Save

In Sudan’s Eastern Borderland: Frontier Societies of the Qwara Region (ca. ad 600-1850)

In: Journal of African Archaeology
View More View Less
  • 1 Incipit-CSIC, Avenida de Vigo s.n., 15705, Santiago de Compostela, Spain
  • | 2 Independent Archaeologist
Full Access

Abstract

The Sudanese-Ethiopian borderland has seen interaction between state and non-state peoples for at least two millennia. However, little is known about these interactions from an archaeological point of view. Our research project intends to cast light on this topic by looking at the lowlands of nw Ethiopia. Surveys conducted during three field seasons in the Metema and Qwara regions – in the Atbara-Dinder watershed – have allowed us to document different cultural traditions that are related to Sudan in medieval and post-medieval times. Here, we present the data and discuss the relevance of the findings to understand border dynamics from the mid-first millennium ad onwards.

Abstract

The Sudanese-Ethiopian borderland has seen interaction between state and non-state peoples for at least two millennia. However, little is known about these interactions from an archaeological point of view. Our research project intends to cast light on this topic by looking at the lowlands of nw Ethiopia. Surveys conducted during three field seasons in the Metema and Qwara regions – in the Atbara-Dinder watershed – have allowed us to document different cultural traditions that are related to Sudan in medieval and post-medieval times. Here, we present the data and discuss the relevance of the findings to understand border dynamics from the mid-first millennium ad onwards.

Introduction: Frontiers, Peripheries, Margins

The borderland between Sudan and Ethiopia is scarcely known archaeologically, a situation that contrasts with the northern regions of both countries. Compared to the core areas of state formation (Edwards 2004; Phillipson 2012), studying the tribal zone between them seems a dull undertaking a priori, since they lack their rich material culture and complex social developments. Yet there are at least three good reasons to investigate the borderland. First, the very persistence of small-scale stateless communities from deep Prehistory to the present is a phenomenon that deserves close scrutiny, as it is proof of a striking capacity to resist dominant societies (Fernández 2003; González-Ruibal 2014) and of a complex scenario of interactions (Jedrej 2006; Brass 2015). The concept of stateless societies, however, requires some clarification. Some of the groups of the Sudanese-Ethiopian borderland have had traditionally an essentially egalitarian political organization (among males): this is the case for most slash-and-burn agriculturalists. But not for pastoralists, who do not fit neatly into clear-cut sociopolitical categories, as Brass (2015: 258) reminds us. We also have to bear in mind that stateless societies are not necessarily alien to the idea of the State or radically isolated from it. Jedrej (2006: 208) has argued, referring to the Hadjeray and Nuba, that “instead of a duality of states and societies, social relationships of alliance and hostility were realized as a hierarchical network of ranked positions. In other words, diverse relationships of power and domination penetrated from the elite in the centre out to the peripheries, creating one political field”. This political field of interaction could well be labeled “borderland”. Furthermore, neither the absence of state or egalitarianism can be equated with lack of complexity. The term “horizontal complexity” better describes the nature of many non-state societies (McIntosh 1999: 9). This complexity can be expressed through ritual positions, corporate institutions that curtail power, intricate horizontal forms of integration and interdependence, etc. (McIntosh 1999).

figure 1
figure 1

A. State and borderland in Sudan-Ethiopia. From the late seventeenth century onwards the extension of the tribal zone has undergone only limited territorial modifications. B. Sudanese-Ethiopian borderland with place names cited in the text and published sites of the Post-Meroitic to Funj times.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

Another reason that makes peripheral regions worth studying is the role played by borderlands in the politics and economy of the different states of the Horn of Africa (Pankhurst 1997; Brass 2015). This role, however, has been hardly investigated, in part due to the absence of written sources. This has led to a crude characterization of peripheries as mere extractive zones, providers of prestige goods for the core regions. Finally, the relative closeness of the borderland communities to the state favored appropriations and hybridizations that are very interesting for understanding processes of cultural contact more generally (Lightfoot & Martínez 1995; Ylimaunu et al. 2014). A good example of these processes of appropriation is what anthropologist Wendy James (1977) has termed “Funj mystique”. It refers to fragments of traditions, genealogies and myths that can be associated to the Funj Sultanate of Sennar (1505-1821) and that have been present among the stateless groups of the borderland until very recently.

Between 2013 and 2016, we conducted three archaeological field seasons in the lowlands of Metema and Qwara (nw Ethiopia) to address the processes of interaction between the borderland communities and the neighboring states during the last two millennia. The region, which is located between two major tributaries of the Nile – the Atbara and the Dinder – has a very low population density and the same seems to have been the case in the past (Fig. 1). The Atbara valley, in particular, proved to be devoid of people between the time of the last hunter-gatherers and the 19th century. Sites are much more abundant from the Rahad valley to the south. Overall, our survey produced 69 archaeological and historic sites, spanning from the msa to the Second World War, of which 40 sites of different types can be dated between the mid-first millennium ad and the late 19th century. A total of 23-25 settlement sites appeared in the lowlands of Qwara, depending on whether we consider some artifact scatters as single sites or part of large sites, plus two more in Metema, (Fig. 2A and B). These will be used in the present article to outline a cultural sequence for the region during the last millennium and a half.

We will employ three key concepts. First, we consciously use the term “frontier” to refer to the eastern Sudanese borderland. As Baudel & Van Schendel (1997: 213) note, “frontier commonly refers to the territorial expansion of nations or civilizations into ‘empty’ areas”. Sudan’s eastern borderland, as we will see, is a thinly populated territory (though not empty) that has been perceived as an open space for expansion and exploitation. As such, it is a land of migration as well and we believe that many of the communities that we have documented archaeologically are groups of migrants – either permanent or seasonal. This is not surprising: all the present inhabitants of the lowlands of Qwara have arrived from elsewhere during the last 70 years or so, as we could gather from interviews with members of the different ethnic groups now residing in the region (Gumuz, Dats’in, Amhara and Tigreans). Apart from expansion and migration, frontiers are also related to three other important phenomena. Conflict between state and non-state peoples and amid tribal groups is perhaps one of the most generalized outcomes of frontier situations (Ferguson & Whitehead 1992). At the same time, frontiers can also be the space for middle-ground cultural encounters, that is, forms of contact between states, empires and other ranked societies, on the one hand, and less hierarchical groups, on the other, that develop in relatively equal terms (Gosden 2004: 30-32). Yet middle ground encounters do not imply necessarily a situation of perfect symmetry. There is usually a more powerful actor and a less powerful one (this is the case, for instance, with the relationship between the Funj sultanate and its periphery), but there is more balance than in colonial or predatory situations. Symbiotic relations, which are typical of ecological border zones, can also be characterized as non-symmetrical middle ground encounters. This is the kind of relations that develops between mobile pastoralists and agriculturalists or sedentary communities and hunter-gatherers. A third phenomenon that is related to frontiers is ethnogenesis (Chappell 1993) – but also its opposite, ethnocide (Hall 2000: 241). The latter has been more characteristic of the last two hundred years, whereas ethnogenesis has undoubtedly existed since prehistoric times. Migrations, conflict, middle-ground encounters and ethnogenesis are all characteristic of African frontier societies (Kopytoff 1987).

figure 2a
figure 2a

Map of the surveyed area in the Qwara lowlands and the southwestern corner of Metema. 1. Jebel Halawa (Funj); 2. Selferedi-Kuter 4 (Gelegu Tradition-Funj); 3. Mirt Rahad (Funj); 4. Gumuz Badimma (Medieval-Funj); 5. Dera Hassan (Funj). Only the sites discussed in the present article have been included in the map.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

figure 2b
figure 2b

Detail map of the area surveyed in the middle Gelegu valley. 1. Mirt Gelegu; 2. Zoha-Funj; 3. Bermil-Gelegu; 4. Zoha-Gelegu; 5. Bermil-Funj; 6. Abba Sheña; 7. Tach Gerara; 8. Gerara 1-3; 9. Mahal Gerara 1-4; 10. Tebeldiya 1 and 2; 11. Dengersha; 12. Amidla; 13. Jebel Mahadid-2; 14. Jebel Mahadid 1; 15. Shimeligir. Note that numbers 8 and 9 in Figure 2B include more than one site each.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

The other couple of concepts that need clarification are “periphery” and “margin”. Note that we are not using these concepts in a derogatory way, taking for granted that cores are creative and dynamic and peripheral regions passive recipients of others’ inventions, as African cultures have been often perceived in archaeology and other disciplines (Wynne-Jones & Fleisher 2015). In fact, our intention is precisely the opposite: to reclaim the relevance of African peripheries and margins, not only for understanding long-term social processes in the continent, but elsewhere as well.

For periphery we understand an area that lies on the outskirts of a state but is an integral part of it; that is, it partakes of the political, economic and cultural practices that develop at the core of the state. Peripheries often lie within the official political boundaries of a polity, but usually retain a large degree of autonomy. Margins, instead, are beyond the state, both geographically and politically, and they are perceived as remote and alien also in the cultural imagination of state peoples. They are not part of the state territory, either de facto or de jure, although (rhetorical) claims on margins are sometimes made by expansive polities. Political, cultural and economic influences from the core arrive indirectly and often as faint echoes. However, margins can play an important economic role, as suppliers of strategic goods that are at the base of the political economy of the state. In our case, these include gold, ivory, ostrich feathers, incense and slaves. Furthermore, margins often reflect, in a sort of domino effect, major events occurring at the state core, such as the collapse of a regime, political fragmentation and processes of empire- and state-building. From this point of view, the history of the Sudanese-Ethiopian borderland can be described as the transformation of a margin into a periphery, a process that has not yet come to an end (Markakis 2011).

In the Margins of Medieval Sudan (ad 600-1300): The Gelegu Tradition

The first occupation by pottery-making, food producing communities that has been documented in the nw Ethiopian lowlands so far occurs during the second half of the first millennium ad. At this time, a series of small settlements appear between the Rahad and Dinder rivers, whose alluvial plains are some of the main natural pathways connecting the Sudanese Gezira and the Ethiopian foothills. Seven sites belonging to this tradition have been found, six of them in the Gelegu valley, a tributary of the Dinder, thus the name of the tradition (Fig. 2B).

Characteristics

Some common elements can be pointed out in relation to the landscape and the organization of space (Fig. 3). In all cases, the sites are not located more than 1.5 km away from the main river (Gelegu or Dubaba). The surface occupied by the settlements is always small, ranging from as little as 1000 m2 (Selferedi-Kuter 4) to a maximum of 5000 m2 (Bermil, Amidla). The smaller sites (1000-3000 m2) are located on the vertisols of the alluvial plain of the Gelegu river (Selferedi-Kuter 4, Zoha, Mahal Gerara). The larger settlements (4000-5000 m2) lie on the base of prominent hills (Bermil, Amidla, Jebel Mahadid). This can be related to the different functions of the sites, the larger ones being perhaps more permanent or at least reoccupied more frequently. The small surface of the sites suggests that they were used by small groups of a few nuclear families, which is consistent with seminomadic communities in the region. Thus, the pastoralist Rufa’a transhumant camps consist of 10 to 15 households, whereas the Fulani herding unit comprises 4 to 8 households (Ahmed, A.G.M. 1973: 49, 52). Each household is made up of a nuclear family (husband, wife or wives, small children and unmarried daughters). Therefore, we can calculate an average camp population of 25 to 50 people among the Fulani and double that figure for the Rufa’a. The Fulani probably provide a better analogy for our sites, considering their size.

figure 3
figure 3

Settlement pattern in the Bermil area: the large site is on the slope of the hill and a smaller site is located on the alluvial plain. Both Funj sites, instead, are on the alluvial plain.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

The density of surface finds is very high in all cases and this evinces an equally dense occupation of the space. The sites are well delimited by artifact scatters and no secondary concentrations have been found that may be indicative of dispersed compounds. No structures were identified, either on the surface or in the test pits that were carried out in five of the sites (Selferedi-Kuter 4, Amidla, Mahal Gerara, Jebel Mahadid, Bermil). The sites on hills (jibal), which are sandstone formations, have been heavily abraded by the action of the rains and no original levels were preserved: in most sondages, either bedrock or saprolite appeared only a few centimeters under the surface. The alluvial sites presented a better panorama.

We conducted five test pits in Mahal Gerara (four of 1×1 m and one of 2×1) and three in Selferedi-Kuter 4 (two of 1×1 m and one of 1.5×3 m). Both sites are located in-between the slope of a bar of silt and the vertisol of the alluvial plain. Unfortunately, Mahal Gerara was completely disturbed by plowing and the unstable behavior of the vertisol. As a result, the occupation level was all mixed up and the charcoal that we submitted for radiocarbon dating furnished a contemporary date. In Kuter 4, we documented three levels with archaeological materials in Sondage 3, along a stratigraphic sequence of 50 cm (Fig. 4). These levels were identified as distinct stratigraphic units from slight changes in soil characterization (e.g. black spongy silty soil SU03, less dark, grittier, more compact SU02) and the distribution of archaeological materials. However, since the limits of stratigraphic units were not so easily recognizable during the process of excavation, due to the nature of the vertisol, we also subdivided the layers in artificial spits of 10 cm, in order to avoid mixing up artifacts from different layers. The most recent level is of the Funj period (see below) and the two others pertain to the Gelegu Tradition. The occupation level proper lies at the interface between the two lower layers (Stratigraphic Unit 02 and 03), at a depth of 30 cm, where bones and sherds appear flat in situ and in larger quantities.

figure 4
figure 4

Stratigraphic profile of Sondage 3 in Selferedi-Kuter 4.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

The material assemblage documented is limited, mainly pottery (Fig. 5). This shows strong resemblances across the seven sites, although there are some minor differences (the parallels for the material will be discussed in the next section):

  • The pastes are of well levigated clay, homogeneous composition and with fine to very fine mineral (quartz) temper (less than 1 mm). Orange color predominates, although the frequency of orange vs. dark-colored sherds varied from some sites to others.

  • Surfaces are polished and often burnished.

  • The ceramic assemblage is massively composed of open forms – basins, bowls and dishes (90%).

  • Rims are often decorated with finger or nail impressions (14 to 33% of the assemblage depending on the site).

  • Pots are of medium size, with 60% of the vessels having a diameter comprised between the 20-29 cm range and only 15% of 30 or more cm.

  • The walls are relatively thin, with 60% of the pots having a wall thickness comprised between 0.6 and 1 cm.

Apart from pottery, few other artifacts were retrieved: a few grinding stones, which do not appear in all sites (they are absent in Bermil and Amidla) and are rare in the rest; a stone weight from a digging stick (Zoha); spindle whorls (Amidla, Bermil, Mahal Gerara, and a stray find), and a few faunal bones and river shells (Kuter 4, Mahal Gerara). We will discuss the spindle whorls in the next section, since they are relevant for providing a chronology. The faunal remains that could be identified belong to cattle. No lithic artifacts (apart from the few mortars and the digging stick weight) appear unambiguously associated to these sites, thus indicating that they were occupied by fully iron-using peoples. The digging stick weight is not informative from a chronological point of view, as similar pieces are known from the Mesolithic to the present (Gascon 1977). In this context, however, it can bear some cultural significance, because stone weights are uncommon in Western Ethiopia, unlike in Eastern Sudan (e.g. Fernández et al. 2003: 252, fig. 41).

Chronology

The Gelegu Tradition is dated by two radiocarbon samples obtained from the only site that provided organic material in context: Selferedi-Kuter 4. Some artifacts, however, have parallels in the Sudan that contribute towards defining the chronology for this tradition. As for the radiocarbon dates, all three occupation levels of Kuter 4 provided charcoal samples that could be successfully dated. The Stratigraphic Unit 01, as has been pointed out, furnished a Funj chronology consistent with the material and will be dealt with later. The two other levels (Stratigraphic Unit 02 and 03) yielded virtually the same date, thus showing that the site had a short occupation. SU02 was dated to 788± 22 bp (cal. ad 1217-1273) and SU03 to 728±31 bp (cal. ad 1223-1299).1 The site was probably occupied around the mid-13th century ad. It is likely that this actually marks the final phase of the Gelegu Tradition, as we will see. Some materials tally well with a medieval chronology. Thus, sherds have been found both in Kuter 4 and in Jebel Mahadid pertaining to bowls with dark polished surface decorated with incised lozenges (Fig. 6, n. 3-5). These have very good parallels in the Christian period of neighboring Sudan: identical forms and decorations have been found in Christian sites in the Gezira, including Soba, the capital of the Kingdom of Alodia, and the church of Saqadi (Crawford & Addison 1951: lxv, 27, 28; Welsby & Daniels 1991: fig. 133; Fernández et al. 2003: fig. 17). Bands of incised crosshatching, as documented in Kuter 4 and Jebel Mahadid, are also very typical of the Christian period (Welsby & Daniels 1991: fig. 125-126).

figure 5
figure 5

Pottery of the Gelegu Tradition. 1-2: Globular bowls with incurved rim; 3-8: Deep bowls; 9-12: Shallow bowls; 13-14: Basins. Provenance: 1-4, 12: Bermil: 5, 9, 10: Amidla; 7-8: Zoha; 6, 13: Selferedi-Kuter 4; 14: Mahal Gerara.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

The beginning of the tradition is more difficult to date, because we were unable to find organic material from an undisturbed archaeological level in any of the four other sites belonging to this tradition where we conducted test pits. Therefore, we have to rely on archaeological parallels. A Post-Meroitic date is suggested by the bowls with stylus-impressed lozenges or quadrangles found in Mahal Gerara (Fig. 6, n.1 and 2). This kind of decoration is well documented in the Sennar area for the Post-Meroitic period (ca. 350-600 ad). We have parallels in Sennar itself (Arkell 1934: 110; Addison 1935: 290, pl. vi, 14) and Abu Geili (Crawford & Addison 1951: pl. xxxviiib). Post-Meroitic and early medieval parallels also exist for the bowls with rims decorated with incised crosses or zigzags (Fig. 6, n. 10-11) (Arkell 1934: 106; Welsby & Daniels 1991: fig. 131, n. 217; Edwards 1998: fig. 6.27, 6801; fig. 6.28, 058/1; Sidebotham et al. 2008: 126, fig. 21, n.2; Abd el-Rahman 2007: fig. 2, grs.008).

figure 6
figure 6

Sudanese style decorations: 1: Stylus-incised lozenges infilled with stylus-incised cross-hatching (Mahal Gerara); 2. Stylus-incised lozenges infilled with stylus or spatula impressions (Mahal Gerara); 3-5: Incised lozenges (Kuter 4: 3-4; Jebel Mahadid: 5); 6-7: Roulette (Amidla: 6; Zoha: 7); 8: Incised zigzag (Kuter 4, Mahal Gerara); 9, 12: Crosshatched band (Kuter 4: 9; Jebel Mahadid: 12); 10-11: Incised crosses on rim (Mahal Gerara); 13-16: Punched and gouged (Bermil).

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

Other interesting parallels can be found in seven sherds from Bermil decorated with stylus punctuates (Fig. 6, n. 13-16). This decoration, described as “gouged” and “punched” by Balfour Paul (1952: fig. 6, 7-11), was documented in Umm Sunt, north of Sennar. The site, or sites, included settlements and cemeteries dated between the Post-Meroitic and Christian periods (Edwards 1991); as well as a Funj site documented later (Fernández et al. 2003). The stylus punctuates appear in the few sherds belonging to globular pots with incurved or straight rims. The decoration was made with a gouge-shaped stylus in five cases and by punching in the other two. The gouge-decorated sherds, in particular, are remarkably similar to the Umm Sunt ones.

All these Post-Meroitic/Christian elements represent a small fraction of all the material recorded during survey and test-pitting. Some might well have been imported from the Sudan. There is, however, a type of vessel that is abundant in the area and locally made, but has perfect parallels in the neighboring country: the large bowls with finger-printed decoration along the tapered rim (Fig. 7), which represent between 14% (Amidla) and 33% (Bermil) of all rims documented in the sites belonging to the Gelegu Tradition. In fact, they can be considered the type fossil for this tradition. Identical bowls and basins with finger-printed rims have been found in Soba where they are called Type J (Welsby 1998: 92-93, figs. 40 and 41). They are common in levels dated from the 6th to 9th centuries ad (Welsby 1998: 21) and they appear associated to perishable structures (post holes and pits) belonging to round and rectangular huts and fences. Some finger-printed rims have also been retrieved in the Post-Meroitic necropolis of Gheresli, south of Sennar (Abd el-Rahman 2007: fig. 2, grs.009; fig. 6). The Post-Meroitic burials of Gabati, which have yielded bowls typologically similar to both ours and Soba’s, is radiocarbon-dated between the 5th and 7th century ad (Edwards 1998: 206).

figure 7
figure 7

A sample of finger-impressed rims from Bermil.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

More generally, the pots that make the bulk of the ceramic record (basins, bowls, and open forms more generally) coincide with the prevalent forms in the typology of Post-Meroitic and early Christian sites (ca. ad 350-1000) known along the Blue Nile (Abu Geili, Saqadi, Soba) and even further north (Edwards 1998: 178-193; Adams 1986: 416-423) and are indicative of similar cooking methods and diet (Fig. 8). The prevalence of bowls and plates indicates that baking, rather than boiling, stewing and brewing, predominated as a form of cooking.

A chronology around the mid-first millennium ad for the beginning of the Gelegu tradition is also suggested by four spindle whorls (Fig. 9), two of which at least have identical Post-Meroitic parallels (Crawford & Addison 1951: lviii, 26 and 27). Frustoconical and ellipsoid spindle whorls appear during the Meroitic period (Näser 2004: 255-257), become extremely popular during Post-Meroitic times (Crawford & Addison 1951: 102-104; McCann, E. 2004: 93; Wolf & Novotnick 2005: 25), are rarer during the early medieval period (Welsby & Daniels 1991: 149-150; Budka 2007: 67-69) and disappear during the Islamic era, when they are substituted by flat whorls and reused sherds. The spread of this item is probably related to the growth of cotton during the late Meroitic period (Fuller 2013: 172-173). The introduction of saqia irrigation around those dates in northern Sudan (Ahmed 1999: 305; Edwards 2004: 202-203) is associated to greatly expanded cotton fields and by the mid-fourth century the fiber was a key element in the Meroitic economy (Fuller 2013). The appearance of spindle whorls in the Ethiopian foothills is probably linked to this expansion of the cultigen during the mid-first millennium.

figure 8
figure 8

Comparative typology of the Gelegu Tradition pottery with Sudanese sites: Abu Geili (Post-Meroitic-Early Medieval?); Saqadi (Early Medieval); Soba (Early Medieval).

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

figure 9
figure 9

Spindle whorls. 1: Bermil; 2: Amidla; 3: Gerara 2; 4: Mahal Gerara.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340011

Discussion

The best typological parallels for the Gelegu Tradition place it in the Post-Meroitic and Early Christian period, roughly between ad 600 and 1000. However, our two radiocarbon dates from a single site lie beyond this range (ca. ad 1250). We would argue that the tradition does start during the mid-first millennium and stretches up to classic Christian times.

Who were the people of the Gelegu Tradition? The material assemblage shows strong resemblances with the Sudanese material culture known from the sites of the Blue Nile. At the same time, it is not an exact replica of that material assemblage. We hypothesize that the people of the Gelegu Tradition were part of small communities moving along the main watercourses from the southern Gezira into the foothills of the Ethiopian plateau. The original territory of these communities probably lay between the Blue Nile and the Dinder, where several Post-Meroitic and Christian sites have been reported (Chataway 1930; Dixon 1963: 234; Ahmed & Ahmed 2000; Bashir et al. 2012). No assemblages have been fully published, however, to make systematic comparisons with ours, although some available materials show remarkable similarities (Bashir et al. 2012).

It is likely that the people of the Gelegu Tradition were semi-sedentary pastoralists moving seasonally with their herds, probably at the beginning of the dry season, from eastern Sudan towards the greener pastures of the Ethiopian foothills: Metema receives an average rainfall of over 900 mm, compared to just 200 mm in Sennar. They would probably return to the Sudan before the start of the rains in May, so as to avoid tripanosomiasis and other diseases that are lethal for cattle.

Proof of these seasonal movements could be the site excavated by Bashir et al. (2012), Azazo ROSE 5, near Roseires, where seasonality is inferred from the existence of multiple levels of occupation, marked by layers of ash and hearths (Bashir et al. 2012: 133). The phenomenon of transhumance between Sudan and Ethiopia is well attested today in the area (Osman 2009; Schlee 2012). Pastoralists from a variety of ethnic groups (Fulbe, Masalit, Daju, Beni Amer) return to the same places every year in the Qwara and Metema lowlands during the dry season and stay for three or four months (around February-May). We only have evidence for herding from the site of Kuter 4, where several bones belonging to bovines were preserved and were big enough for identification. Transhumance today involves primarily cattle, because goats and sheep are much less demanding in terms of diet.

Seasonal mobility could perhaps explain the strange ceramic repertoire of the Gelegu Tradition, which is extremely restricted, with the large majority of vessels belonging mainly to one category (open bowls) with different sizes (dishes, bowls, basins). The striking scarcity of grinding stones and mortars, which are also small, stands in contrast with their overabundance and large size in sites of later chronologies. This could perhaps be understood in relation to the semisedentary or seminomadic character of the Gelegu people. Another point that would corroborate this hypothesis is the very small surface covered by the settlements, of which five out of seven have 3,000 square meters or less.

The question is why the herders of the southern Gezira decided to move into the Ethiopian foothills precisely when they did. We have to look at events in the Sennar area in the Meroitic and Post-Meroitic periods to understand the process. At some point during the first half of the first millennium ad Sennar became a periphery of the Meroitic Kingdom, with an outpost in or near the present town (Arkell 1934; Addison 1935; Dixon 1963, Edwards 1991; Ahmed 1999: 293; Brass 2015: 276). The reason for its creation was undoubtedly the exploitation of the resources along the Blue Nile and the Ethiopian foothills: gold, ivory, slaves and incense (Brass 2015: 276, 282).

Recent surveys and excavations have shown that the Meroitic influence extended further towards Ethiopia than originally thought, along the Blue Nile and Dinder valleys (Eisa & Welsby 1996; Ahmed & Ahmed 2004; and see Fig. 1B): Post-Meroitic and Christian sites have been found as far south as Fazogli and Galegu, almost on the Ethiopian border (Ahmed & Ahmed 2004: 187; Bashir et al. 2012: fig. 1). With the collapse of the Meroitic Kingdom, the Sennar periphery probably became autonomous (Brass 2015). The relative abundance of Post-Meroitic sites documented around Sennar (Abd el‐Rahman 2007) confirms the growth of the former dependency of Meroe. This might have led to overpopulation and the displacement of some groups along the alluvial plains of the Dinder, Rahad and Blue Nile towards the Ethiopian foothills, which were hitherto only visited by hunter-gatherers, as far as we can tell from the many sites with microlithic industries found along the watercourses and the total absence of pottery-yielding sites dated before the first millennium ad (González-Ruibal et al. 2015). The movement of people was probably motivated also by the increasing aridification of Sudan during the Post-Meroitic period, which would have led farmers to encroach into pasture lands and forced the displacement of pastoralists (Fuller 2013: 214). Droughts along the Nile valley are also recorded several times between the 9th and 12th centuries (Edwards 2004: 212-214). In sum, it is possible that the arrival of peoples into the Ethiopian lowlands during the second half of the first millennium ad is provoked by a variety of factors affecting the Gezira, including aridification, loss of pastures, overexploitation of cultivable lands and overpopulation. This might have favored the appearance of transhumant pastoralism as a mitigation strategy.

Between the Kingdom of Alodia and the Sultanate of Sennar (ad 1300-1650): The Jebel Mahadid Tradition

At some point between ad 1300 and 1400, the Gelegu Tradition disappears from the archaeological record and is replaced by another tradition that we call Jebel Mahadid for its most significant site. Five sites have been documented that belong to this tradition: Gerara 1, Tach Gerara, Jebel Mahadid 1 and 2 and Shimeligir. Of these, only Tach Gerara, Jebel Mahadid 1 and Shimeligir are substantial settlements. The other two are small artifact scatters (ca. 1000-2000 m2) on the banks of the Gelegu, perhaps temporal occupations for the exploitation of river resources or river pastures.

Characteristics

It is of course impossible to characterize settlement patterns or spatial organization from just five sites. However, there are some new elements that are very conspicuous and that evince important sociocultural changes. For the first time, permanent structures are recorded. They fall into three categories: defensive, domestic and ritual. Only Jebel Mahadid and Tach Gerara have the three kinds and the best and most numerous examples by far come from the first of these sites, which we will describe in some detail.

The occupation of Jebel Mahadid spreads along the base of a rocky elevation of the same name, which emerges 300 m to the west of the Gelegu River (Fig. 10). The archaeological site occupies three strath terraces that extend from the pediment of Jebel Mahadid to the Gelegu river. The lower and middle terraces have yielded materials belonging to the Gelegu Tradition. The site of the phase under discussion occupies mostly the upper terrace, but also part of the middle. The maximum length is 450 m (ne-sw) and the maximum width 150 m (nw-se), which make this the largest site discovered to date in the entire Qwara-Metema lowlands. The defensive element is composed of a rubble parapet that follows the limit of the upper terrace for around 300 m. It is basically an accumulation of unshaped stones. From the inside of the settlement it raises around 0.5 m only, but from the outside is two or more meters high, so it would have been a useful deterrent in case of attack, especially if combined with a palisade or brush fence. Similar parapets delimit the site of Tach Gerara, although in this case they are perpendicular to the natural terraces and the defensive purpose is less obvious.

We consider as ritual structures at least three different features: oblong or oval stone platforms, fan-shaped platforms attached to hut foundations, and rectangular structures. In Jebel Mahadid, 52 structures were documented, of which 28 are hut remains (some with trapezoidal platforms attached), 23 oblong platforms and the remaining one a rectangular feature. The function of the non-domestic structures is difficult to ascertain, but a ritual use can be hypothesized.

This is undoubtedly the case with the rectangular structures of Jebel Mahadid and Tach Gerara, which are located in isolation at the entrance of the sites. They are both made of big stone blocks delimiting a space of around 3 × 1.5 m, which is filled with burnt clay. In the case of Jebel Mahadid, a fan-shaped stone platform is attached from the south. Inside the rectangular structure we found an almost complete incense burner.

Regarding the oblong structures, the situation is not so clear. We excavated one of them at Jebel Mahadid (Fig. 11, n.1). These platforms are less than 2 m long and are identified as tombs by the local communities. Indeed, they are very similar to Post-Meroitic/early medieval cairns and tumulus (e.g. Edwards 1998: 103-104) and some of our structures may actually be graves. The excavation at Jebel Mahadid, however, only revealed a carpet of stones lying flat directly on the ground, although carefully arranged, and was unable to identify anything resembling a ditch or pit beneath. When we dismantled the platform, we discovered broken grinding stones, handstones, small pottery sherds and tiny bone fragments amid the blocks composing the structure, none of which tells much about the purpose of the structure. The two structures recorded in Tach Gerara are bigger (the largest has a diameter of 6.5 m), circular rather than oblong, and the function might have been different.

figure 10