Version classiqueVersion mobile
OpenEdition Books

Les bâtisseurs sur les deux rives de la mer Rouge

Jean-François Breton

Chapter 5. Earth and wood architecture

Texte intégral

5.1. Earth and clay architecture

1Our knowledge of mud-brick architecture is still defective in South Arabia. The archaeological research from 1950 to 1990 focused on stone buildings, temples, hydraulic works, houses and palaces. In all these buildings, mud-brick walls were secondary, being limited to internal platforms faced with ashlar masonry or with rubble. The city-walls were known by their megalithic masonry (as at Tamna‘ South Gate) or by their heavy masonry (as at Barâqish or Ma‘în). Their internal structure remained unknown for a long period, since dunes had covered up long sections (as at Tamna‘ or at Hajar Yahirr), late construction had settled above them (as at Barâqish and Shabwa) or stone walls had been recently plundered (as at Mâ’rib). Moreover, as few city-walls have been excavated, there is relatively little information regarding the internal structure of brick walls. Mud-brick buildings were later discovered at Raybûn, at Shabwa, at Jujâ, and at Makaynûn (Hadhramawt). The available evidence now suggests that mud-brick architecture was widespread all over South Arabia, from the “Iron Age” to the coming of Islam.

2Generally defective, too, is the study of the bonding of the ancient mud-bricks walls in the region. If there were detailed descriptions from which to assemble a corpus of sizes, bonds etc., this would provide useful chronological indicators. Accordingly it is instructive to give a preliminary assessment of mud-brick works in the region, viz:

  1. Bricks can be made of virtually any earth conveniently available; burnt bricks are non-existent.
  2. Dimensions vary widely and the variation is not yet fully understood.
  3. Massive mud-brick platforms were used as foundations (fortifications, temples).
  4. The city-wall is typically built with mud-brick inside the wall; this is doubled on the exterior with square rubble, cyclopean or ashlar masonry,
  5. Mud-brick houses were frequent in the South Arabian cities during the “Iron Age” (very approximately from 12th cent. BC to 7th cent. BC), especially in Hadhramawt.

5.1.1. Mud-bricks

  • 1 Wright 1985: 354-355.

3As elsewhere in the Ancient Near East, mud-bricks occur both square and rectangular. In the rectangular form the proportion of length to breadth is very variable; no simple rational proportion like 2:1 occurs. It is tempting but probably too simple to say “that the dimensions of the moulded bricks should be standardized and that the variations should have some chronological or regional significance”1. Only large-scale production of bricks in city walls could be standardized. But little information is yet available.

  • 2 Breton 1994: 21.

4Brick walls of fairly massive dimensions are common in military architecture, but as very few of them have been excavated, brick dimensions are available in only a few cases. At Barâqish, bricks are 45 x 27 x 9 cm; at Hajar Yahirr: 45 x 27 x 9 cm; at Shabwa: 28 x 24 x 8 cm (al-Hajar) and 50 x 32 x 8-9 cm (south gate); and at Mâ’rib: 30 x 25 x 9-10 cm or 25 x 20 x 9 cm2.

  • 3 Breton-Darles & Roux 2010.

5In the centre of Shabwa, early levels (12th cent. BO to 6th cent. BO) display a completely mud-brick domestic architecture3. The bricks are made of a fine, homogeneous and compact silt, including nodules of scattered mortar (0.1 cm to 1 cm), occasionally charcoal, gravel (0.5 cm to 2 cm), or infrequently stones or sherds (5 cm). The colour of the silt is generally beige. During the early phases 9 to 11, the bricks are generally of a more light-brown colour. The moulded bricks are cut rectangularly, 7 cm to 11 cm thick, most commonly close to 10 cm thick. The disassembling of wall MR102 (MR = Mur = Wall) showed for each brick a smooth face and another one strongly marked by deep finger impressions. This latter face is always downwards oriented and is laid on the brick below it. Two categories of bricks exist, distributed chronologically. The larger type (45-53 cm by 36-43 cm) is frequent in the early levels (phases 10 and 11: approximate 1 1 th-8th cent. BC) while the smaller type (34-40 cm by 24-29 cm) belongs to later phases. But some elements of small size also exist in deep levels (MRI 13, phase 9).

  • 4 Roux 1992: 322-326.

6At Shabwa, mud-brick of identical dimensions are laid in the phase 2 walls of the extra-muros building n° 13. In this building, the most frequent size is 36-40 cm in length by 24-27 cm in width, probably of a late period4 (fig. 5.3).

5.1.2. Mud-brick foundations

7In fortifications, mud-brick is commonly used for the foundations. The stone city walls stood on heavy brick-work foundations, probably divided by cross-walls into internal compartments. It can be suggested that these compartments were adapted to the layout of the stone walls which they supported, so that the greatest weight, just under the heavy masonry wall, would come on the most solid masses of brick-work. In Ma‘în, only a meter-high brick foundation is visible. Such brickwork platforms are also commonly found under the heavy stone platforms of palaces and houses. Manner of construction

8At Shabwa, some walls have been carefully studied from phase 11 (10th-8th cent. BC). The raw brick walls from these archaic levels are generally built with only one row of bricks. Only two specimens are found with a double row of bricks (MR62 and MR25a), respectively 75-80 cm and 71 cm thick. These two walls show bricks both in stretchers and headers. By contrast, simple walls, between 35 cm and 45 cm thick with an average of 38-40 cm, consist of bricks laid only in headers. Wall MR61 (phase 11) has elements laid in stretchers. Wall MR62, preserved only in two courses of bricks and 70-80 cm thick, is composed of brick rows laid in stretchers. Wall MR67, related with wall MR62, some 40 cm thick, is preserved in three courses of mud-bricks also laid in stretchers with a total height of 32 cm.

9In a later period, approximately 6th-4th cent. BC, wall MR20 is preserved in four mud-brick courses 0.70 m-0.80 m thick. Wall MR79, 1.65 m long and 0.32 m thick, is preserved only in a single course of 6 to 8 cm in height, of mud-bricks laid in headers. The bricks are jointed by a fine and homogeneous muddy silt of a brown colour. The mortar of the vertical joints is very thin (0.1 cm to 0.3 cm), indeed sometimes imperceptible or even absent. In the late phases 6 and 5, the silt joints are of a grey colour and thicker, from 1 cm to 1.5 cm (MR25a, MR25b, MR7, and MR80), sometimes up to 3 cm thick as in wall MR77. The disassembling of wall MR 102 shows that the thickness of the joints in the outer face is 0.2 cm thick, while the thickness inside the wall varies on average between 0.5 cm and 1 cm. Modes of flooring

10During phase 5 (1st-2nd cent. AD) at Shabwa, some mud-brick pavings have been discovered. There are also two mud-brick floors inside one house from the same phase, and the whole construction rests on a massive earth foundation (French: chape de terre).

11Floor 82, 76 cm wide and 15 cm thick, is made of bricks of silt, homogeneous and beige in colour, with some small nodules of mortar and a little charcoal. The floor consists of two superimposed brick courses (lower course: 8.5 cm; worn upper course: 5-6 cm). Floor 84, 14 cm to 16 cm thick, is also composed of two mud-brick courses superimposed. The joints, made of a similar silt, are 1 to 4 cm wide. A massive clay floor (n° 85) exists in the two rooms of the house. This floor is built on a loose and thick level of 40-50 cm, made up of wind deposits mixed with powdery ash.

  • 5 Roux 1992: 327.

12In the extra-muros building n° 13 at Shabwa, identical massive clay floors and mud-brick paving exist in many rooms. As with the clay floors n° 82-84, these rooms are built on an unstable fine sand substrate5 (fig. 5.3). A mud-brick paving exists in rooms 5 and 6 during the second phase of the building. It is made of two layers of square mud-bricks, 10 cm thick, resting on a 1-2 cm thick clay layer.

5.1.3. Mud-brick massifs in city walls

13Most of the fortifications combine stone and mud-brick masonry. Most of these city-walls have an internal brick wall of massive dimensions faced with stone walls (fig. 5.1). This mud-brick wall could be very thick: 2.70 m at Hajar Yahirr, 3 m at Shabwa (general), 3.5 m (at Shabwa southern gate), 2-3 m in Barâqish, 3.70 m at Mâ’rib (southern sector), 10 m at Mâ’rib (western sector), ca. 4 m at Hajar Kuhlân/Tamna‘ (southern gate) and ca. 5 m at al-Binâ’ (NE sector).

14Some information on the technique used for building these fortifications is available. At Shabwa, the southern gate is built with two parallel brick walls of varying height according to the slope of the ground. An internal earth filling supports a horizontal terracing on which stands the central brick wall. A transversal wall indicates the possibility of compartments in the brickwork; they were most probably intended to be filled with earth or sand to form a level platform upon which to construct high walls.

15The original height of these brick walls is still an open question. At al-Baydâ’, the preserved upper limit of courses of stone headers gives the exact height of the mud-brick wall: 2.20 m. At Ma‘în, the brick wall is estimated at 4 m high (west curtain), at Shabwa 4 m (at al-Hajar), at Barâqish ca. 5-7 m (behind a 13 m high stone wall). According to this height, it would have been possible to create a wall of trapezoidal section rather than rectangular section without loss of stability, thereby saving considerable material. But this was not done. The Barâqish retaining walls, near Nakrah temple, are rectangular, probably because they were completely doubled by a vertical stone wall. Although mud-brick is a strong material and a durable one in an arid climate, it is susceptible to erosion by running water and storms. Therefore it may have been protected by a coursed rubble stone wall. No use of buttressing has been recorded. The upper part of these brick walls was used as a covered way. The walls may have been mortared and plastered; in the highest fortifications (such as Barâqish and Ma‘în), they were two storeys high.

5.1.4. Mud-brick platforms

16Mud-bricks are also used to build massive platforms. As most of these were later covered by a wall or by paving, they are often invisible from the surface. The Shabwa excavations have provided some examples of these mud-brick platforms. The largest one is situated in the Royal Palace of Shabwa under the central courtyard connecting buildings A and B. This courtyard, approximately 22.50 m by 28.50 m, rises some 5 m above the level of the “main street”. It was entirely built on a high mud-brick platform. Soundings in the NW corner of the courtyard show the bonding system: bricks were laid in stretchers along the northern side of the stone platform of building A. These soundings prove that the mud-hrick platform existed before building A. On the southern side of the Palace, another (or part of the same) massive mud-brick building was probably erected between the 11th and the 9th century BC.

17In the “Main temple” of Shabwa, massive mud-brick platforms were erected at different places. The (preserved) highest part of the sanctuary, located at the south of the walls (M = Mur = Wall) Μ1, M2 and M3, is composed of three juxtaposed massifs of mud-brick, used as foundation for the highest terrace (now vanished). The mud-bricks, large-sized (40 cm x 25 cm x 8 cm), arc laid out in alternating stretchers and headers. South of the four stone bases, there is also a late mud-brick platform, 14 m long by 3 m wide and 1.60 m high, which was paved with limestone slabs (structure 13). This mud-brick platform was used as the foundation of a late stone staircase (structure 14) built against the main wall (M1) (fig. 8.16).

  • 6 Sedov 2005: 95.
  • 7 Hansen et al. 2004: 62-63.

18In the Hadhramawt temples, mud-brick platforms were built over previous buildings. At Raybûn 1, an early mud-brick construction was later levelled to make a mud-brick platform to support the stone platform of the temples6. At Jûja, the ruins of an early temple were used for the foundation of the “Platform temple”; the old walls were cut down to a height of 0.5 m, and the interior was filled with clay and fragments of mud-bricks. Pure sand was laid on top and used as a foundation for the outer mud-brick walls. Hansen indicates the 4th-3rd cent. BC as a possible period for this “temple”7.

19There are other internal mud-brick podiums which should not be considered as “real platforms” but as a mud-brick core with stone slab facings, such as the Bâ-qutfa podium (2 m x 1.70 m and 0.70 m high preserved), attributed to the 6th cent. BC, and many other examples found inside cellae (Sûna, Mashgha, Raybûn, etc.).

5.1.5. Mud-brick buildings on stone platform

20Some mud-brick buildings on a stone platform have been recorded but never carefully studied. Most of the examples are either from Hadhramawt or from Najrân, where domestic architecture (“houses”) is attested from ca. the 7th cent. BC.

  • 8 Sedov 2005: 89.
  • 9 Sedov 2005: 136.

21Here is a short list of such buildings, published by Α. V. Sedov, without any information about the building techniques. At Raybûn I, the “early sanctuary” is “a structure, originally erected on the place of the temple... built from mud-bricks”8; it is dated not before the 6th-5th centuries BC. Approximately at the same period, at Raybûn V, a mud-brick building was erected on a stone socle of ca. 9 m x 8 m and 1.5 m high. The height of the mud-bricks projecting above the socle is ca. 1 m. An entrance to the house faced north, where a porch (?) adjoining the middle part of the socle is still preserved. At Raybûn XIV, a building with mud-brick walls erected on a stone socle, 10 m x 11 m, shows remains of walls ca. 2 m high. Near Shibâm, the al-‘Uqrân temple is described by Α. V. Sedov as “probably made of mud-bricks”, but “several fragments of limestone facing slabs...scattered in its ruins” demonstrate that the temple was not built only of mud-brick9.

22Similar buildings can be found in Najrân. The ancient city of al-Ukhdûd shows some 100 stone platforms, large in size (from 8 m by 9 m to 12 m by 14 m), some of them preserved to a height of ca. 3 to 4 m. These stone bases are built in brown or beige granite blocks, most of them coursed square, and some in uniform coursed rubble.

  • 10 Except some soundings by Murad-Zarins 1983.

23Most of these buildings remained unexcavated10 until the recent renewal of the work by the Saudi authorities. A long section of the east side has now been excavated, and a number of stone platforms, which are part of the defensive system, were cleared. According to the ceramics found in deep levels, 8th-7th cent. BC would be a possible date for the earliest stone platforms. Evidently, the floor(s) above the stone platform did not fully consist in framework, as none of their beams have been discovered. Thus it might be concluded that the floors were made essentially of mud-bricks, possibly with timber reinforcements. Would forests have been so rare (except palm-trees) that frameworks were impossible, or would there have been a specific building tradition?

5.1.6. Terre pisée11 buildings (adobe; French: structures en terre massive)

  • 11 “Terre pisée as used in English and French can be of ambiguous reference either connoting the mate (...)
  • 12 Wright 1985: 360.

24The existence of adobe construction is still problematic. It is not clear that this system of in situ earth construction was really developed as an alternative to mud-brick. Adobe is not always easy to identify. It may be distinguished from mud-bricks because it is usually built up in “registers” about one meter high and in runs of limited length12.

25At Shabwa, excavations in the centre of the city have yielded a certain number of such massive earth structures. They all belong to phase 10a dated approximately from the 10th to the 7th century BC. These are three rectangular structures which are not mud-brick buildings (Sb 63, Sb 64 and the “southern block”). The material used is rather different from the mud-brick: very compact and beige in colour. It is so homogeneous that there are no remaining traces of how this construction in adobe was carried out.

  • 13 Breton-Darles & Roux 2010: 47-50.

26In order to shape the clay, it was made malleable by adding water, which would dampen it and so avoid collapse. The structures were built rapidly with little preparation beforehand, apart from letting the mixture of clay, water and fibres rest for two or three days before being used. The major advantage of this technique is it that makes possible to build without earth foundations or stone foundations because the “terre pisée” is not vulnerable to erosion due to its “massive” character. Two micro-morphological samples were taken and subjected to analysis of materials (SB 65a and SB 107)13. Description (fig. 5.2)

  • 14 Breton-Darlcs & Roux 2010: 44-46.

27The massive building Sb 6314, the remains of which measure 0.84 m in height and 0.53 m in width, is a compacted terre pisée wall, modelled by hand, composed of fine and homogeneous silt, very compact. No trace of stratification was observed in it. This wall was mainly observed in the northern berm on the southern side of the sounding. It is associated with a mud-brick wall (MR65a) some 0.57 m high.

28The massive construction Sb 64 is a large rectangular structure similar to Sb 63. The walls are vertical. Built on top of structure Sb 63, it is set back by 3 cm at the western side. It is clear that the two structures Sb 63 and Sb 64 were built before wall MR65a and remained for a while in the open air exposed to the elements. In the same sounding, a quadrangular clay massif (Sb 107) is built on a previous wall which was levelled. Ranging between 12 cm and ca. 62 cm high, this massif has a horizontal base and vertical faces. It consists of a mixed earth of fine brown silt, homogeneous and slightly compact, including (occasionally) charcoal, gravel, bone, sherds and stones. This platform Sb 107 served to level and stabilise the ground, thus making it possible to build a new house whose brick wall MR94 delimits the eastern side.

5.1.7. Mud plaster

  • 15 Wright 1985: 420.

29A mud-brick wall without a facing of plaster is unsatisfactory not only from the point of view of durability but also rigidity. Adding “a stiff coat of plaster facings on both faces increased the rigidity of the mud-brick structure”15. The standard plaster for brick construction is mud plaster. Such plasters were discovered in al-Huraydha “farmsteads”. In al-’Oqm, near Shabwa, one of the excavated houses had walls that were plastered and then painted with a red colour. In Jûja, fragments of painted plaster are found in abundance in the “early temple” along the walls both outside and inside, decorated with lines, dots and other geometric figures, parts of zoomorphic and anthropomorphic figurines and letters.

30Besides the obvious presence of mud-brick walls in Shabwa, dated approximately 13th-8th cent. BC, this excavation revealed the existence of structures in massive terre pisée used at the same time as mud-brick constructions. These were either an independent terre pisée massif, or a foundation, or even a base supporting a mud-brick construction. Bricks in Tigray

31According to publications, the use of unbaked bricks is virtually unattested at Axum. It is highly probable that mud-brick was used in domestic architecture but this has never been documented by archaeologists.

32D. Krencker noted some brick walls in the foundations of Ta’akha Maryam, in the southern aisle, but they were not excavated. Except for some ovens at Dongur, walls at Menabiet, a threshold at Adults, and a possible heating system at Dongur, no other unbaked brickwork has come to light. Some fired bricks have been excavated in the Brick Vaulted Structure, in the Tomb of the Brick Arches, and an arch of brick at the west end of the corridor in the Mausoleum. Bricks are 27 x 28 x 6.7 cm (in the Tomb of the Brick Arches), 20 x 20 x 7 cm (in the Brick Vaulted Structure) but few are standardized. These bricks are mould-made, fired, red-brown in color and set in lime mortar. All these are tomb structures, and all the brickwork consists of arches or vaults incorporated in otherwise stone buildings. Probably all brickwork is roughly contemporary, in the fourth century AD.

5.2. Wooden architecture

33A number of inscriptions attest to the use of wood in South Arabian buildings. They bear witness to the fact that this now-woodless region once supplied a variety of timber for the “normal” demands of building construction. Therefore, as trees were probably not so abundant (Pliny strangely records that Roman soldiers were walking in the shade of trees), the use of wood in building may have represented to some degree a free choice between alternative construction materials. As its use became more frequent over time, mainly during the 2nd to 6th cent. AD in South Arabia and at Axum, it might also reflect the ethnic or cultural connections of the builders.

34How was the wood used, as most of its remains have vanished? Careful surveys in Shabwa show evidence of posts; buildings n° 109 and n° no display at their surface rows of regularly set posts; building n° 73 has well-preserved beams inside the construction.

  • 16 Seigne 1982.

35In the village of Mashgha, in Wâdî’Idim (eastern Hadhramawt), three ancient houses, numbered 1, J and K, have a stone platform topped with remains of beamed walls. All the walls (some 0.60 m thick) show regular deep imprints of wooden posts, long-beams and cross pieces (fig. 5.4). When these houses were destroyed by a violent fire, the mud-bricks kept the exact place of the original wood pieces16 (fig. 5.5). Later, in 1980-1983, careful excavations of the destruction layers of the Royal Palace of Shabwa unearthed some 200 beams, charcoalized during the two burnings of the palace and thus having kept their original dimensions, features and decoration.

5.2.1. Wooden frameworks

36The wealthy houses of ancient South Arabia, from Najrân to Zufâr, were all built the same way: a massive stone platform supporting walls with a wooden frame with mud-brick filling. What is important is not the quantity of wood inside the walls but the framework system. This is a repetitive mode of building, the same from the ground floor all the way to the top.

  • 17 Ground plate (French: sablière basse); long-beam (French: sablière); cross piece (French: traverse(...)

37On the top of the upper course of the platform, cross pieces17 are laid at regular intervals (ca. 0.30-0.40 m); as “headers”, their length is equal to the thickness of the wall (0.60-0.80 m). Two ground plates are fixed, on each face of the timber frame, and later a second set of cross pieces are laid upon them. Each upper cross pieces ends in four mortises, two at each end, to set vertical posts, generally 1.20 m-1.50 m high. These posts are then connected, with similar mortises, to cross pieces, supporting further wall plates (or head plates). An upper row of cross pieces supported upper long beams and posts. This framework, made with long beams/cross pieces/posts, was repeated (once as floor and once as ceiling), in order to create a ca. 3 m high ground floor, and then repeated again to build a second and a third floor. The interior of the framework was later filled with rubble and earth; the exterior was mud-plastered but the wooden posts were left visible.

  • 18 Seigne 1992: 136-143.

38The Shabwa royal palace provides many details, due to the huge number of preserved beams: complete framework in situ in building B (northern side)18 (fig. 5.6) and hundreds of pieces coming from building A. The framework consists of cross pieces, long-beams and posts. They are assembled in such a way as to make a regular timber-frame, all over the building, firmly interlocked by mortises and tenons. This heavy (vertical) stacking system of wood has no diagonal braces and no counterbraces. Horizontal beams such as long-beams are enormous, 0.25 x 0.25 m or 0.30 m x 0.30 m in section; posts are 0.35 m x 0.20 m in section. Cross pieces were laid every 0.30 m/0.40 m, each one supporting ground plates.

39The assemblage techniques are now very clear:

  1. Long-beams were assembled with lap-joints (in situ in building B);
  2. Horizontal pieces were assembled using rectangular mortises and tenons;
  3. Cross pieces had two trapezoidal mortises at each end; the posts had corresponding trapezoidal tenons. This careful assembly includes a large variety of building elements, but no angle posts and no diagonal braces. Although no angles have been discovered in situ, it may be assumed that they resulted from one timber-frame entering another one, without lap-joints or corner posts. The junction between two timber-frames was never carefully assembled; neither angle posts nor lap-joints were used, so probably the angles were a weak point in the building system. Another weakness is the lack of diagonal wooden pieces assuming the form of triangles.

40The Shabwa Royal palace displayed most of its framework on the outside: posts were meant to be seen, just like openings. As in the wealthy houses of all the South Arabian cities, stone inset slabs (only 6 cm thick) covered the mud-plastered intervals between the wood pieces. All these slabs were decorated with rectangular panels having chiseled margins and pointed dressed centres. This slab system was used at all periods, from archaic buildings (like the Bâ-qutfa temple, 7th cent. BC?) to late ones (Hajar ibn Humayd: 3rd cent. AD?). The Sanaa Museum also has a large number of such slabs coming from Himyarite sites on the Highlands (2nd-6th cent. AD.)

41Wooden ceilings have generally left no traces; but here again, the Royal Palace of Shahwa excavations have provided unexpected information. In the courtyard were discovered a set of beams and planks (both long and short), all assembled together with tenons and mortises (fig. 5.8). These planks were assembled vertically, transversally or lengthwise to build coffered ceilings, each coffer-panel supported by pillars of the upper-storey portico (fig. 8.23). The centre of each coffer was composed of thin translucent alabaster slabs to provide natural lighting for the portico. This unique type of coffered ceiling in South Arabia, dated approximately mid-3rd cent. AD, has many parallels in Graeco-Roman Egypt (Alexandria) and the Parthian Empire (Nysa).

5.2.2. Posts

42Wooden posts or pillars are used for vertical supports when a larger floor space is required than would be possible by spanning from wall to wall.

43Some buildings in Shabwa and in Tamna‘ had rooms with wooden pillars, thereby making it possible to increase the rooms’ dimensions. In the Royal palace (building A), in the eastern and western sides of the main gate (or porch), there are two large rooms, each with three wooden pillars in a row, resting on quadrangular limestone bases. In Tamna‘ TT1, the US archaeologists noted: “Two meters north from the southern limits of these walls in the four most easterly, are a series of stone ‘pillars bases’ averaging 1 meter square; they occur as separate elements in the walls. These, as are the walls, are of rough stones”. These pillars rested on internal stone walls parallel to the central corridor. In the nearby Temple of Athirat, the stone bases are atop walls of the platform, and in a few cases at the intersection of two walls. The walls of the first floor consist in a regular framework and in some places in pillars used as reinforcement of the mud-brick walls.

  • 19 Sedov 2005: 102 (figure 33: Temple of Dhât Himyam dhât-Kafas: building 2) and 103 (figure 43).
  • 20 Sedov 2005: 103-104 and fig. 34. Temple of Sayîn dhû-Mayfa’ân general plan in Sedov 2005: 95, 116- (...)

44The hypostyle hall in the Hadhramawt temples occupies the major part of the buildings. Generally two rows of wooden pillars divide the hall into three naves; in some temples four rows of pillars divide the hall into five naves19. Each row consists of five to six wooden pillars on prismatic-shaped stone bases (0.34 m-0.40 m x 0.34 m-0.37 m in size and 0.5 m high). Pillared galleries were also discovered in the Raybûn temples (Building 4 in Raybûn I, and the Temple of Sayîn dhû-Mayfa’ân in Raybûn XIV) with wooden pillars on a limestone base20. A pillared gallery probably existed on the terrace of the tower-houses just over the staircase.

5.2.3. Doors and porches

45The one essential structural fitting, the door, is always made in wood. The use of doors was kept to a minimum in the tower-houses for defensive reasons, and not for scarcity of wood.

  • 21 Darles 2005: 167-168, fig. 11-12.

46The Shabwa Palace offers a complex example of a monumental wooden porch, probably built in the middle of the third cent. AD. When the palace was destroyed by fire in the late 4th cent. AD, the framework disappeared but the mud-bricks were burnt all over, thus revealing the original position of the long-beams, cross pieces and posts21 (fig. 5.7). Above a massive stone foundation, four main beams formed the ground floor of the gate: the beams were 1.60 m long x 0.20 m high x 0.25 m wide. On each side of the door, ground plates, cross pieces, upper wall plates, and posts formed the framework of the walls. The door jambs consist in three associated posts, ca. 0.20 m apart, with regular setbacks (of ca. 0.20 m). This is in fact a door frame with three recessed “steps” on all four sides (horizontal as well as vertical) and with the actual door-opening in its depth. This is the exact plan of the doors in medieval Ethiopian churches. In the corners, projecting ends of crossbeams (colloquially known as “monkey heads” in Ethiopian architecture) laid at right angles to the line of the wall were set in order to strengthen the wall and also, in some cases, to provide anchoring for internal structures. Although the door itself has vanished, three bronze locks were found, and fragments of a stone lintel decorated with ibexes (Ataq Museum, Shabwa).

5.2.4. Windows

47Windows were inserted in the wooden frame of the façade. Small window openings can be suggested in the first floor, and wider windows in the upper floor(s) but not exceeding the distance between two posts (0.30-0.40 m in the royal palace of Shabwa). This is of course a reconstruction; nothing of the actual windows survives.

5.2.5. History of timber frameworks

48Although still uncertain, recent archaeological discoveries allow us to outline the history of wood frameworks in buildings. An archaic inscription RES 3946, mentioning the repair of the royal palace Salhîn in Mâ’rib, distinguishes the platform (or the rock?) and its upper part (whether in wood or in mud-brick).

  • 22 Sedov 2005: 98.
  • 23 Sedov 2005: 188.
  • 24 Sedov 2005: 113.

49Temples in Hadhramawt, excavated by Α. V. Sedov’s mission, show a continuous use of wooden frames over many centuries. In Raybûn I city, the early “Rahban temple, building 2” (“House of the priest”) consists of a stone platform supporting a timber framework. “Vertical wooden pillars and cross beams formed the framework, and the space between bars was filled with mud-bricks and mud. The exterior and the interior of such walls were usually faced with polished limestone slabs, plain or decorated, or they were covered with clay and cement plaster”22. Although no detail of the framework is available, it is worth noting that charcoal from the burned wooden framework of its internal walls yielded a C14 calibrated age of 804-775 BC. This is the earliest testimony of wooden frames in Hadhramawt. Later, in Raybûn V, the Kafas/Na’man temple, building 1 (“refectory”), consisting of a hypostyle hall with wooden frameworks and 25 wooden pillars, is dated 8th-6th cent. BC (762-628 BC; 596-574 BC; 558-194 BC)23. During the same period, Raybûn I, Rahban temple, building 4 (“main sanctuary”) provided charcoal from burnt wooden frameworks of external walls, similarly dated 760-620 BC or 600-90 BC. Later, in Raybûn V, Kafas/Na’man temple, room 3 of building 1 (“refectory”) provided charcoal from burnt wooden beams of the roof (stratum of destruction), dated 400-338 BC. Many charcoal samples from its wooden framework provided a ca. 4th-3rd cent. BC dating. In the Mayfa’an temple in Raybûn XIV, building 1 (“the walls of which were most probably constructed in mud-bricks supported by a wooden framework”24) included a niche with a wooden framework in the monumental podium whose charcoal samples are dated 174 BC-82 AD. These constitute clear evidence of timber frameworks in Raybûn, from the 9th cent. BC to the 1st cent. AD.

50Most of the temples in Hadhramawt were built the same, with a stone platform topped with a framework. The Bâ-qutfa temple had a small cella built with such a framework, with cross pieces, long-beams and posts. Limestone slabs set in the external and internal faces of the walls bore inscriptions, most of them from the 7th-6th century BC. At Shabwa, one of the earliest pieces of evidence for frameworks is building n° 44, C14 dated to the 5th-4th century BC (520 to 350 BC). It is probably a temple with a central courtyard surrounded by rooms built with wood. Building n° 74 with its framework is dated 350 to 30 BC; building n° 77 is 60 BC to 220 AD.

51In the Shabwa 2002 excavation, the building US 29 with remains of cross pieces and long beams is dated 380-45 BC. The early huge timber framework of the Royal Palace was destroyed for the first time around 227 AD, then rebuilt between 230 and 250, and definitively destroyed by the mid-5th cent. AD. There is thus a continuous centuries-long tradition of wooden frameworks at Shabwa. Similar conclusions should probably be drawn from the current excavations of Tamna‘ houses. At Mâ’rib, Awâm Temple has a wooden framed portico (“Annex”, north side) but Bar’ân Temple seems to have no frameworks.

  • 25 Faris 1938: 42.

52On the Yemeni Highlands, the Himyarites seem largely to have used wooden frameworks. As only a few excavations have been carried out or published, one must rely on texts and inscriptions. The Yemeni writer al-Hasân Ahmad al-Hamdâni, in the Eighth Book of al-lklîl, speaks about Bayt Hanbas castle and its woodwork and doors; he further observes “The castles (Damigh) were destroyed by the Abyssinians... They burnt their woodwork because of their great size”25.

53On the west coast of the Red Sea, in Tigray, similar techniques of wood show some close similarities with the South Arabian wooden frames. Archaeologists have discovered similar systems in the Sabaean, pre-Axumite and Axumite excavations. The contacts between Ethiopia and the Sabaean kingdom go back to the 9th-8th century BC.

5.2.6. Tigray

54D. Krencker carefully studied the wooden assemblages in the Axumite style, mainly in the medieval churches of Tigray. He saw wooden assemblages built with long beams, cross pieces and posts creating frames which were then filled with mud or mud-bricks. But he could not demonstrate the use of these frames in actual Axumite buildings, because of lack of excavations. He could only study the medieval doors and windows in churches with the projecting heads of cross pieces (called monkey heads). Much later, in the sixties, F. Anfray emphasized in Grat Real Guebri/Yeha the use of beams as reinforcement for the walls. Wood in early periods (8th-6th cent. BC)

  • 26 Anfray 1997: 20.

55In Grat Real Guebri/Yeha, excavations highlighted the use of wood as early as the 7th-8th century RC: beams calcined in the walls, charcoal and burned earth. Wooden posts appear in the north of the portico Τ4 and in the western wall, inserted in the wall to a depth of 20 cm. In the walls of Τ4, timber exceptionally was used as a reinforcement or a framing. During the pre-Axumite period wood was certainly used in the framing of the openings and for the posts supporting the ceiling (beams and boards) but, as we see at Grat Real Guebri26, it did not constitute the framework of the walls. There, beams appear in large numbers: posts, long-beams, and cross pieces; their heads appear under the wooden joists. This assembly gives cohesion to the masonry but does not constitute a real framework. Grat Real Guebri thus attests to the use of wood in pre-Axumite architecture. The comparison with the use of wood for construction South Arabia leaves no room for doubt about the technical interactions. Wood in the Axumite period (2nd-7th cent. AD)

56The excavations of the ‘Enda Mikâ’el and Ta’akhâ Mâryam palaces clearly demonstrated that large rooms or halls were supported by wooden pillars with wooden capitals at regular intervals.

  • 27 Munro-Hay 1989: 138.

57At Axum, S. C. Munro-Hay excavated a building near Enda Sem’on in the IW II trenches27. The walls preserved beam imprints distinctly, showing long-beams square in section, 24 x 16 cm, and spaced at 48 cm intervals. Transverse beams or cross pieces probably preserved the round shape of the trees, ca. 12 cm in diameter, spaced about 50 cm apart, directly above the long beams.

58It is to be noted that in Axumite buildings, long beams are not joined with the cross pieces: they are laid simply laid directly on top of them. There is no assemblage between them. Wooden posts are very rare (probably only at Yeha) but again there is no assemblage between them and the long-beams. This is a major difference from South Arabian techniques of wood construction.


1 Wright 1985: 354-355.

2 Breton 1994: 21.

3 Breton-Darles & Roux 2010.

4 Roux 1992: 322-326.

5 Roux 1992: 327.

6 Sedov 2005: 95.

7 Hansen et al. 2004: 62-63.

8 Sedov 2005: 89.

9 Sedov 2005: 136.

10 Except some soundings by Murad-Zarins 1983.

11 “Terre pisée as used in English and French can be of ambiguous reference either connoting the material (compressed plastic earth) or the mode of its application-rammed between forms (piser = tasser)” (Wright 1985: 416). Terre pisée refers to handfuls and lumps of clay piled up together, not formed into bricks.

12 Wright 1985: 360.

13 Breton-Darles & Roux 2010: 47-50.

14 Breton-Darlcs & Roux 2010: 44-46.

15 Wright 1985: 420.

16 Seigne 1982.

17 Ground plate (French: sablière basse); long-beam (French: sablière); cross piece (French: traverse); post (French: poteau); wooden framework (French: pan de bois, armature).

18 Seigne 1992: 136-143.

19 Sedov 2005: 102 (figure 33: Temple of Dhât Himyam dhât-Kafas: building 2) and 103 (figure 43).

20 Sedov 2005: 103-104 and fig. 34. Temple of Sayîn dhû-Mayfa’ân general plan in Sedov 2005: 95, 116-117.

21 Darles 2005: 167-168, fig. 11-12.

22 Sedov 2005: 98.

23 Sedov 2005: 188.

24 Sedov 2005: 113.

25 Faris 1938: 42.

26 Anfray 1997: 20.

27 Munro-Hay 1989: 138.

© Centre français des études éthiopiennes, 2015

Conditions d’utilisation :