Al-Ḥabis, from the great temple of Petra. (Michael Fulton)
Every stronghold was unique. The shape, design and features of each were influenced by the purpose of the structure, the surrounding landscape, the resources of the commissioner and the building traditions of those who constructed it. Although many of the small administrative towers built by Frankish figures in the twelfth century are similar, no two are the same. Larger castles provided more opportunities to incorporate architectural features to further enhance their defensibility. But no matter how strong a castle was, it was always necessary for defenders to make adequate preparations, and often incumbent on them to take action to thwart the attempts of besiegers to gain entrance.
Rarely were defences of any design beneficial without defenders, and defenders needed to eat. In the event of a siege or close blockade, defenders would live off nonperishable foodstuffs (including grain) that had been set aside, stored and kept dry in preparation for just such an event. How much was stockpiled depended on estimates of how many defenders there might be and how long a siege might last. In addition to feeding defenders and support personnel, both day-to-day and during sieges, animals also had to be fed. Many large castles accommodated cavalry forces, necessitating huge quantities of fodder for both war horses and service mounts.
Food was important but water was a more immediate concern. Wells were the most reliable sources of fresh water and were reasonably common along the coastal plain. Unlike in Europe, however, wells are rarely found in inland castles, which often crown peaks or ridges high above the water table. A rare and impressive example can be found in the citadel commissioned by Saladin at Cairo: the shaft extends down almost 90m, wrapped by a staircase that winds its way down to the water level.
In some regions, including the Sawād, springs are relatively common and would have provided fresh water for certain strongholds. At Montreal, a tunnel was dug from inside the castle down through the conical hill, upon which the stronghold is perched, to an underground spring below the castle, providing those above with a secure and reliable water source. ʿAtlit was surrounded by the sea on three sides, but there were apparently freshwater springs in the castle’s fosse, while fresh spring water was used to fill a wet moat around the small stronghold commissioned by Tancred at Baysān. Although most strongholds were built near a water source of some kind, most relied exclusively on cisterns during periods of siege.
In most areas of the Levant, rain rarely fell through the summer months, compelling commissioners to incorporate large cisterns into the plans of their strongholds. These were designed to collect rain water during the wet winter months and store it through the remainder of the year. The tanks were coated with lime-based hydraulic plaster and were kept as dark as possible, limiting the growth of microbes. Because water was so important, elaborate collection and channelling systems were developed at some castles. At al-Ḥabis, a small Frankish fort on a rocky peak overlooking the basin of ancient Petra, the cistern is the best preserved element of the castle and would have been the most important – today, the region receives less than 200mm of rain each year. For security, and to ensure access, a cistern was constructed or excavated below most keeps, ensuring it would not be thirst that compelled a garrison’s ultimate surrender. This was the case at small rural and urban towers, such as Le Destroit, Tarphile (Khirbat al-Manhata) and Jubayl, as well as at larger castles, including Safed and Montfort.
Cisterns were rarely a stronghold’s main water source during periods of peace. Towns developed near natural sources of drinking water and castles were often sited in proximity to some kind of stream – a source of both water and possibly income if the current proved strong enough to power a mill. Taking Montfort as an example, four cisterns have been discovered so far in the upper castle: one below the keep, two under the central range (one of which presumably serviced the castle’s kitchen), and one below a large tower at the western end of the upper castle, around which the outer wall of the upper ward was later built. As the castle grew, so too were cisterns added, as the number of people they were required to support presumably also grew. The tanks were filled by the annual winter rains, which today yield more than 750mm of water. Below the castle, on its northern side, a perennial stream, fed by a number of springs, flowed through Wādī al-Qarn. Although inaccessible during a siege, this was probably the castle’s main source of day-to-day water and the power source for a watermill.
Al-Ḥabis, from the great temple of Petra. (Michael Fulton)
Ṣubayba, reservoir. (photo courtesy of Steve Tibble)
Saone, cistern. (nzonumentsofsyria.com)
At some large castles, including Crac, Kerak and Beaufort, a large open water reservoir, known as a berquilla, served as both a defensive obstacle against mining and a working water source, probably used for tasks like watering livestock. While the reservoirs at Kerak and Beaufort were outside the main defences, that at Crac was between the inner and outer walls, fed by an aqueduct that passed through the outer wall.
A stronghold’s primary line of defence was its walls. Whether these were the walls of a solitary tower or a grand circuit of city walls, they were the main barrier against those trying to gain entrance. At almost all sites in the Levant, both Muslim and Frankish, walls were built of stone. Unlike in Europe, where the majority of defences were still made of wood in the twelfth century, stone was the construction material of choice, due to its availability and the reciprocal shortage of suitable timber, and the existing building traditions oriented around its use in the Near East. Brick was used at some sites along the Euphrates, including Qalʿat Jaʿbar, Raqqa and al-Raḥba, but it was very rarely used in the regions to the west and north more directly affected by the influence of the Franks.
Towers, which had been built along curtain walls since antiquity, were designed and positioned with increasing mindfulness through the twelfth century. These were semi-independent fighting platforms, sometimes stronger and typically taller than neighbouring curtain walls, allowing a tower’s defenders to dominate any besiegers who made it to the top of an adjoining stretch of wall. A small doorway, typically lockable from the inside, often provided access from these towers to the tops of the neighbouring walls. At some sites, no such doors were provided, isolating each section of the enceinte should it fall to a besieger, but restricting the defenders’ ability to circulate around their defences. Towers that projected from the line of the adjacent walls also allowed archers and slingers on top, as well as any archers shooting from embrasures in the flanks of these towers, to target besiegers near the base of the walls to either side. For this reason, town gates were often flanked by projecting towers. If mural towers were spaced relatively close together, this added firepower could compel besiegers to attack the towers themselves, rather than enter the killing zones between them.
At the top of most walls was a parapet and fighting platform. The parapet, or battlements, was a thinner wall rising from the outer edge of the main wall, which was often crenellated – the iconic use of alternating solid merlons and open crenels – to provide protection and facilitate an active defence from the top of the wall. The fighting platform, often a wall-walk, or chemin de ronde, allowed defenders to circulate along the top of a stronghold’s walls, helping them to move quickly to confront threats as they materialized along different fronts.
By the late twelfth century, some merlons were pierced by embrasures, providing further protection for archers. Early Ayyūbid examples can be found at ʿAjlūn, and every merlon crowning al-ʿĀdil’s new citadel at Damascus, dating to the early thirteenth century, appears to have been pierced. The large round donjon at Margat and quadrangular keep of Chastel Blanc, built by the Hospitallers and Templars respectively, are crowned with pierced merlons, while there is evidence that these were used by the Templars along at least some sections of the battlements of Tortosa and ʿAtlit. It is hard to determine how widely crenellations of this style were employed, as the battlements of a stronghold were typically the first part to be slighted or dismantled by stone robbers, leaving relatively few merlons in place today. Sections of surviving battlements at Saone and Jubayl, which predate the earliest known pierced merlons, and the similar solid merlons found at Montfort confirm that this simpler, traditional style was not replaced, but remained in use at some sites through the thirteenth century.
Some towers and walls had a double-level parapet: a lower gallery, servicing embrasures or machicolations, with a traditional open wall-walk and battlements above. This provided space for an additional level of defenders without the costs of adding another level to the tower. In the twelfth century, the Franks employed double-level parapets around the tops of some of Saone’s towers and at the central keep of Jubayl. The large circular donjon built by the Hospitallers at Margat, probably dating to the early thirteenth century, received similar battlements. Al-ʿĀdil employed this design feature extensively during his fortification efforts in the early thirteenth century and the Mamlūks appear to have shown a similarly strong preference for this style of parapet, famously building such around the outer southern defences at Crac.
Crac des Chevaliers, western end of the outer southern defences. (Courtesy of Denys Pringle)
In order to provide more positions below the top of a wall from which archers could shoot at besiegers, towers and sometimes curtain walls were provided with embrasures – narrow windows that splayed inwards. It was often impractical to extend embrasures through the full thickness of fortified walls, so to give archers a more comfortable and effective space in which to operate, these were often accessed via casemates – vaulted projections of space that typically extended from the inward face of a wall about half way through its thickness. Some casemates were little more than 1m wide, while others were particularly large, such as those in the outer wall of ʿAtlit, each of which might have accommodated two archers. Elsewhere, casemates were not used, as in the four inner corner towers of Belvoir; at the first level of each tower, an exceptionally deep embrasure extended through the entire thickness (3.5m) of each of the tower’s four walls. To allow archers to shoot at low targets, such as a person approaching the base of the wall below, the bottoms of some embrasures sloped downwards as they tapered towards the outer plane of the wall, allowing for an opening (or loophole) that extended well below the feet of the archer.
ʿAtlit, outer wall embrasure (after Johns).
Ṣubayba, casemates and embrasures of the southern rounded tower. (Michael Fulton)
During the late twelfth century, it became increasingly fashionable to pierce not only tower walls but also curtain walls with embrasures. The eastern wall of Saone and northern wall of Kerak, the most approachable fronts of both strongholds, were provided with embrasures along their lower levels, while traditional battlements ran along the top of each wall. At both Belvoir and ʿAjlūn, dating to the final third of the twelfth century, vaulted ranges were built against the curtain walls, not unlike those along the northern wall of Kerak. These allowed defenders to access embrasures in the outer walls from internal spaces, while providing a very broad fighting platform for additional defenders on the roof above. A similar system appears to have been employed when Al-ʿĀdil rebuilt the citadel of Damascus: he constructed a new line of walls beyond the earlier Seljuk trace, vaulting the space in between to provide a gallery that serviced mural embrasures.
Belvoir, embrasures of one of the inner towers. (Michael Fulton)
Belvoir, exterior of one of the embrasures of the inner enclosure. (Michael Fulton)
Saone, embrasures along the outer eastern wall. (Courtesy of Denys Pringle)
Kerak, line of casemates built by the Mamlūks along the outer western wall. (Michael Fulton)
While the embrasures at the citadel of Damascus are spaced fairly evenly, those along the outer walls of Belvoir are not, despite the fairly symmetrical design of the castle. At ʿAjlūn, the southern and eastern walls of the inner castle are provided with a number of embrasures, but when outer ranges were added along both of these fronts (before 1214), it seems embrasures were only provided in the southern wall, suggesting this was regarded as the more vulnerable front. At Qalʿat Ṣadr, also dating to the late twelfth century, embrasures appear to have been placed around most, if not all, sections of the castle’s outer wall.
In the early thirteenth century, long stretches of casemates, either open to the air or accessed from galleries or structures built against the curtain wall, became more common. The outer wall of ʿAtlit incorporated a continuous gallery at the upper level, interrupted only by the three outer towers. When the iconic glacis was constructed around the southern and western sides of the upper castle of Crac, a passage was left, creating a mural gallery that serviced more than a dozen embrasures. This was similarly done around the keep at Tortosa, when the Templars added a talus to their earlier tower-turned-keep. At both Ṣubayba and Safed, lines of mural embrasures were incorporated into sections of the castles’ outer defences, which date to the early and mid-thirteenth century respectively. Regularly spaced embrasures may also have been provided along the 1km stretch of Caesarea’s town walls by at least the time Louis IX had finished strengthening them in the early 1250s. These embrasures were arranged at the top of the talus, as was done along the outer defences of Belvoir, while embrasures ran directly through the steep talus of Belvoir’s inner defences.
ʿAjlūn, level 2 with exposed sections of level 1 (after Yovitchitch and Johns).
Tortosa, keep (after Pospieszny and Braune).
One of the simplest ways to provide additional protection to a walled enclosure, whether a tower, a castle or a city, was to add a fosse. Also known as a ditch or moat, these could: completely encircle a stronghold, as was done at ʿAjlūn, Qalʿat Ibn Maʿān and numerous other citadels and towns; surround three sides, if the stronghold had been built at the edge of a plateau, as at Belvoir, Chastel Neuf, Arsūf and many coastal towns; or, cut straight across the most approachable front where a stronghold was naturally defended by the topography on its remaining sides, as can be seen at spur castles like Saone, Montfort, ʿAkkār and Shughr-Bakās, and citadels like those at Caesarea and ʿAtlit, which enjoyed the protection of the sea on three sides.
Regardless of its shape, a fosse served many functions: it increased the relative height of the walls, making frontal attacks and gaining entrance by escalade more difficult; it complicated the use of siege towers and rams, requiring the ditch to be filled before these engines could be pushed up to the walls; and it made the task of mining more difficult. If the fosse were cut into the bedrock, it forced miners either to tunnel into the rock, often at a level significantly below the foundations of the wall above, or to fill the ditch in order to work against the stronghold’s masonry fortifications. The rock-cut ditches at Chastel Neuf and Saone are perhaps the most impressive elements of these strongholds, while that at Belvoir was an integral part of its concentric design, and it was the northern ditch at Kerak, between the castle and the town, that ultimately frustrated Saladin’s siege efforts in both 1183 and 1184.
Chastel Neuf, fosse. (Michael Fulton)
Wilbrand of Oldenburg: the castle of Beirut, 1211
[Beirut] is an extremely large city, sited on the sea, having a very pleasant territory. The Saracens, struck with extreme fear of our men and preparing to flee [in 1197], destroyed its walls and withdrew to defend themselves in the city’s castle, which is very strong, and was then. However, on approach of the chancellor Conrad, of pious memory, with the whole army of the Germans, so great a fear invaded those sons of iniquity that, fleeing the Teutonic fury, they left behind to our men the undamaged castle with all its contents. It is now owned by a certain John [of Ibelin], a very Christian and vigorous man. And, as has been said, it is a very strong castle. For on one side it is defended by the sea and a high precipice of rock and on the other side it is encompassed by a ditch, walled and so deep that in it we saw many prisoners cast down as in a deep prison. This ditch is overlooked by two strong walls, on which very strong towers have been erected against the assaults of machines, and their large stones are bound together at the joints with large iron bands and hard braces. In one of them, which is being newly built, we saw a very ornate hall.
(Adapted from Wilbrand of Oldenburg, trans. Pringle, p 65.)
A ditch was incorporated into the original planning of many strongholds and, if dug into the bedrock, the stone excavated from the fosse provided plenty of building material. In addition to the ashlars that were often used to face a stronghold’s outer walls, a considerable amount of rubble was also needed. Both Frankish and Muslim masons typically constructed fortified walls by building two facing walls of masonry blocks and filling the space between them with a rubble and mortar core. Thus even if the ashlars were sourced from a quarry slightly further from the stronghold, the stone removed from a fosse could be used for the cores of these walls and to build simpler interior structures. Where strongholds were not founded on the natural bedrock, the scarp and counter scarp were often faced with masonry, which helped combat erosion, reducing the frequency at which debris had to be cleared from the fosse, and allowed some with a nearby water source to be flooded. The revetment can still be seen clearly at the coastal strongholds of ʿAtlit, Caesarea and Arsūf.
Kerak, northern fosse. (Michael Fulton)
Wet moats were more common along the coast, although still relatively rare in the region, and most appear to have been developed after the Franks arrived in the Levant. Caesarea’s impressive town defences were quickly overcome by the Mamlūks in 1265, but the citadel was separated from the remainder of the town by a fosse, filled with sea water, which temporarily halted the advance of the Mamlūk besiegers. The ditches at ʿAtlit and Nephin, strongholds that similarly extended out into the sea with ditches that cut them off from the remainder of the mainland, may also have been flooded, or were perhaps floodable. At Damascus, water from the Barada River was diverted in order to flood the fosse of al-ʿĀdil’s new citadel, and attempts may have been made to flood the much longer ditch that encircled the city’s town walls. The small castle at Baysān, founded in the early twelfth century, is a rare example of an interior castle with a wet moat. Situated southwest of the Sea of Galilee, at the southern edge of the Roman-Byzantine city of Scythopolis, just south of Belvoir, the castle’s fosse was lined with hydraulic plaster and flooded by a spring. The small castle complex of Qāqūn, in the plain southeast of Caesarea, is similarly described by one source as a very strong tower ‘surrounded with ditches filled with water’.33
Caesarea, defences improved by Louis IX and earlier Roman wall (with topography).
The advantages of a wet moat were essentially exaggerated at island strongholds. The sea castle at Sidon, developed in at least three stages during the thirteenth century, was never really challenged before it was abandoned in 1291. At Damietta, the Tower of the Chain enjoyed similar protection from the waters of the Nile that flowed around it. The Tower of the Flies, which guarded the harbour of Acre, was less frequently tested but successfully repulsed at least one attack in 1190.
A bridge was usually required to cross a stronghold’s ditch. These were often wooden and intended to be temporary – easy to remove or destroy if the stronghold were attacked. This was apparently the case at Kerak, where the confusion and panic of people fleeing to the castle in 1183 led to the collapse of the bridge over the northern fosse. At some sites, a support was left when the rock-cut fosse was excavated, as was done at Saone, where a slender, needle-like pier was left in the eastern ditch, and at Nephin, where a similar, if much lower, example is preserved. Elsewhere, masonry piers might be constructed, the remains of which can still be seen at Arsūf and ʿAjlūn, and in the ditches surrounding the town defences of Edessa, ʿAtlit and Caesarea. Although there is some evidence of drawbridges, the surviving architectural evidence suggests these were not particularly popular. One example can be found at the thirteenth-century sea castle at Sidon. When raised, the bridge, which otherwise extended to a pier at the end of a permanent masonry bridge to the mainland, blocked the main gate and left the castle completely surrounded by water.
Sidon, sea castle (after Kalayan).
Saone, eastern fosse. (Courtesy of Denys Pringle)
Visitors to citadels built on tells were often guided along an impressive entrance ramp. These were typically narrow and dominated by the stronghold above, making them deadly killing zones. The iconic ramp leading to the citadel of Aleppo approaches perpendicular to the line of the stronghold’s walls, exposing would-be attackers to a long climb out in the open if they chose to use it to cross the ditch. At Qalʿat Najm, the entrance ramp ascends perpendicular to the neighbouring walls but a few metres out from the curtain, leaving a final section to be bridged at the top and allowing the whole approach to be dominated from the walls to the left of anyone climbing the ramp.
One way to add protection to walls that were fairly easy to approach was to employ a talus: sloping masonry built against or incorporated into the design of the lower section of a wall. These can be found in many Frankish, Ayyūbid and Mamlūk building phases dating to the thirteenth century, although a few earlier examples date to the twelfth century. At some sites, a talus was added to existing walls or towers, as was done at the citadels of Tortosa and Bosra, the castles of ʿAjlūn and Montreal, and the town walls of Caesarea. At others, such as Belvoir and the keep at Montfort, a talus was incorporated into the design of the walls. At Crac des Chevaliers, an exceptionally high talus was incorporated when the inner castle was rebuilt by the Hospitallers following the earthquake of 1202. The sloping masonry appears to begin below the foundations of the wall, climbing up the natural hill, like a glacis, but then continues to rise and joins the wall more than halfway up. The outer stones were cut in such a way as to allow rounded towers to rise seamlessly from the sloping masonry. Following the siege of Margat in 1285, during which the outer southern defences of the castle were undermined, a similar combination of a talus with an integrated rounded tower was built by the Mamlūks.
The primary function of a talus was to dissuade mining. The additional masonry forced the extraction of a greater quantity of stone if the sappers tried to remove the lowest courses, while its mass threatened to collapse a tunnel dug below it, leaving the curtain or tower wall behind standing. Taluses also kept assault forces back from the natural base of a wall, placing them in more exposed positions.
To further secure a stronghold built on a natural or artificial hill, the slope leading up to it might be revetted – forming a glacis – inhibiting any would-be attackers’ ability to climb up the hill. Nūr al-Dīn appears to have been responsible for constructing the glacis around the gate at the northern end of the citadel of Shayzar, while others were employed at Qalʿat Najm, Qalʿat Jaʿbar, al-Raḥba and Ṣarkhad. The citadels of Aleppo, Hama and Homs were also provided with glacis by at least the time they were refortified under the Ayyūbids. Frankish strongholds were less frequently built on tells and fewer made use of a glacis; however, that at Kerak is perhaps the most iconic feature of the castle.
Caesarea, talus around the town defences. (Michael Fulton)
Shayzar, northern glacis. (photo courtesy of Denys Pringle)
Kerak, eastern glacis. (Michael Fulton)
Once a besieger made it to the base of a wall, he typically gained a measure of protection. Although archers in projecting towers could still shoot at him from the sides, the defenders directly ahead were less of a threat – those along the parapet would need to lean out over the wall through a crenel to shoot at or drop something on anyone at the base of the wall. Although a talus or glacis might limit the amount of leaning required, this was still a very dangerous position for defenders. What was needed was a safer way for defenders to attack besiegers directly below them. One architectural device that facilitated this was the machicolation.
At their simplest, machicolations could be murder holes: spaces left in a vaulted ceiling through which attackers on the upper level could drop or shoot things at those below. While these were helpful defences for entrance passages, slots were frequently preferred directly in front of gateways, often contained in a recessed arch, ensuring there were no blind spots in which attackers could hide. For exterior walls, corbels were used to build out from the natural plane of the wall, typically at the level of the parapet, leaving spaces between them through which defenders could target anyone at the base of the wall below.
The earliest surviving machicolations date to the eighth century and can be found over the gate of Qaṣr al-Ḥayr al-Sharqī, a fortified Umayyad palace about 95km south of Raqqa and the same distance northeast of Palmyra. As was done here, machicolations were most commonly used to defend gateways through the following centuries. At their simplest, these could be little more than two corbels with a single opening between them, as were employed over each of ʿAtlit’s six outer gates, above the gates in the northwestern tower of Montfort’s upper ward and the outer northwest gate, and above the entrances of many other towers elsewhere. These often took the form of a box, with masonry rising from the corbels and then returning to the natural plane of the wall at about the height of a person. Larger gates might be defended by such box machicolations with two openings, supported by three corbels, as was done at Qaṣr al-Ḥayr al-Sharqī and those on either side of Cairo’s Bāb al-Naṣr. Others were dominated by three slots, such as the main outer gate at Margat and those employed along the entrance passage at the citadel of Aleppo. Elsewhere, even longer boxes were used, increasing the number of corbels and openings.
From about the start of the thirteenth century, the use of machicolations increased and the boxes once found almost exclusively over gates were constructed by both Franks and Muslims to defend the bases of walls and towers. Like those built centuries earlier at Qaṣr al-Ḥayr al-Sharqī, box machicolations were used by the Hospitallers around the outer walls of Crac. Surviving examples consist of small projections supported by three corbels, leaving two open spaces between them. Others, built by the Mamlūks following the castle’s capture in 1271, make use of between two and five corbels.
Box machicolations similar to those built by the Franks at Crac were constructed around the same time by al-‘Ᾱdil and al-Ẓāhir Ghāzī, most famously at the citadels of Damascus and Aleppo respectively. These have three or four corbels, while some, which wrap around the corners of certain quadrangular towers, have six or seven. The Tour du Garc¸ on (Burj al-Ṣābi’), built by the Hospitallers to overlook the coastal road below Margat, had two two-corbel machicolations along each side of its parapet, aligning with the embrasures at each of the two levels below, and a three-corbel arrangement wrapping around each corner. Many box machicolations that have survived, both Frankish and Muslim, have a small loophole in the outward face, allowing a person inside to see, if not shoot at, what was further away from the base of the wall below.
ʿAjlūn, slot machicolation over the inner gate of the outer gateway. (Michael Fulton)
Montfort, machicolation over the inner northwestern gate. (Michael Fulton)
Montfort, looking down through the machicolation over the outer northwestern gate. (Michael Fulton)
Crac des Chevaliers, Mamlūk box machicolations. (Courtesy of Denys Pringle)
Under the Mamlūks, continuous stretches of machicolations, rather than boxes, were constructed around the battlements of some strongholds. These were often incorporated into the designs of double-level parapets. The corbels of such an arrangement are still evident along the outer southern defences of Crac. Although this might be considered a superior design, and it would gain particular popularity in southern Europe in the fourteenth and fifteenth centuries, box machicolations remained common in Mamlūk architecture. They were employed when the southern defences of Margat were rebuilt after 1285, at Montreal when a number of towers were rebuilt in the 1290s, and when the Mamlūks rebuilt the citadel of Jerusalem.
Through Columns
Machicolations might help keep attackers away from the foundations of a wall, but there was little that could be done if sappers managed to tunnel into or under the wall. Like a talus, the use of through columns was a defence against mining: classical granite or marble columns were incorporated into the masonry of walls, aligned through their thickness in order to bond the inner and outer faces with the core. Like medieval rebar, they helped support and spread the weight of a wall, holding up undermined sections that might otherwise collapse.
Bānyās, through column. (Michael Fulton)
Arsūf, through column. (Michael Fulton)
Columns were frequently used in this way along the coast, as at Alexandria, Ascalon, Caesarea, the sea castle at Sidon and Jubayl, as well as at inland sites such as Cairo, Bānyās, Bosra, Baalbek, Shayzar, Apamea, Diyār Bakr and even Qalāwūn’s tower in the centre of Crac’s outer southern wall. They were most often employed at sites where there had been a Roman presence, or columns were otherwise readily available. Most surviving examples date to the thirteenth century; however, earlier examples are not uncommon. In addition to thwarting the efforts of miners, through columns added more general strength, providing protection against the effects of earthquakes and at some sites reinforcing areas where structural strain may have been most significant.
If attackers were compelled to go through a stronghold’s defences, rather than over them, gates could be an attractive place to attack, as these were natural portals in an otherwise solid enclosure. This led to a steady process of developing gateways over the centuries leading up to, during and after the crusades. The size and complexity of gateways were determined by the opposing forces of convenience and security. The day-to-day activities of strongholds encouraged numerous, simple and wide gateways, as these were the main avenues for people and goods entering and exiting the fortified perimeter. The inherent vulnerability of entrances, however, urged the use of only a few narrow gateways with numerous defences.
Large castles and citadels typically had one or two significant entrances, smaller strongholds only one, and larger town defences often had at least one in each cardinal direction. Most town gates were arranged so as to allow the passage of large carts, but this made them more exposed to the approach of rams, penthouses and other siege engines, so were less common at castles. The vulnerability of primary gates led to the development of posterns, which allowed defenders to launch rapid sallies against their attackers without opening the main gates.
The main element of a large gateway was typically a set of wooden leaf doors, which opened inwards. These were fixed in stone sockets at the top and bottom of the doorframe, and secured with a bar or beam that was placed across the doors or slid across from a pocket in the wall on one side. Thirteenth-century Frankish entrances often incorporated a portcullis, the use of which had increased through the twelfth century as fortifications developed. Their use is clearly discernible by the slots found on either side of a gateway, in which the portcullis slid up and down. Although they were less often employed at Muslim strongholds built in the region, three gates at Ṣubayba (those in Ayyūbid towers 3, 8 and 11) incorporated a portcullis. Many entrances, both Frankish and Muslim, also benefited from the security of a slot machicolation, often placed in a recessed arch in front of the main doors.
Compared to larger main gates, posterns were much smaller and lightly guarded, usually reliant on the strength of the surrounding fortifications. These were typically sealed with a single door, opening from one side and secured with a sturdy locking bar. City gates were normally the widest, facilitating the hustle and bustle of trade that sustained the urban community. These were usually positioned between two flanking towers, a tradition long predating the arrival of the Franks, or placed in gate towers, as was the most common arrangement for twelfth- and thirteenth-century castle gates.
The gates of both Frankish and Muslim castles were often smaller than those of comparable European strongholds. Whereas European castles were the primary residences of the nobility, giving them a more palatial quality, castles built in the Levant generally had a more militaristic function – many Frankish and Muslim rulers chose to reside or spend considerable time in the region’s cities, entrusting custody of their castles to deputies or castellans. Additionally, the armies that campaigned in the Levant were regularly larger than those operating in Europe at the time.
One of the simplest ways to strengthen a gateway was to construct a bent entrance, by placing a second gate, aligned perpendicular to the existing one, in front or behind the first. This created a chokepoint, forcing anyone entering to make a 90 degree turn within an environment controlled by the defenders. While the use of two gates in any alignment provided a killing zone, a bent entrance might disorient attackers and frustrate the use of rams against the inner gate, forcing such cumbersome engines to be rotated and then restricting the space in which they could operate. If the outer gate were placed ahead of the main line of defences, perpendicular to the walls behind, it would be shielded from artillery and besiegers attacking it would be forced to arrange themselves in the shadow of the adjacent curtain wall.
Caesarea, outer gate of the eastern gateway, featuring a slot machicolation, portcullis groove, lower socket for a leaf door and pocket in the wall for a locking bar. (Michael Fulton)
Bent entrances had been employed in the region by the architects of certain Byzantine strongholds and are frequently found at the smaller citadels built in north-western Syria. Earlier Roman-era forts usually had straight entrances flanked by a tower on each side; this design was copied by the planners of the palatial Umayyad ‘desert castles’, by the Muslim builders of Kafr Lām and Caesarea, where a new set of walls, with straight entrances, was built within the old Roman line of defences, and by the Fāṭimids at Cairo. Straight entrances were similarly preferred in Latin Europe on the eve of the First Crusade. But as fortifications developed, bent entrances came to outnumber straight designs around the end of the twelfth century in the work of Frankish, Armenian and Ayyūbid builders.
Towers were often used to create bent entrances, most commonly by placing an outer gate in one of the tower’s flanks and an inner gate in the rear wall. At ʿAtlit, a gate was placed in each flank of the three outer towers, and each pair of outer gates then led to a single inner gate at the back of the tower. Similarly, the inner flanks of the towers or protrusions at each end of Kerak’s northern wall contained a gate. Bent entrance towers can also be found at Belvoir, Saone, Baghrās, Tortosa, Margat, Bourzey, Ḥārim, Safed, Ṣubayba, Crac des Chevaliers, Qalʿat Najm, Apamea, Baalbek and elsewhere. The three town gates of Caesarea were similar bent entrance towers, as were the Zion Gate and Tanners’ Gate at Jerusalem, while the addition of a barbican ahead of the northern St Stephen’s Gate was only slightly more elaborate. To add further protection, another tower could be placed next to the gate tower on the side of the external entrance, forcing anyone travelling through the gate to pass by the embrasures and defenders of the neighbouring tower. This combined the advantages of a flanked straight gate with those of a bent entrance tower, and was employed at Shughr-Bakās, Bosra, the east gate of the citadel of Damascus and a number of Aleppo’s town gates, including Bāb al-Naṣr, Bāb Anṭākiyya, Bāb al-Hadīd and the later Bāb Qinnesrīn.
Bent entrances.
Numerous bent entrances were built under the Ayyūbids at Cairo. Bāb al-Qarāfa is a simple bent entrance tower, while Bāb al-Barqiyya has a slightly more complex entrance portal, commanded by a projecting tower, behind which the gate eventually leads. This system can be found in a more developed form at Bāb al-Jadīd, where a second tower is placed on the other side of the entrance portal, creating a bent entrance flanked by two towers. These are all more sophisticated than the simpler, if typically larger, Fāṭimid gates, such as Bāb al-Naṣr and Bāb al-Futūḥ, which contain a straight entrance portal flanked by two towers. ˙ ˙
Longer approaches could be employed to increase the period of time that adversaries were under pressure before they reached a gate. At Ascalon, the northern Jaffa Gate was recessed into the town, forcing those who sought entrance to pass first a round tower to the east and then a large quadrangular tower to the west, while another tower dominated the back of this pocket, which led to a bent entrance towards the west. At some sites, the topography could be used, forcing an approach up a predictable route dominated by the stronghold and its defenders, as was often the case at sites built on tells. To further control the progress of intruders, the approach could be enclosed behind an outer gate, creating a barbican, as was done at Qalʿat Ṣadr and Montreal. The Franks strengthened the northern gate of Jerusalem with a barbican, creating a bent entrance in the process, and a similar barbican controlled access to Toron. At ʿAjlūn, a small bent barbican was added beyond the gate of the original quadriburgium and another was built outside the later outer gate in the thirteenth century. The Mamlūks similarly developed the southwest gate of Safed in the second half of the thirteenth century, constructing a barbican that extended north from the original Frankish gate.
Cairo, Bāb al-Barqiyya.
The most famous entrances, however, are those at Crac and the citadel of Aleppo. The former, begun by the Hospitallers and developed by the Mamlūks, began and ended with a 90 degree turn, with a 180 degree hairpin in the middle, while the Lion Tower, at the apex of the hairpin, contained another bent entrance that provided access to the space between the inner and outer southern walls. Al-Ẓāhir Ghāzī’s contemporary gatehouse at the citadel of Aleppo forced visitors to turn 540 degrees as they navigated the serpentine series of 90 degree corners. Designed to resist or detain besiegers as long as possible, most barbicans were provided with embrasures and other defensive elements.
ʿAjlūn, outer gate and barbican. (Michael Fulton)
ʿAjlūn, inside the outer barbican. (Courtesy of Fraser Reed)
Safed, southwestern gate and barbican. (Michael Fulton)
Aleppo, citadel gatehouse (after Yovitchitch).
While gates kept assailants out, posterns facilitated sallies and helped defenders interrupt the siege activities of their assailants, buying time before relief arrived. If left unobstructed over a sufficient period of time, a reasonably well supplied besieging army could expect to eventually undermine most defences or starve defenders into submission. If a stronghold had only one gate, a single point of ingress and egress, besiegers could concentrate their forces against it, effectively blockading the stronghold without surrounding it. By integrating posterns into the designs of fortifications, builders compelled potential besiegers to extend their lines and spread their forces in order to oppose the threat of a sally from each postern. Unlike gates, posterns were generally large enough for only one person to pass through at a time, ensuring that besiegers could not force their way in during the course of regular siege operations or follow a group of defenders back in after a botched sally.
Posterns can be found in some defences predating the crusades and proved integral at certain early twelfth-century sieges, like that of Mosul in January 1105. Their use, however, increased considerably from about the late twelfth century, as the size of both armies and strongholds grew. Each was positioned to take advantage of the surrounding defences and local topography. At Caesarea, a postern was placed in each of the three landward sides of the city’s walls, accessed by a staircase that led down through the wall and talus to a small gate at the bottom of the ditch. The perceived advantage provided by these posterns evidently changed between the final stage of their construction in 1251 and the town’s capture in 1265, as those in the eastern and northern walls had been sealed by the time the Mamlūks arrived.
Belvoir was provided with at least five posterns – a sixth may have been located in the destroyed and as yet unexcavated outer northwest tower. Three can be found in the castle’s other outer towers, accessed by stairs running down to small gates, tucked into the recesses in the surrounding talus and hidden from view, just above the base of the fosse. The small addition to the east side of the outer northeast tower contains another, and the fifth is located in the lower barbican, which would have facilitated attacks against any force assailing the main gate. The latter two may also have allowed a messenger to escape down the steep eastern slope in times of crisis, as the topography would have made blockading this side of the castle difficult. Collectively these provided numerous points from which the garrison could strike at any besiegers who entered the ditch, forcing their opponents to commit significant forces to guarding each of these points of egress. The larger sally made by the garrison during the winter of 1187/88, however, would have been launched through the main gate. This was a cavalry force that undertook what was effectively a raid during a period of distant blockade – the besiegers being more concerned with relief forces reaching the castle than those in the castle striking out offensively.
While Belvoir’s system of posterns was built around principles of opportunity and deception, a different, if equally secretive, approach was taken when the Mamlūks developed the western defences of Ṣubayba’s outer bailey. When the tower in the middle of the western wall (tower 11) was enlarged, what had been the outer gate in the south flank of the tower was turned into the entrance to a passage that led down a long flight of stairs to a small postern in the north flank of the enlarged Mamlūk tower. Unlike the Ayyūbid gate, the Mamlūk postern is unimposing: a small portal left between the large ashlars and obscured from the outside by a slight rise in the topography before the ground drops away into the surrounding valleys. The staircase behind is quite wide, perhaps twice as wide as those leading to the posterns at Caesarea and Belvoir, with a lofty rather than low ceiling, which would have allowed a significant body of men to gather in this space before exiting the small postern, and possibly in the concealed dip just outside. This would have facilitated sallies that might more accurately be classified as counterattacks.
Caesarea, closed postern in the northern town wall. (Michael Fulton)
Belvoir, staircase to the postern in the western mural tower. (Michael Fulton)
Belvoir, postern in the outer southwestern tower. (Michael Fulton)
Before it was rebuilt in the early thirteenth century, Crac des Chevaliers had a single line of walls. The main gate, positioned along the east face of the castle, was flanked by two small towers, and a small postern was located at the north end of the castle. Following the castle’s development, the main gate became the final part of the grand entrance ramp and the northern postern became the entrance to the latrine tower – this tower was either a final addition before the great earthquakes of 1202 or the first afterwards. A postern was left in the flank of the latrine tower, which is conspicuously close to the small northern gate that was built when the outer wall was added. Flanked by two towers, this outer northern gate was one of two secondary entrances to the Frankish castle; the other was positioned in the flank of the outer southeastern tower, which provided access to the southern outwork via some kind of bridge. Both of these secondary gates, as well as the grand eastern entrance, were developed by the Mamlūks in the immediate aftermath of the siege of 1271. When Qalāwūn’s large quadrangular tower was added a decade later, a small postern was incorporated into its flank, accessed through a passage leading through the otherwise solid lower part of the tower.
Ṣubayba, postern in tower 11. (Michael Fulton)
Crac des Chevaliers (with topography).
The expansive defences of Saone were likewise provided with gates at practical points. The main entrance, at the north end of the eastern wall, was accessed from the plateau to the east of the castle. Visitors and assailants who might attack this gate were forced to cross a narrow bridge supported by the needle-like pier that had been left when the great eastern fosse was excavated. The gate itself is flanked by shallow rounded towers and provides access to the various layers of the castle’s eastern defences, which acted like an elaborate barbican. A secondary entrance can be found in a simple quadrangular gate tower along the castle’s southern side, the outer gate of which can be found in its western flank. The outer western bailey, which encloses the tip of the spur upon which the castle was built, was provided with two gate towers, one leading to the valley on the north side of the castle and the other to the valley on the south side. None of the castle’s gates is particularly large.
Saone (with topography).
As took place at Belvoir around the start of January 1188, larger sallies, including all those involving cavalry, were launched through main gates, rather than posterns. Such sallies included those undertaken by cavalry forces during the sieges of Arsūf (1265) and Acre (1291), although Acre (and almost certainly also Arsūf) had smaller posterns incorporated into its defences.
The continued use of some posterns in the second half of the thirteenth century and the sealing of others speaks to the particularity of each site and the perceived threats each might be expected to face. Like those that were closed at Caesarea, what appears to have been an Ayyūbid postern in a southern section of curtain at the citadel of Bosra, possibly dating to al-ʿĀdil’s early fortification efforts there, was sealed when a talus was added to the southern and eastern sections of the fortified theatre later in the thirteenth century. Evidently it was believed a talus would add more to the defensibility of the site than the existing postern.
An assortment of factors and influences drove the development of this period’s fortifications. Although different castles developed in different ways, almost all were the product of ongoing refortification efforts. Where abrupt shifts in the design, scale and style of these structures are evident, as can be seen around the end of the twelfth century and start of the thirteenth, this was typically a reflection of the changing sociopolitical context, rather than a response to the drastic improvement of one or more siege technologies.
The keep was a uniquely Frankish defensive element, employed in many regions of Europe. In the Levant, most isolated towers or keeps were built in the kingdom of Jerusalem, where pre-existing defences were comparatively scarce. Rather than a deliberate design feature of a larger stronghold, most keeps began as isolated structures, often administrative towers. Many were surrounded with an outer wall later in the twelfth century, creating early concentric castles, as at Latrun and Belmont (Suba). Although most towers received no outer wall, this was the only addition to many others, while a select few would become the original kernels of much larger castles. To the north, Jubayl, Chastel Blanc and other towers were built and developed in the region around Tripoli, while small Byzantine citadels were often used in lieu of keeps further north. At Saone, the original Byzantine fort remained the centre of the larger castle, which was developed by the Byzantines and then the Franks. The large tower built mid-way along the outer eastern wall is often referred to as the Frankish ‘keep’; however, this might better be designated as a donjon (as the term is used here): a particularly strong mural tower akin to those built at certain sites in western Europe from about the late twelfth century.
Quite large towers can also be found in Muslim architecture. These, however, served a different purpose and were often constructed as part of a larger building phase. Al-ʿĀdil’s large towers were some of the biggest built in the early thirteenth century, but they were often constructed as a series of equally impressive mural towers, no one being central or considerably stronger than the rest. The large circular tower built by the Mamlūks at Safed is perhaps the best example of a Muslim keep. Built within the trace of the Frankish walls, perhaps over an earlier Frankish tower, little remains of this structure. Large Muslim towers were more commonly built along the trace of a curtain, as was done when the outer defences of Ṣubayba, Crac and Montreal were refortified after 1260, 1271 and 1296 respectively. Some of these towers were used as opulent residences, while others were designed with purely military considerations.
Chastel Blanc. (Courtesy of Denys Pringle)
Enclosures
For wealthier Frankish patrons, such as the kings of Jerusalem and military orders, the enclosure castle was an alternative to the isolated tower. These were essentially an outer perimeter of walls without a central keep. The three castles built around Ascalon under Fulk were of this type: each was quadrangular with a tower at each corner, one tower being larger than the rest to provide a final position of resistance. Later in the twelfth century, the Hospitallers built Coliath to a roughly square plan, with four corner towers and an additional tower mid-way along three of its walls, with the main gate in the fourth. The order also built Belvoir, which was essentially a small enclosure castle with four corner towers inside a larger one with added mid-wall towers. In the 1170s, elongated enclosure designs were employed at Crac and Jacob’s Ford. The former may have had a single tower in the centre of the southern wall and excavations have revealed just one tower in the southeast corner of the latter; neither projected far, if at all, from the adjacent curtain walls. Regardless of their shape, ranges of buildings typically lined the inside of the outer walls of these enclosures, providing internal spaces and a broad fighting platform above. Among the early Ayyūbid strongholds, Qalʿat Ṣadr was built as an irregular enclosure, while the initial phases of ʿAjlūn and Ṣubayba were built to a square plan with a tower at each corner. Unlike many built by the Franks, which had an open courtyard of sorts in the centre, the middles of these Muslim castles tended to be covered.
Twelfth-century quadriburgium enclosure castles.
The square quadriburgium design, adopted by both Frankish and Muslim builders, provided an efficient use of space and was simple to plan and build. The design was inspired, directly or otherwise, by Roman and Byzantine examples, as can be found in the Byzantine remains of Saone and Bourzey. Further south, a more immediate influence may have been certain early Islamic structures, such as Kafr Lām and Māhūz Azdūd, which had been built with Roman-inspired plans long before the First Crusade.
The development of the concentric castle – one built with multiple lines of defences, each enclosing the next – is closely linked with the crusades, although perhaps to an unwarranted degree. The principle of concentricity is often overemphasized, and regarded as some kind of necessary design innovation by historians trying to present linear models of castle development. The addition of an extra line of defences along or around an exposed front was perhaps the easiest and most obvious way of increasing the defensibility of a system of fortifications. Accordingly, most castles with concentric elements were the product of gradual enlargement and refortification; as local rulers accumulated wealth and chose to invest it in a stronghold, they increased the structure’s defensibility, but also their prestige and hold over the surrounding region. Although concentricity is often a point of fixation, few castles, with the obvious exception of Belvoir, were designed with truly concentric plans from the outset.
Certain towns in Palestine, including Ascalon, Jerusalem and Tyre, already had a forewall beyond the main curtain by the time of the First Crusade and a number of Frankish towers received outer walls in the twelfth century, as did some larger castles. Rarely, however, were these outer walls built in the same phase as the main line of town walls or original tower – although a central tower, which became a keep once surrounded, could be used as a final point of refuge, most were initially designed to be a first (and only) line of defence. By the time it was captured by Saladin’s forces in 1189, Montreal boasted two lines of walls, while some other large castles, including Safed, may also have had multiple lines of walls by this time, although traces of them are hard to identify. In the Syrian Coastal Mountains, many Byzantine citadels, including Saone and Bourzey, became the innermost core of larger strongholds with multiple lines of defences. With the concentration of wealth and power in the early thirteenth century, the means and incentive to add outer walls to the region’s most formidable strongholds increased.
A concentric design forced besiegers to commit greater resources if their aim was to undertake an active siege, while providing defenders with multiple opportunities to sustain their defence. The defenders’ ability to fall back, regroup and continue fighting as the attackers overcame certain obstacles also provided the defenders with considerable leverage when negotiating terms of surrender, as they could typically do so from a position of reasonable strength, even if they no longer controlled their outer fortifications. This ensured that few concentric castles fell to the sword; similarly, many town defences were taken by force but rarely were citadels captured in this way. Such layered approaches were clearly adopted when the outer walls of ʿAtlit, Arsūf and Crac were designed – each closely follows the inner wall, which dominates it from a higher elevation behind.
Beyond adding strength, the construction of an outer wall also created a larger defensible area. For towers, such as Jubayl, Baysān, Chastel Blanc, Chastel Rouge, Le Destroit and many others, the outer wall provided not only an additional line of defence but also protection for important outbuildings, such as stables, and perhaps livestock. The outer walls of many twelfth-century strongholds may have had fairly limited defensive value; often thin and devoid of towers or other defensive elements, they were useful against bandits and small raiding parties, but ineffective against a larger or committed force. This was not the case at Belvoir.
In 1168, the Hospitallers bought an estate from Ivo Velos. Here, on a plateau overlooking the Jordan Valley from the west, the order constructed Belvoir, the region’s most symmetrical concentric castle. Designed as essentially a quadriburgium within another, the castle was built in a sequence of phases, but almost certainly to a predetermined plan during a single, continuous construction campaign. The castle’s exceptional plan and stout walls reflect its vulnerability – it was easily approached on three sides, leading it to rely on its layered design, rather than the surrounding topography, to a greater extent than any other stronghold of comparable size in the region.
Belvoir. (Courtesy of Michael Eisenberg)
Belvoir (after Biller and Baud).
The castle’s focus on layered lines of defences and its liberal use of embrasures do not represent a response to the development of any particular siege weapon, but rather to the growing threat posed by Nūr al-Dīn and the manpower reserves that he could commit to an invasion of Frankish territory. Events of the 1170s and early 1180s would demonstrate that contemporary castles, and their designs, were sufficiently strong to resist concerted sieges so long as a Frankish field army was available to offer relief. Belvoir’s designers, however, had probably observed the delicacy of this balance in the 1160s, during Amalric’s preoccupation with Egypt, and were thus taking steps to reduce the castle’s reliance on the arrival of a timely relief force.
With Amalric and the army of Jerusalem in Egypt, Nūr al-Dīn had been able to capitalize on his victory over the northern Franks at the Battle of Ḥārim in 1164, taking Ḥārim and then moving on to take Bānyās before Amalric returned. In 1167, with Amalric once more in Egypt, Nūr al-Dīn struck into Frankish territory first from Hama, taking Ḥalba, Arima and Chastel Blanc before withdrawing, then, after Ramaḍān, he invaded Galilee from Bānyās, occupying Chastel Neuf, which its defenders abandoned. Between these campaigns, rapid attacks by Nūr al-Dīn and Shīrkūh had captured Munayṭira and the cave strongholds of Tyron and al-Ḥabis Jaldak in 1165. Belvoir’s plan, conceived or at least settled upon after the campaigns of 1167, was a response to these threats.
The castle’s defences worked in three ways: their complexity was an initial deterrent against any attack; their solidity was protection against a sudden surprise attack; and their sophistication was intended to frustrate attackers and prolong siege efforts long enough for relief to arrive. The strength of the castle’s various elements is often overshadowed by its exceptional plan. It had a considerable rock-cut ditch on its three approachable sides and a complex entry system that was tucked away along its least accessible front. It was built directly on the bedrock and its walls were uncharacteristically strong, most about 2.5–3.5m thick at the first level and built of the hard local basalt excavated from the fosse. It had numerous posterns, many of which were concealed, and embrasures in the lower level of its inner and outer walls, while more may have once lined the level above, and a crenellated parapet would have crowned its various walls and towers. Belvoir was much more than just a concentric castle.
What made Belvoir revolutionary was not only its symmetry, but that its outer walls were probably regarded as its main line of defences. If the outer fortifications fell, the inner structure was designed to provide the defenders with a secure position from which to negotiate favourable terms, rather than attempt to hold out indefinitely. The doors of the inner castle all locked from within the surrounding ranges and towers, rather than the central court; thus, if a gate were forced or the walls scaled, the defenders could continue to hold out even with the very centre of the castle in enemy hands, reducing the chances it would fall to the sword. When the castle was taken in January 1189, Saladin had been forced to undermine the outer barbican, the least assailable section of the castle. Revealingly, as soon as this fell, the defenders quickly sought terms, securing their lives and safe passage to Tyre before the inner structure was attacked. Rather than the plan of Belvoir, which was not copied elsewhere in the Levant, it was the size and strength of the castle’s defences that were the real indicators of things to come.
Outworks are more often associated with early modern fortifications, but at least two Frankish castles made use of a secondary fortified detachment to strengthen a particularly approachable or exposed front. At Crac des Chevaliers, a triangular outwork, surrounded by a rock-cut ditch and perhaps once crowned with a wooden palisade, was developed ahead of the stronghold’s southern front. Although this is generally believed to have been the last element added by the Franks, the spacing of the castle’s outer southern towers suggests it was likely planned at the same time as this outer trace of defences. There are striking similarities with the outwork at Château Gaillard, constructed by Richard I of England in the 1190s, but that at Crac is nowhere near as elaborate. When the castle fell to the Mamlūks, focus was removed from the outwork and replaced by Qalāwūn’s large quadrangular tower, placed in the centre of the outer southern wall. The outwork’s ditch remained a defensive obstacle, but it seems the Mamlūks had no intention of defending it, as the Hospitallers had done during the siege of 1271.
At Beaufort, the Templars built what may have been a more defensible outwork, perhaps more akin to that at Château Gaillard. Often identified as the Templar ‘citadel’, it was constructed between the order’s acquisition of the castle in 1260 and Baybars’ capture of the stronghold in 1268. Very little is known about this defensive feature, as its remains have not yet been exposed by archaeologists. The outwork was probably developed to command the area south of the castle; however, the Templars proved incapable of defending it during the siege of 1268. Opting to evacuate it one night, the Templars gifted the Mamlūk besiegers an ideal position from which to set up some of their artillery, which then bombarded the castle from the shelter of the detached stronghold. Unsurprisingly, the Mamlūks regarded the outwork as a liability and destroyed it after occupying the castle. As at Crac, the Franks’ efforts to strengthen an already mighty castle proved to be too little too late in the face of Baybars’ dominant Mamlūk forces.
Crac des Chevaliers, eastern side of the outwork from the outer wall. (Courtesy of Denys Pringle)
The construction, expansion and refortification of each stronghold were unique, guided by a particular set of motivations and resources. However, certain trends can be seen in the ways that strongholds developed from the early twelfth century through to the late thirteenth. Notably, many castles and certain citadels became larger and stronger, trends most apparent in those that were built or considerably rebuilt around the start of the thirteenth century. There were two interconnected reasons for this: the nature of the threat that strongholds could expect had changed, and so too had their owners.
The consolidation of western Syria under Nūr al-Dīn, and Saladin’s later incorporation of Egypt and much of the Jazīra, meant that far larger forces could be mustered against Frankish castles. These resources may have allowed for the eventual wearing down of the Franks, as appears to have begun in the 1160s and 1170s, and would later take place in the late thirteenth century after the Mamlūks had come to power. But the battle of Hattin changed everything, accelerating the loss of Frankish territory. Frankish rule was effectively preserved for another century thanks to the arrival of the Third Crusade and Saladin’s death, which divided Muslim Syria to a degree unseen in five decades.
The old adage that ‘the Franks used walls to do the work of men’ is most often applied to the twelfth century; however, it more accurately reflects the situation in the thirteenth. The Franks were a small and vulnerable force at the start of the twelfth century, but few sizeable strongholds were built during this period. The growing threats facing the Franks in the late twelfth century are revealed in the investment committed to Belvoir, while Frankish weakness around the start of the thirteenth century is evident in the even greater strongholds that followed, such as ʿAtlit, built from 1218, and Crac, rebuilt from 1202.
Whereas small towers and enclosure castles had been common in the twelfth century, nearly all of the castles developed by the Franks in the thirteenth century were much larger and stronger, built on high spurs or small peninsulas that extended into the sea. This was a response not to advancements in siege technologies, but rather to the new sociopolitical environment. The battle of Hattin had curbed the offensive capabilities of the Frankish principalities and thus their ability to disrupt siege activities. With little chance that a large or rapid relief force could be relied upon, Frankish castles had to be able to resist larger forces for longer periods of time. If a stronghold were captured, control of the surrounding region was lost.
Muslim strongholds underwent a concurrent transformation, but this had as much to do with a desire to project authority as it did with increasing defensibility. In the wake of Saladin’s death, his successors struggled for control of the Ayyūbid realm, using their strongholds as bases of power against the avarice of their rival relatives, and as projections of their power and legitimacy.
Patrons
The crown was responsible for commissioning many of the larger strongholds built in the kingdom of Jerusalem during the early twelfth century. Most were then entrusted to barons or the military orders, who went on to enlarge them. In the other Frankish principalities, most of the large castles were also held by the baronage and military orders, although the majority were of earlier Byzantine, Armenian, Arab and Kurdish origins. As the military orders inherited and bought lands with increasing regularity from the mid-twelfth century, they were able to put the weight of the resources they acquired in Europe behind the maintenance and development of their strongholds. This became evident following the battle of Hattin: Safed and Belvoir, belonging to the Templars and Hospitallers respectively, held out longer than most other castles in Palestine, and the Templar keep at Tortosa was the only stronghold Saladin besieged but failed to take during his campaign through western Syria in 1188, during which he opted to bypass the Hospitaller castles of Crac and Margat. During the thirteenth century, huge amounts of money were poured into these castles, such that only the military orders were capable of investing. Although most of the orders’ castles had been founded in the twelfth century or earlier, some, such as ʿAtlit and Montfort, were not built until the thirteenth century.
Most Muslim fortification efforts of the early twelfth century, like those of the Franks, were undertaken organically, parts of piecemeal processes of investment and improvement. After extending his authority across much of western Syria, Nūr al-Dīn commissioned broader fortification campaigns following a series of devastating earthquakes in 1157 and 1170. Beyond his principal power centres, Aleppo and Damascus, his work touched Shayzar, Qalʿat Najm, and other secondary strongholds. Although these works appear to have been necessitated by factors beyond Nūr al-Dīn’s control, they effectively stamped his mark across his realm. Others took more overt measures to express their authority. Saladin’s new citadel at Cairo not only secured his hold over the city, but was a physical manifestation of the regime change he had headed. Following his death, Saladin’s heirs undertook similar fortification campaigns, which both entrenched their power and served as outward expressions of their authority. Critically, Muslim power was once more divided, leaving significant rivals vying for influence.
The defences commissioned by Nūr al-Dīn and Saladin are considered fairly traditional, keeping to established forms and scales relative to the more monumental and ostentatious elements chosen by later Ayyūbids. Common motifs or styles were often employed by these latter figures, which served as clear markers of their rule. One of the building signatures of Al-ʿĀdil, Saladin’s brother, was scale: the towers he constructed at the citadel of Cairo were up to five times larger than those of his brother – certain rounded towers that he constructed there, which enclose smaller towers built under Saladin, are almost 20m in diameter. Before Saladin’s death, Al-ʿĀdil had already rebuilt the citadel of Ḥarrān, placing an enormous near-circular polygonal tower at each of its four corners. The citadel of Raqqa, rebuilt under Al-ʿĀdil in the 1190s, has a similar quadrangular plan and large corner towers. The exaggerated size of these towers appears to have been motivated by a desire to instil a sense of grandeur in those who viewed them, rather than as a response to any particular military threat.
Cairo, citadel.
Cairo, citadel towers Burj al-Ramla and Burj al-Ḥaddād (after Creswell).
From the start of the second decade of the thirteenth century, Al-ʿĀdil continued to employ enormous towers but settled on a more consistent plan: most were quadrangular and faced with large blocks with drafted margins and rusticated bosses. Such towers were used at the citadels of Damascus and Bosra, as well as atop Mount Tabor, and at Jerusalem and the citadel of Cairo, where work was overseen by his sons al-Muʿaẓẓam ʿĪsā and al-Kāmil respectively. Despite the assertions of some historians, the scale of these towers was not a necessary reaction to the advancement of any siege technology; proof of this can be found at most other sites, where quite small towers remained popular throughout the century.
Al-Ẓāhir Ghāzī, one of Saladin’s sons, patronized a similar building programme to that of his uncle, Al-ʿĀdil, developing not only the citadel of Aleppo but other citadels that fell under his influence. Secondary figures, who sought to gain further power and autonomy by leveraging their position between the major Ayyūbid princes, used strongholds in the same way. Al-Mujāhid Shīrkūh II, the grandson of his namesake, had inherited Homs before Saladin’s death, and consolidated his position during the struggles that followed by rebuilding the citadel of al-Raḥba in 1207. In 1227, he strengthened the citadel of Homs and around 1230 rebuilt the castle of Shmemis, 4km northwest of Salamiyya, and constructed a new castle on a hill 2km northwest of Palmyra, which replaced the fortified Temple of Bel as the community’s citadel. These projects may have encouraged the fortification programme launched by his neighbour, al-Muẓaffar Maḥmūd of Hama, around the same time.
Al-Raḥba. (monumentsofsyria.com)
Al-Muẓaffar Maḥmūd was the grandson of Taqī al-Dīn [Umar, Saladin’s ambitious and capab le nephew. He was originally usurped as emir of Hama by his brother, al-Nāṣir Qilij Arslān, upon the death of their father, al-Manṣūr Muḥammad. Although Hama was traditionally in the sphere of Aleppo, their father had supported al-Ādil against al-Ẓāhir Ghāzī, even sending al-Muẓaffar to assist al-Kāmil, al-Ādil’s son and successor in Egypt. It was thanks to the later support of al-Kāmil that al-Muẓaffar regained Hama. The shaky and delayed start to his rule probably contributed to al-Muẓaffar’s decision to refortify the strongholds of his realm. Around 1232, he reinforced the defences of Hama, rebuilt the citadel of Maʿarrat al-Nuʿmān and strengthened Montferrand.
Under the Mamlūks, Muslim rule was more centralized and many refortification projects reveal clearer attention to military considerations. The ease with which the Mongols seemed to overrun western Syria in 1260 impressed the need to improve certain strongholds. Many castles and citadels had been slighted by the Mongols before they withdrew, providing a natural opportunity to rebuild them. Some sites, such as Ṣubayba, were improved by adding very different elements, including larger quadrangular towers as well as rounded ones. Elsewhere, as at the citadel of Damascus, the defences built by the Ayyūbids were replicated, right down to the masonry style. This site-by-site pragmatism operated in conjunction with a simple policy when it came to captured Frankish strongholds: those along the coast were slighted, so as not to be transformed into bridgeheads from Europe if retaken by crusaders, but significant interior castles were developed and employed as armouries and seats of local power. These great interior castles, including Safed, Beaufort, Crac and Margat, were improved, presumably ensuring they were defensible should another great Mongol invasion materialize.
The designs of Mamlūk towers, like all others, reveal a range of guiding influences. Some, such as the outer southern towers at Crac and Margat, are exceptionally solid, while the expanded southwest tower of Ṣubayba and similarly large tower added to the citadel complex at Baalbek are bristling with embrasures. Others show more apparent domestic functions. Even in those towers designed for quite obvious defensive purposes, the interior can be elegant and the masonry fine, as in the rounded towers at Crac and Ṣubayba. At some sites, efforts were made to match the existing dressing. At Crac, the Mamlūks’ smooth masonry blends almost seamlessly with the earlier Hospitaller work, while blocks with rather irregular bosses were used when the southern donjon/palatial tower at Shayzar was expanded, matching the existing Ayyūbid structure.
Ṣubayba (after Deschamps and Hartal).
Montreal, inscription on the north tower. (Michael Fulton)
Perhaps because the authority of the Mamlūk sultan was relatively unchallenged, it is hard to discern many signature styles, as various tower shapes and sizes were employed by both Baybars and Qalāwūn. Instead, the Mamlūks favoured bold inscriptions. The use of inscriptions to dedicate towers and other structures had longstanding precedent in the region, but those of the Mamlūks were often quite brazen, placed in prominent positions so as to leave little doubt in the minds of visitors as to who had commissioned the work. The large tower that now dominates the southern end of Kerak, along with Burj al-Ẓāhir, one of the town’s towers, were striped with bold inscriptions, as was done around some of the large towers built at Montreal. At Crac, the rounded towers rebuilt in the 1270s at each end of the castle’s outer southern wall display prominent inscriptions below the parapet, as does Qalāwūn’s quadrangular tower between them.
Thicker walls, perhaps 1m thicker on average, accompanied the appearance of larger castles and towers. Many curtain walls dating to the thirteenth century were no thicker than earlier twelfth-century examples, while the thirteenth-century towers with the thickest walls were typically quite tall or contained vast internal spaces, which required substantial structural support.
The thickness of fortified walls built in the twelfth century ranged from site to site and wall to wall; however, the average thickness was about 2.5m. The walls of Belvoir, built from 1168, are 2.5–3.5m thick, while those of Jacob’s Ford, dating to 1178–79, are about 4m thick. These were two of the last castles built in the lead-up to the battle of Hattin, but while impressive, this thickness was not unprecedented. The midtwelfth-century Frankish ‘keep’ in the centre of the outer eastern wall at Saone has walls 4.5–5.4m thick. The walls of the keeps at al-Burj (Tantura), Jubayl and Tortosa, all dating to the twelfth century, are 5m, 4m and 3.5–5m thick respectively. As in Europe, it was not uncommon for the walls of such keeps to be thicker than the outer walls subsequently built around them.
Walls built in the thirteenth century were less frequently at the thinner end of the spectrum, rarely less than 2m thick, but neither were they uniformly thicker. Taking Montfort as an example, the walls of the keep, built in the late 1220s, are exceptionally thick, up to 7m in some sections, although a mural passage or gallery is evident. The outer walls of the central domestic range, built behind the keep not long after, are a more modest 2–2.3m thick. The walls of the cistern tower at the west end of the upper castle, which had at least one and perhaps as many as three levels above, are slightly thicker, about 2.8m. Between the central domestic range and the cistern tower, the great hall rested on two barrel vaults and rose another two storeys above. The vaults sprang from a dividing wall 2.5m thick and surviving sections of the southern wall, although unexcavated, indicate the hall rested on outer walls between 3m and 5m thick. A rounded wall, 3.1m thick, encloses the western end of the upper castle, rising from the bedrock further down the spur. At the northern end of this wall stands the inner northwestern gate tower, the walls of which are just 1.1–2m thick, the thinnest wall being the outward north wall. An outer wall enclosed the northern and western sides of the upper castle, running north from the keep at the east end, down the slope and around an outer bailey, rejoining the west end of the upper castle at an unknown spot. The original outer wall was just 1.8m thick and was crowned with a parapet 0.7m thick, with merlons 1–1.5m wide every 0.7m. The Teutonic Knights were in the process of replacing this wall with a new one, only 1.9m thick, when the castle was taken by the Mamlūks. The rebuilding efforts had progressed from west to east, and included a rounded tower with walls similarly 1.9m thick, but had not reached the outer northwest gate when work was interrupted.
Montfort (after Pringle).
Work at Montfort began around 1227 and ended no later than 1271, when the castle was captured and then destroyed by Baybars, leaving a period of no more than forty-five years for the various building phases to have been completed. The thickness of the walls built during this relatively brief window is revealing. While the walls of the keep are exceptionally thick, those of the central range behind it, which would have been exterior walls at one point, are not. Likewise, the walls of the outer bailey are quite thin by thirteenth-century standards. The curved wall at the west end of the upper ward is noticeably thicker, but it was not provided with any embrasures in its lower level(s), suggesting it might postdate the wall of the outer bailey, and that its thickness was influenced by structural requirements, not defensive ones. Likewise, the simple northwestern gate tower at the north end of this wall may also postdate the outer wall. The gate in the north wall of the first level is protected by a machicolation at the third level; however, the tower boasts only a single embrasure, which extends through the north wall of the second level. The apparent effort to rebuild the outer wall is the most revealing: although this new wall is significantly higher, it is hardly thicker, confirming that walls more than 2m thick were not always considered necessary even in the third quarter of the thirteenth century.
Admittedly, Montfort was neither the strongest nor the most sophisticated castle built by the Franks in the thirteenth century; however, it was the principal stronghold of the Teutonic Knights and under near continual development. So why then are some of Montfort’s walls so thin, while those of other contemporary castles are so thick? The castle most often associated with thick walls is ʿAtlit. Excavations and surveys at ʿAtlit were last carried out in the early 1930s and precise measurements of many of its walls have still not been published; nevertheless, the tremendous scale of the castle is immediately apparent.
ʿAtlit was founded in 1218. Built on a little peninsula, it replaced the small Templar stronghold of Le Destroit, a tower and enclosure complex dating to the twelfth century, about 1km to the northeast. The castle’s walls are extremely stout and were built with the soft local kurkar (fossilized sand originally washed into the Mediterranean from the Nile), quarried from a ridge less than 1km inland from the coast. The three towers of the outer line of defences have walls about 5m thick at the first level, but the eastern (outer) walls of the second level are only around a third as thick. This thinning was done to create what was essentially a continuous eastern casemate in the middle and southern towers, each of which serviced six embrasures, and two casemates in the northern tower, each with two embrasures. The curtain walls connecting ʿAtlit’s three outer towers are about 6m thick, solid at the first level with a gallery above, and an open fighting platform probably ran along the top at the third level. The mural gallery, which halved the thickness of the wall at the second level, provided access to casemates 3m wide, which were spaced less than 3m apart. Behind this line of defences, an inner wall dominated the outer fortifications and was in turn overlooked by two massive towers, more than 34m high with walls more than 5m thick at their bases.
Le Destroit, with ʿAtlit in the background. (Michael Fulton)
ʿAtlit, second level of the outer defences, first level of the inner defences (after Johns).
The castle did not impress a young T.E. Lawrence, better known later as Lawrence of Arabia, who remarked, ‘The strength of Athlit was brute strength, depending on the defenceless solidity of the inner wall, its impassable height, and the obstacle to mining of a deep sea-level ditch in the sand and rock before the towers. The design is simply unintelligent.’34 Lawrence’s assessment seems overly critical; however, his observations relating to the castle’s stoutness are accurate. The extremely thick walls of the outer line’s lowest level, its slight talus and use of extremely large ashlars were a defence against undermining. Above, the second level bristled with archers shooting from casemates, while another level of archers was supported along the top of the wall. Massed frontal attacks and sapping were clearly what the designer(s) of this line of defences feared most.
Behind, the two great three-storey towers of the inner wall were aligned between the outer towers, but the surviving east (outer) face of the north tower reveals evidence of few embrasures, suggesting these were quite palatial structures. The third level, which accounted for half the height of each tower, contained opulent rib vaulting, indicating they may have been added in a second building phase. A central pier helped support the vaulting of the second and third levels, but not that of the first below. Left floating, the pillar and all of the weight that it bore from the lofty levels above was supported only by the vaulting of the first level. Accordingly, it seems structural considerations contributed to the incredibly thick walls of these towers – their thickness allowed for their considerable height and magnificent third levels. The scale and grandeur of these towers would no doubt have impressed those within as well as those who viewed them from the outside.
Al- ʿĀdil’s building works display a similar preference for the ostentatious. At the citadel of Damascus, the massive mural towers he commissioned contain three levels, each a single open room, supported by walls 3–5m thick. Revealingly, it is the flanks of these towers, at either end of the long axis of the vaulting within, that were often the thickest, not the outer face, which would bear the brunt of an attack.
It seems the evident trend towards the thickening of tower walls was, at least in part, a result of the preference shown by certain early thirteenth-century figures, who possessed deep pockets, for awe-inspiring towers with equally impressive open spaces within. This can also be seen in parts of Europe. For example, as keeps in England and France became larger and taller through the second half of the twelfth century and first half of the thirteenth, so too did their walls become thicker in order to support these increasingly grand structures. Where defensive considerations more clearly contributed to the construction of exceptionally thick walls, as with the outer wall at ʿAtlit, this was typically done to frustrate miners and bring more defenders into the fight to help resist frontal attacks.
Ironically, many of the same principles that guided the designer(s) of ʿAtlit’s outer defences, and those of other Frankish castles in the aftermath of the battle of Hattin, can be seen in Mamlūk refortification efforts following the battle of ʿAyn Jālūt, as the threat of another Mongol invasion continued to loom. At Crac, Baybars commissioned a rounded tower at the southwest corner of the outer line of defences, where his miners had broken into the castle in 1271. The chosen design is telling: it has a solid base with a slight batter, creating a small talus, upon which sits a single elevated internal level, consisting of a single room that is arranged in an octagon around a central octagonal pier. A casemate extends into each of the four outward faces of the room, through walls 2.9m thick. This design was both sturdy and accommodated numerous defenders. A similar semi-circular tower, with walls 3.6m thick, was built along the outer southern line of defences at Ṣubayba: it has a solid, battered base and a single internal level, the southern half of which is arranged in a half-octagon while the northern half is rectangular, a shape mirrored by the central pier. Six casemates in the outward faces of the room provide access to embrasures. Another round tower was built at the east end of Ṣubayba, in front of the Ayyūbid citadel. The interior is once more octagonal, although lacking a central pier, and casemates extend into the five outward faces of the walls, which are 3.8m thick.
Mamlūk towers with central pillars (after Deschamps and Yovitchitch).
Although the shape of Mamlūk towers varied considerably, these design elements – an emphasis on solidity and providing protected spaces for archers – remained prevalent. Qalāwūn’s tower at Crac has a single internal level, at the height of the battlements crowning the adjacent walls. This is essentially a gallery: eight bays of groin vaults wrap around a large central pillar; the seven external bays provide access to nine casemates, servicing three embrasures in each of the tower’s outer faces. The same principles can be seen at Montreal, where six large towers were built around existing Frankish ones, each with a sturdy, if not solid base, upon which sat a level liberally provided with embrasures.
The desire to maximize the number of shooting positions within a tower at times overcame the advantages of an otherwise solid lower level. At Baalbek, the large tower built under Qalāwūn at the southeast corner of the Roman Temple of Bacchus is bristling with shooting positions. This philosophy was extended to the wall built around the western and southern sides of the larger temple complex, which had embrasures at two levels, and possibly a double-level parapet above. At Ṣubayba, the first level of the tower built by the Ayyūbids in the southwest corner of the outer bailey (tower 9) had eight embrasures; this increased to fourteen when the tower was enlarged by the Mamlūks, and numerous embrasures are discernible in the surviving masonry above. Below, additional shooting positions were created when two parallel pairs of galleries were incorporated into the sublevels that were built to compensate for the sloping ground around the tower.
Ṣubayba, outer southwestern tower (after Hartal).
From the late twelfth century, the thickness of some curtain walls also increased. Although this balanced the weakening caused by perforating the masonry with embrasures, it seems the main reason behind the thickening of many walls was to provide a broader fighting platform on top, or a mural gallery within, as at ʿAtlit. At Ṣubayba, the wall running north from the outer southwestern tower was thickened prior to 1260, creating spacious casemates similar to those at ʿAtlit behind earlier embrasures in the original wall. If the goal had been to strengthen these walls, much narrower casemates might have been employed, suggesting the primary objective of this thickening was instead to broaden the wall-walk above. This was also done along the wall leading away from the tower to the east, where the wall was thickened just enough to create springers. These allowed for the construction of a vaulted passage between the wall and the large reservoir behind the tower, which would have supported a broad fighting platform above.
Despite the increasing strength and sophistication of many castles, it can be easy to get carried away emphasizing the importance or prevalence of thick walls. The town walls of Caesarea, developed by Louis IX of France in 1251–52, were taken in less than a day by Baybars’ forces in 1265. Weeks later, these same forces, strengthened by the arrival of additional siege equipment, struggled for a month to take the town defences of Arsūf, some exposed sections of which are just 1.2m thick. Although defended by a Hospitaller garrison and boasting walls 3–3.5m thick, the castle surrendered only a few days after the town was captured. Elsewhere there was no progressive thickening of walls. At Beaufort, the thirteenth-century defences added by the Ayyūbids and Mamlūks appear to be no thicker than the walls of the twelfth-century keep (2.5m thick) and the stout enclosing wall to the south (3.6m thick). The twelfth-century town walls of Bānyās are 2.4–2.5m thick, similar to the Ayyūbid curtain walls of Ṣubayba and thinner than the 3.4m-thick Frankish walls of Tiberias built in the twelfth century. For added context, the Lascarid outer wall built around Nicaea in the first quarter of the thirteenth century is about 1.6–2m thick, considerably thinner than the main Roman wall built in the third century, which is upwards of 3.6m thick. All of these pale in comparison to the wall that surrounded Jerusalem around the eighth century BC, exposed sections of which are 6.4–7.2m thick.
Ṣubayba, topmost northeastern shooting chamber of the outer southwestern tower. (Michael Fulton)
Tower Shapes
Since antiquity, many military architects, including Vitruvius, had advocated the use of round or polygonal towers, rather than simpler quadrangular ones. Rounded towers had certain advantages: they were harder to undermine, as they had no vulnerable corners and weight was distributed more evenly; they provided defenders with a more complete field of view, accommodating embrasures in any direction; and they were more effective at deflecting artillery projectiles. They may also have been able to absorb the blows of siege engines more effectively, directing forces towards the centre of the tower, similar to how an arch supports a vertical load. However, rounded towers required more skill to build and the space within could be harder to use efficiently, leading some rounded towers to have rectangular interiors, as can be seen at the inner southwestern tower at Crac, the large ‘donjon’ at Margat and many of the mural towers built by Saladin at Cairo.
Roman engineers had built both quadrangular and rounded towers in the Levant, as had the Umayyads and early ʿAbbāsids, who employed rounded towers at Aqaba (Ayla), Qaṣr Kharana, Qaṣr al-Mshatta, the Roman theatre at Caesarea and Kafr Lām. Polygonal beaked towers were comparatively rare, although Byzantine examples can be found at Ankara, in Anatolia, and further away at Salona. Quadrangular towers were used with increasing frequency under the ʿAbbāsids and Fāṭimids, while Byzantine and Armenian builders employed both shapes, although rounded towers have become particularly synonymous with the fortifications of the latter.
At the time of the First Crusade, rectangular stone keeps were becoming popular in some areas of western Europe, although they were still vastly outnumbered by timber castles. The crusaders brought this tradition with them to the Levant, and almost all towers built by the Franks in the twelfth century were quadrangular. The rounded towers along the outer eastern wall of Saone are rare exceptions, suggesting their shape was determined by earlier Byzantine foundations or the influence of Armenian builders.
Rounded towers became more popular in the thirteenth century, although quadrangular towers were constructed at least as frequently by both Frankish and Muslim builders. Under Saladin, quadrangular towers were used at ʿAjlūn, both shapes were used at Qalʿat Ṣadr, and numerous rounded towers were employed during the refortification of Cairo. Although Al-ʿĀdil built a number of significant rounded towers, most of those he and his fellow Ayyūbids constructed had quadrangular plans. Under the Mamlūks, no clear preference was shown and often towers of both shapes were used at the same site. Both rounded and rectangular towers were built under Baybars at Ṣubayba, while Qalāwūn added quadrangular towers at Crac and Baalbek, and a rounded tower at Margat. Both rounded and quadrangular plans were also employed when Montreal was refortified in the final years of the century.
Although it is often suggested that the Templars preferred quadrangular towers and the Hospitallers favoured rounded ones, these characterizations do not extend far beyond the orders’ most impressive strongholds, which include ʿAtlit and Tortosa, Crac and Margat. Regardless of who built them, rural towers and those of smaller secondary castles were typically quadrangular. Although the advantages of rounded towers are often praised by historians, the continued use of quadrangular towers speaks to the limits of these benefits.
Fine ashlar masonry, ideally making use of a hard type of stone and finely cut blocks with little room between them, was often considered ideal to confront threats like sapping, as this made it difficult to damage the facing stones or pull them out of place. In practice, the availability of materials and skill, as well as the preferred aesthetic, had a profound influence on what walls eventually looked like. Volcanic types of stone were usually quite hard, but they could be just as difficult to work with. The basalt masonry at Margat is much rougher than that found at Belvoir, while the hard volcanic stone used by Frankish masons at Kerak provides a sharp contrast to the much finer limestone work of the Muslim builders who followed them. Along the Palestinian coast, the local kurkar was easy to work with but exceptionally soft, leading to the development of a shelly mortar that was harder and more resilient to the elements than the stone itself. In many regions, limestone was a popular choice. Relatively available, it offered a nice balance: it was fairly hard but also reasonably easy for a skilled stone worker to shape. Together with the thickness of a wall, the type of stone and quality of the masonry are fairly obvious indications of its strength. But what of its dressing – the finish or style of tooling left on the exterior of each stone?
For more than a century, archaeologists have studied the masonry of Frankish and Muslim fortifications with hopes of better understanding who was responsible for which building phases and what styles were preferred by certain figures. Both Frankish and Muslim builders, as well as their Armenian counterparts to the north, regularly made use of ashlars with smoothly drafted margins and raised centres, or bosses, a style which had been employed in the region since at least the Hellenistic period. All parties also used smoothly dressed ashlar, Frankish examples of which are commonly distinguishable by diagonal tooling, traces of a method of shaping stone imported from Europe. Masons’ marks, symbols incised on blocks by the men who shaped them, were long thought to be employed only by Frankish masons; however, such marks can also be found at certain Muslim sites, including ʿAjlūn and the citadel of Bosra, in both Ayyūbid and Mamlūk building phases.
A common explanation put forward for the continued use of, and even preference for, bossed masonry was that it was an effective defence against artillery. This, however, was not the case. If these bosses were intended to turn aside or deflect an incoming projectile, the implicit amount of force would be so meagre that even a sustained bombardment could hardly be seen as a threat to walls anywhere from 1.5m to over 4m thick. If the objective were simply to place more material between the point of impact and the rear of the wall, a simpler, and more reliable, solution would have been to avoid shaping a boss at all, leaving as much of the original stone in place and saving the time and effort of tooling the margins. Simply put, building thicker walls would have been a far more effective solution, and a period of prolonged bombardment would naturally create a rough and ‘deflective’ wall surface.
Bossed masonry was instead employed for practical and aesthetic reasons: such blocks were faster to shape than smooth ashlars and together they produced a rustic yet clean finish. At Jacob’s Ford, where walls were erected as quickly as possible, bossed masonry was used on the inner as well as outer faces of the exterior walls, suggesting most blocks were shaped at the quarry to a standard course height and then brought to the castle where they were employed wherever they were needed. The argument of expedience works at some sites but at others the bosses are quite flat and even finely shaped rather than rusticated.
Ayyūbid bosses tended to be quite irregular, except where smooth decorative bosses, resembling horizontal cylinders, were employed. These finer stones are often found embellishing the corners of towers, as at the citadels of Damascus and Bosra. Although Mamlūk masonry varies considerably, at sites such as Kerak and Ṣubayba their work is clearly distinguishable and much finer than that of their predecessors, often employing cleanly cut limestone blocks, either smoothly dressed or with neat, fairly flat bosses. Frankish sites reveal an assortment of boss styles, from the smooth examples at ʿAtlit and Saone to the rougher ones at Belvoir and Tortosa.
Like their dressing, the size of facing stones varied considerably from site to site and building phase to building phase. While colossal stones over 1m in length were employed at ʿAtlit, much smaller blocks, many just 20cm long, were used at Caesarea and Arsūf, similar coastal sites, 22km and 57km to the south respectively, where the same type of stone was used.
As the size of besieging forces grew, allowing attackers to send more men against the defences of a besieged stronghold, the importance of active defensive measures increased. These were often facilitated by architectural elements. Embrasures and crenellated parapets were the simplest, providing protection for archers and other defenders confronting advancing attackers, while posterns permitted rapid or secretive sallies. Regardless of which features were present, a truly active defence meant taking the fight to the besiegers and disrupting, delaying or defeating their attempts to gain entrance.
Sallies, attacks launched by defenders against their besiegers, were an integral part of siege warfare: a means through which the besieged could seize the initiative. These were generally small-scale operations, which often targeted a particular group of besiegers or the siege equipment they were employing. Due to the difficulties associated with destroying siege engines from within a stronghold, this task was often delegated to a sally party. The attack that followed was often launched rapidly, a quick strike to catch the besiegers unprepared, allowing them to burn the engine, whether a siege tower, trebuchet or penthouse, and withdraw before additional besieging forces could arrive to support their comrades.
At times, sallies by daring or desperate parties of defenders could have drastic consequences. Effective sallies broke Baldwin II’s siege of Bālis in 1122, the Sicilian siege of Alexandria in 1174 and Saladin’s siege of Tyre at the end of 1187. Taking pre-emptive measures, a force of defenders disrupted the crusaders who were intent on attacking Mount Tabor in 1217 and a sally from Margat deterred a more concerted Mamlūk siege around 1280. Less conclusively, a sally by the defenders of Aleppo in 1176 compelled the besiegers to content themselves with a blockade, eventually resulting in a truce, while a sally by the defenders of Dieudamor, on the brink of starvation, broke the besiegers’ blockade and extended the siege, which ultimately lasted almost a year before it was concluded in 1230. Although typically less decisive than these examples, the impact of sallies was still significant.
Along with the potential benefits, significant risks accompanied each sally, as it would implicitly involve opening a gate and attacking part of a much larger force. In 1101, Tancred’s ability to intercept and crush a sally left the defenders of Latakia so weakened that they were compelled to seek peace. According to William of Tyre, Bānyās was captured in 1157 when a party of defenders could not close a gate as they returned from a sally, allowing the besiegers pursuing them to gain entrance to the town. Later, during the siege of Alexandria in 1167, William asserts that Nūr al-Dīn’s forces, led by the 30-year-old Saladin, were reluctant to sally out against their Frankish besiegers, fearing an uprising in their absence by the Egyptian residents of the city, who had no more love for their ‘defenders’ than for the Franks. Although this highlights another risk associated with conducting sallies, the predominantly Sunni population of Alexandria was probably more sympathetic to the Syrian defenders than William lets on.
One day our men took counsel and decided to make a general sally on all sides with horse and foot, to burn the buches. So my lord the master of the Temple and his men, and Sir John [of Grailly and Sir Otto] of Grandson and other knights went out one night from the Templars’ sector (which ran from the seaside to the Gate of St Lazarus), and the master ordered a Provencal, who was viscount of the bourg of Acre, to set fire to the wooden buches of the great engine of the sultan. They went out that night, and came up to these buches, but the man who was supposed to hurl the Greek fire was afraid when he threw it, and it fell short and landed on the ground where it burned out. The Saracens who were there were all killed, horsemen and footmen. But our men, both brethren and secular knights, went so far in among the tents that their horses got their legs tangled in the tent ropes and went sprawling, whereupon the Saracens slew them. In this way we lost eighteen horsemen that night, both brethren of the Temple and secular knights, though they did capture a number of Saracen shields and bucklers and trumpets and drums. Then my lord and his men turned back towards Acre.
(Adapted from Templar of Tyre 491 [255], trans. Crawford, p. 107.)
Few fortifications could deter a determined and experienced group of sappers if they were allowed to go about their work uninterrupted. When sappers chose to tunnel down to the foundations of a stronghold, and the entrances to their tunnels were protected against sallies, defenders had few options but to countermine: to dig their own tunnel, hoping to intercept that of the besiegers. During Saladin’s brief siege of Beirut in 1182, his archers aggressively showered the city’s defenders with arrows, allowing his miners to begin working undisturbed. By the time Saladin withdrew, after a siege of perhaps only three days, the defenders had already begun to countermine. Nine years later, as the siege of Acre ran into its third year, and the arrival of Philip II of France and Richard I of England tipped the balance in favour of the besiegers, the crusaders launched new sapping initiatives. A tunnel excavated by the defenders intercepted the work of French sappers below the Accursed Tower. The meeting resulted in an agreement that both sides would withdraw and the Muslims sealed their tunnel behind them. Another countermine intercepted Richard’s sappers working below a different tower, but the meeting was far less cordial and the defenders, who gained the upper hand in the subterranean fight that ensued, forced the crusaders to abandon their mine.
Fifty years after Saladin’s siege of Beirut, John of Beirut led a force to relieve his titular capital, then besieged by imperialist forces. Although too small to break the siege, a part of his force managed to enter the castle and bring much-needed assistance. The defenders then countermined, driving the imperial sappers from their mine and regaining control of the fosse, destroying the defences that the besiegers had set up there. Free from this threat, the defenders maintained their initiative, sallying out against the besiegers and successfully burning a number of their engines.
In 1265, the Frankish defenders of Arsūf gained some short-lived success following the excavation of a countermine during Baybars’ siege of the stronghold. Tunnelling out from the castle, they were able to set fire to the brush that the Mamlūks had used to fill the fosse, using casks of fat to make sure the fire took hold and then fanning it with bellows. Just weeks earlier, and with greater success, the defenders of al-Bīra had similarly tunnelled into their fosse, allowing them to burn the material the Mongols had used to fill it. The following year, at the siege of Safed, the Franks again countermined against Baybars’ sappers. On this occasion, their tunnel was above that of the besiegers. The Franks dropped into the Muslim tunnel from above and a bitter hand- to-hand fight took place. Although it was a setback for the Mamlūks, this was only one of a number of mines that they had opened and the defenders of Safed, like those of Arsūf, were later compelled to seek terms. At the siege of Acre in 1291, the Franks similarly attempted to countermine against the Mamlūk sappers undermining the Tower of the Countess of Blois, one of a number of sapping parties working against the northeastern section of the city.
Countermining was not typically the first choice of besiegers, as it was dangerous and involved potentially weakening their own defences. If deprived of other options, however, it could be extremely effective.
One of the most effective defensive weapons was cash. In the same way that besiegers could entice a garrison to surrender with generous terms or offer its leader an alternative command, defenders could offer the besiegers a payoff in exchange for lifting their siege. During the First Crusade, the ruler of Tripoli persuaded Raymond of St Gilles to give up the siege of ʿArqa in 1099 with a payoff, albeit after the Franks had become discouraged and most were inclined to continue their march towards Jerusalem. Other coastal towns placated the crusaders by opening their markets to them and even offering gifts to encourage them to move on as quickly and peacefully as possible. This remained a popular defensive measure in the decades that followed, particularly in the fractured political landscape of the early twelfth century.
When Raymond of St Gilles returned to the Tripoli region following the disastrous Crusade of 1101, he once more accepted a large payoff. Agreeing not to attack Tripoli for the time being, he left to besiege other towns in the surrounding area. When Baldwin I took a force against Tyre in 1108, he opted not to launch an aggressive siege, in part because he lacked naval support, and eventually accepted a payoff of 7,000 dinars to lead his army away from the city. After taking Beirut in the spring of 1110, Baldwin made a brief appearance before Sidon, departing without attacking the town in exchange for an increase in the city’s annual tribute from 2,000 dinars to 6,000. This had probably been a feint, which ended up paying off for the Frankish king: Edessa was then under siege by Mawdūd, suggesting Baldwin had no real intention of besieging Sidon before moving to relieve Edessa. Like the coastal cities, smaller regional powers could secure their autonomy, at least for the time being, and deflect attention with payoffs. The rulers of Shayzar, who commanded a small Arab enclave sandwiched between various larger Frankish and Muslim powers, bought off Tancred for 10,000 dinars in 1109, paid Shams al-Mulūk Ismāʿīl of Damascus to break his siege in 1133, and persuaded the Byzantine emperor John Comnenus, who was receiving little help from his Frankish allies, to lift his siege and depart with a large sum in 1138.
The Franks frequently accepted payoffs to lift sieges around Aleppo. Following the death of Riḍwān in 1113, Aleppo suffered internally for half a decade, allowing the Franks to make considerable gains until Īlghāzī ibn Artuq took control of the city. The Franks pushed their advantage while Īlghāzī consolidated his hold over Aleppo, refusing a payoff to lift their siege of ʿAzāz and then extracting a significant tribute as a condition of a broader peace that was arranged after the town had fallen. Although the balance of power shifted in 1119, in the aftermath of the battle of the Field of Blood, Joscelin I of Edessa was able to besiege Buzāʿa in early 1121. After burning a section of its walls, perhaps indicating they had been undermined, he was bought off and withdrew. After Īlghāzī’s death the next year, the Franks launched another series of campaigns into Aleppan territory through 1122, accepting tribute from some regions while rejecting a payoff from Bālis, which turned out to be a mistake when a sally by a group of Turkoman defenders later broke the Frankish siege. In 1123, a combined Frankish force moved to relieve Baldwin II, who had seized the castle of Kharpūt from his jailers. Learning that Balak had retaken the castle while they were still on their way, the Franks turned on Aleppo and raided its suburbs for three or four days, accepting payment to leave according to one source. When control of Aleppo was once more contested in the autumn of 1127, Joscelin took advantage of the situation and appeared in force before the city. Although he probably had no intention of attacking the formidable city directly, his presence was sufficient to encourage a payoff.
Order returned to Aleppo when Zankī took control of the city in 1128. Up to this point, the Franks had taken a share of all Aleppan revenues right up to the city gates, but this would soon change. Zankī quickly went on the offensive, taking al-Athārib and investing Ḥārim, compelling the Franks to offer him half of the town’s revenue to end the siege a nd accept a truce. Suspiciously, Ibn al-Athīr gives a similar account of an attack made by Nūr al-Dīn twenty-seven years later. In 1149, Nūr al-Dīn encamped outside Antioch and opened serious negotiations for the city’s surrender; however, he was ultimately forced to settle for a payoff. The same year, Joscelin II, now only titular count of Edessa, bought off the Turkish forces besieging him in Turbessel, benefiting from the leverage provided by the Frankish army marching to his relief under Humphrey II of Toron, constable of the kingdom of Jerusalem.
As the twelfth century progressed, and Nūr al-Dīn continued to consolidate his power, the stakes and payoffs became larger. In late 1168, Amalric accepted a promise of 1,000,000 dinars not to besiege Cairo. Payment, however, was delayed long enough for Shīrkūh to arrive, compelling the Franks to withdraw towards Palestine in January 1169, taking with them only an initial deposit of 100,000 dinars. Nūr al-Dīn died in 1174, prompting the Franks to attack Bānyās. The siege lasted around two weeks before the Franks accepted a payoff of cash and prisoners. By accepting the peace, the Franks also mitigated the risk of a large Mesopotamian or Egyptian army invading western Syria under the pretext of supporting Nūr al-Dīn’s young son. In late 1177, with Saladin attending to affairs in Egypt, the crusading Philip of Flanders, accompanied by Raymond III of Tripoli and Bohemond III of Antioch, briefly attacked Hama before investing Ḥārim. The latter siege dragged on for four months through the winter until the Franks were bought off by al-Ṣāliḥ Ismāʿīl ibn Nūr al-Dīn of Aleppo, who then besieged the castle himself. Exhausted, the garrison quickly surrendered.
Payoffs were comparatively rare during the thirteenth century: the Franks were rarely in a position to undertake significant siege operations and it was position rather than cash that mattered most to the various Ayyūbid princes as they struggled for greater power at the expense of their relatives. During the Fifth Crusade, the Franks refused what was probably the largest payoff offered during the period of the crusades. In 1219, during the siege of Damietta, al-Kāmil proposed to return Palestine to the Franks if the crusaders would agree to leave Egypt. Perhaps confident that they could seriously threaten Cairo, or wary that al-Kāmil’s offer excluded Kerak and Montreal, the deal was declined. Fear that the Franks might conclude just such an agreement led al-Kāmil’s brother, al-Muʿaẓẓam ʿĪsā of Damascus, to slight most of the fortifications west of the Jordan. Much as the Franks had exploited the weakness of the Fāṭimid towns along the coast in the early twelfth century, it was the Franks who found themselves in this position following the rise of the Mamlūks. Although Baybars showed himself willing to negotiate with the Franks, allowing them to buy his good will, he more often sought land than cash and rarely showed himself willing to negotiate terms that involved his withdrawal once hostilities had begun.
* * *
Although different defensive elements and systems came into fashion as the popularity of others declined, the design of each stronghold, rebuilding initiative or repair effort was influenced by a unique set of variables and considerations. Certain similarities and stylistic preferences can be seen among the defences commissioned by particular rulers, but these were incorporated only so far as local conditions and resources, including building materials, cash, skill and manpower, allowed. A variety of architectural elements might be employed, depending on the nature of the threats the stronghold was expected to face, but active defensive measures were often required to counter a more determined adversary.
Defence was, to a degree, an act of desperation, a fight for survival, and there were no guarantees that terms would be offered or honoured by besiegers – Zankī had the defenders of Baalbek executed in 1139, despite having promised them safe-conduct. Fortifications were thus planned pragmatically, if also artistically in some instances. But sieges, if they ever materialized, were brief periods of time in the history of these structures. Defences were thus also designed around creating spaces intended to fulfil various functions. In the thirteenth century, this led to the construction of thicker walls, as towers were built larger and taller with much greater volumes of open space inside. There was no simple formula that determined the design of fortifications, much as there was none that guided the conduct of sieges.