Vous êtes sur la page 1sur 21

Afr Archaeol Rev (2009) 26:305325 DOI 10.

1007/s10437-009-9061-5 O R I G I N A L A RT I C L E

Variability in Eritrea and the Archaeology of the Northern Horn During the First Millennium BC: Subsistence, Ritual, and Gold Production
Peter R. Schmidt

Published online: 26 January 2010 # Springer Science+Business Media, LLC 2010

Abstract Around the margins of Asmara, Eritrea, hundreds of sites dating to the early and mid-first millennium BC have been documented. They range from single family dwellings to small and large hamlets, small and large villages, and small towns. We call these Ancient Ona sites, using the Tigrinya term for ruin. Our findings testify to significant subsistence, ritual, and economic variation within a region of 12 by 17 km: (1) different subsistence strategies in the well-watered, open basin to the west of Asmara (emmer wheat, barley; cattle) compared to the uplands north and east of Asmara (lentil, teff; goats/sheep); (2) ritual events, marked by stone bulls' heads and a huge ash deposit at Sembel Kushet, that brought people together in rites of passage and intensification during Meskel-like ceremonies, including ritual exchange; and (3) the exploitation of gold north of Asmara among heterarchically organized communities that exchanged specialized products within this region. Rsum Aux environs dAsmara, en Erythre, des centaines de sites datant du dbut et milieu du premier millnaire avant notre re ont t recenss. Ils comprennent des habitations unifamiliales, des hameaux de taille diverse, des villages petits et grands, ainsi que des petites villes. Nous appelons ces vestiges sites Ona anciens, en nous inspirant du terme Tigrinya dsignant une ruine. Nos trouvailles dmontrent une diversit considrable en matire dconomie, de moyens subsistance, et pratiques rituelles dans une rgion de 12 par 17 kilomtres: 1) Les stratgies de subsistance employes dans le bassin ouvert et bien arros a louest dAsmara (amidonnier, avoine, bovins) diffrent de celles en usage dans les hautes terres au nord et lest
P. R. Schmidt (*) 20924 NE 132nd Ave, Waldo, FL 32694, USA e-mail: schmidtp@ufl.edu


Afr Archaeol Rev (2009) 26:305325

dAsmara (lentille, teff, chvres/moutons); 2) Les vnements rituels sont reprsents par des ttes de taureaux en pierre et par un dpt norme de cendres Sembel Kushet; ils rassemblaient les gens dans des rites de passage et rites dintensification similaires ceux des crmonies de Meskel; 3) Au nord dAsmara, l'or tait exploit par des communauts htrarchiques impliques dans des changes de produits spcialiss dans la rgion. Keywords Eritrea . Ancient Ona . Pre-Aksumite . Ritual . Gold . Bulls heads

Introduction In the view of some scholars, Pre-Aksumite means a period of time in the northern Horn when there is evidence for Sabaean influences from the Arabian Peninsula (Drewes 1962; de Contenson 1981; Schneider 1976; Anfray 1968, 1990), even using the term Sabaean to designate a developmental period in the mid-first millennium BC. As David Phillipson (2009) has observed, Sometimes Ethio-Sabaean and Pre-Aksumite were seen as successive stages, sometimes both were attributed to a Pre-Aksumite period. These influenceswhether in the form of appropriated material culture used by local elites or more direct contact with the Sabaean worldare defining attributes that mark off this time period from other post-Holocene cultural periods in the northern Horn. Mere reference to the Pre-Aksumite has taken on symbolic powerbecoming a trope for an era of foreign influence, an era when local cultures accommodated and co-opted such influences to transform into the proto-Aksumite and then into what we know as the political-geographic entity Aksum. The use of Pre-Aksumite has introduced a problematic terminology in the archaeology of the northern Horn by privileging cultural influences external to the Horn and thus submerging local innovations and developments. This paper takes a path away from the homogenizing effects of tropic categories that typify archaeological classification. By examining the archaeology of the Greater Asmara region, I step back from the homogenization of local developments under the Pre-Aksumite label and instead shed light on the regional distinctiveness of the archaeological remains around Asmarailluminating how and why early first-millennium BC developments there diverged from other areas in the northern Horn (for map of the Northern Horn including major sites see Fig. 1). The first step in this process came when my colleagues and I designated the archaeological culture documented from 800 to 350 BC around Asmara as the Ancient Ona Culture (Curtis and Schmidt 2008; Schmidt et al. 2008b) precisely because it did not fit into the earlier tropic classificationlacking evidence for any Sabaean influences or dominant influences from the south. Incorporating the use of ona by Tringali (1965)the Triginya word for ruinsthe Ancient Ona designation leaves behind some vexing cultural characterizations while also opening intellectual space to focus on variation within the northern Horn. Comparisons between the coeval early to mid-first millennium BC sites of Ethiopia and the Ancient Ona of Greater Asmara certainly show that there are affinities between these two zones within a larger region (Fattovich 2009) but importantly also affirm there are noteworthy differences that

Afr Archaeol Rev (2009) 26:305325


Fig. 1 Map of the northern Horn, showing principal sites discussed in this issue (Luisa Sernicola)

testify to unique developmental trajectories deserving to be understood free of biasing labels.1 Among the developments that distinguish the greater Asmara sub-region from other parts of the northern Horn during the early and mid-first millennium BC is the more detailed and nuanced knowledge we now have for subsistence practices across a geographically bounded research universe. Heretofore, subsistence practices that made up quotidian economic life have either been ignored in favor of a focus on exotic and elite artifacts or have been discussed only for a relatively tight cluster of sites at Aksum (Bard et al. 1997; Boardman and Phillipson 2000; Phillipson 2000). There is little evidence in the northern Horn for how communities varied across a sub-region in their subsistence practices during the first millennium BC. The research around Asmara corrects this deficiency while simultaneously providing a better understanding of daily economy linked to topography, water, soil, and climatea more finely grained analysis that allows us to see possible specializations in the subsistence economy of individual communities. Moreover, it is now apparent that in matters of ritual life and ideology, the greater Asmara research throws considerable light on the question of local belief systems and ritual practices that differ significantly from what we know about other sub-

In chapter 10 (D'Andrea et al. 2008) of The Archaeology of Ancient Eritrea (Schmidt et al. 2008a), we use Ancient Ona to replace references to the Pre-Aksumite of northern Ethiopiaan extrapolation that is not germane because of differences discussed here. The use of Ancient Ona (800350 BC) is restricted to the greater Asmara area.


Afr Archaeol Rev (2009) 26:305325

regions of the northern Horn. In particular, the presence around Asmara of ritual objects known as bulls' heads allows us to develop deeper insights into how cattle functioned in the ritual life of first millennium BC people at a very local level and in a manner quite different from northern Ethiopia or even elsewhere in Eritrea. The ritual complex that I have interpreted and documented around Asmara appears in our mapping of quartz bulls' heads to have operated within clear social boundaries and to have been part of local exchange networks on the Asmara plateau (and a few neighboring areas off the plateau), suggesting that ideas that link these bulls' heads with either rock art depictions or other representations of cattle in Arabia or the Horn only serve to erase the distinctively local quality of the phenomenon (e.g., Fattovich 2004; Buffa and Vogt 2001). There are also particular exploitative practices, such as the mining and processing of gold in the first millennium BC, that help to explain some of the demographic clustering that occurred in the northern and western parts of the Asmara plateau another developmental trajectory that differs from surrounding sub-regions. This again illustrates that knowledge about local variations helps to make representations about the first millennium BC that avoid homogenizing conclusions, and that account for the wide variety of natural and cultural differences that occurred in the macro-region that we call the northern Horn. I propose here to make finer grained examinations of subsistence, ritual life, and exploitative economy on the Asmara plateau to establish a comparative framework that goes beyond the commonalities that have arisen in the literature heretofore.

Subsistence: Insights Into Sub-regional Practices I want first to examine how subsistence practices varied significantly across a 12 17 km research universe around Asmara (Fig. 2; see Curtis and Schmidt 2008 for survey details; Curtis 2009). This discussion is based on paleobotanical remains recovered through flotation after having been systematically sampled from excavation units and all features with ashy deposits. The contexts for such recovery were particularly favorable, with hearths, kitchen floors, and ash discard areas providing the greater proportion of seed remains. Here, I discuss five sites from which significant quantities of crop remains were recovered: Sembel, Mai Chiot, Mai Hutsa, Ona Gudo, and Weki Duba (D'Andrea et al. 2008; Table 10.1). The first key observation is that food crops differed significantly across the sub-region, with quite different crop repertoires in the open, better-watered plains west of Asmara (Ona Gudo, Sembel, Weki Duba) compared to the rocky uplands north and east of Asmara (Mai Hutsa and Mai Chiot). The open plains have wide expanses of fields within 2 km of permanent water and also hold small numbers of minor rocky outcrops. This is an ideal setting for plow agriculture with the use of oxen, conducive for planting and harvestingalso using oxen for threshinglarge expanses of highly productive grains. Table 1 shows the range of crops documented at sites such as Sembel, Ona Gudo, and Weki Dubaall located on more open terrain. These communities relied mostly on emmer wheat (Triticum dicoccum) and hulled barley (Hordeum vulgare), with barley forming a

Afr Archaeol Rev (2009) 26:305325


Fig. 2 Sites of different cultural periods, colored coded as per legend. Each survey grid represents 1 km2 (Matthew C. Curtis)

very large proportion of the grains in the deeper deposits at Sembel, the most extensively excavated site located above one of the few perennial streams and its associated rich bottom lands. Some 97% of the 115 hulled barley grains recovered from Sembel came from multiple hearths in middle strata in the southern kitchen area of Room B dating to approximately 600 BC (Table 1; Fig. 6.16 in Schmidt et al. 2008b). While the gross frequencies of hulled barley at Sembel are distinctive, barley was also the dominant grain in the basal deposits at Weki Duba, where 77%


Afr Archaeol Rev (2009) 26:305325

Fig. 3 Excavated sites around Asmara (Matthew C. Curtis)

of the grains recovered from a kitchen floor were hulled barley. Similarly, at Ona Gudo, the bottom seven excavation units held 95% of the barley documented at this site. These results suggest that barley was the crop of choice during the first settlement on the open plains. Emmer wheat, also a widespread part of the Near Eastern grain repertoire, was not as popular in the gross frequencies but nevertheless was a key staple at these same three sites. It was found mostly in the earlier depositions dating between 800

Afr Archaeol Rev (2009) 26:305325 Table 1 Raw frequencies of seeds at excavated sites around Asmara Site Sembel Ona Gudo Weki Duba Mai Hutsa Mai Chiot Hulled barley 131 20 13 0 3 Emmer wheat 19 3 4 0 0 Bread wheat 2 0 0 0 0 Lentil 1 10 6 2 9 Linseed 29 20 33 3 5


Tef 0 0 0 0 1

and 600 BC,2 an observation amplified by the virtual absence of emmer wheat in the later deposits, suggesting a fundamental shift in its role in the diet through timea decline that is mirrored in the diminishment of emmer wheat cultivation in the Ethiopian highlands and other regions (D'Andrea and Haile 2002; D'Andrea et al. 2008). In contrast to the open, lower plains are the rocky uplands where smaller patches of fertile soil prevail among many ambas (mesas), schist outcrops, and other steep and rocky gradients. Mai Hutsa to the north of Asmara and Mai Chiot to the east of Asmara are the two excavated sites located in this topographic/physical zone. Of the five sites discussed here, these two uplands sites yielded no specimens of emmer wheatfixing that crop exclusively to the open bottom plains. Subsistence practices in this rocky zone close to the eastern escarpment may have been influenced by micro-climatic phenomena, such as the fogs that flow over the eastern escarpment and hang over the adjacent plateau. The distinctiveness of the uplands is also emphasized by the absence of identifiable hulled barley at the Mai Hutsa site and the presence of only three barley grains scattered through the deposits at Mai Chiot. The dichotomy between the two zones is thus underlined by the virtual absence of emmer wheat and very low frequencies of barley in the upland rocky zone sites. As Mai Chiot is a more extensively excavated site, it is not sample size that accounts for these clear contrasts. The staple bread of Ethiopia and Eritrea is enjera, fermented flat bread made from teff (Eragrostis tef). Reliable dating of teff in the northern Horn has eluded archaeologists for some time. The presence of teff in Pre-Aksumite deposits at the Kidane Mehret D site in Aksum has been set aside as unreliable because the seeds appear to have been contaminated by later deposits (Boardman 2000: 365; Phillipson 2000: 372; contra Marshall and Hildebrand 2002: 212), yet teff was abundant during the Aksumite period when it was the most common grain in the earlier phases (Boardman 1999, 2000: 366). Teff has also been documented at Beta Giyorgis, dated to the Aksumite and perhaps to Proto-Aksumite times (Bard et al. 1997). In the Greater Asmara excavations, one grain of teff was recovered from a sealed deposit in the Mai Chiot site in the rocky uplands. The context from which it was documented compact, red burned earth associated with faunal remains and linseedsuggests the
As D. Phillipson (2009) notes, radiocarbon dating between the eighth and fifth centuries BC is problematic because of variations in atmospheric C14 during that time. However, we have reliable AMS dating for the beginning and multiple dates towards the end of this problematic period, allowing us to reckon relative time through fine grained analysis of stratigraphic evidence (e.g., Mehari 2008).


Afr Archaeol Rev (2009) 26:305325

presence of a hearth. Though limited to one grain, the absence of disturbance and the secure dating to the second half of the first millennium BC is important for definitively documenting teff in the northern Horn diet at this time and the setting in which it occurredAsmara's rocky uplands. We also have strong evidence for cultivation of lentil in the first millennium BC around Asmara. In fact, we find that lentil is only one of two cropslinseed being the otherthat is not restricted to either the open plains or rocky uplands. Rather, it is found in all excavated sites, appearing in upper deposits at Sembel, in relatively large numbers in a series of ash deposits and hearths at Ona Gudo dated to 770350 BC and 800 400 BC (95% cal; Curtis 2005; D'Andrea et al. 2008), and at Weki Dubamostly in deeper deposits associated with ash and grindstones. In the rocky uplands, lentil occurs at Mai Chiot in deposits dated towards the terminal Ancient Ona (500350 BC) from both middle and deeper deposits; and at Mai Hutsa lentil is found in hearth debris dating to 830420 BC (95% cal; Curtis 2005). This evidence points to lentil being an important pulse and an integral part of the diet throughout the Ancient Ona period. An important oil crop, linseed (Linus usitatissimun) is also found in all Ancient Ona sites in the greater Asmara area and thus mirrors the evidence found at the Kidane Mehret D site, Aksum, in terms of general importance in the diet. The presence of linseed at all sites however must be examined within the context of the two different zones in which sites are situated. The Mai Chiot and Mai Hutsa sites in the rocky uplands held only eight of the 90 seeds documented or about 9% of the total collection, whereas the Sembel, Ona Gudo, and Weki Duba sites held 91% of the documented specimensa very large proportion associated with the open, lower plains. Noteworthy at Weki Duba was the association of many linseed specimens with thickly sooted red ware pots midway through the depositional sequence; many other specimens were located in a dense ashy layer on a basal kitchen floor and in the interstitial spaces in the stone floor, dated to ca. 700 BC. Importantly, at the Sembel site, there are many lower grindstones with a high polish associated with linseed deposits, suggesting that grinding to produce oil was an important component of the subsistence economy and perhaps even for intraregional and interregional trade of linseed oil. Barley and emmer wheat are clearly primary crops found throughout the depositional sequences of the western, open sites of Sembel, Ona Gudo, and Weki Duba. These two crops are not a prominent part of the uplands diet. Though lentil was also found throughout the Asmara plateau, it is more important in the diet of open plains sites such as Ona Gudo and Weki Duba, but not Sembel. This suggests a varied patternnot associated with preservation issues (D'Andrea et al. 2008)that may be related to specialization of different communities, perhaps linked to the heterarchical qualities of settlements and intraregional trade that balanced specialized grain production, each community making contributions to a wider regional distribution system (see Crumley 1995; McIntosh 1998). Linseed, too, was found throughout the plateau, but was more common in those communities bordering the open plains, whereas teff has been documented thus far only in a rocky upland setting. The rich paleobotanical record at Aksum shows a significant shift in crops between the Pre-Aksumite and Aksumite periods, a shift that incorporated additional African domesticates such as finger millet and sorghum as well as Near Eastern crops such as lentil, pea, and grass pea (Bard et al. 1997; Boardman 1999: 144). We do not have sufficient settlement evidence for the Aksumite period in the

Afr Archaeol Rev (2009) 26:305325


greater Asmara sub-region to inquire into possible parallel shifts there, although (as indicated above) lentil is much earlier than at Aksum. Different subsistence practices, closely linked to soil types, proximity to water, and topography are significant reminders that variation at a sub-regional level such as the Asmara plateau is central for understanding variation across the entire region. While some growing or consumption of linseed, barley, and lentil apparently occurred in the uplands, these communities appear to be more restricted in their productive capacity and crop repertoire, especially in the virtual absence of emmer wheat. In contrast, the lower and open plains show a plethora of evidence for the barley and emmer wheat staples as well as linseed, though not always lentil. This finely nuanced system of production, perhaps balanced through exchange via kinship networks, alerts us to the likelihood that such variations express the heterarchical qualities of specialized contemporary communities and that such patterns may also have prevailed in other sub-regions of the northern Hornsomething to keep in mind before we rush to characterizations based on any sub-region to typify a larger region.

Subsistence and Domestic Animals The previous observations about intraregional variations also extend to pastoral life through the consumption of animals, since the Ancient Ona Culture was an agropastoral economy. The differences that appear in food crops are mimicked in the different husbandry and butchery practices that distinguished the people of the open plains from those of the rocky uplands. Cattle consumption was greater in the open areas where oxen would also have been used for cultivation and threshing. In contrast, goats and sheep were far more prominent in the rocky upland areas, suggesting clear divisions between ecological/topographic zones. If we look at the sub-region as a whole in terms of total frequencies of domestic animals represented in our collections from six excavated sites, then there appear to be more small ungulates (mostly Capra/Ovis) than large ungulates (mostly Bos). But this generalization is potentially misleading for the sub-region because the picture is skewed by the much larger number of small ungulates at Mai Chiot and Mai Hutsa in the rocky uplands. Such a scenario does not applyusing MNIto the communities located near the intensively cultivated areas in the open plains such as Sembel Kushet and Ona Gudo, where Bos/large ungulates dominate most of the faunal profiles (Fig. 4a and b). Sembel is less definitive, with equal numbers of small and large ungulates, keeping in mind that the large ungulates have much higher nutritional value. The contrasts between uplands and open plains communities take on greater vividness when we compare Mai Chiot from the uplands and Ona Gudo from the plains as proxies for the two zones in the sub-region. A more detailed discussion of the distribution of different body parts (Shoshani et al. 2008) shows that a contrast between uplands and plains sites appears only when body part frequencies are examined for sheep, goats, and small ungulate-sized specimens. Mai Chiot provides good evidence for an integrated or even-part distribution of meat cuts, as does Mai Hutsathe other uplands site. In the three communities in the open plains, there is a greater frequency of axial cuts (vertebrae and ribs). If we envision the sub-region as an integrated economy, then

314 Fig. 4 a and b. Histograms showing the contrasts between the plains (Fig. 1.2a) and rocky uplands (Fig. 1.b) sites to show dominance of cattle in the open plains and sheep/goats/small ungulate-sized animals in the uplands (Shoshani et al. 2008)

Afr Archaeol Rev (2009) 26:305325

the higher frequency of axial cuts in the lowlands may represent exchange coming from the uplands where the better cuts remained for local consumption along with other axial cuts; such exchange occurred over several kilometers and up to an hour's journey by foot. This integrated perspective on intraregional exchange may also help us to understand some of the crop specializations that we see in the paleobotanical recordwith surplus grains such as processed emmer wheat, barley, and linseed or linseed oil (and possibly beer made from barley) moving from the open plains to the uplands in exchange for easily transportable cuts of sheep and goat meat. RitualThe Search for Meaning The second topic that invites examination for cultural variation within the sub-region is the production and distribution of what have been described as ritual bulls' heads (Tringali 1965, 1967, 1978; Schmidt and Curtis 2001; Schmidt and Naty 2008). Mostly stone objectsalong with some ground potsherd examplesmade by both chipping and grinding (Fig. 5), these artifacts are found in a variety of Ancient Ona sites concentrated along the northern and western zones of the Asmara plateau. Their general classificationusing the triangular frontal view of Bosdoes not capture the variation found in this category of artifact. Many objects are rounded, elongated,

Afr Archaeol Rev (2009) 26:305325 Fig. 5 Ground stone bulls' heads made from metavolcanics, volcanics, and pottery (upper left), excavated at the Sembel site


stubby, and depart from a triangular norm. Some are diminutiveonly several centimeters in diameter, while others resemble 1215 cm wide, roughly finished blanks. There is significant variety in the care and quality of craftsmanship as well as the types of materials used and the scale of the finished product (Schmidt and Naty 2008). Yet within this large range of variation there is little doubt about which objects fit within the bulls' head category. Tringali seemed genuinely perplexed about what to call these objects, also referring to them as crescents, a term that elicits comparison to the Sabaean crescent and circle. In earlier treatments, Fattovich (1997a, b, 2000) favored the idea that images of bulls in the northern Horn are linked to the Arabian Peninsula, but he does not specifically suggest that these particular objects of the Ancient Ona were the result of such contact and cultural influence. He does object however to our interpretation that they should be treated as a local phenomenon and not seen as derivative from other regions such as the Arabian Peninsula: Such an extreme approach may please some Eritrean intellectuals and politicians, and those who worship at the altar of political correctness (Fattovich 2008: 347), a trenchant critique that diminishes the importance of these significant findings rendered in stonenot found elsewhere in the northern Horn let alone the Arabian Peninsula. Any argument for an Arabian connection would be misleading in this setting; the complete absence of any Near Eastern artifacts associated with bulls' heads in Ancient Ona communities and their contiguous multipurpose sites incisively affirms that these objects belong exclusively to local expressive culture. An appraisal of the objects represented in Fig. 5 shows that most are highly stylized representations of bovids, though they do not capture the finer features of bovids. As we search for intentionality in these collections, we find that the exceptions to the norm provide deep insights. The excavation of such an artifact (Fig. 6) at Sembel provides some clues: a ground-stone object with a hump between opposed, tapered points may be recognized as representing a humped Zebu (Bos indicus)lending confidence that the intended representation is indeed a bovid. The wide variation in style and the quality of craftsmanship may indicate a level of craft skill that was not highly developed, or a varied familiarity with production techniques. The absence of standardization may point

316 Fig. 6 A drawing of the ground stone bull's head in the bottom center of Fig. 4, a representation of B. indicusthe earliest known representation of B. indicus in Africa

Afr Archaeol Rev (2009) 26:305325

to variation associated with different social groups, but the possibility that these objects were produced by relatively unskilled young people also emerges as significant. We first thought, following Tringali's observations, that these objects were limited to mounds covered with architectural stone that mark the Ancient Ona settlements of various sizes. Our comprehensive and systematic survey of 90 km2 on the Asmara plateau (Curtis and Schmidt 2008) shows however that they are also common on some of the multipurpose but low-visibility sites of the Ancient Ona period. One site, located in Survey Unit 118 southeast of Asmara and designated as K09, had 12 bulls' heads on its surface, whereas the nearby K01 site had none (Fig. 2). Other sites with bulls' heads but without evidence of permanent settlement are those with large quantities of quartz float processed for gold, such as those located immediately to the northwest of Asmara and east of the Hara Hot gold mine (Figs. 2 and 3). The presence of bulls' heads on these large multipurpose sites (no architectural evidence but plentiful grindstones, lithics, and pottery) and the clear differences in the frequency of bulls' heads among large multipurpose sites suggests that activities associated with these objects happened in specific spaces outside the core stone-built communities and for very specialized purposes, possibly for sequestered ritual instruction of neophytes at initiation schools. Tringali located bulls' heads about 40 km to the south at Gura near Dekemhare and about 38 km north of Asmara towards Karen (Tringali 1978: 62, 98, Fig. 32), though most are confined to the Asmara plateau. They are not known from any other region of the northern Horn. While there have been many speculative ideas about what these objects signify, they are clearly significant for understanding ritual relationships that may once have prevailed on the Ancient Ona landscape. Using strong inference that draws on Alex Naty's ethnographic details on the Kunama (Schmidt and Naty 2008), I argue that these objects once functioned as ritual promissory notes, as devices symbolically promising cattle as the first of several steps in bride-price payment, and as an integral part of a rite of passage also associated with rites of intensification. Naty informs us that the Kunama, a NiloSaharan-speaking people who now live in the western lowlands, have explicit oral traditions about their identity with the highlands, and practicing a form of initiation where young men make clay figurines of cattle, later giving them to the family of their betrothed as symbols of the cattle bride price they will pay upon consummation of the marriage some years later (Schmidt and Naty 2008). Naty also points out in his testimony that these Kunama ritual events are rites of passage because they occur at a particular time (age 68) in a young man's life cycle, but they are also rites of intensification because they are conducted close to the harvest season when

Afr Archaeol Rev (2009) 26:305325


fecundity of the land and the reproduction of society are foremost concerns. Given that many different families in different communities may have been engaged in such rituals, large numbers of people would have gathered together while using such symbolic capital. The bulls' heads on the Asmara plateau are concentrated in the core zone of the open plains, where cultivation of the primary staple food crops once prospered and where larger permanent towns were located. The most significant concentration of bulls' heads has been documented at Sembel Kushet (Figs. 3 and 7), where some 850 of the objects were recorded in the 1970s by Tringali (1978: 62) and where our student teams also observed several dozen on the surface (also see Habtemichael 2000). A second smaller concentration occurred just 1 km to the southeast at the Sembel site. What captures our attention on the Asmara plateau is the variety and proportion of bulls' heads at Sembel Kushet and their dense concentration in a large and deep ashy area. This zone is not the result of domestic ash dumpinga process that usually entails distribution to agricultural fields in the sub-region. Rather, the large number of these ritual artifacts and the distinct boundaries of the ashy zone ask for a more comprehensive explanation, one that also accounts for its distinctiveness on the Asmara plateau. In the rich historical and ethnographic corpus of the Horn, one particular rite emerges as a compelling inference: the
Fig. 7 Map of Sembel Kushet. The key characteristic is the ashy zone in the western part of the site, where large numbers of bulls' heads have been documented


Afr Archaeol Rev (2009) 26:305325

annual Meskel celebration (Feast of the exaltation of the Holy Cross) that now occurs September 27, but which in pre-Christian times appears to have occurred during another time of the year, possibly March (Ullendorff 1968) when, after the coming of Christianity, it would have directly conflicted with Lent. Now a celebration featuring officials of the Orthodox Church, this sacred rite focuses on a huge bonfire (demera), with the future health and prosperity of the immediate region predicted by the direction in which the last burning logs fall, or in some areas, the direction in which the smoke blows (Fig. 8). In many parts of the Horn it is linked to feasting on sacrificial oxen and it is a celebration of the fertility of women and cropsa rite of intensification and renewal that brings together large numbers of people within a region, acting as a social glue in what are heterarchically or horizontally organized communities. This pre-Christian fertility rite (Ullendorff 1968), long ago wrapped into church liturgy as a means of capturing its potency, leaves a large deposit of ash. These days the charcoal is removed and used by participants who mark their foreheads with Christian crosses as gestures of purification; the virtual absence of charcoal from prehistoric deposits suggest a possible parallel function in which bodies may have been colored with charcoal. The performance of Meskel in a central location, in exactly the same place as previous years, robustly suggests that the large ashy deposit on the western side of the Sembel Kushet site was repeatedly usedover hundreds of yearsfor a similar purpose. The great profusion of ritual items shaped like oxenat the same locale further suggests rites of passage and rituals of intensification in which fertility and human reproduction were featured: in this instance with exchange of symbolic bride-price objects during a large public ceremony in which oxen are the most prominent feasting food. Though ground-stone bulls' heads are by far the most common type found at Sembel Kushet, chipped quartz specimens also occur, leading us to ask if they were manufactured elsewhere but subsequently taken to Sembel Kushet for exchange.

Fig. 8 A Meskel celebration in Asmara, 2001

Afr Archaeol Rev (2009) 26:305325


If this hypothesis about the ritual role of Sembel Kushet is to be sustained, then we might also expect to find the production of such ritual items in other contemporary communitiesin readiness for the general exchange at Sembel Kushet and/or local exchange. The closest excavated stone town is at Ona Gudo (Fig. 3), only 2 km to the northwest of Sembel Kushet and across open plains: 17 bulls' heads were documented at Ona Gudo in one test excavation, eight of them chipped from quartz (Curtis 2005). The other site with excavated quartz bulls' heads is Mai Hutsa, 9.5 km northeast of Sembel Kushet in the rocky uplands. Mai Hutsa is coeval with Sembel Kushet and has a suite of nine excavated bulls' heads, six of which are chipped from quartz (Fig. 9; Curtis 2005; Schmidt and Naty 2008). The large numbers of quartz bulls' heads at Mai Hutsa come from several contiguous excavation units, indicating a specific manufacturing arealikely using large, roughed-out blanks for these quartz artifacts, some of which are quite robust, such as the middle specimen in the upper row of Fig. 9. To gain a finer grained perspective, we examined the quartz debitage from these excavation levels and found a significantly higher percentage of angular waste 2 cm or larger, congruent with the larger flake scars on these artifacts (Schmidt and Naty 2008; Teka and Okubatsion 2008). The large size of the finished objects, the absence of fine edge work, and the presence of a higher proportion of large debitage are archaeological signatures in the lithic record for the production of such symbolic capital, certainly a new direction for lithic studies in the Horn. We must consider whether communities such as Mai Hutsa and Ona Gudo produced specialized bulls' heads for exchange at ceremonial centers like Sembel Kushet. To obtain a better idea of the range of quartz bulls' head and their possible distribution networks into areas without such excavated evidence for their production, we examined phase I survey evidence from the Asmara plateau (Curtis 2005; Curtis and Schmidt 2008; Schmidt and Naty 2008). We found that these quartz ritual products reached 4 km south to Daro Pawlos in the far southwestern sector of the research region (Fig. 2; survey unit 154, five specimens), where metasedimentary and metavolcanic materials prevail, but without quartz. Survey unit 142, 3 km to the south of Sembel Kushet, yielded four chipped quartz bulls' heads, while 23 km to the northwest, there were three and two chipped quartz bulls' heads in units 88 and 98, respectively. If we go further afield to survey units 31 (one specimen), 33 (one
Fig. 9 Quartz bulls' heads excavated at Mai Hutsa (Curtis 2005; Schmidt and Naty 2008)


Afr Archaeol Rev (2009) 26:305325

specimen), 35 (one specimen), 17 (two specimens), and 28 (one specimen), we see that the densities change to a gradient that is proportional to the 710 km distance from Sembel Kushet (Fig. 2). May these latter specimens have derived from a nearby community such as Mai Hutsa? If so, then one would expect higher densities given the proximity of Mai Hutsa to these survey units. Rather, it appears that exchange was mediated by the ritual and redistributive hub of Sembel Kushet, where feasting and large gatherings accompanied rites of passage and intensification around a Meskel-like demeraleading to the spread of these objects around the northern, western, and southwestern plateau. Surface and excavated examples show clearly that bulls' heads are virtually absent from the eastern rocky uplands zone. Mai Chiot lacks bulls' heads on its surface; no specimens were excavated there during what was the second largest excavation around Asmara. Mai Chiot was a short-term occupation dating to the terminal Ancient Ona and may post-date the ritual complex, or it may simply have been outside the social orbit of such ritual activity. The uniqueness of Mai Chiot, with its unique crop repertoire, its different domestic animal diet, and its uniqueness as the only community to rely partly on wild game in its dietall of these attributes when compared to the communities in the open plains mirror its separation from the bulls' heads ritual complex. This suggests further that the bulls' heads cult may have been socially and geographically focused, an observation that parallels and affirms other significant variations within the sub-region.

A Point of Continuity: Gold and the Exploitative Economy When discussing quotidian economic life such as subsistence practices, one should avoid the temptation to submerge other variations in economic life perhaps less definitive in some areasfor activities that may have contributed to the relatively large demographic size and prosperity of communities on the Asmara plateau. I now turn to the mining and the processing of gold around Asmara by Ancient Ona and later communities. The exploitation of gold is attested to by a variety of evidence, including the presence of material culture such as Ancient Ona pottery and particularly tina cupssmall terracotta vessels with vertical walls (Fig. 8.6 in Schmidt et al. 2008c: 185) observed in mines that were reopened during Italian colonial rule (Tringali pers. comm. May 19, 2002). There are also reports by early miners, among them F. H. Hatch, an English geologist, who in 1902 reported that deep mines with pottery and stone hammers were observed at both Schmugle and Medrizien before they were reopenedmines that became the largest producers during the Italian era (Blackburn 2003: 9). Other reports of ancient workings are found in both Italian and British records (Barnard 1941; Usoni 1952). An Italian geologist on a mission to investigate the Hara Hot and other mines in the 1930s reported that the Hara Hot areanow to the northwest of the Teacher Training Collegefigured prominently in local stories about Portuguese mining in the area (Usoni 1952: 138; Figs. 2 and 3; survey unit 56). This oral tradition possibly refers to mineral exploration undertaken by Portuguese living at Debarwa (about 35 km southeast of Asmara) during the sixteenth and seventeenth centuries. It is independently repeated by C. G. Barnard,

Afr Archaeol Rev (2009) 26:305325


a British military administrator of the Italian mines during World War II. He reported that the area around Hara Hot was exploited by Portuguese missionaries and came to be known as the Portuguese caves (Barnard 1941). We believe, from the physical evidence available today, that the principal mine before Italian involvement at Hara Hot was a conical hill, now a mined-out crater with internal shafts, appearing on maps as the Gradino or No. 3 mine (Schmidt et al. 2008c: 182; Fig. 8.3). Among the more intriguing reports is Usoni's mention of the Graziani mine several hundred meters from the central processing area at Hara Hot. He observed that about 150 m to the west (sic), or southwest, there were Abyssinian tombs of Mekeleans or Aksumite tombs (Usoni 1952: 139). This tomb shaft was evidently used to access an underground galley in the later Italian mine. There is no basis for assigning an Aksumite date to the tombit could as well have been Ancient Ona but we now know that there was an Aksumite presence in the sub-region that pertained directly to gold mining, both at the BE01 site excavated in 2003 (Schmidt et al. 2008b) just 2 km to the northeast as well as further afield at Emba Dehro, where an early Aksumite or Proto-Aksumite community was situated on an elevated overlook just east of the Medizien mine (Schmidt et al. 2008d). Using our intensive survey results and Usoni's detailed drawing of the Pozzo Graziani, once marked by a steel elevator tower, we have isolated what we believe to be the shaft (left of the elevator tower in Fig. 10) that once housed the tomb and that subsequently led to three of four galleries in the mine. The dominant presence of Ancient Ona communities dating to the 800350 BC era suggests a higher probability that the burial remains date to that era rather than the later Aksumite period. In any event, the evidence from the BE01 site in Adi Abieto and the remains at Emba Derho indicate that regardless of the specific age of the Hara Hot tomb, the area had been exploited for gold by both Ancient Ona and Aksumite period miners (Fig. 3). Tringali's original research around Asmara led to
Fig. 10 A sketch of the Graziani mine section at Hara Hot (after Usoni 1952); the tomb shaft is likely the surface shaft to the left of the elevator tower (Schmidt et al. 2008d)


Afr Archaeol Rev (2009) 26:305325

his discovery and documentation of fragments of what he called crucibles in mines of the Ancient Ona period. While we have not relocated these objects in his collections now housed in the National Museum, he affirms that the tina we have documented are indeed what he labeled as crucibles (Tringali pers. comm. May 19, 2002). We have yet to observe any residuessooting outside or other residues insideindicating these artifacts were indeed used as crucibles.3 Tringali's thorough familiarity with the material culture of the Ancient Ona, however, inspires confidence that his observation of these objects in abandoned mines underwrites the idea of gold exploitation by Ancient Ona peoples. Tringali (1965: 152) was the first to notice the proximity of Ancient Ona communities to ancient gold mines. Our investigations confirm his prescient intuition and further suggest that the largest towns of the Ancient Ona are those found contiguous to the gold fields of the central, northern, and northeastern part of the basin. Most are located in the open plains near schist/quartz outcrops with documented veins. Ancient communities such as Ona Hashel and Ona Gudo, for example, were located next to productive ancient mines (e.g., site MB03). Large multipurpose sites lacking architectural debris, especially those situated near towns and mines, often contain large amounts of quartz float, marking the processing of vein quartz for gold. The presence of an ancient tomb in a deep burial shaft, along with tina vessels in some reopened mines, strongly suggests that the prosperity of the communities on and fringing the open plains was linked not only to their agricultural productivity but also their proximity to and exploitation of very ancient gold workingsanother characteristic that amplifies our understanding of the Ancient Ona, allowing us to see more clearly the broader regional affinities and differences when we focus on local developments across a region.

Concluding Thoughts Variation is often overlooked in our search for broadly applicable generalizations in archaeology. Though the Ancient Ona communities of the Asmara plateau can be characterized as agropastoralists who grew mostly barley and emmer wheat and who favored goats and sheep over cattle in their diet, such generalizations erase the particulars of local histories so critical for understanding why significant differences, say, in crop repertoires, arose over a 710 km distance in the Eritrean highlands. That the crops grown in the western and open plains of the Asmara plateau appear so different in scale and dietary profile from those of the communities located in the rocky uplands is of significant consequence. Soils, topography, and climate vary over very small distances in the highlands of Eritrea as well as other sub-regions of the northern Horn. These finely nuanced differences come to the fore with significant clarity on the Asmara plateau, demanding that we better understand the exchange networks that compensated for these natural imbalances. The absence of emmer wheat at Mai Chiot and the minimal presence of barley must be measured against the plethora of these grains as well as the production of oil from linseed at Sembel, Ona Gudo, and Weki Duba. The agricultural production of the open

Trace-element analysis on these objects is needed, once archaeology in Eritrea re-opens to free inquiry.

Afr Archaeol Rev (2009) 26:305325


landswith their much higher productivitywas matched by the ideal conditions for raising and herding goats and sheep on the rocky uplands, not a suitable enterprise near open fields of grain. The cattle pastoralism practiced by agropastoral communities on the open plains would have required a herd management system with stock required to go outside the sub-region during the growing season, only returning to pasture on the stubble of the fields after harvest was completed. Management of cattle herds favored the harvesting of older animals once their productive lives as oxen or milk cows had been exhausted, while the communities of the rocky uplands practiced a parallel approach with their goat and sheep herdsfavoring older animals that had greater body massa cultural preference seen even today. The key observation to emerge from these variations is the presence of portable, axial body parts of sheep and goats in the open plains sites. This is likely the result of distribution of meat from the not-very-distant rocky uplands through kinship networks for grain products, linseed oil, or even beer from the western communities. All of this attests to specialist contributions by communities organized according to principles of heterarchy, rather than hierarchy in economic and political organization, excepting the ritual cycles and their episodic expressions of hierarchy led by ritual officials who likely acted with significant political legitimacy as they also carried out their religious duties. Variations in ritual practices expand our knowledge about the values associated with cattle in Ancient Ona culture, values that emerge when we examine how Sembel Kushet differs so significantly from any other Ancient Ona site. A major place of ritual practice, the ashy zone of the western side of Sembel Kushetreplete with huge numbers of ritual bulls' heads and filled with deep deposits of ash appears to have been the locale for major ceremonial activities with affinities to contemporary Meskel rituals. As a center for important rites of passage and rituals of intensification, Sembel Kushet also functioned as the social glue in a redistributive system of ritual items used in symbolic exchange for brides, a process intimately tied to principles of reproduction and fertility, linking human reproduction and the land. These integrative activities, drawing together diverse and specialized communities with overlapping economic interests and social connections, kept secular authority at bay and provided sufficient unity to allow variation to prevail until such time that the hierarchical communities of Qohaito and Mataraabout 100 km to the southacted as population magnets, when they rose to prominence through their harnessing international trade in the mid- to late-first millennium BC. The production of quartz bulls' heads provides an unusual window into how local production of symbolic capital may now be recognized in the archaeological record. The physical distribution of quartz bulls' heads in particular opens understanding of exchange networks that were cemented together by episodic and annual ritual events at Sembel Kushet. This ceremonial network did not extend to the entire plateau, but was confined to social groups that shared other common beliefs and practices, among them the exploitation of gold. The communities living in prosperous small stone towns, villages, and hamlets were situated in proximity to gold workings, where miners left behind their material traces. We cannot hope to understand why some sub-regions in the northern Horn may show greater demographic complexity in the first millennium BC unless we examine ritual networks and solidarities, local exploitation of precious minerals, and participation in local exchange, be it meat, salt, grain, oil, beer, gold, or a variety of other products.


Afr Archaeol Rev (2009) 26:305325

Anfray, F. (1968). Aspects de larchologie ethiopienne. Journal of African History, 9, 345366. Anfray, F. (1990). Les anciens thiopiens. Paris: Colin. Bard, K. A., Fattovich, R., Manzo, A., & Perlingieri, C. (1997). Archaeological investigations at Bieta Giyorgis, Aksum, Ethiopia. Journal of Field Archaeology, 24, 387404. Barnard, C. G. (1941). Report on the Mines and the Mining Industry of Eritrea, June 1941. British Military Report of 21 pages on deposit at the Ministry of Energy and Mines, Asmara, Eritrea. Blackburn, G. (2003). Independent Geologists Technical Report: Asmara Project, Hamasian Goldfield, Eritrea. September Eritrea. Online at infomine.com/index/pa145277, accessed September 15, 2007. Boardman, S. (1999). The agricultural foundation of the Aksumite Empire, Ethiopia: an interim report. In M. van der Veen (Ed.), The exploitation of plant resources in ancient Africa (pp. 137148). New York: Kluwer. Boardman, S. (2000). Contributions on archaeobotany. In D. W. Phillipson (Ed.), Archaeology at Aksum, Ethiopia, 19731997 (pp. 268270). London: British Institute in Eastern Africa and Society of Antiquaries. 363-8, 412-14. Boardman, S., & Phillipson, D. W. (2000). Archaeobotany. In D. W. Phillipson (Ed.), Archaeology at Aksum, Ethiopia, 19731997 (pp. 468470). London: British Institute in Eastern Africa and Society of Antiquaries. Buffa, V., & Vogt, B. (2001). Sabir cultural identity between Saba and Africa. In R. Eichmann & H. Parzinger (Eds.), Migration und Kulturtransfer (pp. 437450). Bonn: Habelt. Crumley, C. L. (1995). Heterarchy and the analysis of complex societies. In R. M. Ehrenreich, C. L. Crumley & J. E. Levy (Eds.), Heterarchy and the analysis of complex societies (pp. 15). Arlington: American Anthropological Association. Archaeological Papers, no. 6. Curtis, M. C. (2005). Archaeological Investigations in the Greater Asmara Area: a regional approach in the central highlands of Eritrea. Unpublished PhD dissertation, University of Florida. Curtis, M. C. (2009). Relating the Ancient Ona culture to the wider northern Horn: discerning patterns and problems in the archaeology of the first millennium BC. African Archaeological Review, 26, 327350. doi:10.1007/s10437-009-9062-4. Curtis, M. C., & Schmidt, P. R. (2008). Landscape, people, and places on the ancient Asmara plateau. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 64108). Trenton: Red Sea. DAndrea, A. C., & Haile, M. (2002). Traditional emmer processing in highland Ethiopia. Journal of Ethnobiology, 22, 179217. DAndrea, C. A., Schmidt, P. R., & Curtis, M. C. (2008). Paleoethnobotanical analysis and agricultural economy at early first millennium BCE sites around Asmara. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 207216). Trenton: Red Sea. de Contenson, H. (1981). Pre-Aksumite culture. In G. Mokhtar (Ed.), UNESCO General History of Africa, vol. 2 (pp. 341361). London: Heinemann. Drewes, A. J. (1962). Inscriptions de lthiopie antique. Leiden: Brill. Fattovich, R. (1997a). The contacts between Southern Arabia and the Horn of Africa in late prehistoric and early historical times: a view from Africa. In A. Avanzini (Ed.), Profumi d'Arabia (pp. 273286). Rome: LErma di Bretschneider. Fattovich, R. (1997b). The near East and Eastern Africa: their interaction. In J. O. Vogel (Ed.), Encyclopedia of Precolonial Africa: archaeology, history, languages, cultures and environments (pp. 479484). Walnut Creek: AltaMira. Fattovich, R. (2000). Aksum and the Habashat: state and ethnicity in ancient northern Ethiopia and Eritrea. Working Papers in African Studies 228. Boston, MA: Boston University African Studies Center. Fattovich, R. (2004). The pre-Aksumite state in northern Ethiopia and Eritrea reconsidered. In P. Lunde & A. Porter (Eds.), Trade and travel in the Red Sea region (pp. 7177). Oxford: Archaeopress. Fattovich, R. (2008). Book review: the archaeology of ancient Eritrea. International Journal of African Historical Studies, 41, 346348. Fattovich, R. (2009). Reconsidering Yeha, c. 800400 BC. African Archaeological Review, 26, 275290. Habtemichael, D. (2000). Reconstructing pre-independence Eritrean archaeology: a case study and analysis of bull heads (Senior research paper in Archaeology). Department of Archaeology, University of Asmara.

Afr Archaeol Rev (2009) 26:305325


Marshall, F., & Hildebrand, E. (2002). Cattle before crops: the beginnings of food production in Africa. Journal of World Prehistory, 16, 99143. McIntosh, R. J. (1998). The peoples of the middle Niger. Oxford: Blackwell. Mehari, A. (2008). Ceramic analysis of Sembel site: cultural stratigraphy. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 163177). Trenton: Red Sea. Phillipson, D. W. (2000). Archaeology at Aksum, Ethiopia, 199397. London: British Institute in Eastern Africa and Society of Antiquaries. Phillipson, D. W. (2009). The first millennium BC in the highlands of northern Ethiopia and south-central Eritrea: a re-assessment of cultural and political development. African Archaeological Review, 26, 257274. doi:10.1007/s10437-009-9064-2. Schmidt, P. R., & Curtis, M. C. (2001). Urban precursors in the Horn: early first millennium BC communities in Eritrea. Antiquity, 75, 849859. Schmidt, P. R., & Naty, A. (2008). Bullss heads and enigmas: strong inference and interpretative puzzles in Eritrea. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 235245). Trenton: Red Sea. Schmidt, P. R., Curtis, M. C., & Teka, Z. (eds) (2008a). The archaeology of ancient Eritrea. Trenton: Red Sea. Schmidt, P. R., Curtis, M. C., & Teka, Z. (2008b). The Ancient Ona communities of the first millennium BC: urban precursors and independent development on the Asmara plateau. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 109177). Trenton: Red Sea. Schmidt, P. R., Habtemichael, D., & Curtis, M. C. (2008c). Ancient gold mining north of Asmara: a focus on Hara Hot. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 179187). Trenton: Red Sea. Schmidt, P. R., Haile, M., & Shoshani, J. (2008d). The Emba Derho site: an Aksumite period tomb north of Asmara, Eritrea. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 247274). Trenton: Red Sea. Schneider, R. (1976). Les dbuts de lhistoire thiopienne. Documents pour servir lhistoire des civilisations thiopiennes, 7, 4754. Shoshani, J., Ghebregiorgis, M., & Schmidt, P. R. (2008). Interpretations of faunal remains from archaeological sites on the Asmara plateau of Eritrea. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 217233). Trenton: Red Sea. Teka, Z., & Okubatsion, D. (2008). Lithic artifacts from archaeological sites in the Greater Asmara area. In P. R. Schmidt, M. C. Curtis & Z. Teka (Eds.), The archaeology of ancient Eritrea (pp. 189205). Trenton: Red Sea. Tringali, G. (1965). Cenni sulle ona di Asmara e dintorni. Annales dthiopie, 6, 143161. Tringali, G. (1967). Necropoli di Curbacaiehat (Asmara). Journal of Ethiopian Studies, 5(1), 109114. Tringali, G. (1978). Necropoli di Cascase e oggetti sudarabici dalla regione di Asmara (Eritrea). Rassegna di Studi Etiopici, 26, 4798. Ullendorff, E. (1968). Ethiopia and the Bible. Oxford: Oxford University Press for the British Academy. Usoni, L. (1952). Risorese Minerarie delAfrica Orientale: EritreaEthiopiaSomalia. Rome: Jandi Sapi Editori.