Evidence for Late Pleistocene net-hunting in southern Somalia.

Humans have been farming for about 12 000 years, a period which was preceded by hundreds of thousands of years of living a hunter/gatherer existence. It is logical to assume, therefore, that the social networking skills which made modern civilization possible developed largely before the event of farming. The advent of specialist hunting techniques such as net-hunting, which require advanced planning and co-operation between groups of hunters, and which can produce large amounts of meat in single events, are thought to have been a significant step in the development of Human social networks, connected to the emergence of social hierarchies and concepts of land ownership.

The lifestyle of Homo sapiens in East Africa from our emergence as a species onwards has been heavily investigated for decades, although curiously, despite the importance placed on the role of hunting as a driver of Human evolution by many early archaeologists, the techniques used to hunt have been relatively neglected, with an assumption that big game hunting was the dominant behaviour. Recent studies, however, have suggested that while big game hunting was important on the grassy plains of East Africa, in the coastal forests of the region, remote capture devices, such as snares, traps, or nets, used to target small Animals have been important for at least 125 000 years.

Given that such methods are thought to be closely associated with the emergence of social networks, the nature these hunting practices, and the ways in which they evolved over time, must be seen as an important part of the Human story. Examination of Later Stone Age sites in the semi-arid environment of southern Somalia, have shown that nets were being used to intensively hunt Dwarf Antelopes between about 26 000 and about 6000 years ago.

In a paper published in the journal Archaeological and Anthropological Sciences on 2 December 2023, Mica Jones of the School of Archaeology at the University of Oxford, presents the results of a study of Animal bones from the Late Pleistocene Guli Waabayo rock shelter in the Buur Heybe inselberg cluster, and the implications of this for the use of net-hunting by the Later Stone Age hunter/gatherers of the area.

Inselbergs are hills made from hard volcanic rock, which persists when other deposits are eroded away, typically forming elongate structures. In Somalia such structures are known as 'buurs', and in the inter-riverine region of southern Somalia, where rainfall is typically between 400 mm and 600 mm per year, clusters of these buurs can help to trap nutrient-rich soils and collect rainwater in pools and ponds, providing favourable habitats for a range of Plants and small Animals.

A previous study of the Rife Range Site at Buur Hakaba found that Later Stone Age hunters there specialised in hunting Dwarf Antelope, probably with nets, leading to the development of a more sedentary lifestyle and the holding of territories in the Early-to-Middle Holocene, a period of increased rainfall. However, the small number of remains recovered from this site, combined with an absence of dates, limits the usefulness of this study. The Guli Waabayo rock shelter at Buur Heybe, about 25 km to the northeast of Buur Hakaba, has yielded a much more extensive collection of Animal remains, combined with a series of robust dates, covering a period of 20 000 years, potentially enabling a good test site for hunting practices in the region before the start of the Holocene.

Map of inter-riverine southern Somalia with locations of the Buur Heybe and Buur Hakaba inselberg clusters. Jones (2023).

More than a hundred rock shelters and other archaeological sites have been identified in the inter-riverine region of southern Somalia since the 1930s. However, only three of these sites have been subject to any organised excavation work. The Gogoshiis Qabe site was excavated in the 1940s by the Italian archaeologist Paolo Graziosi. In the 1950s British archaeologist John Desmond Clark investigated the Guli Waabayo and Rife Range sites, establishing that there was a degree of cultural continuity between the two. Following Clark's activities, no further organised archaeological work was carried out in the region until the 1980s, when Steven Brandt and the Buur Ecological and Archaeological Project returned and carried out further excavations at all three sites. These excavations were brought to an abrupt halt with the onset of the Somali Civil War in 1989, although, with the consent of the Somali Academy of the Arts and Sciences, much of the excavated material was exported to the US, where it has been held in the collections of several institutions. 

Jones examined the material brought back from Guli Waabayo to establish the identity of the small Antelope bones present; these had previously all been assigned to the genus Madoqua (Dik Diks), but this is not the only genus of small Antelopes in the Horn of Africa, and, since different Antelope have different ecologies, the identity of these bones has implications for the techniques which would have been effective when hunting them. Next Jones looked at the overall assemblage, and calculated the proportion of remains coming from each type of small Mammal, and then the proportion of each age-set within the overall sample.

Studying hunting methods used to on large prey can be achieved by looking at the weapons involved, or the use and modification of the landscape by the hunters, such as cliff jumps or kite structures. However, the study of how small game is hunted requires less direct approach, as such Animals are often captured with snares, nets, or traps made from perishable materials such as plant fibres, wood, or leather. In order to address this Zooarchaeologists have developed methods of analysing the ways in which people in the past hunted small game, based upon analysis of the variety of remains.

The composition of collections of small Animal remains will vary depending on the hunting method used. Studies conducted in places where net-drives (driving Animals into net traps) are still used, such as the Congo Basin, Australia, and the southwestern United States, has shown that this method will tend to target only one or two species, but will trap almost all members of those species within the target area. This method is most useful for targeting small Animals in forested, bushy or rocky habitats. Pursuit hunting and setting of individual traps such as snares both tend to produce a more diverse range of prey, which will be reflected in the bone assemblage left behind. 

Furthermore, while net hunting captures individuals of all ages, both snare hunting and persuit hunting tend to produce a high proportion of remains from a single age group, since the young of small Animals generally lack the weight to trigger snares designed for adult Animals, and Humans hunting with bows or clubs will generally take a higher proportion of slower, juvenile Animals.

The excavations carried out at Guli Waabayo comprised 11 one metre by one metre pits, each excavated to a depth of 2.5 m. These uncovered a sequence of three Middle Stone Age/Later Stone Age technologies, although, the lower part of the exposed sequence could not be dated due to an absence of Animal remains. Dates were obtained from Ostrich shell fragments for the upper 1.5 m of the sequence, demonstrating at least sporadic occupation of the shelter from 26 000 to 6000 years ago. Two distinct Later Stone Age technologies could be seen within the sequence, the Eibian, which was used during the arid Marine Isotope Stage 2 from about 29 000 to about 14 500 years ago, and the Bardaale, used in the wetter African Humid Period from about 14 500 to about 6000 years ago.

Analysis of the faunal remains found at Guli Waabayo showed that Later Stone Age hunters across the Pleistocene/Holocene boundary targeted a diverse range of game, including large and small Mammals, Birds, Reptiles, and Fish, although they had a clear preference for Mammals massing less than 20 kg. This preference for small Mammals seems to have increased slightly in the African Humid Period, although this change in climate appears to have made relatively little difference to the site's occupants, probably because the African Humid Period was a lot less pronounced in southern Somalia than in other areas.

Of the bones present at Guli Waabayo, 3104 could be identified as coming from small Mammals, with 2111 identifiable to a lower taxonomic level. Of these 1263 have previously been identified as coming from Dwarf Antelope, with the assumption that they originated from Dik Dik. However, this assumption was never actually put to the test, and there is another form of Dwarf Antelope present in the Horn of Africa, the Suni, Neotragus moschatus.

Jones selected all of the first phalanges assigned to Dwarf Antelopes in the Guli Waabayo collection, and compared them to first phalanges from Dik Dik and Suni in the collection of the Field Museum in Chicago. The size ranges of the first phalanges did not overlap, and all of the specimens from the Guli Waabayo collection conformed with the Dik Dik form, although it was not possible to tell which species of Dik Dik they belonged to.

The Dik Dik genus, Madoqua, first appears in the fossil record in the Miocene. There are four living species within the genus, Madoqua guentheri, Madoqua saltiana, and Madoqua piacentinii are endemic to the Horn of Africa, while the fourth, Madoqua kirkii is found from Somalia south to Tanzania and eastern Uganda, as well as on the west coast of Southern Africa in Namibia and Angola. All species favour rocky environments with low thicket vegetation, with the inselbergs of southern Somalia providing an excellent environment, that today supports a large population of Dik Dik, which live in monogamous pairs, with each pair occupying and defending a territory with an area of about 1 km².

A Salt's Dik Dik, Madoqua saltiana, in the wild. David Castor/Wikimedia Commons.

Dik Dik are not widely hunted in Africa today, although in places they are targeted for their meat and skins, which are used to make gloves and cloaks. The Hadza people of northern Tanzania occasionally hunt Dik Dik with bows and arrows, while the Mukogodo people, who lived in the area around Mount Kenya in the eighteenth century are believed to have hunted Dik Dik both with snares and by stalking. 

Modern Somali pastoralists have been recorded to chase down Dik Dik, which seems surprising, given that these Dwarf Antelopes can reach speeds of about 42 km per hour, but is possible because they are so reluctant to leave their territories that they will often run around the perimeter when chased, allowing much slower Human pursuers to keep them in sight until they become exhausted and are easily captured. In the 1940s and 1950s several agropastoralist groups in Somalia were reported to capture Dik Dik in net hunts.

During these hunts, the men used nets about 5 m in length, strung between 2 m long poles. These nets would be used to create semi-circular traps, into which Dik Dik could be driven by men and Dogs, before being dispatched with bows and arrows, a methodology which could conceivably have been used by Later Stone Age hunters in the same region.

Having established the Small Antelope remains at Guli Waabayo came from Dik Dik, Jones next sorted the remains into age groups, based upon examination of complete and partial tooth rows, as well as loose lower 3rd molars and 4th premolars. The dental aging of modern Dik Dik has not been extensively studied, but it was still possible to separate the remains into juvenile, older juvenile, adult, and older adults, based upon the relative eruption and wear seen in the teeth. 

Madoqua mandibles of different age sets from Guli Waabayo. Jones (2023).

Dik Dik remains dominated the faunal assemblage at Guli Waabayo in both Marine Isotope Stage 2 and the African Humid Period. Roughly twice as many remains assignable to Madoqua were found in Marine Isotope Stage 2 selection as in the African Humid Period selection, but this was also true of other Animals present, ab probably reflects the fact that Marine Isotope Stage 2 lasted about 14 000 years, while the African Humid Period lasted only about 6000 years. Nevertheless, the proportion of the assemblage made up by Dik Dik bones did increase, from 55.2% in Marine Isotope Stage 2, to 71.9% in the African Humid Period.

Small Mammal frequencies from Guli Waabayo, presented as % (number of identifiable specimens). Abbreviations: MIS 2, marine isotope stage 2; AHP, African humid period; ka. thousand years ago. Jones (2023).

In both intervals, the age-profile of the Dik Dik remained fairly constant, with all age groups present, but adults making up almost half of the assemblage in both cases, while the other three age groups each made up between 13.3% and 21.3% of the assemblage, although Jones notes that the sample size is very small for the African Humid Period, which may mask greater variation than is recorded.

Dik-dik mortality profiles from Guli Waabayo, presented as % (minimum number of individuals). Abbreviations: MIS 2, marine isotope stage 2; AHP, African humid period; ka, thousand years ago. Jones (2023).

There was a relatively high proportion of Carnivore remains in both Marine Isotope Stage 2 (4.4%) and the African Humid Period (5.4%). This could potentially indicate that the site had at times been occupied by Carnivores (for example Hyenas), and that they would have been responsible for some of the other remains present. Signs of such denning behaviour are considered to include juvenile Carnivore bones, gnaw marks on the bones of other Animals. Of 88 identified small Carnivore bones present in the total collection from Guli Waabayo, only three had unfused epiphyses (two from indeterminate Carnivores and one from a small Felid). In addition a mandible was identified with an erupting third molal (a sign that it came from a juvenile Animal), which could be assigned to a Dwarf Mongoose, Helogale sp.. No bones could be attributed to juvenile medium or large Carnivores, such as Hyenas or Lions, and no neonatal Carnivore bones of any type were found. A single Dik Dik bone from Marine Isotope Stage 2 showed signs of gnawing, but this was clearly by a Rodent Three Dik Dik bones from the African Humid Period also showed signs of gnawing, with two of these again clearly having been gnawed by Rodents. Conversely, none of the Dik Dik bones showed signs of butchery with tools, although this would not typically be expected with the bones of an Animal this small. Signs of burning, another clear indicator of a Human predator, were found on 12.3% of the Marine Isotope Stage 2 Dik Dik remains, and 7.8% of the African Humid Period Dik Dik bones.

Jones's investigation confirmed that Later Stone Age hunters had extensively targeted Dik Dik, small territorial Bovids, during both Marine Isotope Stage 2 and the African Humid Period, something in line with previous research at the site. However, unlike previous researchers, Jones did not concentrate on the toral diversity of remains at the site, but specifically upon the most abundant item, the Dik Dik, with a view to understanding the importance of these Antelopes to Later Stone Age hunting communities in the area. The evidence strongly suggests that the people here became specialist hunters of Dik Dik during the arid Late Pleistocene. 

This is different to the pattern observed at the Riffle Range Site, where a specialisation in net-hunting of Dik Dik appears to have developed during the African Humid Period, alongside a more general shift towards a more settled lifestyle and the permanent holding or territory by Later Stone Age groups. At Guli Waabayo the proportion of Dik Dik among the Animal remains increased in the African Humid Period, suggesting that they became a more important resource during this time, but they were clearly already a major prey species in Marine Isotioe Stage 2. This suggests that the change in climate did not provoke a major change in hunting methods or other behaviours at Guli Waabayo. This raises the possibility that the apparent change in behaviour at the Riffle Range Site 12 000 years ago may be actually indicate that the site was only sporadically inhabited prior to this, or simply that earlier remains have not been preserved. An increase in usage of the Riffle Range Site during the Holocene appears to be the more likely scenario, but why that should be is unclear.

The hunters of Guli Waabayo appear to have taken far more Dik Dik than any other Animal, even similar sized small Mammals, such as Hares or Hyraxes. Thus suggests that they had developed a specialised hunting technique which enabled them to reliably capture large numbers of Dik Dik, bur which did not work as readily on other prey. Furthermore, the Dik Dik remains have a distinct age profile, with all age groups present but adults being the most abundant. Based upon studies of modern Dwarf Antelope hunters, the most likely explanation for this is that communal net drives were used to target Dik Dik on a regular basis, but that other hunting methods were also used to catch other prey.

The selection of faunal remains present at Guli Waabayo makes it likely that both snare and pursuit hunting were also practiced, but the large proportion of Dik Dik in the remains cannot be explained by this. Snares will tend to target a range of Animals of the same size, including small Carnivores, which are present in the Guli Waabayo assemblage, but do not tend to capture juvenile Animals, which are typically too small to trigger them. Had snares been the primary hunting method at Guli Waabayo, then the overall assemblage of Animals would have been more varied, without any single species dominating, and juvenile Dik Dik would have been largely absent. Pursuit hunting, in contrast, tends to produce a far higher proportion of juvenile Animals, which are easier to run down, but again does not tend to produce an assemblage dominated by a single species. 

The bone assemblage from Guli Waabayo was accumulated over thousands of years, making it unlikely that all of the Dik Dik remains there are the result of a single hunting method. Nevertheless, it does appear likely that net hunting played a significant role in the life of Later Stone Age hunters in the region.

The most egalitarian hunter/gatherer societies tend to invest little effort in the manufacture of material goods. However, as societies become more complex, an increasing emphasis is placed upon resource management, and delayed-return technologies, which ensure the future supply of food. Such technologies include both devices used to store food, and to capture Animals for consumption, and are in turn associated with a decrease in mobility, with groups tending to remain in one place for longer. In environments with strong seasonal variations in the availability of food, this can lead to the development of flexible political systems or large resource-sharing networks. 

The Holocene Kansyore people of the Lake Victoria Basin are known to have used pottery and Fish weirs d to exploit seasonal Fish stocks, a specialization which resulted in long-term occupation of sites, and presumably a sense of ownership of those sites. The connection between subsistence, land use, and technology during earlier periods in East Africa is less clear, although the specialisation in hunting Dik Dik in the pre-Holocene Later Stone Age at Guli Waabayo suggests that these people had developed a lifestyle in Marine Isotope Stage 2 which was tied to long term resource management within a single environment, and the use of specialist equipment, in the form of nets, which would have required considerable investment in time to construct and maintain.

The traditional view of Late Pleistocene hunters in East Africa has been one of highly mobile groups which placed an emphasis on the hunting of large game. However, the evidence from s coastal and forested regions suggests that small game may have been a far more important resource than was previously thought, to at least some groups of hunter/gatherers. The use of delayed-return hunting technologies, such as snares, traps, or nets, would have allowed more settled populations in coastal and forested environments, enabling a diversity of survival strategies to survive alongside one another in East Africa during the Late Pleistocene. The evidence for specialist Dik Dik hunting at Guli Waabayo adds to this understanding, giving an example of a community surviving in a semi-arid environment by adopting a more settled lifestyle and a specialist method of hunting.

Most hunting methods target one Animal at a time, but net drives can trap numerous Animals at the same time, potentially leading to surpluses of meat following the hunt. Among modern communities which practice net hunting in the Congo Basin, hunts are typically communal activities, involving men, women, and children, and the re-enforcement of social bonds. In Australia the organisation of net-hunts has been associated with the emergence of political hierarchies, while in the southwestern US the organisation of communal net hunts for Rabbits has been associated with both forming co-operative bonds between different groups, and the establishment of temporary leaders.

No direct evidence of netting has been found at Guli Waabayo, however, the pattern of remains suggests that the hunters here were using communual net-drives to capture large numbers of Dik Dik during the Later Stone Age. Such hunts would likely have resulted in considerable amounts of meat becoming available at certain times, enabling the surplus to be shared among large extended communities.

If this were the case, then Dik Dik hunting would have enabled large communal gatherings, promoting social cohesion between groups and promoting further co-operation. However, the faunal data is not enough in itself to make many assumptions about these ancient societies, and further analysis of Human burials from Guli Waabayo and Gogoshiis Qabe, combined with studies of lithic tools from these sites and the Riffle Range Site, should help us gain a better understanding of these ancient East African communities.

See also...

Follow Sciency Thoughts on Facebook.

Follow Sciency Thoughts on Twitter.

Thirty seven Cholera related deaths in Somalia so far this year.

Between 1 January to 10 July 2022, a cumulative number of 7796 cases of Cholera including 37 associated deaths (a case fatality ratio of 0.5%) have been reported from 25 drought-affected districts, according to a press release issued by the World Health Organization on 20 July 2022. Cholera is endemic in Somalia with recurrent outbreaks reported since 2017. 

The current Cholera outbreak is taking place in the context of other ongoing outbreaks, high rates of malnutrition, and the escalating drought, all of which are exacerbating the fragility of Somalia’s humanitarian situation and have led to large displacement of people, placing additional pressure on an already over-burdened and underperforming healthcare system. Currently, the country has limited capacity to respond to the outbreak, increasing the risk of serious public health impact.

From 1 January to 10 July 2022, according to the Ministry of Health Somalia, a cumulative number of 7796 cases of cholera including 37 associated deaths have been reported from 25 drought-affected districts in Banadir region, South-West State and Hirshabelle State. The districts reporting the highest number of cases include Baidoa (2033), Daynile (1080), Jowhar (825) and Afgoi (681). In addition to being the drought-affected districts, many of these districts also host Internally Displaced Populations in camps resulting from three decades of conflict and the escalating drought situation.

Geographic distribution of Cholera cases in Somalia (%) from 1 January to 10 July 2022. World Health Organization.

Over half of the cases (53.5%) were among children below two years of age, with males and females equally affected. Severe dehydration was reported in 27% of cases.  None of the cases reported having received .Oral Cholera Vaccine  Oral Cholera Vaccine campaigns were carried out in Somalia in 2017, 2018 and 2019.

As of 10 July, a total of 677 stool samples were collected and analysed (by culture) by the National Public Health Laboratory in Mogadishu. Of these, 145 samples (21.4%) tested positive for Vibrio cholerae.

The number of cases reported in the first six months of 2022 has exceeded the number of cases reported in 2021 in the same drought affected districts, when a total of 6205 acute watery diarrhoea/suspected cholera cases including 39 deaths (a case fatality ratio of 0.63%) were reported. In 2021, the outbreak started following flash floods.

Number of suspected Cholera cases in Somalia reported from 1 January 2021 through 10 July 2022. World Health Organization.

Cholera is an acute enteric infection caused by ingesting the Bacterium Vibrio cholerae, a Gram-negative, comma-shaped Gammaproteobacteria, related to other pathogenic Bacteria such as Yersinia pestis (Bubonic Plague), and Esherchia coli (food poisoning), which present in contaminated water or food. It is mainly linked to insufficient access to safe drinking water and inadequate sanitation. It is an extremely virulent disease that can cause severe acute watery diarrhoea resulting in high morbidity and mortality, and can spread rapidly, depending on the frequency of exposure, the exposed population and the setting. Cholera affects both children and adults and can be fatal if untreated.

SEM image of Vibrio cholerae Bacteria. Kim et al. (2000).

The incubation period is between 12 hours and five days after ingestion of contaminated food or water. Most people infected with Vibrio cholerae do not develop any symptoms, although the Bacteria are present in their faeces for 1-10 days after infection and are shed back into the environment, potentially infecting other people. Among people who develop symptoms, the majority have mild or moderate symptoms, while a minority develop acute watery diarrhoea with severe dehydration. Cholera is an easily treatable disease. Most people can be treated successfully through prompt administration of oral rehydration solution.

Cholera can be endemic or epidemic. A Cholera-endemic area is an area where confirmed Cholera cases were detected during the last three years with evidence of local transmission (cases are not imported from elsewhere). A Cholera epidemic can occur in both endemic countries and in non-endemic countries. Uninfected dead bodies have never been reported as the source of epidemics.

The consequences of a humanitarian crisis – such as disruption of water and sanitation systems, or the displacement of populations to inadequate and overcrowded camps – can increase the risk of Cholera transmission, should the bacteria be present or introduced.

A multisectoral approach including a combination of surveillance, improving access to clean water, sanitation, and hygiene promotion, rapid access to treatment, risk communication and community engagement and oral cholera vaccines is essential to control cholera outbreaks and to reduce deaths.

See also...

Follow Sciency Thoughts on Facebook.

Follow Sciency Thoughts on Twitter.

Turtle shells, Frankinsence, and Myrrh: Trade between the nomads of Somaliland and the empires of the ancient world.

Nomadic pastoralists are known to have been important to the economics of the ancient world, operating both within political units such as the Roman Empire and beyond their borders, where they formed an important part of international trade networks. Despite this, these communities have received little attention from historians and archaeologists, compared to more settled groups. This is particularly true of pastoralists living in arid regions of Africa and the Middle East, who traded with, and were economically bound to, entities such as the Roman and Parthian empires, but spent much or all of their time beyond the borders or control of these states. In North Africa the spread of the Roman Empire was linked to a decrease in fueding between pastoralist and settled communities, and integration into a more formal trading system. Beyond the borders of Roman-controlled Egypt pastoral groups appear to have traded with Roman frontier towns, mining communities in the interior, the Aksumite Kingdom on the Red Sea coast, and traders from the Indian Ocean. 

Some archaeological work has been dedicated to pastoralist communities in northeast Africa during the last millennium BC and the first millennium AD, the period during which these communities are thought to have emerged and developed complex systems of interactions with other economies and the natural environments they moved into. This has revealed that these peoples often had significant economic status, with grave goods from nomadic burials frequently containing high-value goods from distant locations. 

In a paper published in the January 2022 edition of the American Journal of Archaeology, Alfredo González-Ruibal and Jorge de Torres of the Institute of Heritage Sciences of the Spanish National Research Council, independent archaeologist Candela Martínez Barrio, Manuel Antonio Franco Fernández, also of the Institute of Heritage Sciences of the Spanish National Research Council, Adolfo Fernández Fernández of the Faculty of History at the University of Vigo, Pablo Gutiérrez de León Juberías, again of the Institute of Heritage Sciences of the Spanish National Research Council, José Yravedra Sainz de los Terreros of the Prehistory Teaching Unit at the Complutense University of Madrid, Michela Gaudiello of the Department of African Studies at Warsaw University, and Somali archaeologist Ahmed Jama Dualeh, present the results of archaeological exploration carried out on first millennium AD sites on the coast of Somaliland between 2018 and 2020, and the insights these provide into the links between the nomadic pastoralist peoples inhabiting the area at the time, and the wider global economy of the ancient world. Somaliland was a province of Somalia until 1991, when it unilaterally declared independence from Somalia at the outbreak of the Somalian Civil War. Few outside countries recognised this independence, but it did serve to protect Somaliland against much of the chaos that enveloped the rest of Somalia.

As well as the trade networks of northeast Africa and the Roman world, the nomadic pastoralists of the Horn of Africa were connected to the Indian Ocean trading system, which reached from northwest Europe to China, and included states such as the Roman and Byzantine empires, the kingdoms of Meroe and Aksum in Northeast Africa, the kingdom of Himyar in South Arabia, the Parthian and Sassanian empires in the Middle East, and the Middle kingdoms and the Gupta empire in India. This trade network began to develop around the end of the last century BC, and is considered to have been fully operational by the middle of the first century AD, surviving until the 19th century, when European settlers took control of much trade in the region. This trade network allowed for the movement of both goods and ideas over huge distances, and played a major role in the development of cultures around the Indian Ocean, and the way in which they interacted with one-another. As with other trade networks, nomadic groups are likely to have played a major role in the operating of this system, although they have been, to date, largely overlooked.

 

Maps showing the area of study: top, map of Northeast Africa and the Western Indian Ocean around 300 AD showing major state actors (all capital letters) and nomadic groups (initial capital); bottom, map of the study area in the northern Somali coast showing places mentioned in the text: (1) Zeila; (2) Berbera; (3) Siyaara; (4) Xiis; (5) Ceelayo; (6) Qandala; (7) Daamo; (8) Ras Hafun; (9) Gugux; (10) Yiroowe. González-Ruibal et al. (2021).

Whilst the Horn of Africa could seem like a backwater in such a global trading network, it is located at the (strategically important) entrance to the Red Sea, as well as being the conduit for valuable goods from the African interior, such as ivory and incense. Outside of the archaeological evidence, our main source of information about trading around the western Indian Ocean in early Roman times is the Periplus Maris Erythraei, an account of this trade apparently written by an anonymous sailor who plied these routes. In this document, the Horn of Africa is referred to as the 'Far Side Ports', with a number of trading centres listed; Avalites which is thought to be modern Zeila, Malaô, probably Berbera, Mundu, probably Xiis, Mosyllon, Arômatôn Emporion ('Spice Market'), probably modern Daamo, and Ôpône, probably Ras Hafun. The first archaeological excavations in the area were carried out by the French traveller Georges Révoil at Xiis, and produced a large amount of Roman glassware, supporting the Periplus Maris Erythraei's account of the importance of the area. No further excavations were carried out until the 1970s, when British archaeologist Neville Chittick carried out excavations at Daamo and Ras Hafun, south of Cape Guardafui, where he found pottery from Rome, Parthia, and India, dating from the first to third centuries AD (Daamo) and the second to fifth centuries AD (Ras Hafun). Pottery thought to be ancient on origin has also been found at the Nugaal River mouth, 250 km to the south of Ras Hafun. No evidence of coastal ports trading with the ancient world has been found south of the Nugaal River, though Hellenistic and Roman coins have been found in Mogadishu and Buur Gabo.

Previous archaeological work on trade around the western Indian Ocean has overwhelmingly concentrated on urban centres and larger states, but, as with other such trade networks, nomadic groups are known to have been part of this system, and in areas such as northeast Africa, to have been important players. Since 2015 archaeological in Somaliland, has attempted to better understand these groups, and has along with Puntland (another area acting as a de facto breakaway state from Somalia, and the area Neville Chittick explored in the 1970s), has provided the most easterly evidence for commerce with the ancient world in Africa to date.

González-Ruibal et al.'s work builds upon previous work by staff from the Department of Archaeology of Somaliland, who have surveyed known archaeological sites across the territory and produced a map of Somaliland's archaeological heritage, although this work did not add to the short list of sites with known links to the outside world left by Georges Révoil and Neville Chittick. González-Ruibal et al.'s work began around Xiis, where the best documented evidence of external trade has previously been found, and was intended to carry on their for several years. However, this work was interrupted by Cyclone Pawan in late 2019 and the COVID-19 pandemic of 2020, causing González-Ruibal et al. to shift their attentions to the area around Berbera, enabling a comparison between the two different regions.

In 2018-19 González-Ruibal et al. carried out an extensive survey of the coastal plain at Xiis. The main site here comprised over 300 structures, interpreted as tombs, which were surveyed and mapped using GPS positioning. Surface material not associated with any structure was also recorded, with extensive material being found on the islet of Xiis and on Ma’ajilayn Hill, which overlooks the site. Four tombs were excavated in 2019, with each stratigraphic layer being documented photogrammetrically. Human remains were subject to osteological analysis by physical anthropologist Candela Martínez Barrio both in situ, and at the Museum of the Department of Archaeology in Hargeisa.

 

General map of Xiis and inset detail of main site, with zone letters (A)–(E). González-Ruibal et al. (2021).

The Berbera region, 210 km to the west of Xiis, was first surveyed in 2015-17, producing a wide range of archaeological sites, dating from the Neolithic to the 19th century, but nothing from the first millenium AD. In 2020 González-Ruibal et al. returned to this area, surveying inland as far as Burao, 116 km from the coast. This survey revealed several pre-Islamic necropolis, with those that produced evidence of long-distance trade being subjected to detailed surveying and mapping.

Faunal remains were sent to the Complutense University of Madrid for analysis by zooarchaeologist José Yravedra, and a selection of pottery sent to the University of Vigo, where they were studied by Roman pottery expert Adolfo Fernández, but all other material was studied in Somaliland, and all material eventually deposited with the Department of Archaeology in Hargeisa.

In the Periplus Maris Erythraei the port of Mundu is described as a place 'where vessels moor fairly safely at the island that lies very near the shore'. This description fits well with modern Xiis, where a rocky island (also called Xiis) lies about 200 m offshore. The local landscape is dominated by the Ma’ajilayn Hill, which is flat-topped and steep-sided, providing good views of the surrounding landscape and seaways. The necropolis here has been known since the 19th century, but was not subjected to systematic surveying until 2017, when archaeologist Carl Phillips carried out a survey for a construction company planning a new road in the area. Phillips' survey supported Georges Révoil's assertion that the area was trading with the wider ancient world during antiquity, as did a preliminary survey by González-Ruibal et al. in February 2018.

González-Ruibal et al.'s survey found a total of 296 tomb-structures. These were distributed in five zones, A and B to the west of Ma’ajilayn Hill, C opposite Xiis islet, and D and E to the northeast of Ma’ajilayn Hill. These tombs are most densely packed in zones D and E, where González-Ruibal et al. concentrated their work. Further tombs can be found along the coast to both the east and the west, with a notable site at Shalaw, where a stele with a Sabaean inscription has been found.

 

Shaláw Sabaean writing, Sanaag. Mire (2015).

The tombs at Xiis fall into two broad categories, circular tombs (78% of the total), generally arranged in clusters, and square tombs 18% of the total), generally arranged in rows, with a small number (4%) of indeterminate tombs, which are most likely badly weathered circular tombs. Excavation of both types of tomb has, to date, only produced single burials (no square tomb has been excavated at Xiis to date, but Neville Chittick excavated a similar tomb at Ras Hafun in 1976). 

 

Square structures: (1) at Ras Hafun Main site, based on an unpublished plan by Neville Chittick; (2) and (3) at Xiis, Zone E. González-Ruibal et al. (2021).

There are 39 square tombs at Xiis. These are typically 7-8 m across, with a perimeter of black volcanic stones, and a central cairn made from sandstone or coral. Seven of these tombs have 2-12 stelae (upright standing slabs), which might be a sign of South Arabian influence, as such standing stones are commonly associated with tombs there. Similar square tombs are known from elsewhere in Somaliland, Puntland, Somalia, and Kenya.

 

One of the square cairns at Xiis. González-Ruibal et al. (2021).

Circular tombs at Xiis are 2.5-10 m in diameter although they are generally smaller where they are more densely packed. Elsewhere, similar tombs can be larger, with circular tombs around Berbera often exceeding 10 m in diameter. These tombs can be 2-3 m high, although smaller tombs are generally only 0.5-1 m high. They have larger stone blocks at their base, providing stability, and rocks of contrasting colour (black volcanic and light-coloured limestone and coral) are used to achieve chromatic effects. 

 

View of the main site of Xiis (Zones D and E). Visible are doughnut-shaped cairns in the foreground (Zone D, Clusters 4 and 5) and square structures close to the sea (Zone E). González-Ruibal et al. (2021).

Most of these circular cairns have a crater or doughnut shape, leading to the term 'looter's pits' being used by Révoil, Chittick, and the local community. However, this appears to be the intended shape of these tombs, and even the central part of the 'pit' is well above the surrounding ground level, with the graves beneath being below ground level. This implies that had the pits been dug by looters, then they would have stopped before reaching the graves, and any grave goods within, and done so multiple times. Moreover, any such excavation by looters would be expected to leave material scattered on the surrounding ground, which is not the case. Similar doughnut- or ring-cairns have been recorded elsewhere in northeast Africa, but are uncommon in Somalia, where they are known only from Xiis, Ceelayo, and Qandala.

 

Tombs of Cluster 4, Xiis. González-Ruibal et al. (2021).

The square tombs are largely restricted to the easternmost part of the Xiis funerary complex (Zone E), while the other areas (Zones A-D) are dominated by circular tombs. There could potentially be a number of reasons for that; the square cairns could have been produced by a different social group, clan, or ethic group, been reserved for members of a different social strata, or have been produced at a different time. González-Ruibal et al. favour the latter explanation, noting that a number of factors favour the idea that the square cairns post-date the circular ones. Firstly, the square cairn excavated by Neville Chittick at Ras Hafun produced goods dated to the third to fifth centuries AD, later than the dates obtained by González-Ruibal et al. for items recovered from circular tombs (first to third centuries AD). Secondly, many of the square cairns appear to have started with a circular shape, then been modified to have a square shape, suggesting that the square cairns are a latter development that somehow evolved from the round cairns. Thirdly, the necropolis appears to have started at the base of the Ma’ajilayn Hill, and spread towards the coast, with the square cairns being closest to the sea, suggesting a later date.

 

Zones D and E of the main site of Xiis showing locations of square versus circular structures. González-Ruibal et al. (2021).

The square tombs of Zone E are arranged in two rows flanking a small valley with an opening towards the sea, forming a sort of funerary avenue, something seen in later medieval cemeteries in Somaliland. Circular tombs are arranged in a series of clusters, each covering about 4500 m², and containing between 8 and 59 tombs. These clusters again may represent different social or kin groupings, but appear to have a chronological order. Clusters 1-5 spread along the base of Ma’ajilayn Hill, with Clusters six, seven, and eight, spreading towards the sea. Investigations of Cluster 3 found a single Roman glass flagon dated to the third century, while investigation of Cluster 4 found numerous Roman amphorae, dated to the first to second centuries, plus Italian terra sigillata, millefiori (mosaic glass), and ribbed glass, all dated to the first century. 

 

Tomb Clusters 1–8 in Xiis. The area marked with an X is an activity area where abundant amphorae, storage jars, and other imports were found. Pink dots mark the locations of excavated tombs. González-Ruibal et al. (2021).

González-Ruibal et al. excavated four of the round tombs at Xiis. 4, one in Cluster 2, two in Cluster 3, and one in Cluster 4.

Tomb 49 was located in Cluster 2, and comprised a simple ring of stones 2.6 m in diameter. Three fragments of green glass were found on the surface, but it was unclear if these were associated with the tomb or were later in origin. Two stratigraphic layers could be found inside the ring; a reddish sand interspersed with numerous small white stones, which covered most of the area, and an oval of greyish sand, which was found to be debris left by tomb robbers. The burial shaft in this tomb was one metre deep, with a Human skeleton in a supine position at the bottom. The only item found in the grave was a Sea Snail shell with a perforation which implies it was used as a pendant, although, since this was found close to the skeleton's feet, it is unclear if it was a deliberate inclusion in the tomb. It is possible that the grave originally contained other goods which were looted, and even that the glass fragments derive from an item that was originally in the grave, but this cannot be asserted with any confidence. Unfortunately it proved impossible to date the bones. The glass fragments on the surface probably date to between the first and third centuries AD.

 

Orthophotos of the process of excavation of Tomb 49, Xiis: (left) cairn; (center) kerb and looting pit; (right) burial pit with skeleton. González-Ruibal et al. (2021).

Tomb 75 is located in Cluster 3, and is a oval cairn 2.7 x 2.0 m, and 57 cm high. This was reasonably intact, but with some evidence that looters had attempted to break into it from its northern edge. The cairn is made up of dark volcanic rock, with a basal layer of stones 60-80 cm in size, upon which the main structure was built of 30-50 cm stones. Removal of these visible stones revealed a layer of lighter coloured coral blocks 40-50 cm thick, and beneath this, another layer of volcanic blocks, again 40-50 cm thick. Beneath this was a layer of white sand with small chunks of coral and seashells, and beneath that a light brown sand, thought to be the original substrate. The white sand filled a pit 1.98 x 0.56 m orientated northwest-southeast, at the base of which was a flat stone at the southeast end of the platform and the remains of a wooden platform at the northwest end, beneath which was a badly decayed Human skeleton.

 

Orthophotos of the process of excavation of Tomb 75, Xiis: (left) cairn; (centre) structure sealing the grave (wooden framework toward the left; stone slab at the right); (right) burial pit and skeleton. Note the bones blackened by contact with decayed wood. González-Ruibal et al.  (2021).

The skeleton was male, about 1.68 cm high, and estimated to have been between 25 and 39 years of age at the time of death, with remarkably good dental health for a skeleton of this age from this time period. It was lying with its head to the northwest, facing south, and the bones were discoloured from contact with the decaying wood. No confirmed grave goods were found in this tomb; three of the shells found in the grave fill had perforations which might imply they were used as pendants, but again their position is ambivalent, making it impossible to know if their placement was deliberate. Not finding any grave goods associated with a burial this elaborate is in itself surprising. Neither the bones nor the wood from this tomb could be dated.

Tomb 120 was also found in Cluster 3, and was a circle of dark volcanic stones 3.7 m in diameter and 1.5 m high. This was made up of blocks of 20-30 cm, resting on a base platform made of blocks of 70-100 cm. Beneath this was a substrate of alternating layers of white and light brown sand, with the while sand having potentially had white gypsum (available locally) mixed into it. The actual grave was found beneath two layers of each type of sand, and comprised a pit 1.60 m long, 0.50 m wide, and 0.67 m deep, with an east-west orientation. The fill of this grave contained a single Clam shell, identified as Anadara uropigimelana, without any modifications. At the bottom of the grave was the skeleton of a child aged 7.5 to 9.5 years, lying supine with the head to the west, facing south. The preservation of the skeleton is poor, with the feet having decayed and left hand being completely decayed, and many other bones partially lost. In addition, it has been partially flattened by the weight of the cairn above.

 

(top) Orthophotos of the process of excavation of Tomb 120, Xiis. (bottom) Section of Tomb 120, Xiis. González-Ruibal et al. (2021).

A range of grave goods were found within this burial. These include a 22 cm tall Roman glass flagon placed next to the head, a a necklace of 24 beads (12 blue glass, six faience, and six bronze), around the neck and an iron anklet on the right ankle. The sex of the skeleton could not be determined, but jewellery is more common in female burials in the Horn of Africa, suggesting that this was the grave of a young girl.

 

Detail of Tomb 120, Xiis, with the remains of the deceased and a Roman flagon to her right by her head. González-Ruibal et al. (2021).

Tomb 153 in Cluster 4 was identified as noteworthy during the 2018 season, and excavated in 2019. The tomb had previously been looted, but had a large amount of material scattered around it, marking it out as being of interest. Because of the damage to the tomb, caused both by the looters and torrents flowing from the slope of Ma’ajilayn Hill, it was impossible to determine its original size and proportions, but it appears to have been a rectangular structure surrounded by a stone ring about 6 m in diameter, beneath which were layers of light and dark sand with stones and sea shells. Only a few fragments of Human remains could be found, including a phalanx.

Despite this disturbance, a surprisingly wide range of goods were found in the burial pit beneath the tomb, which was 1.8 m deep, and contained glass fragments representing at least three monochrome vessels, one of which could be identified as a pillar-moulded bowl, and at least two millefiori bowls. These glass fragments suggest a date from the last century BC or the first century AD. In addition, there were 25 strips in white or red glass paste, thought to have originally decorated a box or casket, and four items of pottery: a small vessel of Mesopotamian Glazed Ware with green glaze, straight neck, and flaring rim; a fragment of Italian terra sigillata; a sherd of a Roman amphora; and a bowl of Egyptian fine ware. There were also two beads, one carnelian and one glass, and a piece of iron. Scattered around the tomb were fragments of many Roman amphorae and South Arabian Organic Storage Jars, though it is unclear if these were removed from the tomb (or a neighbouring tomb) by the grave robbers, or were the results of a funerary feast.

As well as the items found in, or associated with tombs, a number of items were found on the surface at Xiis in locations not directly connected to the tombs. These were found in three main areas, on the Islet of Xiis, the top of Ma’ajilayn Hill, and in an area in the foothills to the southwest of Cluster 4.

The top of Ma’ajilayn Hill produced numerous fragments of cooking pottery and South Arabian Organic Storage Jars, but only a few fragments of glazed ware. In addition, three sections of stone wall, including what appeared to be the corner of a building, were found, although it is unclear if they were linked to the other material at the site. This may imply occasional temporary settlement of the area, which would provide a good view over the surrounding land and sea.

On the Islet of Xiis, a few pieces of pottery were found, including some plain wheel-made items, but no fine ware and no storage vessels, which makes it unlikely goods were ever offloaded here. Again some low walls were found. González-Ruibal et al. suggest this area may have been used as a temporary campsite by merchants arriving by ship.

Finally, an area of boulders, small rock shelters, and outcrops, at the foot of Ma’ajilayn Hill produced a large number of Roman amphorae, South Arabian Organic Storage Jars, a few fragments of wheel-made kitchenware, and a selection of fragments of finer ware, including the remains of various glass vessels, pieces of at least two different millefiori dishes, and the remains of an opaque blue glass vessel with undulating yellow bands, interpreted as a core-formed glass alabastron or aryballos (glass containers used for high value liquid items, such as rare oils or perfumes).

Two of the amphorae fragments had the remains of inscriptions; one large vessel, interpreted as having come from a Koan or Rhodian amphora, has the letters A, Y, and N, while a smaller vessel had the letters Δ and K. A number of locally made items were also found here, including a fragment of a handmade finger-printed rim and three shell beads.

 

Pottery fragment with graffiti on an amphora sherd from Xiis (scale is in cm). González-Ruibal et al. (2021).

Most notable at this site was the discovery of a number of pieces of resin. These are a particularly significant find, as they support the identification of Xiis with the trading centre of Mundu, which is recorded in the Periplus Maris Erythraei as having been an important source for three types of resin, myrrh, frankincense, and mokrutu. These are likely to have come from the Daallo Mountains, to the southeast of Xiis, which are home to three varieties of incense-yielding trees: Boswellia frereana, Boswellia carterii, and Commiphora myrrha. This situation is unique to the Daallo Mountains, nowhere else in the Horn of Africa or South Arabia can these three trees be found growing together, and access to this resource could have made Xiis an important trading centre in ancient times.

 

Resin from the artifact scatter south of Cluster 5, Xiis (scale is in cm). González-Ruibal et al. (2021).

The discovery of such a large amount of imported material at Xiis (where locally made pottery makes up less than 1% of the total assemblage), strongly supports the concept as the location being a trading centre well connected to international networks. The most abundant material found here is Mesopotamian Glazed Ware, being found associated with 46 cairns, followed by various types of glass, associated with 25 cairns, then fragments of Roman amphorae, 19 cairns, and undecorated wheelmade wares, 15 cairns. Seven cairns had glass or stone beads associated with them. Sites away from the necropolis produced large numbers of South Arabian Organic Storage Jars and Roman amphorae, also evidence of imported goods.

Many of the pottery and glass items found by González-Ruibal et al. and Révoil are of high value. Révoil's collecting is known to have been highly selective, collecting high value items to the exclusion of both local pottery and items such as amphorae, but his material still provides evidence of long distance trade, including fragments of at least 11 millefiori dishes and bowls (to which González-Ruibal et al. have added another four), as well as inlays, plaques, and tiles, often highly decorated, which are thought to have originally been set into wooden or ivory boxes or other containers, and most likely produced in Alexandria or nearby. Some palmette-decorated inlays resembles material from the royal tombs at Meroe, dated to the middle of the first century AD. The site also produced blue-green pillar-moulded glass bowls (with González-Ruibal et al. and Révoil finding two each), which can also be dated to the middle of the first century AD.

 

Glass from surface survey in Xiis: (1) plaques and tiles; (2) unguentarium; (3) mosaic glass. González-Ruibal et al. (2021).

The amphorae are mostly of the Dressel 2-4 type, which were produced between the last century BC and the second century AD, and generally used to transport wine. Most of these come from Italy, but there are examples from Cilicia, the Argean region and Egypt. Other amphorae found include examples of the Egyptian AE3, Gauloise 4, and Cilician Agora M54 types, also used for wine. Also present were numerous South Arabian Organic Storage Jars, which were made in India as well as Arabia, and used to transport a range of liquid foodstuffs.

 

A selection of amphorae and storage jars from surface survey in Xiis: (1), (2), (5) Italian Dressel 2-4; (3) Gauloise 4; (4) Egyptian AE3; (6)–(8) South Arabian Organic Storage Jars. González-Ruibal et al. (2021).

This paints a picture of an area with trade links to the Roman Empire, South Arabia, the Parthian Empire, and to a lesser extent, India. Written sources from Antiquity tend to place an emphasis on Roman trade with the rest of the world, but generally took little interest in trade between other regions. This picture was emphasised by Georges Révoil's selective collecting, which placed an emphasis on high value goods from the Mediterranean region, but tended to ignore more common trade goods, but is counteracted by González-Ruibal et al.'s more detailed study, which reveals trade links with a wider sphere.

 

Fine wares from Xiis: (1)–(4) Mesopotamian Glazed Ware; (5) and (6) Italian terra sigillata; (7) Eastern Terra Sigillata A; (8) and (9) Egyptian fine ware. González-Ruibal et al. (2021).

The Berbera region has been trading with the wider world since at least medieval times, and González-Ruibal et al.'s initial surveys found an abundance of sites dating from the eleventh century AD or later, but little earlier material. However, the 2020 survey was able to find some evidence of trade between the region and the rest of the world predating medieval times. The name Berbera derives from Barbaroi, a term used to describe the area in Greek and Roman texts, and which was being used by Arab traders as late as the 14th century. The term Barbaroi has been applied to various parts of Africa historically, but has been used specifically for the Somaliland area since at least the Early Islamic Period. 

 

Map of Berbera with pre-Islamic and medieval sites and shell middens. González-Ruibal et al. (2021).

Historically, the name Malâo in the Periplus Maris Erythraei has been associated with Berbera, although this is much less likely than the association of Mundu with Xiis. Malâo is described as sheltered by an ἀκρωτήριον to the east, a term usually translated as a promontory or cape, something completely absent in the Berbera area. However, Berbera does have a sand spit, which the early twentieth century classical scholar Wilfred Schoff thought made it a candidate for the location of Malâo. González-Ruibal et al. reject this hypothesis, suggesting instead that Malâo refers to Siyaara, 30 km tro the east of Berbera, which does have a distinctive rocky promontory, the only such structure between central Somaliland and Djibouti, and where some evidence for ancient trade has been found. Furthermore, Siyaara has drinkable water available from wells year-round, something which Berbera lacks, and has been an active trade centre since at least the eleventh century AD.

The trading port at Berbera, in contrast, is located on an island, something not mentioned in the Periplus Maris Erythraei. There has been a settlement here since at least the sixteenth century, and probably much longer, although the modern town is built over the older settlement, making it hard to date precisely, though evidence of medieval trading has been found in the area. 

Taking a wider view, however, González-Ruibal et al. consider the entire Berbera region rather than just the modern port, and examine an area running 100 km along the coast from the Issutugan River east to Siyaara, and inland as far as Yiroowe, 130 km from the coast. This area contains no monuments as obvious as the Xiis necropolis, but does contain a number of individual cairns, some of which are thought to be pre-Islamic in origin.

The port of Siyarra is known to have been home to a major seasonal fair from at least the eleventh century till the sixteenth. Surface surveys in this area have produced large amounts of medieval goods, notably imported glass, pottery, and jewellery, but also a few earlier items, notably fragments from three jars from the Himyarite Kingdom, which occupied roughly the area of modern Yemen, between about 100 BC and about 520 AD.

 

Three Himyarite jars found in Siyaara. González-Ruibal et al. (2021).

Siyarra lacks an obvious necropolis, but several cairns are located in the area around the city, at least two of which are clearly pre-Islamic in origin, and two more, one in Siyaara and another on the coast 4 km to the southwest, are located on the shore in such ways that they are conspicuous when seen from the sea, but more-or-less hidden from inland.

 

A pre-Islamic cairn at the site of Siyaara 2 (4 km south of Siyaara), overlooking the Gulf of Aden. González-Ruibal et al. (2021).

Possibly the most significant sites on the Berbera coast are found around Ceel Gerdi, a small hamlet located 22 km to the east of the port of Berbera. Here a series of six shell middens can be found on a six km stretch of coast. Three of these are clearly Neolithic, and have produced pottery material dated to the second millennium BC. The remaining three, identified as Ceel Gerdi 2, Ceel Gerdi 8, and Ceel Gerdi 9, have produced imported artefacts dated to the first millennium AD, with another pottery fragment which may be an imported first millennium AD item, found on the surface in the village itself. These shell middens cover fairly large areas, Ceel Gerdi 2 about 2800 m² in a flat area amid dunes, Ceel Gerdi 8 covers about 1500 m² on a dune top, about 2.7 km to the northeast of Ceel Gerdi 2. Ceel Gerdi 9 is smaller, covering only about 100 m², and is located in a flat area between dunes, to the east of Ceel Gerdi 2 and close to the Neolithic Ceel Gerdi 1. These sites are not stratified, with all material apparently on the surface, which González-Ruibal et al. suggest is a result of nomadic groups returning to the same site, but not the same exact spot, repeatedly over a period of many years. All of these sites yielded handmade pottery, imported pottery, flint flakes and cores, Ostrich egg shells, Turtle shells, Fish bones, and, sea shells, while Ceel Gerdi 2 also produced Animal bones, handstones, pounders, and net sinkers.

 

View of Ceel Gerdi 2 from the northwest, looking inland from the coast. Álvaro Minguito Palomares in González-Ruibal et al. (2021).

Analysis of the shells at part of Ceel Gerdi 2 found that the site was dominated by Spider Conchs, Lambis spp., and True Conchs, Tricornis spp. (both types of large marine Snail), but other areas of the middens appeared to be dominated by other shells, with Ceel Gerdi 8 dominated by the Clam Anadara antiquata, suggesting that tastes or foraging habits varied over time. Fish bones were also common, with bones assigned to large members of the Serranidae (possibly Groupers) were found at Ceel Gerdi 2 and Ceel Gerdi 8, while bones attributed to the Sparidae (Sea Bream) or Lethrinidae (Emperor Fish) were also found at Ceel Gerdi 8. All of these Fish could potentially be caught from the shore, although the Periplus Maris Erythraei describes the people of the Babaroi area as being able to cross the Gulf of Aden to trade with Yemen, implying that they had some form of boat and could potentially have fished from the sea. 

View of shell midden Ceel Gerdi 8 from the southwest. González-Ruibal et al. (2021).

The presence of the bones of Goats or Sheep at Ceel Gerdi 2 suggests that these people were not simple foragers, but also either herding or hunting these Mammals. More perplexing is the presence of larger Bovid bones; the arid climate of the region make it most unlikely that Cattle were being raised here, but large Antelope, such as Oryx, Oryx beisa and Greater Kudu, Tragelaphus strepsiceros, were common in the area until the early 20th century, and could possibly have been hunted. Alternatively, Cattle could have been raised in the mountains of the interior and brought to the coast to be slaughtered. Either interpretation is surprising, as nomadic groups living in this area of Somaliland would have been predicted to have herds of Goats, and possibly some Camels (a Camel metatarsal was found at Ceel Gerdi 2), as was the case with the medieval and postmedieval populations of the area, while large Antelope such as Oryx and Kudu would have been extremely difficult (and dangerous) to hunt with the weapons of the period. This paints a picture of a nomadic community with a mixed economy involving herding, fishing, gathering shellfish, and some hunting. The handstones found at Ceer Gerdi 2 also suggest that some cereals were being ground.

Seven different types of pottery were found in the Ceel Gerdi middens. One of them is thought to have been local, and made up 95% of the material found. This is made from a course brown-to-grey material with abundant large inclusions, from which hand-made bowls were produced and decorated with shell impressions. This is essentially a development of the Neolithic tradition in the area, when similar bowls were made, but with different decoration.

Local handmade pottery from Ceel Gerdi 2 (scale is in cm). González-Ruibal et al. (2021).

Imported pottery items found at Ceel Gerdi 2 include wheel-made one in pink fabric and fine muscovite temper, probably of Mediterranean manufacture, a number of pieces of brick-red pottery, which are compact, hard, and homogeneous in colour, but coarsely moulded and with some muscovite temper, including two fairly complete pots, possibly coming from the Middle Aksumite period; a Himyarite jar similar to those of Siyaara; and a Late Aksumite jug, with black surface and geometric decoration, made from a material with an abundant mineral temper, including volcanic inclusions, which probably comes from eastern Tigray, possibly Adigrat, rather than Aksum itself.

 

 

Imported pottery and glass bangle from Ceel Gerdi: (1) Aksumite jug; (2) Himyarite jar; (3) and (4) handles of unknown provenance; (5) bangle in dark green, almost black, glass, probably Egyptian. González-Ruibal et al. (2021).

Other items at Ceel Gerdi which must have been imported include two pieces of obsidian, a bladelet core and a flake, which probably came from Yemen or northern Ethiopia, and a monochrome dark green bracelet, of a type manufactured in Egypt and the Levant between about 300 and 700 AD. 

These finds do not suggest access to a major international trading network as seen at Xiis, but rather a more limited trade with other areas of the Horn of Africa and southern Arabia, and possibly the Red Sea. The Periplus Maris Erythraei, suggests that the trade goods to be obtained at Malaô were similar to those found at Mundu, incense, and sometimes slaves. It is likely that nomads visiting the coast at Ceel Gerdi could have brought with them incense harvested in the interior to trade, but they could also potentially have traded in goods obtained locally, notably Turtle shells, which were valued in the ancient world and listed as valuable commodities from many Far Side Ports in the Periplus Maris Erythraei. The presence of Turtle remains in the middens at Ceel Gerdi strongly suggests these Animals were being captured here. Ostrich eggs, which are also represented in the middens, could also have been traded; these are known to have been a valuable export from the region from at least the Early Islamic Period.

Exploring inland, González-Ruibal et al. found 22 necropolises with cairns close to natural routes between the coast and the escarpment dividing the coastal plain and the highlands. The largest of these was at Gugux, 65 km from the coast, where 44 burial mounds and a rectangular structure are arranged in a straight line orientated north to south. No surface material was found at this site, preventing easy dating of the locality, but neither was there any sign of any of the tombs having been broken into, which leaves good potential for future archaeological investigations.

 

Map of the pre-Islamic necropolis of Gugux. González-Ruibal et al. (2021).

Another such necropolis was found on the road between the town of Burao and the coast. Investigation of a cairn that had apparently been broken open by looters yielded no material at all, while examination of a 3.5 m diameter cairn not aligned with the main group found a bone protruding at the surface. This was taken for radiocarbon dating, yielding a postmedieval date, suggesting that either the tomb is recent or that the bone, which was found on the upper part of the cairn, is a later intrusion. 

 

A cairn of Galoley necropolis; bones from it were radiocarbon dated to approximately 1649–1916 AD. Álvaro Minguito Palomares in González-Ruibal et al. (2021).

A group of four cairns was found on top of a flat-topped hillock at Yiroowe, 22 km southeast of Burao. Two of these are on the eastern edge of the hilltop, 30 m above the surrounding plain, have diameters of about 12 m, and are made of small stones (averaging about 10 cm). The other two cairns are located on the southern slope of the hill, and are smaller, with diameters of about 4 m. González-Ruibal et al. suggest that the smaller cairns are likely to be younger than the larger ones, since large cairns in this region have never been found to be younger than early second millennium AD, whereas small cairns were still being constructed in at least the 19th century. One of these smaller cairns had previously been broken into, although no material could be found associated with it by González-Ruibal et al. The larger cairns, however, had been disturbed by anthills, bringing some material to the surface. 

 

One of the tumuli at Yiroowe, with tall anthill. González-Ruibal et al. (2021).

The more southerly has several fragments of turquoise glass on the mixed in with its stones, which are believed to have come from a single vessel, although the nature of this vessel could not be reconstructed from the available material. A piece of pottery was found beside the more northerly cairn. This was made from a buff, friable material, and had a crazed green glaze, similar to the Mesopotamian Glazed Ware found in Xiis. González-Ruibal et al. suggest that these cairns probably date to before 700 AD, as later burials in the area would be Muslim, and unlikely to contain grave goods.

Comparison of glazed pottery: (1) sherd from Yiroowe found near the northern tumulus; (2) and (3) Mesopotamian Glazed Ware from Xiis (full scale is 5 cm). González-Ruibal et al. (2021).

Finally, González-Ruibal et al. attempted to locate some of the other Far Side Ports mentioned in the Periplus Maris Erythraei. The port of Zeila has been suggested as a likely candidate for the ancient port of Avalites. However, no pre-Islamic material could be found either around the city or on the nearby island of Sacadin, which has produced large amounts of medieval remains. However, many large cairns flank the natural routes between this area and the Boroma Highlands of the interior, suggesting that this was an important trading area in ancient times.

The port of Mosyllon is described in the Periplus Maris Erythraei as being between Mundu and Arômatôn Emporion. If Mundu corresponds with modern Xiis, as seems likely, and Arômatôn Emporion is modern Daamo, as suggested by Neville Chittick, who excavated Ware and Dressel 2-4 amphorae there, then Mosyllon could be either Qandala, which lies in modern Puntland, or Ceelayo, on the Puntland-Somaliland border. 

Satellite images show a necropolis of about 200 cairns at Qandala, on an escarpment facing the sea. The majority of these are circular or doughnut shaped, as at Xiis, and they range from about 2.5 m in diameter, up to about 17 m (larger than anything at Xiis), though most are smaller than 10 m. Unfortunately, Puntland is not as stable as Somaliland, and González-Ruibal et al. did not feel it was safe to visit this site, although the similarity to the necropolis at Xiis does suggest that the two sites are probably contemporary.

 

Satellite image of part of the ancient cemetery of Qandala (Puntland), marked on the map; probably contemporary with Xiis. Maxar Technologies/Google in González-Ruibal et al. (2021).

Ceelayo has an even larger necropolis, with about 500 tombs extending about 6 km along the coast. These tombs are again predominantly of the circular/doughnut shape, but square cairns similar to those at Xiis and Ras Hafun appear to be present too. Again, the similarities to other sites suggests an age range somewhere in the first half of the first millennium AD.

González-Ruibal et al.'s investigations at Xiis and Ceel Gerdi confirm the ancient accounts of trade in this area, and the earlier findings of Georges Révoil and Neville Chittick, as well as providing new insights into the subject. The site at Xiis appears to have been used between the first and third centuries AD, with trade moving to other locations (possibly quite nearby) after this. At Ceer Gerdi, the material from the middens suggests nomadic groups were visiting this area, presumably at least partially to trade, between the fifth and seventh centuries AD. 

The variety and value of the goods found at Xiis implies that these nomad groups had high purchasing power, something that has also been observed for nomadic groups in other areas, such as Late Roman Egypt, where such groups controlled key trade routes and access to valued materials. Despite this, nomadic groups are often overlooked in both ancient and modern texts, and where mentioned, their importance tends to be downplayed, and the commodities in which they trade dismissed as common items of little value. This perspective was challenged by Georges Révoil's excavations, which showed that the nomads of the Somaliland coast were able to trade for high value items from distant locations, a finding supported by González-Ruibal et al.'s work, which shows that the material recovered by Révoil was not some unique, atypical finding. They suggest that this trade was probably driven by the export of high value incenses.

Furthermore, the similarity of the material found at Xiis to that found at other sites around the Western Indian Ocean, strongly suggests that these nomads were fully connected to the economic and cultural network that linked societies from Rome to India.

The goods imported were not just high value decorative items placed within tombs, as might be assumed from Georges Révoil's work, but also included large volumes of consumable items, such as wine, which have previously been overlooked. The large numbers of empty containers suggest that communal feasting was a part of the culture of these nomads, something seen in other groups living on the fringes of the Roman Empire. The Periplus Maris Erythraei notes that the provision of wine was considered to be important as a way for merchants to foster good relations with Barbaroi in East Africa and India, something that appears to be supported by the Xiis amphorae. González-Ruibal et al. do, however, present an alternative hypothesis for the presence of these amphorae, suggesting that nomadic traders may simply have been transferring the contents of amphorae to animal skins to make transport into the interior easier.

Another noteworthy finding is the individual nature of the burials at Xiis, Ras Hafun, Ceelayo, and Qandala. This represents a significant departure from the Neolithic practice in the region, where communual burials were the norm, as is the case in late first millennium AD cairns in Ethiopia. Whilst more excavations need to be carried out to confirm solitary burials were the usual, or even the most common, practice, this switch does seem to be significant. A long term chronology of burial practices in Somaliland has yet to be developed, but solitary burials appear to be connected to the first few centuries AD, and to the Islamic Period, both of which are associated with the development of long-distance trade networks. González-Ruibal et al. suggest that this might also have been linked to the development of a more stratified society, in which people were aware of social differences, which might in turn be reflected in the rich grave goods seen in some burials at Xiis.

The people at Ceel Gerdi appear to have been subsisting largely on Fish and shellfish gathered from the shore, which may link them to Greek and Byzantine descriptions of 'Ichthyophagoi', a negative image of primitive savages living around the Gulf Aden. Such peoples were depicted as poor, naked, using only stone and wooded tools, and largely dependent on Fish washed ashore in storms for sustenance. However, the fact that people looked primitive to Greek merchants does not imply that they were not capable of engaging in complex trading activities. and the people of Ceel Gerdi were probably trading incense and possibly other items with the Byzantine Empire, Oman, and the kingdoms of Himyar and Aksum.

Modern perspective on nomadic groups in antiquity tend to view the peoples living between Egypt and Somalia as a single homogeneous group. In contrast, Artemidorus, the author of the Periplus Maris Erythraei, clearly viewed the peoples of each of the Far Side Ports as being separate political and cultural entities, although he does not describe these people's cultures, while other ancient sources tended to view these peoples through a lens of negative preconceptions. In the absence of any reliable source of contemporary information, archaeology is the only tool we have to try to understand these (and many other) peoples of the ancient world.

Another noteworthy feature of this research is that the perception of trading always happening at fixed spots, which probably derives from a Mediterranean world dominated by city states (or later largely self-governing cities integrated into larger states), but which does not seem to tie in well with the situation on the Horn of Africa, where, in the absence of a centralised infrastructure that anchored people to specific locations, trading sites could change on small timescales due to changing political or climatic conditions. More localities have now been recorded on the coasts of Somaliland and Puntland than are mentioned in ancient sources, which suggests a variability in trading location unexpected by writers from the Mediterranean region. These locations probably varied with both the movements of nomads responding to seasonal changes and longer term climatic variations which would have necessitated moving to gain access to the best grazing for their herds, and those of sailors on the Indian Ocean, who had to deal with a far tougher climate than those who sailed the Mediterranean, and whose choice of mooring location would be driven by the rhythm of the monsoon winds. This makes it likely that many of the 'ports' listed in the Periplus Maris Erythraei were in fact seasonal fairs rather than permanent settlements, and that these fairs could be moved to new locations if the old ones became unsuitable for some reason.

See also...

Follow Sciency Thoughts on Facebook.

Follow Sciency Thoughts on Twitter.