Version classiqueVersion mobile

Les bâtisseurs sur les deux rives de la mer Rouge

Jean-François Breton

Chapter 6. Fortifications and urban planning

Texte intégral

1Most of the early South Arabian cities are to be found in the corridor, at an average altitude of 800-1,000 m, which separates the highlands of Yemen and Hadhramawt plateau from the Ramlat as-Saba’atayn desert. This corridor is crossed by wâdîs flowing down from the heights, and thus giving birth to oases. The main towns in this region are Shabwa at the mouth of Wâdî Irmâ’/’Atf, Hajar Yahirr at the mouth of Wâdî Markha, Hajar Kuhlân/Tamna‘ in Wâdî Bayhân and Mâ’rib in Wâdî Dhana’.

2Some cities were built deep into the major wâdîs, like Hajar Lagiyya, Hajar Tâlib, and Hajar al-Nâb in the Markha valley, like as-Sawdâ’ and al-Baydâ’ in the Jawf valley, or like Jujâ, Shibâm, and Makaynûn in the Hadhramawt valley. South Arabian cities are built upon a plain of beige-coloured silt. This situation is convenient for agriculture—cities are in the middle of their fields and gardens—and for local transport of foodstuffs and of building materials such as stone and wood. But the risks were high: cities had to face the threat of rising waters, sometimes flash-floods, so violent that they could destroy any building. A second threat consisted in the continuous silting of the neighbourhoods.

  • 1 Ryckmans 1981: 57 and fig. 53a.

3It is highly improbable that waters directly hit the city-walls: canals diverted the waters away. Some cities evidently had moats. H. St. J. B. Philby and J. Ryckmans observed a large moat on the west and south side of Najrân/al-Ukhdûd1. In Wâdi ‘Amd, the city of al-Huraydha had a moat or a reservoir facing the river flood, and Raybûn had a similar moat on its south side. Earth levees would also have been used to protect the city-walls: there is one in Asâhil south sector (with a stone facing), and a similar one in Najrân east sector. At Mâ’rib, a mud-brick wall was erected some meters in front of the west fortification.

4To combat the rising silt level, the only available method was to raise the fortifications and the extra-muros buildings. The Mâ’rib city-wall was probably elevated in this way many times during its long history, as was the Bar’ân temple. But it is difficult chronologically to correlate the rising of the silt level and the successive raisings of the walls over the course of a millennium.

5During the early 1st millennium BC, South Arabian city-walls were built on top of Bronze Age tells. In the Jawf valley, the Kamna fortification sits atop a 9 m high tell in the north-east sector, and a 6 m high slope in the south sector. The Ma‘în fortification also rises in some places 4/5 m high above an earlier settlement.

6.1. Early defensive systems (ca. 9th-7th century BC)2

  • 2 Breton 1994.

6The origins of early defensive settlements are still unclear, as few deep excavations have been conducted in South Arabia. Little information is available concerning villages and city defences in early South Arabia. On the Yemen highlands, only the Hammat al-Qâ Bronze Age fortification has been excavated.

7An examination of the design of urban fortifications requires a broad analysis of its main elements. We will classify the main design elements as: curtains, towers and gates. “Curtain” can mean a barrier wall, more or less used in a military sense, but also has the additional connotation of a stretch of wall connecting defensive features. Thus two systems need to be examined. The first type of “curtain” consists in a series of dwellings arranged in a circumvallation, and the second one in casemate walls.

6.1.1. Casemate walls

  • 3 Wright 1985: 181.

8As few specific excavations have been conducted thus far, it is difficult to distinguish walls of houses from so-called casemate walls. Casemate walls are perhaps more prominent in nonurban fortifications but their varied ramifications and background warrant their being considered here. “To construct a casemate massive wall by two faces of masonry and an infill is a basic technique and further to stabilise it by compartmentalising it with cross walls, i.e. utilisable chambers in its thickness... In this way it can be seen that typologically the casemate wall goes back to the most primitive type of enclosure, a circumferential ring of cabins”3.

  • 4 Breton 1994: 141 and fig. 36.

9Casemates forming part of small-size Iron Age fortifications in South Arabia (9th-7th century BC?) are somewhat problematic at some sites. In Hajar ar-Râ’, in Wâdî Mafâyir, SE of Tamna‘, the defensive perimeter shows two parallels walls with cross-walls, but surface observations could hardly distinguish “casemates” and “houses foundations”. The nearby Hajar al-Barka, in central Wâdî Markha, would be the most interesting case of casemate construc tion4 (fig. 6.1). The south defensive line consists of two parallel walls, 1.70 m apart, each one 1.30 m-1.50 m thick, with cross-walls every 2.30 m-2.70 m. They were then filled in with earth and silt. In the west sector, two parallel walls, 3 m apart, are connected with cross-walls every 4-5 m. Excavation would determine if the 900 m long curtain consists entirely of casemate construction. The same issues arise for the defensive systems of Hajar Kuhaylâ (north of Harîb), of Janâdila (see above), of Hajar Tâlib, etc.

  • 5 Herzog 1983 and 1984.

10Due to lack of excavations, the chronological frame of these two types of curtain remains uncertain. Casemates walls in the Jabal al-Nisiyîn areas (Hajar ar-Râ, al-Barka, Surbân) yielded ceramics of the 8th-7th century BC and a few archaic inscriptions (Hajar ar-Râ) belonging to this early phase of urban settlements. In Palestine, curtains of casemate construction are a feature of the Early Iron Age with a number of examples (Hazor, Gezer) showing similar details, but not restricted to this period5. During the 10th century BC, casemate curtain walls were widely used in Palestine (Tell Beit Mirsim, Tell Qasileh, Beth-Shemesh) and in a number of Negev fortresses (Ramad Matred); they were used again during Iron II (Tell al-Fûl, Beersheba: levels III-I).

6.1.2. The “enclosed settlements”

11The so-called “enclosed settlements” exist all through Arabia and Syria-Palestine although few are described in publications; in Palestine, enclosed settlements of the 12th-11th century BC are known in the Negev. Functional architectural studies in the 1980s made it possible to distinguish enclosed settlements and casemate forts.

12“Enclosed settlements” consist in a series of dwellings arranged as a belt around the circumference of the hilltop. Each dwelling was built separately; their rear walls do not form a straight line from one house to its neighbour. Additional walls connected the dwellings to one another. The outer boundary of the site is enclosed by a single wall. This is not a plan of casemates adjoined by additional rooms but rather a circumvallation of houses, the broad rooms at the back of each creating the illusion of casemate walls. Moreover, there are no houses (or only a very small number) in the centre of the enclosure. The entry consists only in a narrow passage between two houses, which is not a gate with flanking towers. Such multi-dwelling units of the “enclosed settlement” type were suitable to accommodate small tribal groups or families.

  • 6 Breton 1994: 142-146.

13Small-size settlements are found in South Arabia, some 300-500 m in perimeter, as at Dar as-Sawdâ’ (in Mâ’rib oasis), Hajar Khamûma (in Wâdî Khamûma), Hajar al-Barka (fig. 6.1) (in Wâdî Markha)6. Houses are generally small in size (6m to 8m on average); they have a stone platform with an internal grid of orthogonal stonewalls. Large-size settlements with a circumvallation of houses are to be found in Sabaean areas (at Yalâ), in Qatabanian territory (at Janâdila (fig. 6.2), at Tamna‘ and at Hinû az-Zurayr) or in the north (at Najrân). The central part (or tell) at Yalâ has some twenty houses built with dark grey blocks of schist. In 1987, the Italian Archaeological Mission excavated “house A” at Yalâ, belonging to this central ring of houses (fig. 6.3).

  • 7 Breton 1994: 120-122.

14Hinû az-Zurayr, in the Wâdî Harîb, west of Wâdî Bayhân, also shows a continuous peripheral belt of buildings (fig. 6.4). Its defensive perimeter consists of buildings, mainly houses, irregularly built on a large trapezoid (250 m E-W by 200 m N-S). Dwellings are clearly identified by their dimensions (12 m x 10 m), their high-rise granite platforms with an internal grid of partition walls, and some traces of their superstructures. Some larger buildings (storehouses?) in the south and southwest sectors are adjoined or connected by one or two parallel walls. The western and the eastern gates consist in a passage between buildings7.

  • 8 De Maigret & Robin 1989: 288-289.

15The great variety of “enclosed settlements” does not provide a homogenous picture and does not indicate a distinction between fortresses, villages and small fortified towns. In the city of Yalâ, the excavation of house A provided a simple chronological frame: the destruction of the rooms and the upper floor is dated between the 8th and 7th century BC (stratum A), and its construction could be dated 10th-8th century BC? (stratum B)8. It can only be surmised that the early ring of dwellings was planned during this period. Would the early ring of houses at Tamna‘ be dated to the same period? That would fit with the foundation of the “Main Building” (or “TΤI Building”, 8th century BC?).

  • 9 Herzog 1983:41-49.

16Once again, Palestine offers a similar occupational pattern: during the Early Iron Age, the “enclosed settlements” emerged as a new population pattern in Beer-Sheba (stratum VII), in Nahal Beer-Sheba (site E), in Mesad Hatira, and later during Iron Age II (in Samaria, Tell al-Fûl, Beer-Sheba, levels III-I)9. At Beer-Sheba, the former casemate walls were transformed into a belt of dwelling units (Level VII, then IX).

17It is worth mentioning that some of the ring settlements in the Qatabanian areas were in use during the first millennium BC, and sometimes till the 2nd century AD. Hinû az-Zurayr had an early belt of separate buildings units, and its centre was gradually occupied with similar houses, temples or storehouses (fig. 6.4). This defensive line was in use for centuries, from the 7th century BC? (?) on. When the primitive defensive system was extended to the west in order to include the main extra-muros temple, a new line of detached buildings was laid out some 10 meters further on (to the west). The early passage between two nearby houses was replaced by a similar one, approximately 15 m north.

  • 10 Zarins 1983; Breton 1994: 150-151.

18Najrân city (fig. 6.5), locally named “al-Ukhdûd”, or the “Qasr” of H. St. J. B. Philby, is built in the middle of irrigated fields10. The walled city forms a quadrangle about 235 m square. The defensive line consists either in separate buildings connected by a single or two parallels walls, or adjoining buildings. None of them are erected in a continuous line, so that the whole walled perimeter shows an elaborate hut quite haphazard succession of buildings (houses?) alternating irregularly with intervening curtain walls. Projecting “bastions”, at corners and on some sections of the northern and eastern walls, are inserted in the defensive system. Every dwelling had its own individual shape and size (average 12 m by 15 m), a granite platform preserved 2-3 m high, and clay remains of the upper floors. About fifty similar stone platforms were erected in the centre of the city. Later, gates were inserted in this frame between adjacent houses. According to inscription RES 3945 from Sirwâh, Karib’îl Watar destroyed the already existing city in the middle of the 7th century BC. Najrân city was occupied continuously during the first millennium BC and the first half of the first millennium AD and it was never protected by a detached exterior city-wall.

6.1.3. From “enclosed settlements” to fortifications

19Some enclosed settlements were later encircled by a continuous city-wall. Two sites provide the key for understanding this major architectural change.

20Yalâ shows two distinct development phases. The earlier one consisted of a ring of independent dwellings, while during the second phase a fortification was erected all around during the 6th century BC. The first building phase began during the 10th-9th century BC (?), or possibly even earlier (?).

21Hizmat Abû Thawr, located in the lower course of Wâdî al-Kharîd, a tributary of Wâdî Jawf, although not excavated, has yielded three main inscriptions related to its architectural development. The two older ones refer to persons building their houses on the defensive system of the city; the last speaks of a mubarrib building the city-wall of Mnhytm.

22The earliest Sabaean fortifications (8th-7th century BC) show rectilinear sections without any tower. Khirbat Sa’ûd (the ancient Ktlm) (fig. 6.6), located in the middle course of Wâdî Ragwân, 50 km NW of Mâ’rib, is known as one of the major Sabaean city-foundations. It is a 195 m long by 175 m large rectangle with a canted side (west side) and a recess (east side). The thick wall (2.5 m minimum) has a double face with stones set as coursed squared rubble, and an internal stone filling. Long rectilinear sections are reinforced by large buttresses every 13-14 m whose top reaches the same level as the curtains. In some buttresses, mainly in the southern and western sectors, long stones with dedications are inserted bearing the names of the “mukarribs of Saba’”.

23Al-Asahîl (the ancient ‘rrtm), 2.5 km SW of Khirbat Sa’ûd, is a more complex fortification. In the SE and SW sectors, there is a thick rectilinear wall with buttresses, in the Ν and Ε sectors there are short sections of indented wall, and in the W sector there are only cross-walls connected with curtains. The rectilinear wall, as thick as at Khirbat Sa’ûd, has long buttresses (7.50m long) every 12-13 m. In the northern sector the city wall (ca. 2.5 m thick) is built as a succession of ca. 10-12 m long units, each one offset ca. 0.50 m. In this sector, an inscription mentions the name of Karib’îl Watar, son of Dhamar’alî, mukarrib of Saba’. Later inscriptions indicate that three mukarribs built different types of curtains, during the 7th century BC.

24Jidfir ibn Munaykhir (the ancient Khlm), 43 km NW of Khirbat Sa’ûd, in the lower course of Wâdî Jufra, shows similarities with the two previous sites. Its fortification is an indented wall, 1 m to 1.5 m thick, built as a succession of units 28 m long (maximum: 40 m), each one offset ca. 0.70-0.80 m from its predecessor (fig. 6.8). The advantage of this type of construction was not primarily military, except that from the salients a soldier could control long sections of the walls; rather, the indentations serve mainly to provide physical stability. No bastion protects this city, except a late one on the south side.

25These three archaic Sabaean fortifications all bear inscriptions of the 7th century BC, which allows us to date this building style. The standard Sabaean rubble wall is built with two faces of stones set as coursed squared rubble, with the interior made up variously of smaller stones and earth (or mud mortar). However, any distinction between coursed rubble, uncoursed squared rubble and uniform coursed rubble is more of geographical than of historical significance. Unfortunately, there exists no general work on South Arabian walls, and while ashlar masonry has received attention, the bulk of wall building through the 2nd-1st millennium BC has never been subjected to technical analysis.

26The earliest defensive systems, as described above, do not incorporate towers. Tower arose later, as an extension and expansion of the (relatively small) salients. At Khirbat Sa’ûd and al-Asâhil, the thick walls are regularly buttressed for stability purposes, and not for defensive ones.

  • 11 Breton 1994: 60-62.

27One noticeable feature of Sabaean walls is the relatively small number of gates and posterns. Khirbat Sa’ûd has only two symmetrical gates, Al-Asâhil has two (visible) ones, and Jidfir ibn Munaykhir two also: one main gate with a flanking bastion and a small northeast postern11. Clearly the archaic form of gate involves a break in the curtain. For the main gate an inevitable development of the break arises on structural grounds; the break must then be thickened for stability. There is an evident military advantage in channelling entry through a narrow corridor. The Khirbat Sa’ûd gates consist in simple openings extended by narrow interior corridors; the al-Asâhil NW gate has the same gate system, as does Kamna (east gate). The Jidfir ibn Munaykhir south corner gate has a long entry channel with a two-chamber lateral gatehouse. This could be considered as a prototype for Sabaean gates. In Mâ’rib, the western gate has a recessed entry channel, 3 m wide, inside an 8 m thick wall, and the corridor is controlled by a lateral chamber (6.39 m x 1.96 m) built inside the wall masonry, with a small door in the middle.

6.2. Urban fortifications (from 7th cent. BC to 1st cent. AD)

6.2.1. Construction, building techniques

28The “archaic” (Sabaean and Minean) city-wall consists in two elements: an inner mud brick massif and an outer stone wall bonded to the massif. Generally mud brick massifs have not been studied in detail. They are 2 m thick at Barâqish, 3 m thick at Shabwa, 2.20/2.40 m thick at al-Baydâ’, and 3.50 m high at Ma‘în and 6 m to 7 m high at Barâqish. The external ashlar masonry wall is bonded to the mud brick massif with stretchers, laid every ca. 3 m. These stretchers thus form a coherent system for anchoring the stone wall to the mud brick massif.

  • 12 Schnelle 2000 so-called “implantate” type.

29The Sirwâh defensive system consists in a different type of construction (fig. 6.7): long curtains with towers (or salients) on the southeast, long sections without towers (on the north), double walls (on the south), double walls with internal filling12 (on the northwest). All these sections belong to different periods. Two main buildings are included in the city-wall: the Almaqah Temple (the so-called “oval temple”, on the east) and an “administrative building” on the northeast.

6.2.2. Curtain

30The curtains walls of the fortifications built between the 6th and the 4th century BC are much slighter and their trace is generally a regular succession of salients and recesses. To determine the source and motivation of this kind of jagged trace of salients and recesses is not straightforward. As fortifications were made up of a mud-brick core and an outer stone wall, salients with stone bondings would exercise a buttressing effect on the curtain. But there is no direct connexion between the height of the city-wall and of the protruding salients. It seems that, whatever military virtue (greater height) the new walls were designed to possess, the uneven trace was adopted basically for reasons of prestige and aesthetics.

31The ratio between salient (or tower) and curtain changes over the centuries. In archaic fortifications: Mâ’rib has curtains 33/35 m long and salients only 5 m long (ratio 6.5/7), al-Baydâ’ has curtains ca. 19.7 m long and salients 7.7 m (ratio 2.55), Barâqish has curtains 8.5 m long and salients 5.8 m long (ratio 1.45). At Shabwa the older walls have curtains ca. 9 m long and salients 5 m, but in later walls the curtains are 6.6 m long and towers 6.50 m long, and there are even two cases where towers are longer than curtains. This is evidently a reflection of the increasing importance of towers, probably for stability of the city-wall.

32Little information on fortifications is available from Hadhramawt, as most of the cities seem not to have been fortified (Raybûn or Huraydha). In eastern Hadhramawt, Makaynûn has a 220 m by 150 m city-wall generally built in rough coursed masonry. The main phase of its building belongs to the second half of the 4th or early 3rd century BC. In the Ε and SW sections, the best preserved ones, salients alternate with curtains. The salients, showing considerable protrusion in the SE sector (up to 6 m), are smaller and more regular in the SW sector (protruding only 2.60 m to 3 m), and the curtains are only some 4 m to 5 m long. To the north, curtains are generally 3.40 m long, and salients 2.70 m to 3 m long. In this place, the city-wall sits behind an outer wall, one meter thick, with large buttresses.

33As the archaic fortifications have no salients but only buttresses (Khirbat Sa’ûd), the major change consists in the development of salients and then towers.

  • 13 Breton 1994: 50, 95-97.

34Al-Baydâ’, in the middle course of the Jawf valley, some 95 km NW of Mâ’rib, belonged very early to the Sabaean kingdom (fig. 6.9). Its well-preserved succession of 58 salients and curtains, buit in an oval shape, 1,500 m in perimeter, is one of the most prominent city-walls of all South Arabia13. The salients are 7.70 m long, protruding only 2.20 m, while the curtains are 19.65 m long. The curtains and the salients are both the same height, ca. 4.4 m. They are all built in ashlar masonry, with each course of stones having the same height, carefully dressed with drafted chiselled margins, and one of the upper courses has been polished to receive a long inscription. The top of the wall ends in a frieze of pseudo-headers (or dentiled blocks). The face of the wall consists of long stretchers with some lines of vertical headers inserted in the mud-brick masonry. Their height, ca. 2 m high, indicates the level of the mud-brick behind them. Three courses below the top, the wall has a line of regularly set openings, at least 4 to 6 in the face and 2 on each side. These were not used as loopholes but as holes to insert the beams of a wooden platform behind the curtain.

35The Ma‘în city-wall, one of the most important of the Jawf valley, has been largely plundered, so that there remain only a tower and a curtain in the east side and three gates (fig. 6.10 and 6.11). The remaining curtain, 7.80 m high and 1.08 m thick, is built with stretchers 0.76 m to 1.89 m long and headers 0.25 m to 0.38 m wide. The northern part of the wall displays alternating stretchers, serving as a link to the next part of the curtain. In the 18th and 28th courses, there are two vertically separated parallel rows of narrow holes that go all the way through the wall, one 3.60 m above the other. These openings could only have been used to insert beams to support wooden floors. Thus, it is possible to reconstruct a lowest level just above the mud-brick wall (behind the curtain), a second level 3.60 m higher, and finally a third level used as a covered way. Wooden stairs or ladders would have been used to reach each of these levels. This Ma‘în curtain is the best-preserved example of a curtain in the Jawf valley military architecture of the 6th-5th century BC.

36At Ma‘în, there is only one complete tower, which flanks the eastern gate. It is 4.70 m projecting and only 4.30 m wide. In contrast with the nearby curtain wall, it is built with rows of stretchers, some 0.40 m thick. In its angles, alterning headers every two courses provide a remarkable strengthening. The outer face of this tower was not completely dressed, evidently due to the lack of time or money, and the internal face shows undressed quarry stones. This tower has two rows of holes that go all the way through the wall, between two stretchers in the same course, the altitude difference between the two rows being 3.70 m. The openings lie at the same altitude as those in the nearby curtain, so the system of internal covered ways or floors was very similar in layout to the curtains.

37Isolated towers were formerly considered to be few, but surveys demonstrate their extensive use in parallel with the traditional South Arabian defensive system still to be found in Yemeni villages today. Isolated towers were erected some meters in front of the house line in the “enclosed settlements” as in Hinû az-Zurayr (south sector), to protect the entrances. In the Yemeni highlands enclosed villages display just such isolated tower-houses, some of them connected by covered bridges (near Mahwît).

6.2.3. Functions of the city-gates14

  • 14 Breton 1994: 59-77.

38The gates generally fulfilled a threefold function: passage, service and defence.

  • The Shabwa main gate (n° 3), because of its strategic position, is an especially important issue, it opens directly onto the main street and the “main temple
  • The main gates of the cities were used for services and probably taxation. In the western gate of Ma‘în chambers and porticoes were designed for trade and taxes (fig. 6.13). The Ma‘în text n°4, located on one tower, indicates a tax office. At Shabwa, the function of service remains unclear, as no text has been found in situ. The issue of collecting taxes was raised by Pliny: “The incense is carried by camels to Shabwa where only one gate is opened. There, the priests withdraw taxes not by weight but by volume in the benefit of Sayyîn...” (Pliny, Natural History, XII, 63-64). Given the narrowness of the passage entrance (gate n° 4), it would be inconvenient to unload camels and to weigh theirs loads there, and quite likely these operations were carried out at one (or two) of the exterior northern gates. These gates 7 and 8, built at low points between two hills within the city, control the roads from the north, and are associated with specific fortification lines with regular towers and some buildings.
  • The question of the adjacent positioning of the “main gate” and the “palace” is of concern for some South Arabian cities. However, such an association of gate and palace or “Administrative building” is not chronologically well established at Sirwâh and at Shabwa.
  • 15 Breton 2003: 202.

39At Shabwa, the “Royal Palace” assumed the defence of the central northern sector. Its U-shaped building (B), consisting of one floor in the early phase, directly dominates the gate, but its central tower (building A) dominates not only the gate but all of its accesses. In the absence of deep excavations, the different building phases have not been resolved. Some suggestions have been made that there was an early building there, later incorporated in the new city-wall, which shows a 20 m long salient to the northwest15. During the first centuries, the “Royal Palace” occupied a key position in the Shabwa defences.

40Finally, the question remains as to why these fortifications were constructed. With their high walls, which measured 8m in height, and their massive towers, it seems unlikely that any could have been successfully assaulted or destroyed by artillery and, indeed, it is most unlikely that South Arabian armies in the first millennium ever possessed such equipment. The only opportunity would have been to besiege the city and hope to cut its water or food supply in order to force surrender. Alternatively, the function of these walls may have been political or to enhance prestige, since they acted as highly visible urban political landmarks in the lowland oases. The peasants, nomads and shepherds were thus outside the wall, whereas the tribal, religious and commercial authorities were within; in this respect it is significant that the main “caravan cities” such as Ma‘în, Bârâqish, Mâ’rib and Shabwa had the most prestigious fortifications.

6.3. Town planning

6.3.1. City walls and inscriptions

41A central issue is the system of land law at the various stages of urban development, village and then city. Were the walls built on state land or private land, and under what sort of title and estate was this land alienated into public ownership, as city-wall? Was this title registered or was it attested by virtue of long-term possession, i.e. for two or more generations? Or were the city-walls still private property? Some vestiges of information concerning land tenure are available from inscriptions.

42The Sabaean fortifications were founded directly by the mukarribs: they testify to the territorial expansion of Saba’. Their dedications texts are always similar: “X, son of Y, mukarrib of Saba’, has built the tower XXX in the city of Y...”.

43The Minaean inscriptions are different. Bârâqish inscriptions mention the names of the tribe Gab’ân, divided into clans and sub-clans: Zalwamân, ‘Aman, Sha’thum, Bawsân... It appears that members of certain tribes or clans were in charge of building some parts of the city-walls.

  • 16 De Maigret & Robin 1993: 430.

44A detailed study of the Barâqish SW sector provides more information about land tenure16 (fig. 6.12). Based on inscriptions written on the towers and curtains in this sector, it is clear that most of the towers and curtains were built by the members of four sub-clans: Zalwamân, Yafa’ân, Bazr Sahfân, and Sha’thum. These four sub-clans built the towers 1 (?), 2, 3, 4, 5 and 58. Members of the clan Yafa’ân were associated with the sub-clan Sha’thum to build the towers 4 and 5 (inscription RES 3022). Thus it seems that most of the SW Barâqish sector was planned and built by the Sha’thum sub-clan, part of the tribe Gab’ân, during the reign of Abîyada’ Yatha’. This clan was probably settled in this part of the city and contributed to its fortification. The tribe Gab’ân, probably the royal one, can be supposed to have been the main land owner.

45The different sections of the city-wall must have been situated on territory belonging to several tribes and clans; these were in charge of building a part of the city-wall, probably on their own territory. Such an important building site needed the cooperation of the various tribal components. Tribes must have been obliged to conclude agreements to do their part in creating a comprehensive town design and a complete city-wall.

46The process of coordinating the building work can be seen only at Ma‘în and at Barâqish. The eastern sector at Ma‘în was built in 4.75 m to 4.90 m long sections. These lengths, very regular, were probably each designed and executed separately; they may also have been separately financed. But they are all inserted in the general layout of the city wall. The towers were probably also designed in the same way: they display regular 4.70 m-4.90 m long sections. The vertical lines of headers correspond to building sections. Each clan was responsible for building a particular length of the wall, specified in the inscriptions in cubits. In the case of a tower, one, two or three clans (of the same tribe?) had to agree on designing the tower and on financing its construction. The city walls thus reflect the social organization and the economic situation of the kingdom of Ma‘în.

6.3.2. Layout of the cities

47From a socio-economic viewpoint it would seem reasonable that a network of streets enclosing blocks divided up into recognisable building lots must have presupposed a land department administering real properties.

  • 17 Breton 1992: 61-66.
  • 18Sabkha” is the Arabic word for “shallow salt lake” or “wet saline plain”.

48Shabwa, the old capital of Hadhramawt, is situated in the midst of a large triangle of hills. This triangle divides the Wâdî ‘Irmâ into two main branches, the Wâdî Mi’shar in the west and the Wâdî Mikhbad in the east17. These hills, about 50 m high, form a roughly equilateral triangle 900 m on a side. In the south, the al-‘Aqab hill overlooking the Wâdî Mi’shar protects the whole city from floods and silt. In the west, the Qârat al-Hadîda range reaches an altitude of 747 metres. The centre of the triangle is a flat depression locally called al-Sabkha18, at an altitude of 699 m, where four salt mines employed in antiquity still remain in use today. This flat area and the adjoining city are connected to the surrounding irrigated fields by passes through the hills (fig. 6.14 and 6.15).

6.3.3. The fortification system

49It may be assumed that the erection of the Shabwa city-wall began in the 7th-6th century BC. The earliest walls are visible in the Dâr al-Kafîr tower, in the western corner, and in the al-Hajar citadel in the eastern part of the city. As these two spots are rather distant from each other, it is probable that the whole defensive system was planned in advance. Shabwa is nevertheless unique in that its citadel was at first separated from the city and only later connected with it (fig. 6.14).

  • 19 Darles 2003: 215-220.

50In the western sector, the city-wall forms a trapezium with a 350 m long base. This fortification has five gates, which are, from west to east: a narrow gate (Gate n° 1) down the al-‘Aqab range, at the foot of Husn al-Mâ’; a second gate (Gate n° 2) about 80 m northeast of D âr al-Kafîr; the main gate (Gate n°.3) is near the “Royal Palace”; an important gate (Gate n° 4) with a unique asymmetrical plan occupies the northern corner; and a narrow passage (Gate n° 5) stands in the eastern sector of al-’Aqah range19.

51Generally these gates give onto open spaces and not onto regular “streets”, and only Gate n° 3, with its north-south axis, was intended originally as a path leading to the main temple. The case of Gate n° 2 is unclear but Gate n°4 opens onto an irregular space measuring approximately 200 m (east-west) x 100 m (north-south), an area of some 2 ha. This early fortification was later enlarged with an indented wall on the crest of the hills; it has two main gates (Gates n° 7 and n° 8). In the southern sector, additional walls later encircled the silt projection south of al-Hajar hill.

6.3.4. Streets and districts

  • 20 Van Beek 1952.

52Tamna‘, the old capital of Qatabân, fortified with an 1,800 m long defensive system, has four main gates, each one located in an angle. The South Gate, excavated in 1950-1951, is a major “gate-house” associated with a granite glacis, ca. 30 m long and ca. 5 m high (fig. 6.18). The gateway is formed by two flanking towers, 3.30 m high, constructed with massive roughly-coursed masonry, some blocks being as large as 2.50 m long. The considerable quantity of charred wood found in the gateway originated from its upper wooden storeys. Built against each tower on either side of the entrance were long stone benches. The gateway gave access to a large court from which streets opened out leading to the different sections of the city. A number of buildings separated two parallel streets, approximately oriented NW-SE. However, excavations have not been able to provide a clear chronological frame of the urban plan of the whole sector. It seems that early buildings, such as buildings A (House Yafa’am), D and B, demarcated residual empty plots of land which were later used as streets. These scattered houses were then combined with terraces in order to give the houses a more coherently aligned front-line so as to create well-defined streets. For instance, northwest of building A, there is a long stone platform with staircases; north of building B, annexes have been added to create a long straight line. Building C, erected north of and across a narrow east-west street from building D, was approached by two stairways; it occupied a “dominant position” suggesting “that it may have been a government building”20. The planning of this area involved private, religious and public buildings.

53At Shabwa, the main “street”, behind Gate n°3, may have been an early axis, sloping up gradually from Gate n°3 to the “main temple”, from 705 m to 712 m high (fig. 6.16 and 6.17). It did not result from any kind of early town planning but from “empty spaces” between buildings or from boundaries between private territories. It gradually took on the role of the main street of the city.

54The only symmetrical axis opens behind Gate n° 4 onto plots of land at a relatively low altitude (700-707 m) surrounded by a few buildings (91, 92, 94); but here again this axis is only the result of previous building positions. Gate n° 2, NE of Dâr al-Kâfir tower, opens onto a plot of land with few buildings. We can suppose that a narrow lane lay between buildings n°21, 22 and 23. So, in general it is doubtful that these secondary streets constituted any particular pattern. Most probably secondary streets were meant to give access to a district city quarter; and for the rest circulation was by way of residual alleys, only practicable for going to visit the neighbours. The “blind alleys” between buildings n° 101, 102, 103, 104, 105 have a distinct role in the social organization of the local neighbourhood.

  • 21 Darles 1992.

55The western intra-muros sector is covered by a large number of stone buildings, consisting of high stone platforms and covered by debris of their former mud brick and wooden frames21. More than 120 have been recorded over less than 70% of the surface, the remainder having been buried beneath the modern village of Mathnâ’ or destroyed by salt mines. The total number of buildings covering the whole surface is therefore estimated at 200-250.

56In the most recent periods of the city, houses were concentrated in the central and eastern parts of the city. The western districts seem to have been less heavily occupied than the eastern ones, and none of the buildings have yielded any imported Roman ware or glazed pottery. The population of Shabwa seems to have been concentrated more and more in the central and eastern districts, above an altitude of 710-713 m. But the “Royal Palace” (building n° 55) and the neighbouring buildings (n° 52-53, n° 56-57) still remained the core of the administrative city. The “main temple” (n° 40), extending along the high northern slopes of the al-‘Aqab range, dominates the whole city and the main street going downwards to the northern gate (Gate n° 3).

57How did the population settle in Shabwa? We can assume, as far back as the Iron Age and possibly earlier, that families gradually settled there, sharing the main parts of the city among them. Possibly by mutual agreement or by force, they delimited plots of land to set up their homes. These plots were probably large enough to allow for further growth of each family within its boundaries. The urban centre became progressively denser, and large plots of land were divided up by passages between buildings, short lanes and dead ends, for instance near buildings n° 72-84.

  • 22 We refer here to the rules of the central market of Tamna‘ recorded on a large inscription.

58This raises the question of whether or not there was any central marketplace. Could the plain of al-Sabkha have been used as an extra-muros marketplace with its own specific organization and rules?22 If that plain was indeed mainly used by caravans, merchants and people near Shabwa, was the extra-muros temple on the southern slope of the Qarât al-Hadîda range reserved for them? Could the large space inside the city east of the main road, near the Royal Palace, some 50 m x 50 m, be considered as an intra-muros “marketplace”? In this case, a parallel may be drawn with the ’SMR’ central market of Tamna‘ (fig. 6.21). Were there intra-muros markets in different districts inside the city? All this is still unknown.

59In a later phase, and probably not earlier than the 1st century AD, the city’s wealth and the power of its kings transformed the urban landscape. The earlier north-to-south “path” changed into a central “main street”, connecting the northern gate (Gate 3) with the “main temple”. The existing buildings were then combined with terraces in order to bring about some minimal degree of alignment of the buildings, but even in the later periods, the “main street” varied in width from 8.60 m to 11.40 m. At the end of this street, the so-called “main temple” evolved into a magnificent dominant landmark of the city.

  • 23 Sedov in Frantsouzoff 2001: 21 (translation mine). Nonetheless, it is clear that erosion later cut (...)

60At Raybûn, a Yemeni-Russian mission excavated five temples (included Hadrân and Rahbân Temples) and some four houses. The recently published city-map (fig. 6.19) shows that the city had no fortifications and hence no city gates. The buildings seem scattered all over small loess hills, separated by gullies. “The unsystematic placement of the buildings makes it difficult to recover the original orientation of the streets and the location of open spaces; the construction of buildings in this city does not appear to have followed any regular plan”23. These overly brief remarks are not sufficient to specify the urbanization process of the town with any precision.

  • 24 Salles & Sedov 2011.

61The harbour of Qanâ’ (now Bir ‘Alî) (fig. 6.20) has a number of buildings scattered along the seashore24. At the north and northeast, larger houses with “patios” would have belonged to wealthy merchants and smaller houses are to be found in other districts. As far as we know, no regular city planning ever existed at Qana’.

  • 25 Breton 1994: 41-46.
  • 26 Becker 2000: 262-267.

62Hajar Yahirr, the capital of the kingdom of ‘Aswan (?)25, lies in the lower course of the Wâdî Markha. The city wall, about 1,100 m long, is partly covered by sand so that its gates are not now visible. But the use of a magnetometer has made it possible to discover some parts of the city plan under the low Bat sands26. It appears that the city plan consisted in square buildings (circa 10 m x 10 m) lined up along the streets, thus forming a regular pattern. Smaller blocks (or houses) are associated with secondary thoroughfares. This kind of rectilinear town plan seems to occur nowhere else in the Hadhramawt or ‘Awsân kingdoms.

6.3.5. City planning in Tigray

63In our present state of knowledge, no Axumite site was ever surrounded with fortifications; this is doubtless the most original feature of these settlements. Axum and all the cities situated further north (Matarâ, Tokonda, Kaskase, Qohaito, Adulis) or further east (Maqabar Ga’ewa, Agula) appear to show no urban defensive elements whatsoever. But it would doubtless be naïve to believe that peace always reigned over these highlands, that the kingdom of Axum expanded from the 1st century BC till the 7th century AD and that there were no defence systems that left visible traces.

  • 27 Dostal 1983: 99-100.

64We may suppose that the defence of cities devolved on the countryside, by means of towers, advanced lines of small forts, light walls of defence, and control of access points (roads, pathways, valleys, passes, etc.). Such light systems are attested today in certain rural regions of traditional Yemen. Free-standing towers were used less for the protection of families and their property than to warn the inhabitants of an impending attack. the border regions of tribal territories, several villages and field towers were often grouped together to form a unified defence system27.

65That the more or less high tower-houses contributed to a large extent to urban defence is also a conception that was widely implemented in South Arabia. This holds even in cases where the cities were provided with powerful fortifications, as at Tamna‘, Shabwa and Mâ’rib.


1 Ryckmans 1981: 57 and fig. 53a.

2 Breton 1994.

3 Wright 1985: 181.

4 Breton 1994: 141 and fig. 36.

5 Herzog 1983 and 1984.

6 Breton 1994: 142-146.

7 Breton 1994: 120-122.

8 De Maigret & Robin 1989: 288-289.

9 Herzog 1983:41-49.

10 Zarins 1983; Breton 1994: 150-151.

11 Breton 1994: 60-62.

12 Schnelle 2000 so-called “implantate” type.

13 Breton 1994: 50, 95-97.

14 Breton 1994: 59-77.

15 Breton 2003: 202.

16 De Maigret & Robin 1993: 430.

17 Breton 1992: 61-66.

18Sabkha” is the Arabic word for “shallow salt lake” or “wet saline plain”.

19 Darles 2003: 215-220.

20 Van Beek 1952.

21 Darles 1992.

22 We refer here to the rules of the central market of Tamna‘ recorded on a large inscription.

23 Sedov in Frantsouzoff 2001: 21 (translation mine). Nonetheless, it is clear that erosion later cut out gullies and thereby exposed the platforms of the buildings.

24 Salles & Sedov 2011.

25 Breton 1994: 41-46.

26 Becker 2000: 262-267.

27 Dostal 1983: 99-100.

© Centre français des études éthiopiennes, 2015

Conditions d’utilisation :

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search