebuchet appears to have been undertaken by many parties, regardless of who employed this technology first. The weapons chosen and their method of use reflected the resources and style of warfare suited to the different parties, not superior knowledge or ignorance.
Impetuous suggestions that Europeans ‘learned’ how to build significant castles as a result of the crusades remain common. At the same time that the Franks were compelled to construct larger and stronger castles in the Levant, made possible by the growing resources of the military orders, so too were European monarchs financing similarly grand structures, facilitated by parallel trends towards increasing monarchical wealth. Ignoring these underlying factors, some have suggested that participants of the Third Crusade found inspiration or gained new knowledge while in the East. Château Gaillard, the mighty castle overlooking the Seine between Rouen and Paris, built by Richard I in 1196–98, is often highlighted, it being suggested that its design was influenced by that of Crac des Chevaliers. The most glaring issue with this suggestion is that Crac, in the 1190s, looked little like it does now, and nothing like Château Gaillard. Furthermore, Richard did not visit Crac, or any other significant inland castle, nor did any known member of his retinue. Saranda Kolones, built in the thirteenth century, probably by the Lusignan rulers of Cyprus, is one of the few castles with a plan clearly copying that of another castle; although only half the size, and provided with rounded and two pentagonal towers, it is a clear copy of Belvoir, down to its numerous posterns. But if Frankish castle plans were not simply copied or imported, what guided their development: were they the natural product of European architectural traditions, or the result of Eastern influences?
Judgements relating to the superiority of either Frankish or Muslim strongholds are often based on perceptions of influence, particularly those apparent in Frankish castles – were the Franks importing Western building styles or adopting Eastern practices? Emmanuel G. Rey (d. 1916), upon whose work so much later scholarship has been built, pointed to what he interpreted as a mix of traditions evident in the fortifications built by the Franks. After a tour of Frankish castles in 1909, T.E. Lawrence (Lawrence of Arabia, d. 1935) went on to write his undergraduate dissertation on the subject, concluding, ‘In dealing with the twelfth century in the East, Arab influences in architecture may be entirely discounted. Beibars seems to have been the first Arab sovereign to build respectable fortresses.’36 Lawrence had a fairly rudimentary appreciation of the development of these structures; nevertheless, the work was fairly well researched, relying heavily on his own observations, and the challenge he issued – that the architecture of Frankish castles owed little to local innovations and that none was then brought back to Europe – has had a profound impact on subsequent studies.
Later in the twentieth century Robin Fedden (d. 1977) and John Thomson embraced Lawrence’s perspective, suggesting that the Frankish building tradition was superior to those of the Byzantines, Armenians, Turks and Arabs. They poetically wrote, ‘The Frankish architects built with a two-hundred-years’ frenzy, and they built with genius, taste and cunning, leaving the imprint of twelfth- and thirteenth-century France – for theirs was essentially a French venture – strangely and beautifully on the Levant.’37 These sentiments were echoed a decade and a half later by Israeli scholar Meron Benvenisti. More recently, Carole Hillenbrand has associated Frankish castles with those of the Nizārī Assassins and the Armenians, suggesting these were places of refuge for beleaguered minorities. But she has also asserted that the Franks possessed superior technological skills, characterizing Muslim castles as inferior: ‘Crusader castles were built to withstand siege; Muslim castles were not.’ So although ‘the crusader castles were manifestly superior in design and execution’, Muslim rulers and architects had few reasons to borrow from the Frankish arts of fortification.38
In opposition to these views, archaeologist and historian of architecture Wolfgang Müller-Wiener (d. 1991), saw local traditions as the most influential, pointing out that most early Frankish strongholds were built prior to the arrival of the Franks or were constructed by local residents using local materials. He suggested that as time progressed, this influence waned and was overshadowed by the ‘importation of stylistic elements from Western Europe’.39 This is particularly true of the region north of the kingdom of Jerusalem.
Rey’s fairly centrist position found new breath in T.S.R. Boase (d. 1974), who looked favourably on the Muslim work at sites like Kerak and Ṣubayba, observing that the Franks ‘borrowed eclectically from the west and the east, from the present and the past, and they learned from a prolonged and rarely broken experience’. Boase also pushed back against the notion that there were two concurrent Frankish building traditions in the thirteenth century: the Western-influenced ‘Hospitaller’ style, which was more complicated and favoured smooth masonry and rounded towers, and the Eastern-influenced ‘Templar’ style, which was more heavily influenced by the Byzantines and made regular use of quadrangular towers and bossed masonry.40 This was a notion put forward by Rey and picked up on by Lawrence, and it can still be found in many books on ‘crusader’ castles. What Boase and others have since acknowledged is that although this framework holds up well when looking at a handful of impressive castles, it quickly becomes far less persuasive when a broader range of structures is examined.
Historians Joshua Prawer (d. 1990) and Hugh Kennedy, as well as archaeologist Adrian Boas, have also endorsed this notion of a nexus of interactions and influences. Perhaps the most insightful, although already dated, examination of this debate was made by Denys Pringle, fittingly included when he re-edited Lawrence’s thesis in 1988. In the introduction, Pringle emphasizes one of Lawrence’s primary insights: that tower-keeps, the most common defensive structures built by the Franks in the kingdom of Jerusalem in the early twelfth century, were imported from Europe. But, to quote Pringle, ‘it seems unlikely that any definitive answer to the East-West influences in medieval castle-building will be possible, until Crusader, Armenian, Muslim, Byzantine and, one should add, Italian and perhaps Spanish castles and town defences have been subjected to the same kind of scrutiny that English, Scottish, Welsh and French ones have undergone in the last seventy-five years’.41 Pringle has since updated this in a study that highlights the contributions of Armenians.
Many nineteenth- and even twentieth-century examinations of crusader castles were plagued by certain biases, often a prejudice that praised a European building tradition or that of the Byzantines. Others have sought to explain the development of Frankish fortifications as conforming to an overly simple model of development, often ignoring regional, topographical, economic and various political influences. As Lawrence wrote, ‘To consider the Crusading castles in their chronological order is extremely difficult: they are mainly a series of exceptions to some undiscoverable rule.’42 Attempts to define that rule have, to date, proven futile.
Tower-keeps were one element that the Franks certainly imported. Like their early stone counterparts in Europe, these were centres of local administration and defence. The rural towers built in the Levant typically had two storeys, but, unlike contemporary European examples, the entrance to most was located at the first level (ground floor) rather than the level above. This is indicative of the different threats that these strongholds were expected to face. Towers like these were not strong enough to keep out a large Near Eastern army for an extended period of time, but while an elevated entrance provided protection against the forces of a neighbouring European baron, a ground-floor entrance, which was more convenient, was strong enough to resist the less organized robbers and bandits of Palestine. In both Europe and the Levant, these towers often formed the nuclei of larger castles as defences were gradually developed and expanded over time.
The declining popularity of keeps in Europe during the thirteenth and fourteenth centuries may have been influenced by the designs of certain Frankish strongholds. Dating to the late twelfth century and early thirteenth, a number of large Frankish castles, including Kerak, Belvoir, Crac, Jacob’s Ford and ‘Atlit, lacked keeps. The shift towards the use of enclosure designs may have been inspired by Eastern traditions, as keeps were not prominent features of contemporary Byzantine, ‘Abbāsid or Armenian fortifications. Although less popular, the construction of keeps continued after the battle of Hattin, as at Chastel Blanc and Montfort, while other keeps, like those at Tortosa and Beaufort, became the focal points of much larger castles. What caused the shift towards the construction of enclosures is unclear, but it began with the resources to construct these much larger strongholds – of the examples above, Kerak was the only one commissioned by a baron.
Frankish castles that made use of a quadriburgium design were almost certainly influenced by the layouts of similar Eastern examples, such as the small Byzantine frontier forts in western Syria and larger Muslim strongholds of Kafr Lām and Māhūz Azdūd along the Palestinian coast. This design was particularly popular in the kingdom of Jerusalem during the twelfth century, especially at sites commissioned by the crown, such as Scandelion, Ibelin, Bethgibelin, Blanchegarde, Burj Bardawīl and Castellum Regis. Whether influenced by Frankish strongholds like these or not, the basic rectangular enclosure plan became popular in France around the reign of Philip II (r. 1180–1223), and a number of castles with this plan were built by Frederick II in Italy. The design gained favour in Britain slightly later, apparently a result of its earlier spread to Savoy; James of St George, the Savoyard master builder, used slight variants of a quadrangular plan when designing castles like Flint, Rhuddlan, Harlech, Conwy and Beaumaris for Edward I in northern Wales. Although it was never completed, Beaumaris replaced Belvoir as what is generally considered to be the most elegantly planned concentric castle. Regardless of whether a keep was present or not, increasing attention was devoted to the development of additional lines of defences in the thirteenth century.
Considerable ink has been dedicated to the topic of concentric defences, their origin, use and development, but the idea of using multiple lines of fortifications was neither dramatic nor new at any point in history, dating back to prehistoric earthworks. Simply adding a ditch ahead of a wall is, by definition, adding a concentric defensive system; however, what is typically meant by this term is the use of an outer line of walls to surround a primary inner circuit. This is exemplified by the famous Theodosian walls of Constantinople, which date to the fifth century. Despite their advantages, concentric defences were expensive, implicitly requiring more than one line of fortifications, so were relatively rare in the centuries leading up to the First Crusade.
The European motte and bailey castle was a quasi-concentric castle: the focal motte, topped with a keep, was often partly surrounded by the outer bailey. Where the bailey secured an approachable front, it might be considered ‘concentric’; otherwise, the motte component might be regarded as a donjon or akin to the citadel of a town. At many later ‘concentric castles’, multiple lines of defences were employed only along fronts that were left exposed by the surrounding geography. This was especially true in the rough region of the northern Levant, where the Franks and Armenians used the topography to their advantage, at times expanding earlier Byzantine outposts but rarely adding a second line of walls along fronts that were naturally inaccessible. To the south, the outer wall at ‘Atlit runs only along its eastern front, yet it is often labelled a ‘concentric castle’, while the famous Theodosian walls at Constantinople similarly stretch along only the western (landward) side of the city.
Securing a vulnerable front with an extra line of defences was a simple way of increasing the defensibility of a stronghold. Truly concentric designs, however, appear most often where the surrounding topography is reasonably level or the castle sits on a conical hill, leaving all sides equally susceptible to attack. Two complete lines of walls can be found at Belvoir and Crac, and may also have encircled Safed and Montreal, but most other large castles had multiple walls along no more than three fronts, as at Arsūf, Montfort, Tortosa, Baghrās and elsewhere. The issue is that the term ‘concentric castle’ has been abused, leaving it with no clear definition but rather vaguely synonymous with ‘significant’ or ‘strong’.
With few exceptions, truly concentric stone castles were built by the Franks before their European counterparts. While Belvoir is the earliest clear example of a single-phase castle of this type, many other smaller strongholds, such as Baysān, La Fève, Latrun, Bethgibelin, Jubayl and many smaller towers, were completely encircled by an outer wall by 1187 at the latest. At many of these smaller sites, it seems the outer wall was constructed to create a larger defensible space, rather than a significantly stronger one. At larger sites, starting with Belvoir, defensibility was more clearly the primary motivation. In northwestern Europe, a similar trend towards adding surrounding walls was under way by the late twelfth century, as was the case at Dover, Gisors and many other sites. Gravensteen, built in 1180, and Château Gaillard, built in 1196–98, commissioned by Philip of Alsace and Richard I of England respectively, were perhaps the first stone castles with concentric designs constructed in a single phase in Europe. Philip and Richard had spent significant periods fighting in the Levant before commissioning these castles, but whether the advantages of concentric designs were impressed on them during their time in the East is unclear. Despite both men’s links to the Latin East, neither Gravensteen nor Château Gaillard bears any significant resemblance to one of the castles they might have seen while on crusade. This is not unlike the way in which the Franks may have been, subliminally or consciously, influenced by earlier concentric defences they encountered in the region.
Montreal, from the west. (Courtesy of APAAME)
One of the European kings most closely associated with concentric defences is Edward I of England, another crusader. Before becoming king, Edward, who arrived in the Holy Land shortly before his thirty-second birthday, spent around 500 days in Acre between May 1271 and September 1272. The city’s defences seem to have impressed the prince, who financed a new tower along the outer line of walls near the northeastern salient, and they might have provided the inspiration behind his development of the Tower of London, where he commissioned an outer wall and barbican upon returning to England. This barbican is suspiciously similar to that built around the same time at Goodrich by William of Valence, Edward’s half-uncle, who had accompanied Edward on crusade, or his son Aymer, suggesting the same master builder may have directed both projects.
Concentric principles were not employed during all of Edward I’s fortification efforts – far from it. During the king’s conquest of Wales, Rhuddlan, Harlech and Beaumaris were all designed with an outer wall, while the equally impressive strongholds of Flint, Conwy and Caernarfon were not. None of these castles resembles those he might have seen in the Levant, but reasonably similar templates were to be found in South Wales by this point. Gilbert of Clare commissioned Caerphilly, a concentric castle, in 1268 and work was probably completed under his son and namesake around the time Edward was away on crusade. Less than 80km to the west, Kidwelly was also rebuilt with a concentric plan sometime around the late 1270s; as at Caerphilly, the outer defences surround a quadrangular inner enclosure. By the end of the thirteenth century, a broader movement towards the construction of concentric castles was under way, to which Edward’s experiences at Acre contributed.
Despite the increasing popularity of concentric defences, these by no means rendered all others obsolete, nor was it ignorance that led to the continued construction of strongholds without outer walls. Most Muslim castles and citadels were designed with a single line of walls – Ṣubayba is a rare example where a large outer bailey was added around part of the earliest component of the castle. Likewise, many of the strongest castles built in Europe during the thirteenth and fourteenth centuries also employed no more than a single line of walls.
Edwardian castles.
Ṣubayba, upper castle from the outer southwestern tower. (Michael Fulton)
Rounded towers had been constructed throughout the Roman Empire and continued to be built in the Near East following the Muslim conquests of the seventh century. In Europe and the Frankish Levant, however, they were rarely employed again before the start of the thirteenth century; exceptions include the twelfth-century rounded towers at Saone in the Syrian Coastal Mountains and Châteaudun in France. Rounded towers became more popular in medieval Latin architecture from the time of the Third Crusade, commonly used in the works of the kings of England and France. This design became conspicuously more popular at the same time that larger castles became more common; it seems it was the greater resources of those who financed these projects that led to the use of this slightly more complex tower shape. Regardless of where they were first built by Latin architects, their growing popularity in Europe during the early thirteenth century, particularly in France, probably contributed to their increasingly frequent use in the Levant. In part, this was due to European financing. As Müller-Wiener and many others have noted, European architectural influences are quite clear at most large Frankish castles dating to the thirteenth century.
Machicolations
Machicolations were one of the few elements that can clearly be seen to originate in the East and travel via the Latin principalities to Europe. The earliest examples appear in Umayyad architecture and can be found at some ʿAbbāsid and Fāṭimid sites. Some of the earliest Frankish/European machicolations (discounting simple murder holes, which have an earlier origin) were of a recessed slot or buttress style. These can be found around the top of the keep at Château Gaillard in France, and the earlier northern inner tower of Crac, where they served as latrine chutes rather than defensive elements. Box machicolations, and similar variants that were built out on corbels, were widely used by the Franks from the start of the thirteenth century. These gradually made their way across the Mediterranean and found widespread use in southern Europe in the fourteenth century.
Prior to the introduction of stone machicolations, some European strongholds had employed hoarding, which similarly allowed defenders to cover the walls below them. It was probably advances in artillery technology, to which hoarding was particularly vulnerable, that led to the adoption of machicolations in regions where stone was readily available. In parts of northern Europe, where timber was abundant, there remained a preference for hoarding well into the fourteenth century. Machicolations remained important defensive elements until they were eventually made redundant with the introduction of more effective siege guns in the fifteenth century.
Fort Saint-Andreé (Villeneuve-lès-Avignon), main gate. (Michael Fulton)
Embrasures were a common feature of Byzantine fortifications in greater Syria. They were seldom employed in Europe before the First Crusade, but became common as fortifications were increasingly built of stone rather than wood. By the mid-twelfth century, Europeans and Franks, as well as their Muslim counterparts, frequently employed embrasures in towers. The use of lines of embrasures, typically accessed by casemates, along stretches of curtain walls appears first in the East, but this quickly spread to Europe. Lines of mural embrasures were employed when Edward I rebuilt the western section of the Tower of London’s main wall, and at castles such as Caernarfon and Beaumaris.
The construction of lines of mural casemates and embrasures may have been inspired by the earlier use of double-level battlements. This style of parapet had been employed by the Byzantines prior to the crusades and most twelfth-century Frankish castles that made use of a double-level parapet, such as Jubayl and Saone, were located in regions where Byzantine influence remained significant. From the Levant, the use of double-level battlements also spread to Europe, where they were employed, particularly in the south, from the late thirteenth century.
The talus is another feature that may have spread to Europe from the Levant, or may have been developed simultaneously as a response to similar threats. The Frankish glacis at Kerak, talus at Belvoir and Nūr al-Dīn’s glacis at the north end of the citadel of Shayzar date to the mid- to late twelfth century, as might some of the revetment on the sides of fortified tells predating Ayyūbid rule. In the early thirteenth century, the outer towers of ʿAtlit and keep of Montfort were given battered bases, while a more dramatic glacis was employed when the southern and western fronts of the inner enceinte of Crac was refortified. But perhaps the most iconic talus, that at Caesarea, was added during Louis IX’s visit in the 1250s.
In Europe, the keep of Château Gaillard is battered and the western towers of Chinon, also dating to the late twelfth century, were provided with taluses. Battering the bases of towers became a relatively common practice in Europe in the thirteenth and fourteenth centuries. Some, like the tour du Moulin at Chinon, Marten’s Tower at Chepstow and the towers of the outer southern gate and those along the wall to the east at Caerphilly, rise from pyramidal bases into round or polygonal towers, a design not found in the Latin East. The inner southern towers built at Crac by the Franks and the outer southern tower at Margat built by the Mamlūks are the closest comparables, but these rounded towers rise from a surrounding glacis, rather than from a base of a different shape.
The portcullis appears to have been a defensive feature imported from Europe, where it was employed sparsely in the twelfth century. Most early examples are found in England, but from about the reign of Philip II of France the portcullis saw increasing use in France, Britain and the Netherlands. This trend extended to the Latin East, where portcullises became more common in the thirteenth century, coinciding with the increasing scale and sophistication of castles. Although employed by the Franks in many of their large strongholds, portcullises were rarely used by neighbouring Muslim rulers, making the presence of portcullis slots at Ṣubayba intriguing.
Chepstow, Marten’s Tower and straight eastern entrance. (Michael Fulton)
One of the more obvious differences between the strongholds built in the Levant and contemporary fortifications in Europe was the use of spiral staircases. These were common in Europe but so far have been found only at Crac des Chevaliers and Montfort in the Frankish East. They were also used when the Mamlūks expanded the southwestern tower at Ṣubayba, providing access to the two new sub-levels. According to traditional thinking, spiral staircases were designed so that attackers would be forced to ascend or descend in a clockwise direction, permitting defenders to meet them while moving counterclockwise. This allowed a defender to swing a weapon in his right hand across his body, towards the centre of the staircase and any attackers he might meet, while forcing an attacker to resort to an awkward backhand motion. Accordingly, it is possible to postulate the direction from which it was believed attackers were most likely to enter a staircase – would they storm it from the bottom or gain entrance from the top?
Although far from the most celebrated defensive features, the design and placement of staircases were important. If attackers managed to climb up to the top of a wall, they needed a way to get down or were otherwise stuck there; or, if they broke in through a gate or breached a wall, they needed a way to get up to engage the defenders along the parapet. Almost all staircases built by the Franks and Muslims in the Levant were straight and most were constructed within the thickness of tower walls, providing access between the upper and lower levels. A gate often provided access to the adjoining parapet from the upper level of many mural towers. This type of arrangement forced an attacker to gain entrance to the tower in order to change levels.
In addition to having their entrances on different levels, the floors/ceilings of Latin tower-keeps built in Europe and the Levant often differed. In Europe, towers were usually divided by wooden levels, which were cheaper and simpler than stone vaulting. Presumably owing to the general shortage of suitable timber in the Levant, and the local tradition of building with stone, most Frankish towers were built with vaulted ceilings at each level, providing a stone floor for the level above. The regular use of fire as a siege weapon might also have incentivized the use of stone, but this would appear to be a point of encouragement rather than a cause.
Despite the common use of bent entrances in the Levant, these never caught on in Europe to the same degree; the majority of European castle and town entrances remained straight. The preference for straight gates in Europe seems to have been influenced by factors of convenience: it was much easier to bring a wagon or ride in ceremony through a straight gate. This was perhaps a greater factor in Europe because these structures remained the primary residences of the nobility, whereas many Frankish nobles in the Levant came to reside along the coast, entrusting their strongholds to castellans or selling them to the military orders.
The availability of reliable and complete information is the greatest challenge to modelling trends related to the sieges of this period. Despite the number of surviving contemporary accounts, there were probably numerous sieges that went unreported, suggested by the regularity with which sieges, especially smaller ones, are found in some accounts but not others. Contemporaries had a tendency to mention sieges where the side with which they identified was victorious, and more generally to inflate the significance of some successful minor actions and omit or downplay failures. For example, Frankish sources assert that Antioch was besieged by a Mamlūk force in 1262, while Muslim sources unanimously present this as a raid, which resulted in the sack of the port of St Symeon. Similarly, William of Tyre provides a vivid description of Saladin’s attack on Beirut in 1182, while contemporary Muslims present this as a far less significant event.
Even when sources agree that a siege took place, the duration of the engagement is rarely stated; in exceptional cases where an exact figure is provided, it is usually found in only one original account. Sieges are more often described as having started in one month and ended in another, or such a rough sense of timing can be deduced by examining surrounding events. Although helpful, this is far from ideal when trying to discern if a siege that began in April and ended in May lasted less than a week or the better part of two months. Sometimes it can be difficult to determine in which year a siege took place, let alone precisely how long it lasted.
With this in mind, some 342 fairly dateable sieges involving parties influenced by the presence of the Franks can be identified between 1097 and 1291 (see pp. 286–99). The region considered stretches from Cilicia across to the drainage basin of the Euphrates north of Raqqa (as far as Sinjār, Ḥiṣn Kayfā, Mayyāfāriqīn, Mardin and Amida, but not as far as Mosul), and down through western Syria, Palestine and Egypt. Although Frankish armies did not penetrate the eastern reaches of the Jazīra, many who came into direct contact with the Franks hailed from this area. Īlghāzī, Zankī, Saladin and certain Ayyūbid princes all divided their attention between affairs in various parts of this broader region. Of these 342 sieges, the outcome of all but one is clear – the exception being Joscelin I’s siege of the tower at Tell Aran, during which he was injured and never fully recovered.
Due to the bias and often politically charged vocabulary of contemporary sources, a qualitative judgement is often required to distinguish a ‘siege’ from an ‘attack’ and a ‘sack’ from an immediate or otherwise fairly peaceful surrender. A siege here is considered to be a concerted attack against a fortified position from which determined resistance was offered, at least initially. Accordingly, the siege of Sidon by the Mongols in 1260 might be considered an attack or a sack, yet references to attempts to take the two castles is sufficient to classify it as a siege. Likewise, Nūr al-Dīn’s repeated moves against Damascus prior to 1154 might be called sieges by contemporary commentators, but are not considered as such here because his forces did not attack the city’s defences or impose an aggressive blockade. An exception has been made in the case of the ‘siege’ of Damascus in 1148, the climax of the Second Crusade. Although no ladders or projectiles ever touched the city’s walls during this event, the scale and significance of the operation are sufficient to warrant the exception and its inclusion.
When considering the principal belligerents of these 342 sieges, 237 (69%) involved Muslim forces, Frankish figures took part in 91 (27%), Byzantine elements were present at 9 (3%), and 13 (4%) were undertaken by the Mongols. On at least two occasions – the siege of Aleppo in 1124–25 and Bānyās in 1140 – Frankish and Muslim besiegers worked cooperatively, while Frankish and Byzantine forces participated alongside each other in at least six sieges, including the siege of Nicaea in 1097, those of John Comnenus’ campaign in western Syria in 1138 and the siege of Damietta in 1169. From the opposite perspective, 189 (55%) of these sieges were directed against Muslim strongholds, 144 (42%) targeted Frankish defences and 9 (3%) involved strongholds under Greek or Armenian control. Overall, the success enjoyed by both Frankish and Muslim forces from 1097 to 1260 was almost identical – slightly better than 62%. During the period that followed, the Franks undertook no sieges against their Muslim neighbours, while the Mamlūks pushed the overall Muslim success rate up to almost 65%.
Belligerents of sieges.
Two general and very obvious patterns can be seen when looking at the distribution and focus of the period’s sieges. First, the frequency of Frankish sieges declined considerably between the twelfth century and the thirteenth: 81% of Frankish sieges took place before the battle of Hattin and just 13% followed the end of the Third Crusade and Richard I’s departure from the Holy Land in 1192. By comparison, Muslim sieges were more evenly distributed through the twelfth and thirteenth centuries. Second, the proportion of sieges that were directed against coreligionists rose. Although some Muslim figures might have seen the Franks as their principal adversaries, or at least professed this, most were at least as concerned with Muslim rivals: 45% of Muslim sieges were directed against coreligionists. Although 54% of these took place before Saladin’s death, as a proportion of the total number of Muslim sieges these increased from 35% before 1193 to 66% during the following century. In other words, although the overall number of sieges undertaken by Muslim forces against Muslim-controlled strongholds declined slightly, they accounted for two-thirds of Muslim sieges between 1193 and 1291. Frankish sieges are again more dramatically divided in this regard. Aside from Baldwin III’s sieges of Mirabel (Majdal Yaba) and Jerusalem, during the brief civil war with his mother, Frankish sieges directed against coreligionists were restricted to Armenian and Byzantine strongholds up to 1193, collectively accounting for just 12% of Frankish sieges during this period. Following the Third Crusade, however, 75% of Frankish siege operations were directed against fellow Franks.
Sieges against coreligionists.
Length of sieges, Frankish/Muslim/Mongol (blue/red/green outline), success/failure (black/grey fill)
The approximate length of 170 of these 342 sieges can be determined. This is just 50%, and considering that many smaller sieges probably took place that we do not know about, considerable room is left for error. Using the available data, the average siege-length across the entire period was about seventy days. Eight sieges were exceptionally long and lasted more than a year: four accompanied Saladin’s conquests in the wake of the battle of Hattin; two were the defining sieges of the Third and Fifth Crusades; the other two were Tancred’s extended siege of Latakia in 1101–2 and the Ibelin-Genoese siege of Kyrenia in 1232–33. (The extended blockades of Tripoli, beginning in 1102, renewed in 1104 and lasting until 1109, and that of Ascalon, from 1244 until 1247, have not been included because a concerted or close blockade does not appear to have been maintained throughout.) When examining the complete range of these sieges more closely, it quickly becomes apparent that, like the development of larger and stronger fortifications, patterns in the length and success of sieges were influenced more by dramatic political events than technological advancements or the exceptional capabilities of particular individuals.
During the First Crusade, the Franks engaged in six sieges with discernible lengths. These lasted an average of seventy-four days (two were concluded in just two weeks while the siege of Antioch went on for more than eight months). The Franks enjoyed success at two-thirds of these sieges, their failures coming at ʿArqa, where Raymond of St Gilles bit off slightly more than he could chew, and Jabala, where another group of crusaders abandoned the siege as rumours circulated that a large Muslim force was gathering to attack Raymond’s party at ʿArqa. To these can be added Baldwin I of Edessa’s successful siege of Sarūj, although it is unclear how long this siege lasted, and Riḍwān’s siege of ʿAzāz, which was abandoned less than a month after it began.
Between the crusaders’ capture of Jerusalem in 1099 and the point at which Saladin crossed the Jordan ahead of the battle of Hattin in 1187, the length of eighty-seven sieges (42%) can be discerned, lasting an average of forty-three days. Of the larger total number of identifiable sieges during this period, Frankish forces took part in 66 (32%) and Muslim forces in 140 (68%), achieving success on 58% and 60% of occasions respectively. This fairly even picture changes dramatically if the period is broken down.
From August 1099 until 1128, the year Ṭughtakīn died and Zankī acquired Aleppo, Frankish forces engaged in forty-one sieges and Muslim parties in thirty-nine, finding success at 71% and 46% of these respectively. During this period, 31% of Muslim sieges were directed against the strongholds of coreligionists, 67% of which were successful, while just 37% of those directed against Frankish towns and castles ended favourably for the besiegers. It was in the north that Muslim forces found particular difficulties. Ṭughtakīn managed to take Rafaniyya in 1115, but not until Īlghāzī defeated Roger of Antioch at the battle of the Field of Blood in 1119 did the rate of successful sieges in the region increase noticeably, due in part to the effective actions of Balak in the early 1120s.
From 1128 to 1154, the period during which Aleppo was the seat of Zankid power in western Syria, the frequency of Frankish sieges fell, from 1.4 per year between 1099 and 1128 to less than 1 every other year. The Franks’ success rate also dipped, dropping to 50%. Meanwhile, Muslim sieges increased from 1.3 per year to almost 2 per year, 63% of which ended successfully. The decline in the number of Frankish sieges was due in part to their earlier conquest of the coast, aside from Ascalon, which removed potential targets. To the north, regency issues persisted in Antioch, with a king of Jerusalem filling this role for around ten of the twenty-six years between 1128 and 1154. Raymond of Poitiers brought strong leadership to the principality between the regencies of Fulk and Baldwin III, but in Aleppo he faced the might of Zankī, who, unlike Riḍwān, could call on resources east of the Euphrates when needed. Muslim sieges of coreligionists rose slightly, to 41% of all Muslim sieges, largely a result of Zankid and Būrid efforts to establish dominance over western Syria.
From 1154, the resources of Aleppo and Damascus came under a single ruler, to which holding Egypt was later added. During the twenty-year period between Nūr al-Dīn’s acquisition of Damascus and his death in 1174, the Franks undertook only six sieges (two coincided with Thierry of Flanders’ crusade in the late 1150s and four took place during the contest for Egypt in the 1160s). Only two of these (33%) were successful. To make matters worse for the Franks, Nūr al-Dīn’s attention also shifted their way. Of the twenty-six sieges prosecuted by or on behalf of Nūr al-Dīn from 1154, only three (12%) were directed against Muslim strongholds (one being the siege of Damascus in 1154), while 68% of all sieges undertaken under his banner were concluded successfully during this period. This figure might have been better had Saladin, who ruled Egypt on Nūr al-Dīn’s behalf from 1169, been successful at more than one of the five sieges he initiated in this position.
After Nūr al-Dīn’s death in May 1174, Saladin forged an even greater empire, effectively harnessing the resources of Egypt to spread his authority over western Syria and much of the Jazīra. In the thirteen years between Nūr al-Dīn’s death and the battle of Hattin (not including the siege of Tiberias, which spanned the battle), Muslim forces conducted twenty-seven sieges, two-thirds of which were successful. Of the twenty-four sieges prior to the summer of 1183, twenty-one targeted Muslim strongholds (seventeen of these were carried out under Saladin’s banner). The final siege in this series saw Aleppo fall to Saladin, from which point he devoted his attention to the Franks. Despite his resources, Saladin was successful at only three of the six sieges he undertook against Frankish strongholds prior to the spring of 1187, capturing the incomplete castle at Jacob’s Ford (1179), the isolated outpost of al-Ḥabis Jaldak (1182) and the administrative tower at Jinīn (1184). Meanwhile, Latin forces initiated just seven sieges during this period and were successful only once (14%), retaking al-Ḥabis Jaldak in 1182. Among the failed sieges, two corresponded with the crusade of Philip of Flanders and another was the independent Sicilian attack on Alexandria in 1174.
Following Saladin’s decisive victory at the battle of Hattin, the three remaining Frankish principalities were pushed to the brink of collapse. From the start of July 1187 to the surrender of Beaufort in April 1190, Saladin orchestrated the sieges of twenty-one strongholds, all Frankish, taking nineteen of them (90%), failing only against Tyre and the Templar tower at Tortosa. Collectively, the average length of these sieges was about 140 days; however, thirteen (62%) were concluded in less than a month, lasting an average of just eight days. The remainder, excluding the siege of Tyre at the end of 1187, which extended for about thirty-nine days, were fairly passive blockades, lasting an average of almost 450 days, or about fifteen months. Meanwhile, the Franks, who were left almost paralysed, could do little in response. The only concerted offensive action during this period was Guy of Lusignan’s longshot attack on Acre, which was sustained through the help of the crusaders who streamed into the Holy Land in response to the loss of Jerusalem. The siege eventually ended in a Frankish victory almost 700 days after it had begun. Aside from Richard I’s subjugation of Cyprus, the only other sieges prior to the end of the Third Crusade were Richard’s siege of Dārūm and Saladin’s failed attack on Jaffa. Richard maintained a perfect siege record while on crusade, but this, like Saladin’s mixed record, was influenced by the political, economic and social context in which he was operating, not just by his ability and resources.
Although broad trends relating to the frequency and success rate of twelfth-century sieges appear to correspond with the evolving balance of power, siege lengths follow less obvious patterns. From the siege of Nicaea in 1097 to Nūr al-Dīn’s capture of Damascus in 1154, the approximate length of 120 sieges can be determined, lasting on average sixty-four days. Nine sieges lasted more than a hundred days (six executed by the Franks, two by Muslim forces, plus the joint siege of Aleppo in 1124–25). Among these, the Franks achieved victory at Antioch (1097–98), Latakia (1101–2), Tyre (1124) and Ascalon (1153), while none of the protracted Muslim sieges ended successfully. At the other end of the spectrum, Frankish and Muslim forces participated in at least thirty-five sieges lasting less than thirty days, sixteen and twenty sieges respectively, including the cooperative siege of Bānyās in 1140, and both found success at about 60% of these.
The length of most sieges remained relatively consistent between August 1099 and June 1187, although the sources reveal little regarding the length of most of Nūr al-Dīn’s sieges. Using the available data, there was a slight decrease from fifty-five days between 1099 and 1127, to thirty-eight days from 1128 to 1154, back up to forty-four days between 1155 and 1173, then dipping to twenty-six days from 1174 to June 1187. Although the number of Frankish sieges declined through the twelfth century, the Franks remained willing to commit themselves to lengthy sieges if necessary, especially when joined by significant groups of crusaders. Like Nūr al-Dīn, Saladin appears to have avoided lengthy sieges when possible; most undertaken by his forces lasted no more than about a month and a half. The siege of Homs in 1174–75, which was conducted largely in Saladin’s absence, was a rare exception. Most of Saladin’s sieges were much shorter, and nearly all resulted in generous terms as he expanded his hegemony across western Syria during the decade following Nūr al-Dīn’s death. The context of the sieges during the six years that followed the battle of Hattin, and their polarized lengths, which were almost all either very short or very long, averaging 135 days, have been dealt with above in the context of negotiated surrenders.
Distribution of twelfth-century sieges.
During the post-Saladin Ayyūbid period, siege patterns were relatively consistent, with about 60% of both Frankish and Muslim sieges ending successfully. Taking 1220 as a dividing point, which roughly corresponds with the end of the Fifth Crusade and al-ʿĀdilʾs death, there were twenty-five Muslim sieges between 1193 and the end of 1220, and an additional thirty before the battle of ʿAyn Jālūt in 1260. In both groups, 60% were directed against fellow Muslims and the same percentage ended successfully. Of the seventy-four total sieges during this broader period, only ten (14%) were undertaken by the Franks: three sieges accompanied the arrival of large forces of crusaders and targeted Muslim strongholds; seven sieges were directed against fellow Franks, five of which were undertaken on Cyprus. These were generally quite long sieges, averaging about 200 days – four continued for more than nine months. Typically happy to leave the Franks to their squabbles along the coast, and fearful of provoking another crusade, Muslim forces besieged Frankish strongholds on only ten occasions during this period, accounting for just 18% of all sieges conducted by Muslim forces in the region. Among the Ayyūbid sieges with determinable lengths, the average was close to three months – this includes three that lasted more than 150 days, while the average length of the remaining eighteen was approximately fifty days.
From 1259, the Mongols’ presence was felt in western Syria. Often portrayed as a purely nomadic force with a limited siege tradition, Hülagü’s army was successful at eight of the nine clearly distinguishable sieges that it undertook in Syria – the one-day attack on Sidon, often characterized as a raid, saw the sack of the town but failed to bring about the capture of either of its castles. Although the Mongols’ numbers were perhaps their greatest siege weapon, they demonstrated their ability to undertake a set-piece siege when investing Aleppo, employing artillery and sappers through the thirty-eight-day operation. But as quickly as the Mongols had swept into Syria, they seemed to disappear, chased out by the Mamlūks. Although they remained a threat, the Mongols showed little ability to take the strongholds occupied by the Mamlūks along their frontier, failing on three separate occasions to take al-Bīra.
Between the battle of ‘Ayn Jālūt and the Mamlūks’ capture of Acre in 1291, the attack led by Prince Edward of England against Qāqūn in 1271 was the closest that the Franks came to besieging a Muslim stronghold. Two short Templar sieges in 1278, launched against Tripoli and Nephin, both of which failed, were perhaps the only ‘sieges’ undertaken by Frankish forces during this final period. The episodes of fighting in Acre by Italian factions and the military orders are hard to classify as sieges given the available information, even though siege engines were employed at times as various parties attacked the quarters of their rivals.
Distribution of thirteenth-century sieges.
The undisputed masters of the region west of the Euphrates during the final four decades of the thirteenth century were the Mamlūks, and their siege record reflects this. Seventeen of their nineteen sieges (89%) were successful: fourteen of fifteen against Frankish strongholds and three of four against fellow Muslims. The two failures appear to have been tests of strength, early efforts by Baybars to size up the defences of Kerak (1261) and Acre (1263). The average duration of the fourteen sieges with discernible lengths was about thirty-eight days, or just twenty-four days if excluding the lengthy siege of Cursat in 1275, which Baybars delegated to subordinates. Testament to the resources that Baybars could bring to bear, Caesarea (1265), Jaffa (1268), Beaufort (1268), Antioch (1268), ʿAkkār (1271) and Montfort (1271) were all taken in two weeks or less. The mighty castles of Arsūf (1265), Safed (1266), Crac (1271) and later Margat (1285) fell between thirty-eight and forty-five days after siege actions began. Tripoli probably fell in a comparable period in 1189, and Acre, the seat of power of the kingdom of Jerusalem since 1191, was taken in fifty-three days.
Although they provide a convenient overview of certain trends, analytics like these are no substitute for close analyses of the contexts in which these sieges were undertaken and the plethora of variables that made each unique. While the strength of fortifications and topography might be easy to gauge, numerical strength, experience, morale and even weather could have a far more dramatic effect on a siege than the defences and siege engines that so often draw focus.