Open this page in a new tab

Al-Marqab Citadel

Al-Marqab citadel Figure 2

Al-Marqab citadel, as seen by attacking enemy from the south, was mostly built by the crusading Order of St. John (Hospitallers) in the late twelfth century. The hilltop plateau, 350 m above sea level, is vesicular basalt lava of Pliocene age, exposed below the highest towers. Terraces of olive gardens carved in weathered basaltic strata cover the slopes. Banyas city and the Mediterranean Sea are seen in the background on the right

Kázmér and Major (2010)


Names

Transliterated Name Source Name
Qalat Al-Marqab Arabic قلعة المرقب
Al-Marqab Arabic قلعة ا
Margat Latin
Introduction
Introduction

Kázmér and Major (2010:183) identified and dated V-shaped and U-shaped failures, single-corner and symmetrical corner collapses, and in-plane shifts of ashlar masonry walls of the al-Marqab Citadel using historical and archaeological methods in an archaeoseismic study of the castle. The citadel is perched on top of a 350-m-high volcanic mountain, ~2 km inland from the coast, overlooking the town of Banyas and guarding the coastal route. Masonry walls were made with and without mortar, using the opus caementum technology (Roman concrete) ( Kázmér and Major, 2010:185-186). Many of the walls, including the donjon wall have an outer ashlar facade or shell with a poured concrete and rubble interior wall. This site may be subject to a Ridge or Slope Effect.

History of the Site

The first castle of the site is reported to have been built by the local inhabitants in H. 454 (A.D. 1062–1063). After a brief period of Byzantine occupation starting around 1104, it was taken by the Franks (Crusaders) from the local tribes in 1117–1118. The castle seems to have reverted to as-yet-unknown Muslim hands in the 1130s during the civil war in Antiochia. It was recaptured by Renaud II Mazoir in 1140, and then became the seat of the Mazoir family (Deschamps, 1973, p. 260–261). The Mazoirs were one of the highest-ranking baronial families in the Crusader principality of Antioch and were responsible for building most of the earliest surviving structures in the castle. In early February 1187, the Mazoirs transferred Al-Marqab and all their landed properties to the Order of St. John (Hospitallers) due to unbearable maintenance costs related to warfare and damaging earthquakes (Burgtorf, 2007). The Hospitallers turned the castle a military, judicial, and administrative center of the region.

Given the fact that Al-Marqab became the central castle of one of the most influential organizations of the age, it is not surprising that the Hospitaller period was characterized by large scale construction programs that resulted in the erection of most of the surviving buildings seen on the mountain top. The castle was put to the test several times by besieging armies. It was besieged by an army from Aleppo in 1204–1205, and again in 1231. Banyas and the lands around the castle were destroyed by the Aleppines. Attempts on the Muslim side to take the castle twice, once in 1269–1270, and subsequently in 1281 and 1282, ended in failure. Al-Marqab was finally taken by the Sultan Qalawun on 25 May 1285 after a relatively short siege of 5 weeks. The sultan agreed to the peace offer of the garrison to save the castle from further damage, and the destructions caused by the siege were repaired immediately (Ibn-Abdazzahir, 1946). After the complete expulsion of the Crusaders, the castle started losing its importance, and its diminishing role in the Mamluk (1250–1517) and Ottoman periods (1517–1917) is reflected in the reduced scale of later building activities. For a lively description of the castle and the function of various buildings, see Kennedy (1994, p. 163–179). For additional details concerning the history of AlMarqab, the reader is referred to Major et al. (2010).

Maps, Aerial Views, and Plans
Maps, Aerial Views, and Plans

Maps and Aerial Views

  • Fig. 1 - Location Map from Kázmér and Major (2010)
  • Al-Marqab Citadel in Google Earth

Plans

Normal Size

  • Fig. 3 - Plan of the southern portion of Al-Marqab citadel from Kázmér and Major (2010)

Magnified

  • Fig. 3 - Plan of the southern portion of Al-Marqab citadel from Kázmér and Major (2010)

Chronology
Dating Damages

Figures
Figures - Dedicatory Inscription

  • Fig. 12 Dedicatory Inscription from A.H. 684 (9 March 1285 - 26 Feb. 1286 CE) - from Kázmér and Major (2010)

Discussion

The first archaeological excavation in Al-Marqab started in 2007; therefore, a large proportion of the castle fabric is still undated. However, a relative chronology (architectural stratigraphy as understood by Galadini et al., 2006) can be readily established for the buildings studied in this project. Muslim-built portions of the circumferential wall, especially the southernmost tower in the outer enceinte, are decorated with a frieze-like white row of ashlars within the black basalt wall. These blocks bear an Arabic inscription, testifying to its construction by the sultan Al-Mansur Qalawun (Mamluk sultan from 1279 to 1290), who took the castle in 1285 (Fig. 12).

The donjon is certainly a Hospitaller construction and thus dates from after the order acquired the castle in 1187. Besides architectural design and the sheer size of the building, which could hardly have been financed by a private lord, the first results of the geophysical surveys also seem to support this dating. A georadar survey carried out inside the castle chapel (unequivocally accepted to have been the first Hospitaller construction on the site) detected the contours of a rectangular structure. Its position and the thickness of its wall, exceeding 3 m, make it a likely candidate for being the residential tower of the Mazoir family. This assumption is further strengthened by the presence of an old cistern incorporated in the southern walls of the chapel that stands in the center of the conjectured Mazoir tower. The presence of rock-cut cisterns under the main towers of eminent Crusader castles in the twelfth century is very common. Because the defense of the southern part of the castle mountain requires the presence of a dominant building, the substitution of the Mazoir square tower by the chapel must have been closely followed by the building of the new donjon, which is likely to have taken place in the 1190s.

The supposed construction date of the donjon soon after 1187 and the precise dating of the southern corner tower after the successful Muslim siege of 1285 put constraints on dating the earthquake damages.

Earthquake I - 1187-1285 CE

Kázmér and Major (2010:194) report the following:

Earthquake 1 produced the V-shaped extrusion on top of the donjon (60°–240°). This earthquake occurred after the donjon was completed and before the southern tower was built: there are no traces at all of this damage direction on the southern tower. Earthquake 1 occurred during the interval between 1187 and 1285, after Hospitallers took the castle and before Mamluk occupation. A candidate earthquake is that of 1202, this being the largest in the Middle East ever recorded (see Table 1).

While caution must be exercised in assigning damage azimuth to epicenter direction, according to Ambraseys and Melville (1988), the epicenter of the 1202 earthquake was south of Al-Marqab, in the Bekaa Valley, while all major successive earthquakes had their epicenters to the north, near Aleppo (see also Fig. 1).
Kázmér and Major (2010:196) also stated
Earthquake 1 consisted of vibration in SW-NE plane, damaging the donjon and room M3. It was a major event between 1187 and 1285, possibly the 1202 earthquake.

Earthquake II - after 1285 CE

Kázmér and Major (2010:194) report the following:

Earthquake 2 produced the U-shaped damage to the southern corner tower. Additionally an extension of the top of the tower and of the box machicolation occurred in 120–300° direction. We can give only a terminus post quem date: it happened after 1285, i.e., during the Muslim period of Al-Marqab. In addition, a relative intensity of this quake would be lesser than that of the 20 May 1202, since it did not cause any visible damage to the donjon.
Kázmér and Major (2010:196) also stated
Earthquake 2 consisted of vibration in NE-SW plane. It damaged the southern tower + NW corner of the donjon. It was also a major but lesser event than number 1, and it occurred after 1285. Candidates are the 1404 and 1759 events reported in Sbeinati et al. (2005).

Seismic Effects
Earthquake I - 1187-1285 CE

Effect Location Image (s) Comments
V-Shaped Failure/Extension SW sector of the donjon wall

  • There is a spectacular V-shaped extrusion on thedonjon, the main tower of the citadel of 5 m height and 5 m width (Fig. 5). Similar features occur elsewhere in the castle. It seems that if failure were to progress, we would see a wedge-shaped block missing from the wall of the donjon.

    The V-shaped block is shifted toward the SW by ~20 cm. No trace of it can be seen within the donjon. Bounding surfaces are joined before reaching the hall inside.
    - Kázmér and Major (2010)

  • in the upper part of the donjon (Fig. 4C). While the other fractures crossing the residential tower can be observed both on the outside and inside of the wall, this several-meter-long body is only visible from the outside. The bounding fractures thus meet inside the wall. It does not start down from the top of AV either, there is a break in the direction of the foundation, so this cannot be the result of uneven subsidence. Probably during an earthquake, a larger piece broke off from the upper, strongly swaying part of the tower and moved according to the direction of the vibration. - Kázmér (2008)
Symmetrical Corner Collapse A. NE corner of room M3
B. SW corner of room M3

Room M3 sits on top of the vault of the kitchen. It is the sole remnant of a previous, larger cluster of rooms, which might have served as an independent kitchen. Walls that are 66 to 104 cm wide bear a barrel vault. Diagonally opposite corners have suffered symmetrical damages (Figs. 7–8). Fractures that are concave outward have developed. The NE corner collapsed in full, destroying a segment of the vault and portions of the adjacent wall (Fig. 7A). The concave fault developed in the SW corner as well, but only part of the vault collapsed: there is a 2 × 1 m hole in the top of the vault, connected by an arcuate fracture — a would be failure scar — to the still-intact adjacent walls (Fig. 7B). - Kázmér and Major (2010)

Earthquake II - after 1285 CE

Effect Location Image (s) Comments
U Shaped Gap               Top of Mamluk tower
The top of the southern corner tower of the outer enceinte, the outer ring of walls of the Mamluk-built structure, bears a downward-concave failure. Both thin and thick portions of the tower have failed (Fig. 9). - Kázmér and Major (2010)

All Seismic Effects

Effect Location Image (s) Comments
V-Shaped Failure Donjon masonry wall
There is a spectacular V-shaped extrusion on thedonjon, the main tower of the citadel of 5 m height and 5 m width (Fig. 5). Similar features occur elsewhere in the castle. It seems that if failure were to progress, we would see a wedge-shaped block missing from the wall of the donjon.

The V-shaped block is shifted toward the SW by ~20 cm. No trace of it can be seen within the donjon. Bounding surfaces are joined before reaching the hall inside.
- Kázmér and Major (2010)
Single Corner Collapse NW corner of the donjon
Adjoining, possible perpendicular walls have collapsed at their joining. Collapse occurs where both walls are free-standing, i.e., unconfined at least to one side. This partial collapse produces an uneven oblique surface, cutting both walls at an angle (Fig. 6). Although of irregular shape, the pattern of collapse is comparable to a failure plane that can be interpreted as a normal fault. The smoothed surface of the failure is considered the fault plane, where the two directions necessary for geological characterization, strike and dip, can be measured and/or calculated. Because we do not have any evidence for the displacement direction of the hanging wall (fallen fragments have been cleared centuries ago), we assume dip slip. - Kázmér and Major (2010)
Symmetrical Corner Collapse A. NE corner of room M3
B. SW corner of room M3

Room M3 sits on top of the vault of the kitchen. It is the sole remnant of a previous, larger cluster of rooms, which might have served as an independent kitchen. Walls that are 66 to 104 cm wide bear a barrel vault. Diagonally opposite corners have suffered symmetrical damages (Figs. 7–8). Fractures that are concave outward have developed. The NE corner collapsed in full, destroying a segment of the vault and portions of the adjacent wall (Fig. 7A). The concave fault developed in the SW corner as well, but only part of the vault collapsed: there is a 2 × 1 m hole in the top of the vault, connected by an arcuate fracture — a would be failure scar — to the still-intact adjacent walls (Fig. 7B). - Kázmér and Major (2010)
U-Shaped Gap Top of Mamluk tower
The top of the southern corner tower of the outer enceinte, the outer ring of walls of the Mamluk-built structure, bears a downward-concave failure. Both thin and thick portions of the tower have failed (Fig. 9). - Kázmér and Major (2010)
Dislodged Building Blocks Mamluk tower
A large variety of shifted and rotated building blocks (ashlars) are seen at Al-Marqab. A shift within the plane of the wall is spectacularly shown in Figure 10. Heavily protected stone boxes extrude from the top of walls. Open bottoms allowed defenders to pour hot water, oil, or burning tar on attackers climbing the walls.

Box machicolation and adjacent walls on top of the southern Mamluk tower suffered in-plane extension of several tenths of a meter, and open spaces up to 10 cm wide formed between adjoining blocks during ground shaking. Although an indirect observation, this type of damage is confidently assigned to earthquakes, even by the otherwise cautious Ambraseys (2006, p. 1010). Similar open joints are described by Sintubin et al. (2003, their Fig. 5a) and Marco (2008, his Fig. 2F), and have been reproduced by vibration experiments (Vasconcelos et al., 2006, their Fig. 7).

There are many other kinds of damages observed in Al-Marqab: dropped keystones, in-plane and out-of-plane failures of walls, twisted walls, rotated blocks, extruded blocks, etc., which will be treated separately.
- Kázmér and Major (2010)
Subsoil Donjon and later addition
The buildings and walls of Al-Marqab have been erected on the solid subsoil of a several-meter-thick layer of compact Pliocene basalt (Fig. 11). This rock is not prone to liquefaction, even under major earthquakes, and neither is it affected by compaction under changing groundwater level (Ambraseys, 2006). The latter is ~50 m below the citadel, as shown by the location of the public bath on the western hillside

The heaviest possible damages inflicted by pre-gunpowder warfare were created by trebuchets (highly evolved catapults), throwing stone balls up to several hundred kilograms in weight against walls and onto roofs. The southern side of the donjon wall, most exposed to incoming projectiles, bears only minor fractures of conchoidal shape, witnesses of minor hits.

The only really efficient siege tactic, mining, yielded collapse of walls. This method helped Sultan Qalawun’s army to take Al-Marqab in 1285 by undermining the southern tower. No traces of the mine were found. This gravity-induced failure, subsidence, has different geometrical features than those yielded by lateral seismic shaking.

A common source of damage, original construction defects, can be excluded by examining the surviving portions of the southern sector of Al-Marqab citadel. Mortar is still rock-solid in the failed walls. Textbook examples of subsidence are missing. Therefore, a seismic origin of damages is highly probable.
- Kázmér and Major (2010)
Broken Stones - Broken top of a window Donjon W side, adjacent to hall N3
The bridging of practically all openings (windows, doors) is broken (Figure 3A). By definition, the break occurs in the middle of the long stone beams. The earthquake origin of the break has not been proven — the long stone beams, which are thin in relation to their length, are anyway sensitive to tension and break easily. After taking the dimensions (also modeling the weight on the beam), the earthquake origin could be confirmed or refuted by the basalt fracture test. - Kázmér (2008)
Slipped blocks in vaulted openings - Fallen ashlars in arch above gate Western outer gate, SW tower
Some of the stones carved into a wedge shape, but usually fitted together without a binder, tend to slide lower than the others. Displacements of this kind can only occur in arches carrying a relatively small load. If we observe similar phenomena in heavily loaded vaults, they could only have occurred after the load above them collapsed. In arches with a symmetrical load, the keystone slides down, in the case of an asymmetrical load, one of the side stones. It has been proven with both loading experiments and computer modeling that this kind of deformation can only occur in the case of strong earthquakes (Marco 2008: 149-150, with detailed literature references). - Kázmér (2008)
Horizontally dislodged blockstones - Shaken ashlars of machicolation Qalaun tower, S side
(Fig. 3C, D). This displacement is caused by the vertical component of the seismic waves. The rock block, or the load on it increases, so friction is reduced to a minimum and the block moves easily due to the horizontal components (Marco 2008: 150). - Kázmér (2008)
Horizontally dislodged blockstones - Shaken ashlars of machicolation Southern bastion, W side
(Fig. 3C, D). This displacement is caused by the vertical component of the seismic waves. The rock block, or the load on it increases, so friction is reduced to a minimum and the block moves easily due to the horizontal components (Marco 2008: 150). - Kázmér (2008)
Clockwise rotated ashlars Southern bastion, W side
(Fig. 3E, F). A special case of the previous deformation; translation and rotation often occur together. The rotation can be right (clockwise) or left (counter-clockwise) (Korjenkov & Mazor 1999, Fig. 20). - Kázmér (2008)
Clockwise rotated ashlars Qalaun tower, S side
(Fig. 3E, F). A special case of the previous deformation; translation and rotation often occur together. The rotation can be right (clockwise) or left (counter-clockwise) (Korjenkov & Mazor 1999, Fig. 20). - Kázmér (2008)
V-shaped extension SW sector of the donjon wall
in the upper part of the donjon (Fig. 4C). While the other fractures crossing the residential tower can be observed both on the outside and inside of the wall, this several-meter-long body is only visible from the outside. The bounding fractures thus meet inside the wall. It does not start down from the top of AV either, there is a break in the direction of the foundation, so this cannot be the result of uneven subsidence. Probably during an earthquake, a larger piece broke off from the upper, strongly swaying part of the tower and moved according to the direction of the vibration. - Kázmér (2008)
Arch Fractures vault of the chapel along the western wall
perpendicular to the longitudinal axis (Figure 4D). The depicted crack system appears in both fields of the two-bay vault of the castle chapel, along the connection with all four retaining walls. Therefore, it cannot be considered a reflection of a foundation problem, but rather the result of a force that affected all the walls. - Kázmér (2008)

Background info on the structure

Effect Location Image (s) Comments
Structure of Masonry Wall Windward wall of the donjon
Ashlars in the western, windward wall of the donjon are seemingly unsupported. However, their rear side is firmly embedded in Roman concrete, the cementing material of the several-meter-thick wall. Laid initially with mortar, westerly winds and rain have removed much of it throughout eight centuries. - Kázmér and Major (2010)
Structure of Masonry Wall southern window of the chapel
Three-leaf wall as exposed in the southern window of the chapel. An external regular ashlar work served during construction as a mould for casting the core. The core is agglomerate of stones and mortar - Kázmér (2008)
Structure of Masonry Wall external wall of the donjon
Basalt ashlars on the external wall of the donjon. This is only a mould supporting the agglomerate core during casting. Ashlars frequently are not joining each other, or are separated by thin flakes of basalt, unsuitable to bear the weight of the donjon. Rear side of ashlars is embedded in mortar of the core. - Kázmér (2008)

Deformation Maps
Earthquake I - 1187-1285 CE

Deformation Map

Modified by JW from Fig.s 3, 5, and 8 of Kázmér and Major (2010)

Earthquake II - after 1285 CE

Deformation Map

Modified by JW from Fig.s 3 and 9 of Kázmér and Major (2010)

Intensity Estimates
Earthquake I - 1187-1285 CE

Intensity Estimate from Earthquake Archaeological Effects (EAE) Chart

Effect Location Image (s) Comments Intensity
V-Shaped Failure/Extension - Displaced Masonry blocks SW sector of the donjon wall

  • There is a spectacular V-shaped extrusion on thedonjon, the main tower of the citadel of 5 m height and 5 m width (Fig. 5). Similar features occur elsewhere in the castle. It seems that if failure were to progress, we would see a wedge-shaped block missing from the wall of the donjon.

    The V-shaped block is shifted toward the SW by ~20 cm. No trace of it can be seen within the donjon. Bounding surfaces are joined before reaching the hall inside.
    - Kázmér and Major (2010)

  • in the upper part of the donjon (Fig. 4C). While the other fractures crossing the residential tower can be observed both on the outside and inside of the wall, this several-meter-long body is only visible from the outside. The bounding fractures thus meet inside the wall. It does not start down from the top of AV either, there is a break in the direction of the foundation, so this cannot be the result of uneven subsidence. Probably during an earthquake, a larger piece broke off from the upper, strongly swaying part of the tower and moved according to the direction of the vibration. - Kázmér (2008)
VIII +
Symmetrical Corner Collapse - Wall Collapse A. NE corner of room M3
B. SW corner of room M3

Room M3 sits on top of the vault of the kitchen. It is the sole remnant of a previous, larger cluster of rooms, which might have served as an independent kitchen. Walls that are 66 to 104 cm wide bear a barrel vault. Diagonally opposite corners have suffered symmetrical damages (Figs. 7–8). Fractures that are concave outward have developed. The NE corner collapsed in full, destroying a segment of the vault and portions of the adjacent wall (Fig. 7A). The concave fault developed in the SW corner as well, but only part of the vault collapsed: there is a 2 × 1 m hole in the top of the vault, connected by an arcuate fracture — a would be failure scar — to the still-intact adjacent walls (Fig. 7B). - Kázmér and Major (2010) VIII +
This archeoseismic evidence requires a minimum Intensity of VIII (8) when using the Earthquake Archeological Effects chart of Rodríguez-Pascua et al (2013: 221-224). Although this site may be subject to a ridge effect, the effects are severe enough to suggest a minimum intensity of VIII (8).

Seismic Parameters from Kázmér and Major (2010)

Kázmér and Major (2010:196) noted that the donjon walls were up to 5 m thick and thus highly earthquake resistant. The walls also showed no evidence of being damaged by earthquakes after the 1st earthquake. Based on this and Geoffrey of Donjon’s description of damage to the Citadel (Heavily damaged but still functional for military purposes), they suggested that an Intensity of VIII-IX while stating that estimating Intensity as high as IX (9) might be too heavy a statement — we did not observe any buildings yet shifted off their foundations. Kázmér and Major (2010) did not consider the possibility of a Ridge or Slope Effect. The azimuth of displacement for the first earthquake is NE-SW (60°–240°).

Earthquake II - after 1285 CE

Intensity Estimate from Earthquake Archaeological Effects (EAE) Chart

Effect Location Image (s) Comments Intensity
U Shaped Gap - Collapsed Walls Top of Mamluk tower
The top of the southern corner tower of the outer enceinte, the outer ring of walls of the Mamluk-built structure, bears a downward-concave failure. Both thin and thick portions of the tower have failed (Fig. 9). - Kázmér and Major (2010) VIII+
Although the archeoseismic evidence requires a minimum Intensity of VIII (8) when using the Earthquake Archeological Effects chart of Rodríguez-Pascua et al (2013: 221-224), since the wall was only partially collapsed and this site may be subject to a ridge effect, the minimum Intensity might be better estimated at VII (7).

Seismic Parameters from Kázmér and Major (2010)

Kázmér and Major (2010) did not estimate Intensity for the second earthquake however Kázmér and Major (2010:194) noted that the relative intensity of Earthquake II should be less than that of Earthquake I since it did not cause any visible damage to the donjon . Kázmér and Major (2010) did not consider the possibility of a Ridge or Slope Effect. The azimuth of displacement for the second earthquake is NW-SE (120–300°).

Notes and Further Reading
References
Historical Earthquakes reported or suspected to have affected al-Marqab Citadel

Table

Historical Earthquakes Table 1

HISTORICAL EARTHQUAKES REPORTED OR SUSPECTED TO HAVE AFFECTED AL-MARQAB CITADEL (MODIFIED AFTER AMBRASEYS, 2009)

Kázmér and Major (2010)


Chart

Historical Earthquakes Figure 13

Dating of major earthquake damages in the history of AlMarqab citadel. Known earthquakes are listed after Sbeinati et al. (2005); most damaging seismic events are underlined. Double arrows refer to vibration directions as calculated from orientation of failures. These display an earlier, 60°–240° direction as shown by V-shaped extrusion of the donjon and symmetrical extensional failure of room M3 (Figs. 7–8). This is probably due to the 20 May 1202 earthquake. The later, 310°–130°-directed vibration is seen on the southern tower, built during the Muslim period (Fig. 9). It occurred any time after 1285 and may be correlated to the 1404 (and/or 1408?) earthquake.

Kázmér and Major (2010)


Construction Techniques

Most walls of Al-Marqab, both Crusader and Muslim, are one of two types: either stone masonry or opus caementitium, i.e., “Roman concrete” (Lamprecht, 2001) or “ancient concrete” (Ferretti and Bažant, 2006). Stone masonry is characterized by dressed stones, carved rectangular and of standard size, with or without mortar, always without metal anchors. Arches, door, and window ledges, box machicolations, and some wall heads have been constructed this way.

“Roman concrete” or “ancient concrete” is a mixture of sand, lime, and added stone material and is very similar to modern concrete in appearance. Invented by the Romans, the technique survived well into the Middle Ages. Opus caementitium is often combined with traditional masonry, where an outer, visible layer of variously dressed blocks was erected with mortar. This external, regular masonry work served during construction as a mold for casting the core. Poured material served for the inner, invisible parts of the wall (Ferretti and Bažant, 2006; Mistler et al., 2006). Masonry both served aesthetic demands and provided a hard, protective layer to counter weather effects and enemy attacks. This layer often served as framework during concrete pouring only, having no supporting function when concrete hardened (Fig. 4). Walls and vaults of variable thickness, from a few tenths of a meter up to 5 m thickness, were constructed this way.

This multilayer construction technology made walls of AlMarqab castle extremely durable, even without reinforcement. For assessment of earthquake damages, the external masonry is treated as consisting of discrete blocks, while the concrete wall behaves as a cohesive block. For modeling purposes, this kind of wall is treated as poor Portland cement concrete (Ferretti and Bažant, 2006). While this type of wall may deteriorate through the centuries due to creep and fatigue (Anzani et al., 1995, 2009), we can be sure that this was not the case just a few decades after construction.

Damage Mechanisms

Indirect earthquake damage to buildings is caused by ground shaking. If the frequency of earth vibrations is close to the frequency of resonance of the building, excitation will occur, damage will be pervasive, and the building will collapse. If frequencies widely differ, the building will survive, possibly intact (for the spectacular example of the Pont du Gard in France, see Volant et al., 2009). Likely, this is the primary cause why halls with lower proportions, e.g., the Main Hall, collapsed (Major et al., 2010), while tall, stout buildings like the donjon survived each earthquake for 800 yr.

The donjon of Al-Marqab, being of 20 m diameter, 24 m height, and having walls up to 5 m thick, is a robust structure. Height/thickness ratio is h/t = 5, indicating extremely strong and earthquake-resistant construction (Lourenço et al., 2007). We note that Eurocode 8 building codes allow a maximum of h/t = 9 for earthquake-resistant buildings (Anonymous, 2003). In-plan area ratio (Lourenço and Roque, 2006) is 57%, again an overly resistant structure against all kinds of earthquake resonance. Eurocode 8 recommends 5%–6% for regular structures. A minimum value of 10% is recommended for historical masonry buildings (Meli, 1998). For simplicity sake, high seismicity cases can be assumed to be those where design ground acceleration for rock-like soils exceeds 0.2g.

Area to weight ratio (Lourenço and Roque, 2006) is 10.4 m2/MN, i.e., more than 8× higher than recommended (Meli, 1998). Seismological modeling of a smaller tower in Roman Tolbiacum, Germany (8.3 m diameter, 8 m high, having an up to 3.1-m-thick wall), yielded 0.12 m horizontal and 0.06 m vertical displacement at the top of the tower in case of a M > 6.4 earthquake (EMS98 intensity IX) (Hinzen, 2005). Deformation of the Al-Marqab donjon (Fig. 5) was of similar dimensions. A 0.06 m vertical displacement is more than enough to reduce friction between ashlars of the Mamluk tower while extension of the box machicolation and adjoining walls is in progress during shaking.

Permissions

Photos reproduced with permission from Miklos Kazmer (email, 7/17/2022)

Wikipedia page for Margat