Version classiqueVersion mobile
OpenEdition Books

Les bâtisseurs sur les deux rives de la mer Rouge

 | 
Jean-François Breton

Chapter 1. Ethiopia and South Arabia: Geographical similarities

Texte intégral

1.1. Two mountains separated by the Red Sea

1The main phase of the Red Sea widening started about 5 million years ago. The Gulf of Aden sank, allowing the Bab al-Mandâb to be a strait connecting the Red Sea to the Indian Ocean. The level of the escarpments bordering the Red Sea rose again, pushing the mountains of Yemen and the ‘Asîr (just north of Yemen) up to their present height and increasing the downward tilt of the Arabian Peninsula from west to east (fig. 1.1). The central deep trench along the mid-line of the Red Sea started to split, and has continued to do so to the present day. The central Red Sea has widened by 150 kilometers in this last 5-million-year phase of sea-floor spreading.

2A fundamental change occurred in the climate of the Peninsula. The high rainfall over an extended period of more than 2 million years (the Rift period) changed the whole of the landscape. All the Yemen and ‘Asîr mountains were scoured by great rivers, and on their eastern slopes, rainfall draining east created the jawf-Hadhramawt water basin. During the late Pliocene and early Pleistocene, the climate was hot and humid; heavy forest cover blanketed much of Arabia and the fauna probably resembled that of Africa. In Yemen, the old crust was lifted up to 300 meters, and then covered by limestone. The Rift period in the Red Sea was accompanied by intensive volcanic activity producing layers of lava and basalt, some 1,500 to 2,000 meters thick. The western escarpment of the Arabian highlands, dominating the Red Sea, features impressive valleys with very steep flanks, peaks overlooking terraces, and scattered hamlets atop. The central highlands, at an altitude of 2,500 meters, alternate cultivated basins and chains with difficult high passes. The Jabal Nabî Shu’ayb peak, 3,760 meters high, dominates the narrow Sanaa plain (2,300 meters high). These highlands end in a steep western escarpment which dominates the valley of the Jawf and the Mâ’rib area; rainfall creates temporary wâdîs that flow eastwards, converging in the sandy depression of the Ramlat as-Saba’atayn.

1.1.1. Tihâma

3In Arabia, the Red Sea is bordered by the Tihâma plain. It is very narrow in the North, indeed non-existent in the Gulf of Aqaba. Then it widens gradually to 40 kilometers wide near Jizân and up to 60 kilometers in Yemen. The general geographic aspect is always the same. All along the great faults, the mountains fall down to the plain in steep cliffs. Wâdîs have cut deep and narrow gorges along the east-to-west faults, creating large alluvial fans when they arrive at the Tihâma coast. The ragged coastline shows indented bays some kilometers wide and some 30 to 50 meters deep. These bays, locally named “sharm” or “marsa”, have been used since Antiquity as harbours. Some are closed off now by coral reefs or by sand, as at Luhayya or Mokha.

  • 1 A salt dome is also called a diapir.

4Many of the coastal areas in the Red Sea are fractured by minor faults, part of the long collapse of the overall Red Sea faults. About 2 million years ago, parts of the Rift escarpment were lifted above sea level by columns of salt from deep underground, known as salt domes1. The Farasân Islands are an example of these phenomena. They lie some 40 kilometers off the Saudi Arabian coast opposite Jizân, scarcely rising 60 meters above sea level.

5The Red Sea coastline in Yemen shows a succession of sands, narrow mangroves and palm trees around fresh water (near Khawkha). But Tihâma has a dreadful climate. Summer is very hot (average temperature is 33° C, with 45° C peaks) and humidity about 75% to 85%. Winter is cool but short. Rainfall in Tihâma never exceeds 150 mm, 110 mm in Hudayda, 54 mm in Jeddah and 32 mm in Jizân.

6Hadhramawt extends from the Saba’atayn Desert east to Zufâr on the Indian Ocean, covering some 50,000 square kilometers. It consists mainly of Paleocene and Eocene limestone, some 900 meters thick, forming a tabular plateau locally known as the “Jawl”. In the west, the waters running down the Yemeni highlands have cut an impressive mouth flanked by the steep escarpments of al-Abr. Wâdî Hadhramawt flows eastwards collecting the temporary runoff from the tributaries; the most important ones, all running south to north, are Wâdî ‘Amd, Wâdî Daw’ân and Wâdî Bin ‘Ali. The flat wâdî beds heve been occupied by densely cultivated oases since ancient times. Downstream, Wâdî Hadhramawt shrinks to a narrow canyon, and under the name of Wâdî Masila, flows into the Indian Ocean.

7On the western side of the Red Sea, the Ethiopian highlands reach a maximum altitude of over 4,530 meters (Mt Ras Dashen in the Simien Mountains), and extensive high plateaus range between 2,000 meters (Axum: 2,150 meters) and 5,000 meters (fig. 1.2). The highlands are bisected by the Rift Valley and by the Blue Nile and the Takazze rivers. A steep escarpment, 2,000 meters high, separates these mountains from the Red Sea, behind Adulis, the ancient harbour. Tigray owes its initial formation to Tertiary earth-movements accompanied by volcanism. Granite (syenite in Axum) and basalt are the predominant rock, types. The general topography is for the most part the result of erosion, creating gorges and valleys with bare rock, sides, and rugged-in mountains which include mesas (or ambas) and volcanic plugs. Much of the soil removed by erosion has been carried away by Nile tributaries, but some has accumulated in highland basins.

1.2. The monsoon system

8The southern part of the Arabian Peninsula has a distinctive rain system, the monsoon. In summer, a cool and humid wind, oriented north to southeast, blows towards India (fig. 1.4). In the northern Indian Ocean, these winds blow over Yemen and Zufâr, and rains fall over their mountains. In the southern part of the Red Sea, winds blow southwards to the Gulf of Aden. In the Yemeni highlands, rains become strong in July-August, and the core of the “Green Yemen” receives more than 900 mm rainfall annually (with a maximum of ca. 1,000 mm a year near Ibb) (fig. 1.3).

9The situation in the south is reversed during winter. Gold winds from India blow from NE to SW along the southern coastline of Arabia; the same winds then enter the Red Sea where they blow northwards as far as Jeddah. In the northern part of the Red Sea, by contrast during winter the winds blow southeast. The climatic frontier between these systems, with monsoon rains in the south and moderate winter rainfall in the north, lies roughly along the Tropic of the Cancer. Ancient seafarers were well aware of this boundary; their ships had to await reversal of the winds, often mooring for some time.

  • 2 Ozer & Mahamoud 2013; Omondi & Awango 2013.

10On the western side of the Red Sea, highlands, dominate a wide coastal plain stretching over 1,500 kilometers long. The total annual rainfall in the Ethiopian highlands increases from north to south and varies from less than 200 mm in the northeastern plain to more than 700 mm in the southwestern lowlands. While the coastal plains are very dry, some areas on the eastern escarpment receive more than 1,000 mm. The Debub region, in central Eritrea, at an average elevation of 1,600-1,900 meters, receives 500-700 mm annual rainfall during the monsoons, and has an average temperature between 15° C and 21° C. Monsoons are strong on the eastern slopes from July to early September, inducing heavy floods down in the lowlands near the Red Sea. In the southern part of the Red Sea (Kamarân, Hanish and Dahlak Islands), the rain falls mainly during winter (in December) but scarcely at all during summer. In Aden yearly rainfall is estimated at 38 mm, and in Djibouti 44 mm2.

  • 3 Seleshi & Zanke 2004.

11The North of Tigray has a semi-arid climate and variable precipitation from 400 to 820 mm (fig. 1.2). By contrast, in its major part, highland Ethiopia, between 1,700 and 3,500 meters in height, has a climate characterized by significant day-to-night variation in temperature, but especially by the presence of two rainy seasons: a short rainy season in March-April (called belg) and a long season from June till October (keremt). There can be 1,000 mm of rain in the area covering Gondar (1,140 mm), Simien, Wällo, Showa and Wälläga, and more than 2,000 mm in Käfa3.

  • 4 Fattovich & Bard 2000: 13. “Aksum is located on the Tigrean plateau in northern Tigrây... at an av (...)
  • 5 Yule 2007: 14-20.

12So, when the Axumites left their city (2,200 meters altitude and 700 mm rainfall a year)4 or the Adigrat area (2,400 meters altitude) in the early 6th century AD to conquer Sanaa (2,300 meters altitude and 250 mm rainfall) and Zafâr (the capital of Himyar)5 (2,800 meters altitude and 800 mm rainfall), they found themselves in a similar geographical context. They easily adapted to the rough terrain and heavy rainfall and they continued to use terraces so as to maximize the arable surface and the retention of moisture.

1.3. Crops and forests

13The plants under cultivation in South Arabia included sorghum, wheat, and barley but the principal crops were fruits and vines. Date palms thrived in the irrigated zones.

14Prominent among the many species of trees in the desert regions is Zyzyphus spina Christi, a member of the Jujube family, which was used principally for construction but also for medical purposes. Locust trees are also found in the area bordering the desert, the most common species being Tortilis and Hamulosa, with their characteristic flat-topped silhouette. These species were used for the wooden frames commonly found in the buildings of Hadhramawt. The forest cover during the period near the end of occupation of the South Arabian cities was probably similar to that observed today; the trees were severely depleted and were not replaced by any new natural vegetation. The mature standing forests praised by the ancient writers have entirely vanished.

  • 6 Hepper & Wood 1979: 65-71.

15The deforestation was not limited to the area of the cities of the lowlands; the mountains of Yemen underwent the same process. The already-quoted species were frequent there, but also a great number of varieties such as Ficus, Juniperus, Dracaena, Olea Africana, etc., all the normal species which we find at the same altitude in the highlands of Ethiopia. The deforestation occurred for many reasons (construction, domestic uses etc.) so that Junipers do not form more than scattered small groves in Hujariyya, in Ta’izz and in the mountains of ’Asîr6.

  • 7 Fattovich & Bard 2000: 18-19.

16In Tigray, palynologists have investigated some aspects of the vegetation. The open vegetation pattern indicated by sedimentary evidence from the 2nd millennium BC at Maqala and Adigrat is also indicated at Aksum7. There is striking evidence of pollen from non-arboreal taxa. The vegetation of the hills around Aksum was dominated by shrubs and herbaceous plants characteristic of open vegetation and areas of human settlement. Trees were very numerous (Juniperus procera, olea and celtis) in the surrounding plains.

  • 8 Asfaw & Demissew 2009: 86-88.
  • 9 Strelcyn 1973: 197; Bulakh & Fiaccadori 2007: 1 30.

17Finally, one should mention incense trees such as Boswellia papyrifera. A number of recent papers have now established the territory of boswellia8. Botanists have located significant populations of papyrifera (Hochst) in Tigray, Gonder, Godjam and Showa, at heights between 950 m and 1,800 m, on steep rocky or volcanic slopes. The most common name in Amharic would be “mäqär”, idem in Tigrinya, and libânät in Harari9. As for other varieties of Boswellia (Boswellia microphylla Chiov. and Boswellia neglecta S. Moore), they are found in southeast Ethiopia, in Bale, Sidamo, Harerge and towards Somalia. The resin (in Geez: ’atan or labn, doubtless interchangeable terms), of a yellowish color, is well known for its medicinal virtues and for its perfume when burned.

1.4. Man and the landscape

18Yemen and ‘Asîr are considered as the most favoured regions of the Arabian Peninsula, and Yemen even more than ‘Asîr. Maximum rainfall occurs on the western side of the Yemeni mountains during summer. The summer storms precipitate flows of water down the slopes but nature distributes it unequally. The Red Sea front gets the major part, the one-meter annual rainfall which drowns the Himyarite regions of Yarîm. The rain has cut deep valleys with inaccessible ledges and jagged slopes and has carried down debris as far as the Tihâma plain.

1.4.1. Terraces (fig. 1.5)

19Wâdî courses, often following fault lines, form deeply incised narrow valleys in which alluvial sediments collect. The enclosed basins of the western highlands hold deep loess sediments that today are farmed. The limited wâdî sediments provide most of the potentially cultivable soils through the area, which terracing can further stabilize.

1.4.1.1. The “sawaqi” technique

20The so-called “sawaqi” technique consists in a total terracing of the slopes in order to retain the water. Slopes are carefully terraced; retaining walls can reach 5 meters high and the fields may be strips only 5 meters wide.

  • 10 Wilkinson & Edens 1999; Edens 2000.

21During the Bronze Age, these slopes may have served mainly as pasture lands. Rains are more than sufficient to water the narrow terraces, which turn into mud ponds. Since the third millennium BC, all the slopes in the Dhâmar, Yarîm and Sanaa basins have been carefully landscaped by terraces. The aim is to break up the steepest slopes, to prevent erosion, to facilitate infiltration, to retain the soils. At the downhill edge, a low wall retains the waters and the soil, with a narrow opening to evacuate overflow. The terraces form huge steps; their height may reach up to five meters or more and their width can be less than three meters. This traditional system explains the longevity of human occupation of these plateaus. The social organization in tribes (and clans) is rooted in the carefully distributed lands; the tribal system aims at maintaining law and order and thus at maintaining the terrace systems. We know from recent archaeological surveys near Dhâmar that barley, sorghum and cereals have been cultivated since 2300-2200 BC10. By the early 2nd millennium numerous towns had appeared in the intermediate zone between the dry eastern slopes and the wetter uplands to the southwest.

22The environmental history of Tigray is still largely unknown, as only very limited research has been conducted in this area. A major exception is the geo-archaeological investigation conducted by K. W. Butzer at Aksum in the early 1970s. In the 1990s fresh evidence was uncovered near Maqala and Adigrat by Ethiopian and Italian geomorphologists.

  • 11 Fattovich & Bard 2000.
  • 12 Butzer 1998 and 2000.

23According to Fattovich-Bard11, agricultural activity intensified in the Late Holocene, during the humid phase from the mid-1st millennium BC to the mid-1st millennium AD, to sustain a state-level urban society (Pre-Aksumite and Aksumite periods). Demographic pressure increased in the same period and reached a peak in the mid-first millennium AD. According to Butzer12, the first phase of aggradation, tentatively 100-350 AD, points to heavy rainfall with strong periodic floods, on a landscape already partially deforested and degraded because of intensive land-use in the early 1st millennium AD. A second phase of aggradation, tentatively dated 6th to 8th century AD, suggests a period of deep soil erosion due to very intensive land-use and very heavy periodic rains, combined with a progressive abandonment of settlements.

  • 13 French archaeological mission in Wakarida (O. Barge and E. Regagnon).

24Geo-archaeological research recently conducted in Wakarida13 (Eastern Tigray) yields a similar picpicture: intensive land-use in the water basin, and aggradation from the 6th century AD. Near Dongur, 5 km west of Aksum, a series of lines on the terrain could be interpreted as ancient field terraces, partly overlaid by the Dongur palace itself.

1.4.1.2. Mastering the floods

25Flooding down from the Highlands both in Yemen and in Ethiopia, the main wâdis encounter a very different landscape, only 100 to 150 kilometers from their source. This lowland environment is extremely dry, with annual rainfall of less than 100 millimeters; in exceptionally dry years, precipitation levels can he as low as 20 millimeters. The summers are extremely hot, with average temperatures of 31 to 32 degrees Celsius, and a recorded maximum of 43 degrees. Evaporation is therefore very rapid. The region’s sparse rain usually falls in July and August, but the main water source is the hinterlands of the Yemen or Tigray highlands. When rain does arrive, it comes in a torrent, carrying along earth and pebbles in its mighty surge down the stark limestone gorges and into the defiles (fig. 1.6). These flash floods come two or three times a year.

  • 14 Brunner 1997: 35-56

26The genius of the ancient South Arabian peasants lay in their ability to harness the power of these floods. Over the course of several millennia, they gradually began to use simple dikes to hold some of the water, perhaps originally experimenting in secondary wâdis where the volume of the floods was more manageable, no more than a few cubic meters per second. They become familiar with the areas where the slope was gentle and the water slowed, and where the terrain showed promise for irrigation, with meandering streams where water could be captured. They learned how to direct the water through wood or stone sluices towards collecting areas prepared in advance (fig. 1.7). Eventually they were able to direct most of the rainfall toward irrigation zones where the fine sediments suspended in the water would settle, providing excellent nourishment to the soil. The long process of experimentation probably began as early as the fourth millennium BC, and by the end of the third millennium BC14 the peasants had begun to tame more powerful torrents and thus irrigate a relatively wide area. Mastery of the wâdîs Dhana and Markha probably came sometime in the third millennium BC. An undertaking of this scale and sophistication required tight organization and precise management of the flood waters.

  • 15 Breton-Arramond-Coque-Delhuille & Gentelle 1998: 50, fig. 7a-7b: Flash-floods at al-Haraja, Wâdî B (...)
  • 16 Herberg 1986 and 1987; Gerlach 2012.

27The skill of the Sabaeans made it possible to irrigate the fields every year –sometimes more than once a year-without major destruction. The challenge was to channel a sudden gush of water into a large bed that remained dry for most of the year15. Only stonework could hold up the water’s power. Therefore they started with stone bulwarks to break the initial onset of the water, and below these there were long channels of stone (or earthwork) that directed the water toward a precise destination. In Mâ’rib, monumental hydraulic works have been discovered in the very river-bed of the Wâdî Dhana16. After flowing through these monumental structures, the water was then distributed through a series of earthwork canals, each measuring between five and eight meters wide. These fed into large stone sluices fitted with grooves. Farther downstream, narrower canals led into structures designed to distribute the water in different directions (fig. 1.8). The current had to be slowed somewhat as the water approached the fields, in order to limit erosion of the earthen levees, but had to travel fast enough to hold the fine sediments in suspension until the water reached the fields. Side trenches branched out of the central path of each channel into the borders of rectangular (often square) fields whose size was determined by the amount of water available to the area.

28The sediment carried down by the water enriched the soil but also posed a technical challenge, as great volumes of sand and silt accumulated in the fields over time. The rate of siltation is thought to have been approximately 0.7 centimeters per year, or 0.7 meters per century. This is a theoretical average, since floods occur with a certain randomness, and the flood waters were not dispersed equally over all fields. Over time, the entire area around Mâ’rib (fig. 1.9) rose by some thirty meters, while other areas suffered a rise of as much as ten meters at Shabwa (fig. 1.10) or at Tamna‘. The first and easiest solution to this long-term problem was to dredge the canals periodically, but this was only feasible for the first several hundred meters of their trajectory, and dumping the silt outside the canal banks had the unwanted side effect of reducing the area of the surrounding fields proportionately. Clearing the fields was a task of an entirely different order of magnitude, though such projects are thought to have been undertaken in more recent times in certain areas around Shabwa.

1.4.2. “Hydraulic societies”

  • 17 Breton 2013: 39-40.

29An oasis like Mâ’rib or Tamna‘ must have supported thousands of peasant farmers. But agricultural productivity depended on a form of collective organization17. Farmers must have come to an arrangement whereby water was distributed equally, with the fields closest to the breakwaters receiving roughly the same amount of water as those at the farthest ends of the canals. Such an arrangement implies a coherent system of land ownership and a pact of mutual agreement between the various tribes. The situation was unstable over the long term, since more influential tribes would have enlarged their domains or taken possession of the fields closest to the canal heads, which “drank” the most water. The violence of modern-day rivalries over land suggests, if imprecisely, similar ancient conflicts.

30The orderly functioning of an oasis like Mâ’rib or Tamna‘ is thus the clearest indicator we have of the existence of social union and community in ancient times. Sadly, the details of these communities remain blurry. We can suppose that the Mâ’rib farmers chose a “water-master” who was responsible for directing the operation of sharing out the floodwaters, overseeing the distribution of allotments by volume or by time.

  • 18 Breton 1998: 18.

31On the western coast of the Red Sea, due to the specific geographical conditions of the “Afar” depression, ancient irrigation by “flash floods” was mainly impossible. North of Massawa, the seasonal rivers originate from the highlands of the eastern escarpment with wâdîs such as wâdî Laba, May Ule and Wekiro. The first two wâdîs are the source of spate irrigation in Sheeb and the third in Wekiro18, but their history has not been determined.

32The only antique irrigation system that we know of is to be seen at Adulis. Unfortunately, the archaeologists who worked at Adulis established no link between the city and its surrounding fields, and thus were hardly capable of evoking the problem of the supply of the city. Undoubtedly, the population of Adulis lived on the exploitation of its irrigated territory, like many ancient cities of the Red Sea, and not on imported luxury foodstuffs.

1.4.2.1. Adulis irrigation system (fig. 1.11)

33Three main wâdîs (wâdî Aligeddeh, wâdî Kumheyle and wâdî Haddas) constitute a main river system that drains water during the heavy monsoon rains which fall during the summer. They join approximately around the village of Hedele, and concentrate the floods south of the modern village of Foro. The water flow went through a gorge (now covered by a new dam) only about 50 meters wide and then fanned out into several channels. These water channels, cut in the basalt, are still visible on both sides of the river-bed. To the north, there is a channel, 6 to 7 meters wide and some 4 meters deep, which is visible for 200 meters (minimum). It opens on an oval water sluice, and then proceeds northeast towards the village of Afta. From there, a secondary water channel leads to Afta, while the main channel goes southeast. To the south, there is a water gate opening on a rock-cut channel, 3 meters wide and visible for 250 meters, dominating the wâdî bed. Modern water sluices and concrete basins were built in the 1950s. Today a new dam diverts the water some 1, 100 m south, and this is where the flow now runs down east to the sea. This new river bed has destroyed part of the ancient irrigation system.

1.4.2.2. Reconstruction of the antique irrigation system of Adulis19

  • 19 Breton 1998: 32-33.

34Although a large part of the ancient irrigated fields is now covered by sand dunes, it is possible to reconstruct the irrigation system by combining both aerial photographs and field observations.

  1. In the northern oasis, the main channel reaches the city of Adulis and then divides itself into secondary channels running to the east. Thus Adulis was situated in the middle of its ancient fields. The summit of the tell (+25 m) dominated the ancient fields by 5 to 6 meters.
  2. In the southern oasis, the main channel and secondary channels are easily recognizable on aerial photographs.
  3. The central valley: the meanderings of the wâdî are due to the very flat bottom of the valley of Haddas. It is easy to reconstruct at least two of its phases, an ancient one situated in the north, and another (more recent) in the south. Between the two rises a circular rocky eminence, about 1 km in diameter, on which is the modern village of Zula. It is clear that no antique site ever developed there, by reason of the danger of floods.

35It may be calculated approximately that the northern oasis covered 10 km2 (or 1,000 ha) and the southern one 5.5 to 6 km2 (550/600 ha), and so the “double oasis” of Adulis would have extended over approximately 1,500-1,600 ha.

1.4.2.3. Adulis city

  • 20 Mehari & Ghebru 2005: 89-97.
  • 21 Breton in press.

36According to recent investigations (in 2007 and 2008), the city of Adulis covered 38 ha (some 400 m by 600 meters) culminating at a height of + 25 m20. There were no fortifications or city walls. Some buildings were excavated, as early as 1868 (by William West Fellow), in 1907 (by Sund-ström), in 1908-1910 (by Paribeni21) and in 1960 (by F. Anfray), but no complete town plan has ever been reconstructed. The deep Paribeni trench reached the presumed base of the tell, some 10 meters below the city summit, thus approximately +15 m high. In this case, the height of the silt would probably have reached 5-6 meters.

37Numerous instructive parallels can be drawn with South Arabia:

  1. A double oasis on both sides of a river is a well-known phenomenon in South Arabia, in Mâ’rib and in Shabwa for instance.
  2. A double water catchment, cut into the rock and leading to two main water channels, on both sides of the river bed, is also documented in South Arabia (Mâ’rib).
  3. The main city in the oasis is located in the midst of its fields (Mâ’rib, Shabwa, Barâqish).
  4. An oasis extending for 1,600 ha (Adulis) could be compared with Barâqish (1,550 ha) and Raybûn (1,300-1,400 ha), but is still far from Mâ’rib (about 9,600 ha).
  • 22 Schmidt & Matthews 2008: 35; Peacock & Blue 2007: 79-108.

38The main unsolved question is the dating of the Adulis irrigation system. Previous excavations at Adulis yielded some fragments of marble bases of thrones (?) dated mid-first millennium BC (?)22. Pre-Axumite pottery has been found in the southwest sector of Adulis, but it does not seem earlier than the 1st cent. AD. In his deep trench, Paribeni found “Ptolemaic pottery” (jars, amphorae, oil lamps) which he dated to the 3rd cent. BC. Late pottery consists mainly of “Late Roman amphorae 1 and 2” generally dated to the early 5th-6th cent. AD. Should it therefore be assumed that the Adulis irrigation system was in use between the 3rd cent. BC and the 6th cent. AD? Based on the height of silt accumulation (5 to 6 meters), this is only a hypothesis. Did the irrigation system even exist before the 3rd cent. BC? No data is yet available.

Notes

1 A salt dome is also called a diapir.

2 Ozer & Mahamoud 2013; Omondi & Awango 2013.

3 Seleshi & Zanke 2004.

4 Fattovich & Bard 2000: 13. “Aksum is located on the Tigrean plateau in northern Tigrây... at an average elevation of 2,200 m... The aerial photo-mosaic and satellite image clearly show that some hills, with an average elevation ranging between 2,406 m and 2,289 m, surround Bêta Giyorgis (2,240 m in elevation and May Qoho (2,235 m in elevation).”

5 Yule 2007: 14-20.

6 Hepper & Wood 1979: 65-71.

7 Fattovich & Bard 2000: 18-19.

8 Asfaw & Demissew 2009: 86-88.

9 Strelcyn 1973: 197; Bulakh & Fiaccadori 2007: 1 30.

10 Wilkinson & Edens 1999; Edens 2000.

11 Fattovich & Bard 2000.

12 Butzer 1998 and 2000.

13 French archaeological mission in Wakarida (O. Barge and E. Regagnon).

14 Brunner 1997: 35-56

15 Breton-Arramond-Coque-Delhuille & Gentelle 1998: 50, fig. 7a-7b: Flash-floods at al-Haraja, Wâdî Bayhân, in 1990 and 1991.

16 Herberg 1986 and 1987; Gerlach 2012.

17 Breton 2013: 39-40.

18 Breton 1998: 18.

19 Breton 1998: 32-33.

20 Mehari & Ghebru 2005: 89-97.

21 Breton in press.

22 Schmidt & Matthews 2008: 35; Peacock & Blue 2007: 79-108.

© Centre français des études éthiopiennes, 2015

Conditions d’utilisation : http://www.openedition.org/6540