Version classiqueVersion mobile
OpenEdition Books

Les bâtisseurs sur les deux rives de la mer Rouge

Jean-François Breton

Chapter 4. Masonry and architectural elements (9th century BC to 5th century AD)

Texte intégral

4.1. Stone masonry

1Basically the stone masonry of South Arabia can be classified as megalithic, cyclopean masonry, rubble masonry and ashlar masonry. South Arabian masonry used blocks of very large dimensions which could not be carried by one worker; these come under the rubric of megalithic and cyclopean masonry.

4.1.1. Stone work Monoliths or megalithic masonry

  • 1 Wright 1985: 396.

2“Megalithic masonry consists in the use of very large units of stone often with a face area of several square meters. The units are thus so massive that they need a company of men and auxiliary earthworks to get them in place. The essence of the style is that the units comprise in themselves complete or substantial parts of a building element”1.

3The long monoliths appear mainly in the sanctuaries: Mâ’rib (temples of Bar’ân and of Awâm), Barâqish (temple of Nakhrah), Sirwâh (temple of Awâm), Ma‘în, as-Sawdâ’, etc. In these buildings, construction in stone has replaced the earlier construction in wood, but preserving the identical forms. This is especially striking as regards the long thick pieces of wood, i.e. the pillars and especially the beams of the wooden frame; these are now made in stone. The majority of the stones can be considered as true megaliths, as at the sanctuary of ‘Athtar dhû-Risaf at as-Sawdâ’, the weight of whose blocks varies between 2 and 4 tons (fig. 4.1). More exceptional specimens, like the lintel in lumachellic limestone preserved at the extra-muros temple of Ma‘în or like some pillars in the intra-muros temple of as-Sawdâ’, probably exceed even this weight. Techniques of megalithic asssembly

  • 2 Bessac 1998b: 197-198.

4This section concerns chiefly the temples, the main buildings to use large monoliths for their construction. On these large blocks, several vertical assemblages can be identified2.

  1. Cylindrical mortises and metal dowels are used to fix the long horizontal architraves onto the pillars. This mode of assemblage is represented in the sanctuary of Khirbat Hamdân and also in the temple of ‘Athtar dhû-Risaf at as-Sawdâ’ (top surfaces of pillars A to F) (fig. 4.2).
  2. Square mortises and tenons, both cut in stone. The tenon is prepared in the bedding (top) surface of the pillars and the mortise is carved out on the resting (lower) surface of the architrave which sits on top of them. Such an assemblage can be observed in the al-Baydâ’ sanctuary, in the Nakrah temple at Barâqish (stone tenon is 7 cm x 7 cm x 5 cm in height). Many pillars in the Mâ’rib sanctuaries (Awâm and Bar’ân temples), the Mâ’rib “mausoleum” and the Sirwâh temples show similar stone mortise-and-tenon assemblages.
  3. Assemblage by lateral or longitudinal grooves. On the upper edges of horizontal slabs disposed at regular intervals, large grooves are cut which provide the bedding surfaces for other monoliths which lie on top of these and form the upper parts of the building. There are spaces remaining between the slabs which are completely covered over by the next higher course of stones (fig. 4.3). The upper slab is fixed laterally by these grooves, and so it is not necessary to cut the upper slab at all. This process has been particularly well studied on the entrance portico in the temple of’Athtar dhû-Risaf at as-Sawdâ’. Long slabs (n° 64, 68 and 70) with longitudinal grooves exist in the as-Sawdâ’ temple and in the extra-muros temple at Ma‘în.
  4. Assemblage by carving a deep notch at the top of a vertical pillar. This system is analogous to the rabbet joint of construction in wood. The notch is equivalent to more than half the width of the section of the architrave which is laid upon it. Such an assemblage is visible on the porch of the extra-muros sanctuary at Ma‘în.
  5. Assemblage by superposition only (without mortise and tenon joints or cramped joints). This is the simplest and most widespread system. Many architraves and flagstones are laid in this way without any specific joints; their weight alone is enough to maintain them in place. The upper architraves in the intra-muros building (n° 1) at Ma‘în correspond to this definition: no special joints at all are required.

5Horizontal assemblages of two adjacent blocks were made of metal cramped joints. Joints are liable to be pulled apart and thus need assistance in counteracting the stress. This is obtained by the metal cramps or ties. These are fixed across the joint of the stones, so as to draw the stones together. The cramp should be fitted tightly into a mortise cut into the top surface of the two stones forming the joint.

6Bronze cramped joints were in use as early as the 8th cent. BC at as-Sawdâ’ (fig. 4.4). Swallowtail cramps were largely used in the ‘Athtar temple of as-Sawdâ’ between architraves in the porticoes: they are 12 cm long, 6 cm wide and 3 cm thick. Swallowtails cramps are also widely used in the Nakrah temple at Barâqish for assembling architraves: each half-mortise is 13 cm at its base, 8.5 cm at its opening and 4 cm deep (dated ca. 6th cent. BC).

7Except for assemblage by superposition, which is the most commonly used, the other systems are hardly known in ancient architecture outside of South Arabia. Thus the Egyptian buildings, which use similarly large monoliths, display assemblage by superposition. Even wooden swallowtail cramps intended to temporarily maintain the blocks during their installation do not seem to have been used for the monoliths in Egypt. In Greece, metal dowels are fixed in cylindrical holes, primarily to hold together elements of columns, drums, bases and capitals, in particular in the archaic phases of architecture. Such metal fixings in long monoliths, such as lintels and architraves for example, are unknown in Greece. It is the same in Roman buildings. Cyclopean masonry3

  • 3 The term “Cyclopean” masonry refers to huge natural blocks used as such with no dressing.
  • 4 Van Beck 1952.

8The Tamna‘ building that was numbered TT1 (Harîb Palace or Hatîb Temple) by the archaeologists of the American expedition in 1950-1951 consists of two parts: the eastern one (A) and the western one (B), with a central courtyard. Building A is a massive stone platform, trapezoid shaped (32.50 m max. x 25.60 m), some 10.30 m high (on its eastern side), with two symmetrical recesses on each side. It is entirely built in cyclopean granite blocks, laid in stretchers, in headers or (in its recesses) in orthostats. The southwest angle, visible to a height of 4 m, shows blocks some 2 m high, 3 m exceptionally, laid in irregular courses. In the SW recess (recess S3), the blocks are 0.90 m x 1.2 m x 1.25 m long; in the south wall, some blocks are 1.20 m high and 2.50 m long. The bottom of the recesses is built with huge orthostats. For instance, the recess E1 consists of a single 4.30 m long x 2.26 m high x 0.22 m thick monolith. According to G. van Beek, this “massive masonry” can be dated to the 6th cent. BC and the foundation of the platform to the 7th cent. BC4.

9Cyclopean masonry is also attested in Tamna‘ South Gate, consisting of two granite bastions. The western bastion shows one megalith: 1.70 m x 2.80 m x 0.50 m high, and the eastern one a megalith 2 m (now broken) x 0.87 m x 0.30 m; this last block bears the dedication text RES 3881 mentioning two bastions (nmr) Hamarr and Shah built in massive blocks (‘bn) and in balaq (blq) smaller stones. This text probably goes back to the 4th cent. BC. Another text from the northern gate (MAFYS-Tamna‘ 1) mentions “... Saqaraw, its two bastions and its gateway (?), in massive blocks (‘bn), in wood (‘ḍ), in alabaster (?) (mwgl), in balaq stones (blq)”. The Dar al-Kafîr tower at Shabwa shows, in its lower courses, cyclopean limestone megaliths, some 2.60 m long x 1.15 m x 0.40 m.

10The above two examples of massive masonry belong to archaic stone architecture, from the 8th to the 7th/6th cent BC. Later, as at Tamna‘ and Shabwa, smaller blocks were generally used, although sometimes megaliths were still quarried for specific uses (for instance, statue foundations in the Shabwa “main temple”). Part-worked rubble masonry or coursed squared masonry

  • 5 Wright 1985: 397.
  • 6 Wright 1985: 397.

11An important part of South Arabian masonry was part-worked rubble masonry. “Rubble stone may be used for walling in all forms ranging from completely irregular shapes to reasonably squared up blocks. These forms may occur more or less naturally or be effected by a measure of dressing which can range from occasional knocking away of useless excrescences to considerable hammer facing. The size of the units may be anything from what can be picked up easily in one hand to as massive a stone as a man can handle”5. “Whatever the general nature of the stones used the strength of rubble masonry is customarily enhanced by setting larger and regular blocks at angles and jambs and by providing through bonding stones at regular intervals”6.

  • 7 Bessac 1998a: 191.

12In temples, this part-worked or coursed square masonry (French: “petit appareil”)7 appears only in the enclosing walls of the sanctuary of ‘Athar dhû-Risaf in as-Sawdâ’. It is entirely made up of hard oolithic limestone. The dimensions of the stones vary roughly from 5 to 14 cm in height, and 20 cm to 60 cm in length. This is a “petit appareil allongé” consisting of small modules of roughly one cubic dm. If one transforms volume into weight, considering the average density of this type of stone as approaching 2.5 kg/m3, one obtains less than 50 kilograms per block. The dimensions of this rubble masonry are essentially defined by the geological characteristics of the outcrop from which the stones come.

  • 8 De Maigret & Robin 1993: 446, ff.

13This coursed squared masonry is very commonly used in walling the external faces of the mud-brick massif of the ramparts. Excavated parts of such masonry appear in the Barâqish city wall, near the Nakrah Temple; after excavation, it is now visible to a height of ca 10 m8. Square rubble masonry

  • 9 Noel 1968: 26; Bessac 1998a: 192.

14According to building professionals, “square rubble masonry” (French: “moyen appareil”) has courses from 15 to 40 cm high and from 30 to 70 cm long at the maximum9. Most of the constructions of the Jawf exceed this last dimension and thus are to be classified in the massive heavy masonry category (see next section). Nevertheless, one single monument in the Jawf area is included in this category; it is the building with setbacks at al-Baydâ’ (fig. 4.6). Its courses have a regular height of 26 cm, a length from 51.5 to 52 cm for the stones in relief, and 49 to 52 cm for the other stones, including the hidden elements at the bottom of the setback steps; the visible part of the courses measures from 45 cm to 45.5 cm. The relief of the setbacks varies from 10 cm to 10.4 cm. Ashlar masonry or massive heavy masonry

  • 10 Wright 1985: 401.

15“Ashlar masonry is built with finely dressed blocks, cut so true on the bed and rising joints that they can be set either with or without mortar (dry stone) leaving between the joints only a space of a few millimetres (modern specification: 3 mm) or no appreciable thickness (hair line jointing). The dressing of the face is not basic to the definition of ashlar—this can be fair faced, roughly dressed or quarry faced as desired or convenient. In modern practice ‘ashlar’ connotes that the block is cut truly rectangular—i.e. the rising joints are both perpendicular to the beds joints and to the face”10 (fig. 4.7).

  • 11 Breton & Bessac 2002b.

16All the buildings in the Jawf valley are “ashlar masonry” or “massive heavy masonry”, except for the sanctuaries whose pillars and lintels form a special group. In this specific ashlar masonry the courses range from 26 cm to 53 cm in height. However, at each site, small specificities appear. For example, the height of the courses in the ramparts built with firm limestone at as-Sawdâ’ ranges between 40 cm and 53 cm whereas at Ma‘în, where hard limestone prevails, it ranges from 22 cm to only 38 cm high. It thus seems that the firm oolithic quality of limestone allows higher courses than the hard quality limestone. The cold stone, used especially at Kamna11, gives a height varying between 26 cm and 41 cm. The varieties of limestone, changing from one quarry to another, each provide varying dimensions in each city. In addition, the assumption of specific quarries belonging to each city seems thereby to be confirmed.

4.1.2. Internal assemblies and structures in ashlar masonry (Jean-Claude Bessac) Walls made horizontally of a single row of stones

  • 12 Bessac 1998a: 193.

17Walls made horizontally of a single row of stones are by far the most frequent kind of assemblage in the Jawf area, at al-Baydâ’, as-Sawdâ’, Khirbat Hamdân and partially in the Ma‘în ramparts12. Mainly composed of stretchers, these walls also include however, at regular intervals, headers whose crude inner tail is deeply embedded in the internal rubble.

18This inner tail, which is quite long, typically serves to bind together the outer stone wall and the inner mud-brick massif. At Ma‘in, particularly in the southern and western gates, vertical lines of such headers appear at regular intervals of approximately three meters, but only in every other course of stretchers. This system of fixing the external stone wall to the internal mud-brick massif is also to be seen in the lower courses of towers n° 38 and n° 39 in the al-Bayda’ city wall, up to approximately one meter below the level of the holes. These holes were most probably intended for inserting wooden beams for a floor; the line of the holes corresponds to the level of a walkway on the rampart above. With such a reconstruction, the presence of headers at this height would be pointless: not only would a doubling of the wall at this level be structurally unnecessary, but it would narrow the walkway and impede easy passage.

19A crucial factor inheres in the difference in strength of stone walls is brick walls of fairly massive dimensions: relative subsidence may affect the mud-brick walls but not the stone masonry, thus potentially breaking the stone headers deeply embedded in the interior mud-brick massif. In order to avoid such cracking of the headers at the internal face of the outer wall, additional smaller stone blocks were laid (within the mud-brick interior) between two courses of headers to strengthen and support the headers.

20Another system of connexion between the stone wall and the internal rubble masonry is attested in some sluice gates at Mâ’rib. It is composed of pairs of vertical series of headers, laid at three-meter intervals on every other course of stretchers but offset horizontally from one course to the next, so as to form two staggered vertical lines. These alternating headers protrude from the outer surface of the wall and are very carefully imbricated so as to form one very strong stone structure.

  • 13 Wright 1985: 93, fig. 22.

21There are thus significant differences between the Mâ’rib and Jawf buildings. The Mârib sluice walls are so wide that there are in fact two stone faces enclosing a filling of mortar and rubble. The headers are transverse and constitute true internal crossing walls which stiffen the outer walls. G. R. H. Wright, who studied this type of structure, describes it as a Vitruvius “Graecorum structurae13.

22Neither double face walls nor mortar are attested in the recorded buildings in the Jawf.

23A second mode of stone assemblage can be seen in the Jawf: the imbrication of two series of alternating headers at corners of the walls for reinforcing the angles of the towers. Such bonded quoinings, existing in the Ma‘în towers up to their summit, are adapted from an early system used in the city-walls in al-Baydâ’: there, vertical chains of headers are used in the front walls of the towers but not to reinforce their angles. The builders of Ma‘în, on the other hand, made extensive use of these chains in the angles of the city wall. A chronological evolution may be assumed from the archaic al-Baydâ’ fortification (8th-7th cent. BC.) to the Ma‘în city-wall (approximately 6th cent. BC). At Barâqish, such reinforcements of angles also exist at least in some towers.

24The principle of vertical chains of headers deeply embedded into an inner brick-massif is a major characteristic of early South Arabian ashlar masonry. After this “archaic” phase, this principle evolved to a system of reinforcement in the angles, though without giving up its archaic role. Lastly, with the introduction of mortar, the technique was further improved, as can he observed in one of the Mâ’rib sluices.

25Do similar techniques exist elsewhere in the ancient world? To the knowledge of J.-C. Bessac, Egyptian construction does not provide similar examples of angle bonded quoinings. In late Roman buildings, we find vertical chains of stones to reinforce the angles. Consisting in one stretcher and a superimposed header in alternation, they constitute essentially autonomous stiffeners. It is a significant characteristic of South Arabian construction that the vertical chains of headers, isolated in front walls or located as reinforcements in the angles, always introduce a discontinuity into the appearance of the ashlar masonry, which mainly consists of rather long stretchers. Casemate (or emplecton) walls

  • 14 Schnelle 2007.
  • 15 Albright 1958: 218.

26In the Sirwâh fortification, some walls are constructed on the “casemate” principle. Two parallel walls of ashlar masonry are built 4 m to 5 m apart. Every 4 or 5 m, orthogonal cross-walls are built at the same time. Then the casemate is filled, the chips of the upper course of each layer being immediately pushed inside. This is not a late filling of the compartments but an integral part of the construction process14. In the Awâm oval temple at Mâ’rib, the inner and the outer wall of ashlar masonry are tied together by a series of pairs of cross-walls built in rectangular headers, measuring 1 m to nearly 2 m in length, laid in regular courses (fig. 4.5). There are some 60 pairs of these cross-walls, with 0.50-1.00 m between the two walls of a pair and 3.00-3.50 m between pairs15. Walls made horizontally of double rows of stones

  • 16 Breton & Bessac 1998a: 194.

27Walls like this are to be observed in the Kamna rampart, in the western curtain and in the intra-muros temple of Ma‘în16. Although there are some differences in their internal construction, these buildings all show two rows of stone in each course, one internal and one external.

28The most striking example of this technique appears in the eastern curtain wall at Ma‘în, approximately 8 m in height preserved all the way to the top, and made up of two rows of stone (fig. 6.11). The wall is 3.80 m long and 1.08 m thick, and at each of its ends are vertical series of headers.

  • 17 Breton & Bessac 2002b: 150-151.

29These headers are finely dressed on their two (visible) faces. Besides these two vertical series of headers connecting the two faces of the wall, there is one single isolated header at the middle height of the curtain. It shows the upper level of the former brick wall, today washed away. There are two horizontal series of holes used for the beams of two floors17.

30In Ma‘în the intra-muros temple is made of two stone rows. The front wall and the angles of the building are apparently reinforced by series of vertical intersecting headers. The internal structure of these walls shows the absence of any junction between their external and interior faces, and none of these headers really goes through both of the faces. It is thus a false reinforcement made up of pseudo-headers. Although the construction is unusual, one understands that it is easier to build such walls with two independent stone rows: there is no need to take care of a perfect concordance of the two faces neither in the courses, nor in their dressing.

  • 18 Bessac 1998a: 194 and 220, 229; Breton-Bessac 2002a: 152-153.

31This may have been done only to save money, unless it was simply a fraud on the part of the builders. These false structures seem to show that in Ma‘în, it was difficult to design a building without incorporating such reinforcements made up of headers. The tombs of Mâ’rib display interior doubling of the walls. The doubling is made of a very poor tuff, summarily dressed, of the same size as the stones of the exterior face. These exterior faces, made of a high quality limestone, are very carefully dressed with drafted margins all around. This building technique also appears in some places of the city-wall and in the oval wall of Awâm Temple. In the Jawf area, the principle of doubling the walls, although less obvious and rarer, appears in the southern sector of the Kamna city-wall, there with no of mortar and thus probably older18.

  • 19 Bessac 1998a: 194 and 220, 229; Breton-Bessac 2002a: 152-153.

32The Kamna fortification consists of one interior is made up of rough blocks of cold limestone, i.e. very hard and almost impossible to dress correctly, and of an exterior face made of oolithic limestone which is rather easy to dress finely19.

33The similarity between these two “doubling” (French: doublage) techniques is mainly aesthetic and is based on dressing techniques of limestone. Tuff was selected for the internal faces is selected only because it is easy to work and can be quarried quickly. It is important to stress that the speed of extraction is relatively independent of the hardness of the rock. Sometimes these stones, easy to extract and dress, behave very badly under the impacts of tools, such as the cold limestone in Kamna. Walls with setbacks (fig. 4.6)

  • 20 Bessac 1998a: 195-196.

34In all of South Arabia and Tigray, the only example of such a technique appears in one building at al-Baydâ’20. In this perfectly isodomic wall, the system of internal and external assemblage is made of deeply embedded vertical series of header stones 52 cm long, forming regular setbacks in the front wall. Such “pilasters” consist of vertical units 52 cm long, in alternating rows; one row has a stretcher, the row above/below it has two headers whose combined width equals the length of the stretcher. In the angles of the building, the blocks forming the projection widen up to 86 cm. The back of the wall remains roughly dressed, with chiselling only on the lower edge. It is not certain that the rear face of the wall was visible, at least up to its present level. The protruding trapezoidal blocks widen in front to a T-shape. Some chronological conclusions

  1. Ashlar masonry or heavy masonry appears in South Arabia during the archaic Period after around 800 BC. Indeed, examples of walls solidly constructed throughout in dressed blocks occur and can be considered as true “ashlar masonry construction”, as at Kamna.
  2. “Finely dressed masonry” can be considered as posterior to the “later Iron Age” in Palestine during the Israelite rule (post 1000 BC). It appears probably as early as the 6th century BC. The blocks are middling massive and are so finely dressed that they can be set with joints of only a few millimetres as in the Ma‘în or the Barâqish city-walls. The best examples are jointed hair-line drystone masonry (Ma‘în). This South Arabian ashlar masonry seems a local technique.
  3. Various epigraphic references to this masonry, from the 7th century BC to the 4th century BC, can be understood as designating different parts of the walls or different kinds of stone.
  4. Buildings are mainly built entirely in ashlar masonry. Most of the city walls are constructed in this way, from the Jawf to Hadhramawt.
  5. There are no examples of ashlar masonry where the walls are constructed as a series of ashlar-built piers (built on alternate courses of headers and stretchers) connected by panels of rubble fill. Strangely, this type of masonry was developed during the last centuries of the second millennium in Palestine but not in South Arabia.

4.1.3. Dressing from the 8th to the 4th cent. BC (Jean-Claude Bessac)

  • 21 Bessac 1986: 95.
  • 22 Noel 1965: 299-300.

35In general it is not easy to distinguish between dressing for building only (French: taille fonctionnelle) and dressing for decoration (French: taille ornementale)21. This, however, is not the case for the South Arabian dressings in the Jawf valley. In this area, when dressings are carried out merely for decoration, they are realized not on the loose separate blocks but in situ on the already-built monument22. In addition, in most cases, decorative dressing shows drafted margins made only for decoration. Recent observations indicate that most of the in situ dressing was carried out only on ashlar masonry. Below, dressing with metal tools will be studied first, then lithic dressing (dressing with stone tools). Dressing for building

  • 23 Bessac 1998a: 183-184.

36In large part the blocks were dressed into ashlars, often of standardized dimensions so that they could be set in regular courses. In the process of squaring up the block, marginal drafts were first cut around three or four sides of the face; the remaining central part was removed later. This cutting of marginal drafts is done to create a surface free of “twist”, before knocking away the boss. The boss can be cut away either before or after setting the block in the wall. We consider as “functional dressings” all dressings (preliminary or final) aimed at obtaining a smooth surface. Five degrees of functional dressing can be identified in the Jawf buildings23. Rough dressing at the time of extraction with minor improvements (fig. 4.7)

37Because of their originally quadrangular form, sometimes rather regular, some stone blocks can be used directly as they come from the quarry beds with their natural longitudinal splittings (French: joints de stratification) and perpendicular joints. In this case, the stones are set in the condition in which they arrive from the quarry, or roughly squared with a hammer (see the coursed squared masonry of the temple of ‘Athtar dhû-Risaf at as-Sawdâ’ and the internal faces of the eastern walls of Kamna). Some internal faces, like those in the eastern tower of Ma‘în, are left in their natural condition, except for some limited corrections on their upper edges. These corrections consist in removing small projections, using either a kind of hammer with a cutting edge, or a chisel with a narrow round end. Hammer work (French: taille éclatée)

  • 24 Scranton 1941: 24. (French: parement dressé au têtu; Ginouves & Martin 1985: 130.

38The term “hammer work” refers to a crude surface finishing achieved with an edged hammer24 (French: marteau têtu), but still showing some hollows and bumps. Hammer work is done only on stones which are already laid in the wall. Similar to the rough dressing, the hammer work is realized only on vertical faces, once the stone laid in the wall, notably interior faces or hidden faces deep in the foundations. Such hammer work is very common in the Graeco-Roman World where it is often used as decoration on a bossed face (French: bossage), in contrast to South Arabia where it is used only as a preliminary step or on hidden blocks. Rough pointed dressing (French: taille pointée de dégrossissage)

  • 25 Breton & Bessac 2002: 152-153.

39This dressing often is an intermediate stage in the final refining of the dressings. Done with a chisel with a narrow round end, it is carried out after the hammer work but before the execution of a decorative dressing. Sometimes, it can also be carried out as the final dressing of the outer face of a rampart built in hard or cold limestone. The eastern city-wall at Kamna, made in a cold oolithic limestone, is a beautiful example of rough pointed dressing25. Fine pointed face with narrow furrows (French: Taille smillée de dégrossissage)

40This dressing is realized with a “spalling hammer” (French: smille), a small double-pointed hammer. It is a refinement of the preceding dressing, with which it is sometimes associated on the same face. It is characterised by distinctive impacts which, instead of being very short and dispersed randomly in all directions, form furrows approximately 2.5 cm long, roughly parallel and often laid out obliquely to the edges of the stone. Irregular oblique engraved dressing (French: taille ciselée oblique irrégulière)

41In the stone construction of the Jawf, irregular oblique engraved dressing appears only on small surfaces, such as the dressing on the small blocks of hard oolithic limestone in the temple of ‘Athtar dhû-Risaf at as-Sawdâ’. There, the chisel is used only to smooth out irregularities of the natural stone faces. The impacts of the chisel are distributed obliquely, neither oriented in any precise direction nor showing a homogeneous distribution of the blows. Decorative dressings

42Decorative dressings are used lor aesthetic purposes; they are not dictated by technical requirements. Consequently, the definition of these dressings applies only to the visible faces of the blocks. Regular oblique engraved dressing (French: taille ciselée oblique régulière)

43It is a variant of the irregular oblique engraved dressing. All the blows are deliberately aligned obliquely, at an approximate angle of 25°/35° to the edge of the block. It has been identified only in the epigraphic fields of certain 5th-4th cent. BC. inscriptions in the Barâqish city-wall. Pointed dressing with drafted margins (fig. 5.1)

  • 26 Van Beek 1958: 287-295.

44This is probably the most common type of dressing in all South Arabian buildings26. It always has drafted margins showing incised lines perpendicular to all four edges of the block. These drafted margins are generally of rather regular width, very often wider than that of the chisel which cut them, perhaps from 2 to 8 cm. Commonly the vertical drafted margins are narrower than the horizontal ones (sometimes only half the width).

45Pointed dressing occupies all the surface remaining in the centre delimited by the drafted margins. In its technical characteristics it is similar to the rough pointed dressing (French: taille pointée de dégrossissage) but differs in its much finer quality and in its homogeneous distribution of the impacts. The impacts are produced by a chisel with a narrow round end.

  • 27 Martin 1965: pl. XIV, fig. 2. For Thera: Scranton 1941: 78, fig. 17.

46It is still difficult to draw reliable chronological conclusions from this decorative pointed dressing and the various techniques of the drafted margins. This technique is absent from all the archaic sanctuaries, traditionally considered the oldest temples in the Jawf, where the use of the polished lumachellic monoliths prevails, Outside of Yemen, in Greece, pointed dressing with drafted margins is known at least at Taos and at Thera, where R. L. Scranton called it “tooled work27. Specific lithic dressings

47At Mâ’rib as at the Wâdî Bayhân sites, dressings are divided into two main groups according to the tools used: lithic tools (stone tools) employed for the dressing of sandstone vs iron tools used for the shaping of hard limestone. As seen at Shabwa (building n° 7) and Tamna‘ (for the granite stonework), lithic dressing is characterised by a slight convexity of the faces and a rounding of the edges.

4.2. Elements of building

4.2.1. Stone platforms with internal compartments

  • 28 Darles 1992: 110.

48Platforms with internal compartments are generally 9 m x 9 m or 10 m x 12 m in dimension. At Shabwa, one-third of the buildings are approximately 11.60 m x 11 m, another third are 13.3 m x 11.60 m and another third are 14-18 m by ca. 11 m28. At Tamna‘, houses are 9.80 m x 7.20 m and 9.60 m x 7.80 m (houses Β/Α and Β/Β). These platforms presently survive to a height of about 2-4 m.

49All the stone platforms around the Ramlat as-Saba’atayn desert and in the main valleys, but also on the highlands, are built with similar cellular platforms (fig. 4.9). But their geomorphological contexts are quite different. While Shabwa buildings lie in a salty area, sponge-like when dampened by rain; Tamna‘ buildings lie on a silt ground base, and houses built on rock also used the same cellular platforms, (dearly the stone platforms in themselves were considered by ancient masons as strong enough to support very heavy loads. Although the platform is a mixed combination of stone walls, compartments and filling, they were considered as truly being “platforms”, ensuring an equal distribution of forces and of subsidence.

  • 29 De Maigret 2003.

50These platforms are cellular bases which were built to support the superstructures. The platforms are composed of a grid of stone walls delimiting rubble-filled compartments, so as to create an elevated podium. The internal frame consists in two long main walls, generally parallel to the length of the building, and orthogonal walls dividing the platform into compartments: the two main walls are used as foundations for the main corridor and the secondary walls as foundations for the rooms. Tamna‘ house B/B has regular compartments (average 2.20 x 1.40 m), while house B/A has one central compartment 8 m long, two compartments in the southwest (including a large one ca. 6 m), and four in the northeast29 (fig. 4.12).

  • 30 Hamilton 1942.

51A number of foundation-platforms of this kind have been partly excavated in South Arabia, although their nature has not always been correctly understood and they have been mistaken for tombs30, for storerooms or left unexplained. During the 1975-1978 excavations at Shabwa, building n° 52 was excavated to its deepest levels. Its high platform consists of a 1 m thick stone wall, built in regular ashlar masonry, with one horizontal row of long stretchers. In its angles, double headers form regular bonded quoinings, and the walls shows vertical series of headers disposed at intervals every 2.30-2.96 m (fig. 4.10).

  • 31 Breton 1998b: 27-28.

52The platform of Shabwa building n° 52 includes a geometric pattern of partition walls consisting, at its upper level, of 34 compartments, numbered 201 to 234, in regular rows, except in the northwest angle where one expected compartment is missing from the grid (between 216 and 232), and in the south angle where there is a single compartment (225). All these compartments are similar in dimensions (1.30 m-1.60 m), with side walls built of small blocks with one worked face and joined with pink mortar. They have been filled with earth or sand, and contain a few bones and ceramics. In the upper levels, this filling was covered with one or two layers of mud bricks, then with a thick layer of mortar. The three excavated compartments 209, 218 and 229 were all sealed with such mud brick layers. Deep excavations inside compartments 202, 204, 209 and 225 clearly reveal the chronological phases of building. The earliest phase of wall construction consists of two central walls whose foundations lie at a relative altitude of 0 m. Then, the exterior wall of the platform was built at 1.50 m altitude. Inside the platform, new walls, oriented NE-SW, were erected above these central walls, thus forming a grid of compartments filled with regular mud-hrick courses. At the altitude of 3.50 m the frame of orthogonal walls now extended to the entire surface of the platform; the work was later completed to an altitude of 4.50-5 m. The intention of the stone platform was to raise the stone building. At the end of the construction, the whole platform stood ca. 5 m high31 (fig. 4.11).

53In Egypt, during the 8th and 7th centuries BC, some platforms were built in mud-brick with sand-filled compartments. Karnak, house A, has a 14 m x 15 m mud-brick base with 10 compartments, and another building (22.20 m square) also comprises a regular grid of compartments. But, as they mainly occur inside a sacred enclosure, these have been mistakenly thought to be “high temples” or “storage buildings”.

  • 32 Hamilton 1942: 121-122.

54The question of the use of these compartments was recently raised again in regard to the houses of the “Market square” in Tamna‘ (fig. 4.12). It seems, according to A. de Maigret, that most of these compartments were empty and used as “storerooms”. Some of them had plastered walls, and some contained remains of pottery. He suggests that wooden ladders might have been used to go down into these compartments. In one house only (B/E), stone steps inside a compartment led down to the ground. At Shabwa, Hamilton described some compartments of building 72 as “storerooms” but provided no details32. But no similar examples have been documented in the other buildings at Shabwa and at Raybûn. Although these compartments could have been used as storerooms, humidity would prevent a continuous use for storing perishable foodstuffs.

4.2.2. From the platform to the upper floors

55Generally, the walls of the platform are used directly as foundations for the walls of the ground floor; the upper walls are positioned directly over the lower ones, and their thickness must be directly related with that of the foundations. We note however, in certain cases, that the walls of the ground floor are thicker than those of the foundations.

56Two particular cases must be envisaged: pillars built inside compartments, and foundation walls of supporting pillars.

571° The presence of circular pillars built within compartments up to the level of the ground floor seems a characteristic of Axumite architecture. Such pillars indeed appear in Matarâ (building Β central: rooms E, F, Η and J; building C: main building; site EI: main building). Some compartments of the palace of Dongur in Axum contain mason-work pillars, two in the smallest compartments and four in the largest ones. To our knowledge, this type of internal pillar is unknown in South Arabia. One explanation might be the absence of long beams that could cover the whole of the compartments.

582° In the second case, the top of the wall of the base is crowned with stone bases, at regular intervals. In building “A” of the palace at Shabwa, two long walls situated on both sides of the entry porch are surmounted by a series of rectangular bases of limestone, arranged at regular intervals of 2.00 m. They supported wooden pillars which divided the two main northern rooms.

59At Tamna‘, some walls of the platform of the “Main monument” support stone bases about 1 m in section, arranged at regular intervals; also, on both sides of the central corridor, are walls supporting similar stone bases of pillars. At least two rooms, on the western side, were also divided by a series of wooden pillars. In Tamna‘ also, the base of the temple of Athirat (22.70 x 19.40 m) shows a longitudinal set of foundation walls connected to short supporting walls; these latter are all surmounted by bases of rectangular pillars. These pillars arranged in rows determine a series of rooms which line the northern, western and southern sides; the organization of the central part remains uncertain. Houses in Ethiopia

60In Matarâ, in the main building (12.60 x 11.20 m), at least 2.50 m high, only the room C (2.25 x 3.00 m) contains a staircase of eight steps, strengthened in an angle by a mason-work, pile. This room communicates with its neighbor room by a small opening. Room F, much smaller (2.00 x 0.95 m), might have served as a prison reached by a wooden trapdoor.

  • 33 Schmidt & Curtis 2008: 123.

61At Sembel site (near Asmara), there is a house consisting of a few rooms (A-C2) whose walls have partially collapsed inside. According to R Schmidt and M. C. Curtis, one of the distinguishing features of the Sembel structures is that doors from interiors to the outdoors are absent... suggesting that entry and exit must have been through the roof possibly by ladder and trap door in a solid earth/wood roof or via an opening in a grass or wooden roof33. In fact, however, these excavated walls belong to foundation compartments and not to upper rooms at all.

4.2.3. Staircases

62The high platforms were approached by means of a lateral staircase, generally “L” shaped. In Shabwa building 52 the staircase is a 10.60 m long construction, 1.90 m wide at its base, and 2.45 m wide at its upper level (fig. 4.10). Its base consists of three partition walls perpendicular to the northwest wall of the platform, defining four compartments of increasing height filled with mud bricks and earth. As this staircase is relatively light, the walls do not need a deep foundation. Such L-shaped structures are very common at Shabwa (buildings n° 44, 53, 55, 72, 73...) (fig. 4.14), at Hajar al-Dhaybiyya, and at Tamna‘ (houses Β/Α, B/B, B/C, B/H); most of them are lateral staircases leading up to the ground floor of houses (fig. 4.12).

63The sides of the staircases are formed by banisters, higher than the flights of steps. Generally these banisters are topped by blocks decorated with pseudo-header panels. If the staircase comprises a flight of many steps, the landings (the flat resting-places between two flights of steps) are very few in number (no landings at all in the main staircase of Bar an Temple at Mâ’rib, and in Shabwa palace north staircase). But the banister does have “landings” or “steps”: in the Bar’ân temple, each banister is crowned by a frieze of pseudo-header panels with bronze statues upon them.

  • 34 Van Beck 1958: 291-294.
  • 35 Breton, Darles, Robin & Swauger 1997: 53-54·

64In the “main building” at Tamna‘ (TTI), there was a narrow staircase (now looted) built into one of the large recesses of the south wall. It was L-shaped, with a small landing, flanked by a 2.80 m high stone banister (fig. 4.15). Judging by the height of the courses, the size of the blocks and the “smooth proportionately narrow drafted margins on all edges of the face”34, G. Van Beek suggests an early 6th century date (P); more probably it was 4th cent. BG. However, there was also a main monumental staircase situated on the north side of the building: it was 10 m wide, with a lower flight of seven steps, leading to a landing on which were three more steps, originally paved with limestone slabs. The two flanking banisters also had two landings with their northern façades decorated with recessed panels35.

65The vertical face of each of the banisters is decorated with narrow recessed panels, which were a common element on building façades throughout the South Arabian period. The Mâ’rib museum displays a number of such recessed panels (1 m high x 0.30 m wide), some of them originally painted in red. The eastern staircase in the Shabwa palace is flanked by two banisters, each with wide grooves in which decorated recessed panels were inserted.

4.2.4. Propylons

66The temple is generally reached through a monumental propylon, or “entrance lobby”, considered as an independent architectural element. It typically consists of a platform (stylobate) made with large blocks aligned on the entrance axis, with pillars standing in front of it. Their position is marked by rectangular mortises, 2 or 3 cm deep. In some temples, complete pillars are still standing in situ (Bar’ân temple at Mâ’rib, Na’ît), some are broken (Nakrah temple at Barâqish), some have vanished (Tamna‘ TT1), but they are all similar: rectangular shaped, slightly decreasing in volume from bottom to top, topped by square tenons, suggesting that they supported an architrave. In some temples, the stylobate is reached by a lower platform with steps on two or three sides and a flat centre paved with slabs, aligned on an axis with the entrance.

67In the Tamna‘ “Main building” (TT1) the propylon measures 7.75 m (north to south) x 3.35 m (east to west) x 0.95 m high (above the courtyard) (fig. 4.16). It includes eight long monolithic bases, aligned on the entrance axis, of different lengths, 0.20 m to 0.48 m apart, all of them showing two rectangular hollow mortises at both ends to support the pillars. One could reach this platform only by two steps to the west (there were no steps on the other sides). It can be supposed that the front pillars were originally connected with the rear pillars by stone joists, either horizontal or oblique. At Nakrah (Barâqish), the stylobate is reached by a platform at the northwest with three side steps. In some temples of Hadhramawt, the propylon consists of a single block (or two adjoining blocks) perpendicular to the entrance axis, with regular mortises to support the pillars.

68The propylon of the extra-muros temple of Ma‘în is certainly one of the best preserved in all South Arabia (fig. 4.17). The western wall of the temple includes a central gate with a monumental porch, 4.77 m wide. On the west, there are four monolithic pillars 0.51 m to 0.84 m apart (A to I)), supporting a 4.86 m-long architrave. About 1.80 m behind them, on the east, there are two pillars (E and F) 2 m apart supporting three architraves, each one recessed. On the upper architrave lay four long tilting slabs (only three remain in situ). The foundations of the pillars are unknown; their visible height is ca. 4 m. The Awam temple at Mâ’rib also has a monumental propylon, 12 m north of the “peristyle hall”, with 8 pillars whose size ranges from 0.75 m to 0.77 m (north-south) and from 0.68 m to 0.70 m (east-west).

  • 36 Röring 2005: 153-155.

69Some monumental tombs excavated in Area Β at the Mâ’rib Necropolis are fronted by a podium which supported a pillared portico. In Tomb 29, the podium on the north façade consists of eleven limestone courses with two staircases situated on both sides of the podium leading to a five-pillared portico (fig. 4.18). These pillars (two fragments remain in situ) stand ca 1 m apart. They support an architrave of a single stone lintel, topped by large raised regular panels capped with a perpendicular architrave which in turn supported five stone beams. The latter connect the portico with the tomb proper36.

70This type of propylon is a distinctive element of South Arabian architecture. It appears as early as the 9th-8th cent. BC, and archaic temples as at as-Sawdâ’ had a monumental porch. The Mâ’rib monumental tombs show similar design principles: pillars are rectangular monoliths supporting an architrave, and crowns or cornices are decorated with pseudo-header motifs. The Sirwâh propylons of the temple show similar rectangular pillars crowned with pseudo-header motifs (fig. 4.19). Propylons at Yeha

71The propylon is attested in Ethiopia during the archaic period (8th-7th cent. BC). The Grat Beat Guebri/Yeha building, partly excavated by F. Anfray in 1970, is now being reexcavated by the German Archaeological Institute (DAI). The building has a 6 m high podium with an interior frame of walls. The podium supports six horizontal stone monoliths, each in turn supporting a pillar whose original height was around 10 m. The same kind of propylon occurs in the Yeha main temple nearby. On the west, along the façade, a podium stands some 3 m high (10.40 x 5.10 m). Atop it, there are six bases, each supporting a pillar, some 9.20 m (0.90 x 0.75 m). These pillars supported an architrave; on it lay orthogonal beams covering the propylon. Some 5 m back, the propylon opens on a narrow passageway of only 1.20 m in width with a double wing door.

4.2.5. Pillars and columns

  • 37 Sedov 2005.

72Quadrangular stone pillars are the most common in the propylons of the temples. The size of the pillars is 0.40 m x 0.50/0.57 m in section with a 2 cm-deep mortise (Tamna‘ TT1); 0.25 x 0.45 m in section and 5 cm deep mortise (Dhât-Kafash in Rayhûn V); and 0.25-0.28 m x 0.34-0.37 m in section and 3/4 cm deep mortise (Sayîn dhû-Mayfa’an Temple in Raybûn XIV)37. In the Mâ’rih tombs, the size of the pillars is 0.35 x 0.46 m in section (tomb 29) with an original height of 3.74 m. Elsewhere the height of the pillars is 5.58 m (Nakrah temple), 8.3 m (Bar’ân temple), 7.65 m (Awâm temple). For Mâ’rib, Röring calculated a section-to-height ratio of 1 to 11 for the pillars of the Bar’ân (fig. 4.20) and Awâm temples, and a similar ratio for the Tomb 29 pillars (3.74 m high).

  • 38 Griaznevich 1976.
  • 39 Griaznevich 1994: 78-99.

73Stone pillars are used extensively in buildings in the Yemeni highlands. The al-Huqqa, Na’ît, Riyâm, Hâz, ‘Amrân, Ghaymân buildings had roofs supported exclusively by rows of rectangular pillars. A number of stone pillars were documented by Rathjens (Rathjens, 1953) at Ghaymân, Shibâm-Kawkabân, Sûq al-Na’âm; others by Griaznevich38; and some are displayed in the Sanaa museum. Most of these pillars are rectangular in section, and a few octagonal (fig. 4.21). These monolithic pillars end in a capital decorated with row(s) of polished flat stone, or rows of false headers, or horizontal grooves. Specimens of capitals with one to five rows are documented by Griaznevich from various sites north of Sanaa39. Wooden pillars are numerous in the Hadhramawt temples; as none have been discovered complete, their section, size and capitals are unknown. In the Royal Palace of Shabwa, octagonal wooden pillars, ca. 0.30 m in section, supported the porticoes.

4.2.6. Stelae

  • 40 Breton & Arramond 1998: 186, ph. 16.

74Stelae occur as early as the Bronze Age in various buildings. They are generally associated with buildings whose function is still unclear, i.e. whether houses or temples. Many of them have been found in Wâdî Surbân, Wâdî Hammâm or around Jabal al-Nisiyîn. In an-Nusayh (Wâdî Surbân), a rectangular building (13.30 x 9.20 m) consists of two rooms, one with a roughly dressed stela, 0.20 m in section and 1.50 m high in an angle of the room. West of Hajar Yahirr, another building has a 2 m high stone stela, and in al-Duwayr (Wâdî Hammâm) a stela rises in the centre of a room. In Rumahâ building RMH-4, the north gate is flanked by two stone stelae, one 1.80 m high, the other 1.10 m high40.

75The earliest extensive use of stelae is attested in the 8th cent. BG extra-muros ‘Athtar Temple of as-Sawdâ’ (fig. 4.22). On its western façade, two long podiums built on both sides of the entrance are crowned by four limestone trapezium-shaped stelae on the north side, and six stelae on the south side. These stelae, roughly dressed, have neither inscription nor decoration. In the middle of the central passage way, two stelae are implanted in small podiums. Inside the temple, in the southeast angle of the courtyard, there are two trapezium-shaped stelae without any decoration. The precise function of these stelae has never been determined.

76Stelae were also used as boundary stones to demarcate a sacral territory. At the Rayhûn oasis, a ca. 3 m high stone stela, ca. 50 m to the southwest of the Kafas/Na’man temple, was used as a boundary mark. We may assume that the entire territory on which the religious structures were situated could be considered as sacral territory. Similar stelae were used around the Darb as-Sabî settlement, southwest of Barâqish: each of them, ca. 2 m high, was set up to fix the boundaries of the sacral territory.

4.2.7. Benches

  • 41 Robin & Breton 1982: 617-618, 624-625.
  • 42 Schmidt 1982: 73-77, pl. 31-32.

77Benches of different types are recorded in halls, temples and city gates in Sabaean regions. The complex of Lawdh al-Ka’âb, at the foot of Jabal al-Lawdh, comprises a number of large open-air halls with series of tables and benches. Building 1, for instance (98 m long x 37 m wide), comprises two separate halls (fig. 4.23). In the east hall, there are two sets of benches 16.80 m long and ca. 0.60 m wide, and in the west hall a central corridor separates two sets of 24 wide stone benches (or tables, ca. 0.70-090 m by 0.50 m high), each associated with a narrow bench (ca. 0.40 m wide by ca. 0.30 m high). Another hall, ca. 50 m south (Building 2), displays a similar system of wide benches and narrow and lower ones, and two nearby halls also have benches. These halls were probably used during special ceremonies that have been termed “federal union treaties” (French: pactes de fédération des tribus), and thus they might be considered as “assembly halls”. Possibly ritual meals may have been distributed to the members of the tribes during these ceremonies; if so, they would have needed tables41. In this case the higher and wider “tables” should be distinguished from lower and narrower benches. We know that other similar halls, in the Mâ’rib area, had similar benches or tables where meals were probably also provided to tribes. At Dish al-Aswâd, for example, there is an open-air hall (25.50 x 15 m), with a central nave on each side of which are 7 benches, each 6 m long and 0.40 m wide (fig. 4.24)42.

78In the “Early Temple” of Jûja, built between the 9th and the 8th cent. BC, low benches were constructed along the eastern walls of the vestibule (locus 23a) and near the entrance hall. As early as the 8th-7th cent. BC, the extra-muros temple of as-Sawdâ’ has an open-air courtyard with a portico on two sides. Between all the pillars of the two porticoes, there is a stone bench, ca. 0.38 m-0.49 m high and ca. 0.62 m wide. Furthermore, a limestone bench appears on each side of the courtyard, ca. 5.90 m long and ca. 0.40 m high. In the Nakrah Temple of Barâqish, archaeologists found two sets of stone benches and great offertory tables in the side aisles of the central nave of the hypostyle hall. In the Bar’ân temple (Mâ’rib), alabaster benches were built all around the porticoes, each one having an internal hollow in which animal bones were found.

  • 43 Sedov 2005: 106.

79The Hadhramawt sanctuaries have rows of benches usually set inside naves of the hall between the bases of the pillars or against the walls. These benches also existed in other parts of the religious complex connected with the sanctuary, for instance, in the roofed gallery surrounding the sanctuary on top of the monumental platform in the temple of Mayfa’ân (Raybûn XIV). In the Raybûn I, “refectory”, the interior had mud-brick benches (0.40 m high and 0.30 m wide) along its western and northern walls (fig. 4.25). In the temple of Dhât-Himyam, in Raybûn V, the later Building 2 has square-shaped mud-brick seats, ca. 0.35-0.40 m high in the naves, while mud-brick benches of similar height occupied the space between the pillars43. A very characteristic feature of many of these benches is that they bear an inscription. The seats of the benches give the name of the donor. Such very short dedications are engraved on those parts of the temple buildings whose construction or reconstruction was paid for by the donor, for instance the “wall of Dasmân” in the Rahbân temple, and the “bench of Qatham” in the temple of Mayfa’ân. In those temples, on particular parts of the northern wall of Building 3 and on the lateral face of the long bench in the roofed gallery around the sanctuary, the names of these donors are repeated regularly.

80In the Shabwa palace benches are of two kinds: earlier ones (in the rectangular hall or “theatron” in the centre of the courtyard) are low (only 15 cm high) and 40 cm wide, while later ones, built under the porticoes of the courtyard, are 0.45 m high x 0.40 m wide; the latter have vertical slabs decorated with regular recessed motifs, and a limestone seat. The assemblage between the vertical slabs and the base consists of tenons and mortises, just as with wooden works. There were also benches in city gates, such as the Tamna‘ South Gate (fig. 4.26): two lateral benches, ca. 0.50 m high, were built along both sides of the entrance corridor, the west bench 5 m long and the east one 2.70 m.

4.2.8. The originality of South Arabian building techniques

81In conclusion, a number of building techniques characteristic of South Arabia can be highlighted:

  1. The double header system in angles is attested at Ma‘în (in the angles of the northeast tower) and at Shabwa (in the angles of the platform of the Royal Palace and in Dar al-Kafîr tower). Unfortunately, none of these constructions is very precisely dated. One may only surmise that double headers in angles had appeared by the 7th/6th cent. BC. Elsewhere, in the Greek and Persian world, the existence of this type of angle reinforcement is already known in contemporary constructions, or perhaps slightly prior to the “Iron Age”, but they are different from their South Arabian counterparts. Might this double header system have originally been South Arabian, and then spread out to Ethiopia and later to the Arabian Peninsula, even as far north as Petra44.
  2. Vertical lines of headers set regularly in the ashlar masonry walls appeared at least by the 7th/6th cent. BG They are very often associated with mud-bricks walls at al-Baydâ’ and at Ma‘în. This system may be assumed to be of South Arabian origin.
  3. Fine pointed dressings with drafted margins on all sides are very widely used in archaic South Arabian buildings. Although some comparisons exist with Ancient Egypt, this dressing technique is undoubtedly characteristic of Southern Arabian architecture45. It is used also in the Main temple at Yeha (Tigray). Similar dressings observed at Hegra or Petra occur only later, not before the 2nd cent. BC.
  4. The use of a ‘malleable rule’46 is particular clear in the city-walls at Jawf. At this period, the only possible technical parallel in the Mediterranean world would be the ‘lead rule’ which was used by the Greeks to take an imprint of stones having a polygonal shape with rounded edges. But this latter instrument is not sufficiently malleable to reproduce the tiny irregularities which were part of the building technique common at, e.g., Kamna. Hence the South Arabian procedure shows a certain originality”47.
  5. Among the various modes of assemblage, the mortises and tenons cut in the monoliths constitute a South Arabian specificity. The stone mortises and tenons, as in the intra-muros temple at al-Baydâ’ or the Awam Necropolis at Mâ’rib, are dated between the 6th and the 3rd cent. BC48.
  6. The use of prefabricated limestone slabs decorated with rectangles, so as to mimic the appearance of true ashlar masonry, is completely unknown outside of South Arabia.


1 Wright 1985: 396.

2 Bessac 1998b: 197-198.

3 The term “Cyclopean” masonry refers to huge natural blocks used as such with no dressing.

4 Van Beck 1952.

5 Wright 1985: 397.

6 Wright 1985: 397.

7 Bessac 1998a: 191.

8 De Maigret & Robin 1993: 446, ff.

9 Noel 1968: 26; Bessac 1998a: 192.

10 Wright 1985: 401.

11 Breton & Bessac 2002b.

12 Bessac 1998a: 193.

13 Wright 1985: 93, fig. 22.

14 Schnelle 2007.

15 Albright 1958: 218.

16 Breton & Bessac 1998a: 194.

17 Breton & Bessac 2002b: 150-151.

18 Bessac 1998a: 194 and 220, 229; Breton-Bessac 2002a: 152-153.

19 Bessac 1998a: 194 and 220, 229; Breton-Bessac 2002a: 152-153.

20 Bessac 1998a: 195-196.

21 Bessac 1986: 95.

22 Noel 1965: 299-300.

23 Bessac 1998a: 183-184.

24 Scranton 1941: 24. (French: parement dressé au têtu; Ginouves & Martin 1985: 130.

25 Breton & Bessac 2002: 152-153.

26 Van Beek 1958: 287-295.

27 Martin 1965: pl. XIV, fig. 2. For Thera: Scranton 1941: 78, fig. 17.

28 Darles 1992: 110.

29 De Maigret 2003.

30 Hamilton 1942.

31 Breton 1998b: 27-28.

32 Hamilton 1942: 121-122.

33 Schmidt & Curtis 2008: 123.

34 Van Beck 1958: 291-294.

35 Breton, Darles, Robin & Swauger 1997: 53-54·

36 Röring 2005: 153-155.

37 Sedov 2005.

38 Griaznevich 1976.

39 Griaznevich 1994: 78-99.

40 Breton & Arramond 1998: 186, ph. 16.

41 Robin & Breton 1982: 617-618, 624-625.

42 Schmidt 1982: 73-77, pl. 31-32.

43 Sedov 2005: 106.

44 Petra’ citadel with its double headers is not prior to 2nd cent. BC.

45 Van Beek 1958: 287-295.

46 ‘Malleable rule’ is the conventional term for a hypothetical instrument having a plasticity similar to that of clay which would have been employed to reproduce the microshape of the edge of an in-situ block upon another block so as to ensure a perfect fit (Bessac 1998: 53, note 116). Translation mine.

47 Bessac 1998a: 53.

48 Gerlach 2001: 50-51.

© Centre français des études éthiopiennes, 2015

Conditions d’utilisation :


Volume papier