Introduction

In 1852, the British traveler Richard Burton took a short detour from his route through Somali lands to visit the ruins of a city close to the river Abasa. The local tales talked about Kola, a local Galla queen who ruled this city in the sixteenth century and fought against the neighboring town of Aw Boba, until both cities were consumed in the struggle (Burton 1894:146–147). (A derogative term Oromo people, now rejected as it is considered offensive but included here to respect Burton’s original expression in the text.) The following day Burton visited Aw Boba, where he contemplated the tomb of Sheikh Boba and described two mosques and a large graveyard (Burton 1894:150–151), before continuing south to the Marar Prairies and, eventually, Harar. Almost a century later, a British officer working for the commission in charge of delimiting the border between Ethiopia and British Somaliland (Clifford 1936) studied Darbiyah Kola, Aw Boba and several other sites, ascribing them to the medieval period (Curle 1937). Since then, the number of medieval sites located in surveys (Fauvelle-Aymar et al. 2011; Huntingford 1978; Mire 2015; Warsame et al. 1974) in this region and other areas of Somaliland has grown, but this steady increase in numbers has not corresponded to a better understanding of their historical context and role. The sites often exist just as dots on the maps.

This paper analyzes a cluster of these medieval sites located about 20 km north of the city of Borama, close to the border with Ethiopia (Fig. 1). Three of them, surveyed in 2018 along with other sites around Borama (Torres et al. 2018), documented several of the settlements described by Curle. The three settlements are very close to each other and seem to correspond to a coherent territorial unit, bordering a small mountain range, along two parallel rivers and very close to an important caravan route. The material culture found at these three sites is identical, and the imports found at them establish a common chronology of the fifteenth to sixteenth centuries, based on late celadon pieces, speckled wares, and Yemenite White Cream wares (González-Ruibal et al. 2021:11). This chronology places them at a time when this region was integrated into the so-called Sultanate of Barr Saʿd al-Dīn, a state which had its heyday in the first half of the sixteenth century and is still poorly known archaeologically (Chekroun and Hirsch 2020; Chekroun 2023). The chronological homogeneity and geographical proximity of the sites allow us to study them as an interconnected cluster, rather than as isolated pieces. Despite this close relationship, the sites have significant differences in size, ranging from a proper town (Abasa) to a small hamlet (Iskudarka Dayeergalka Kifiile). They allow us to raise topics such as settlement hierarchization and the political control of the territory, which so far have been absent from the archaeological discussions of the region. This paper combines the results of the 2018 surveys, the materials donated by A. T. Curle to the British Museum, and a topographic and cartographic study of the area, to shed some light on the organization of these settlements during a key period of the history of the Horn of Africa.

Fig. 1
figure 1

Medieval sites of western Somaliland and the nearby region of Ethiopia, showing the area of study

Full size image

The Territory

The medieval sites of western Somaliland are located in the westernmost ranges of the Ogo Mountains, an extension of the Karkaar massif which runs parallel to the coast from the border between Ethiopia and Somaliland to Cape Guardafui. In the west of Somaliland, these mountains are organized in several minor, parallel ranges that run northwest–southeast, reaching heights of 1,800 m and separated by parallel valleys with seasonal rivers that eventually turn north and find their way through the mountains into the Red Sea. The fluvial network depends on the rains which are often torrential, creating numerous deep ravines throughout the landscape. The existence of several parallel mountain ranges hinder movement from the coast to the interior, and the valleys become the key pathways for displacements from one mountain pass to another. Medieval settlements are usually located at the base of mountain ranges, close to the valleys of the main rivers but not in their immediate vicinity. Most of them lie along the Sau range, the southernmost part of the Ogo Mountains which runs from Borama to Hargeisa and opens to the big valleys of the Harrawa and the Wajale.

The three sites studied in this paper (Fig. 2) are located further to the northwest but in a similar position, around a small range of hills—the Wenole hills—surrounded by two parallel wadis (the Abasa and the Okholale) tributary to the Durdur, the main river in this part of Somaliland. Significantly, the settlements are not found in the wider valleys but upstream, maybe to avoid the floods that come regularly between March and May with the monsoons. Hasadinle and Iskudarka Dayeergalka Kifiile are located along the courses of the rivers, while Abasa lies between them, in a curve of the Abasa River which acts as a pass between both valleys and leads to the Harrawa Valley and the end of the mountains. The symbolism of the pass was beautifully described by Burton (1894:147), who, after leaving behind Abasa and climbing the ridge, could see the Harar massif on the horizon and felt at the end of his journey. This strategic position as a hinge between the Abasa and the Harrawa valleys probably explains the size and importance of Abasa, as a benchmark in the routes that connected the coast—especially the important city of Zeila—with the interior of the Horn of Africa. Not only was it the route chosen by Burton’s guides to take him to Harar, but it is marked as a caravan path in twentieth-century maps (GSGS 1939). With slight variations the road traced in the 1930s and still in use today follows the valley where Abasa is located.

Fig. 2
figure 2

Location of the three studied sites. The site of Aw Boba has been included as a reference, although it could not be visited during the 2018 survey

Full size image

However, it is necessary to remark that this strategic position does not necessarily imply a key role in the trade that connected Zeila with Harar and other cities in the Horn of Africa. All the nineteenth-century reports and maps (Hornby 1907; Paulitschke 1884) make clear that the main caravan route from the coast ran farther to the west, crossing the mountains as soon as possible after leaving Zeila to reach the territory of what is now Djibouti, taking advantage of the flatter ground and marching to Gildessa, a key trade hub. That was the original route planned by Burton (1894:90), who had to modify it due to political instability in the region. It is likely that traders in earlier periods also preferred this western route through Djibouti, if we consider the low amounts of imports that have been found in Abasa and the other sites, which contrast sharply with the situation in trade hubs such as Fardowsa to the east (Torres et al. 2020).

On the contrary, archaeological, geographical, and ethnographic evidence points out that agriculture could have been the main activity for Abasa and the other sites. The westernmost mountains of Somaliland are one of the few areas where rain rates are significant enough to allow some cultivation (Hadden 2007:11–12), as was noticed by the early British officers who considered the land apt for the cultivation of sorghum. The colonial reports refer to some cultivation activity in the area, but they fail to mention permanent settlements in the early 1900s (Hornby 1907:11, 48–52). Traditional, small-scale cultivation is today widespread in the area on the banks of the rivers, and they make use of relict meanders which preserve fertility and humidity but are not subject to seasonal floods. The emplacement of the medieval sites and the archaeological evidence found at them suggest that agriculture could have had been significant in the economy of these settlements. The position of Abasa and the rest of the sites seems to have been slightly set back in relation to the most active areas of the states of Ifat and Barr Saʿd al-Dīn, in whose territory they were located.

The Town of Abasa/Darbiyah Kola

The site of Abasa is located 9 km to the southeast of the modern village of Boon, close to the river from which the site takes its current name. Surprisingly, the settlement was not located immediately by the Abasa valley, but set back, 2 km to the south, close to a tributary stream and extending over a plain basin surrounded by hills. It is a large site of about 200 houses scattered all along the basin, without evidence of planned urbanism as is common in the medieval sites of Somaliland. However, some architectonic features point to a relatively complex organization of the space in the town, including the presence of two mosques, two graveyards, and a building with wide walls and a monumental entrance that could be interpreted as a stronghold. All these features have made Abasa the most complex settlement studied so far in the region, and support the statement made by Curle (1937:316) of Abasa being one of the main towns of Somaliland.

The surveys and research conducted by the Incipit team in 2018 have been able to identify Abasa with one of the sites described by Richard Burton during his 1854 trip across western Somaliland. Burton (1894:92–93) accurately portrays his journey along the Abasa River, including the description of a city in ruins:

This is the site of Darbiyah Kola,—Kola’s Fort,—so called from its Galla queen. It is said that this city and its neighbor Aububah fought like certain cats in Kilkenny till both were ‘eaten up:’ the Gudabirsi fixed the event at the period when their forefathers still inhabited Bulhar on the coast,—about 300 years ago. (…) Remnants of houses cumber the soil, and the carefully built wells are filled with rubbish: the palace was pointed out to me with its walls of stone and clay intersected by layers of woodwork. The mosque is a large roofless building containing twelve square pillars of rude masonry, and the Mihrab, or prayer niche, is denoted by a circular arch of tolerable construction. But the voice of the Muezzin is hushed forever, and creepers now twine around the ruined fane. The scene was still and dreary as the grave; for a mile and a half in length all was ruins—ruins—ruins.

The name of Darbiyah Kola is today completely lost to both the local population and the staff of the Department of Archaeology of Somaliland, who knew of the site but called it Abasa after the closest geographic feature. A “ruins” sign with the name of Abasa appears in the British maps made in the 1930s when the border between Ethiopia and British Somaliland was established, and that has been the toponym of the site since then. However, an earlier Italian map dated 1894 (Ministerio della Guerra 1894) marks a site named Darbija-Cola in this area, a toponym that is reproduced in a British map printed in 1926 (GSGS 1926) with the name of Darbiya Kola. This information, along with some straightforward coincidences between Burton’s descriptions and the archaeological evidence gathered by the Incipit team in 2018, identify beyond doubt Abasa as Darbiyah Kola, the place described by Burton in 1852. In this article, we will use the name of Abasa to follow the current nomenclature of the Department of Archaeology of Somaliland to lessen confusion with the rest of the bibliography which has systematically used this name for the site.

Close to the border, Abasa was studied during the mid-1930s by A. T. Curle, who accurately described the main mosque of the site and took some photographs of the now-collapsed mihrab (Curle 1937:319, plate III). Seven additional photographs taken by Curle were donated to the British Museum (Refs. Af,B53.15; Af,B53.17; Af,B53.18; Af,B53.20; Af,B53.22; Af,B53.25; Af,B53.26), most of them are shots of the mosque. It is not clear if Curle excavated at the site, but the materials he collected were also donated to the British Museum, where they remain. Since Curle, the site has been regularly visited by researchers: in 1971 the site was visited by the joint Soviet–Somali expedition (Warsame et al. 1974:323), and four years later, by Neville Chittick (1976:128–129). In 2001, a third visit was made by François-Xavier Fauvelle-Aymar and Bertrand Hirsch (Fauvelle-Aymar et al. 2011:37–40), who confirmed the progressive deterioration of the site and learned about a second settlement with the same name which they were unable to visit (2011:40). That site, known as Abasa 2 by the Department of Archaeology of Somaliland, was documented by the Incipit team in 2018, and is presented in this article renamed as Iskudarka Dayeergalka Kifiile—the name of the place according to the information provided by locals during the visit. Finally, Abasa is briefly mentioned by Sada Mire (2020:89, 167, 274), who refers to Burton’s trip and mentions the use of wood in the buildings’ construction.

The Site

Abasa extends over 56.5 ha in a basin running slightly to the north and surrounded by hills to the south and the east (Fig. 3). To the west, the site is delimited by a deep ravine caused by a stream that runs south–north and then turns east to join the Abasa River. This demarcation is not a radical one, though, as some buildings have been located to the east of the ravine. The houses appear widely scattered throughout the basin; in some cases they are 5–6 m apart, although there could have been other buildings built in perishable materials and now lost which once occupied these empty spaces. The whole area is densely covered by shrubs, acacia trees, and cacti, making the identification and recording of the structures difficult. During the 2018 survey, 97 houses and several other structures were located and georeferenced, although the total number of houses is undoubtedly higher and will likely be closer to the 200 estimated by Curle (1937). In addition to these houses, several singular buildings also have been documented, including two mosques located to the southeast of the site, and a possible stronghold to the west. The remains of other unidentified structures are also visible throughout the site as well as some accumulations of ash and archaeological materials that could correspond to dump heaps, some of which were tentatively identified as market areas by Fauvelle-Aymar et al. (2011:40). Two graveyards were located during the survey, to the north and the southwest of the site.

Fig. 3
figure 3

Schematic representation of Abasa/Darbiyah Kola showing the main features

Full size image

The houses of Abasa have a rectangular or square plan, with sizes varying but usually around 20 to 40 m2 with partition walls clearly visible, defining two or three rooms per house. The walls are made of well-laid flat stones of medium size, bound with simple mud mortar. In general, the state of preservation of the structures is poor, and it has dramatically worsened in the last decades, according to what Curle’s (1937) photographs show, but exceptionally some houses still stand about 1.5 m high. In general, the archaeological deposits look well preserved, with only occasional looting holes documented inside some houses.

The Main Mosque

Since Burton’s first description of the building in 1854, the main mosque of Abasa has been described and documented by all the researchers working on the site. Thus, we have relatively good knowledge of its original shape and the progressive deterioration of this remarkable building. The description made by Burton (1894: 92) of a “building containing twelve square pillars of rude masonry, and the Mihrab, or prayer niche, is denoted by a circular arch of tolerable construction” was relatively accurate, with minor errors in the shape of the columns. A more systematic description was made by Curle (1937:319), who also took the first photographs of it (Curle 1937:plate III; British Museum collection numbers Af,B53.15; Af,B53.17; Af,B53.18; Af,B53.20; Af,B53.25; Af,B53.26). In 1975, Neville Chittick drew the first sketch plan of the mosque and confirmed the collapse of the mihrab arch that was standing 40 years before (Chittick 1976:128–129). Decades later, Francois-Xavier Fauvelle-Aymar and Bertrand Hirsch (2011:37–40) confirmed the information provided by Curle and Chittick. In 2018, the Incipit team made the first accurate plan of the mosque using a drone and conducted a small test pit in the interior of the building (Fig. 4).

Fig. 4
figure 4

View of the remains of the main mosque of Abasa/Darbiyah Kola

Full size image

Unlike most of Somaliland’s medieval settlements, the main mosque of Abasa is not located in a central position in the town, but to the east side of it. It is a square building of 18 × 17 m in size, with a slightly trapezoidal shape, and surrounded by a lower wall, undetected in previous surveys, which probably defined a courtyard for ablutions or provided extra space during prayers. The main entrance was to the south, aligned with the mihrab, with three additional access points from the east, west, and north. The interior of the mosque has 12 columns distributed at even spaces in three rows, oriented east–west. Contrary to Burton’s description, the pillars are not square but circular or cruciform, and have slightly different dimensions. The columns in the central line are all cruciform, with the majority of the other two rows made of circular pillars, except for the column located in the southeastern corner (which is cruciform). Horizontal timber beams can be seen inserted in all the pillars (Mire 2020:167), probably used to reinforce the structure, although their use as hooks for draperies has also been suggested (Curle 1937:319). The rows of pillars are not perfectly aligned, showing a somewhat careless construction already described by Chittick (1976:128), who noted that in one case the upper part of a cruciform column was considerably thicker than the bottom.

The mihrab, one of the most remarkable features of the building, is now completely collapsed, but the photographs taken by Curle allow us to make a good reconstruction of its shape and characteristics: it protruded 2 m from the qibla and had a rectangular shape to the exterior (Af,B53.26), while to the interior it was shaped with a series of increasingly lower arches. The mosque lacked a minbar or minaret, two architectonic features that are persistently absent in the medieval mosques of Somaliland, except those documented at Zeila (Fauvelle-Aymar et al. 2011:34). A total of nine niches were documented, distributed regularly along the walls. Regarding the roof, the lack of arches points to a flat roof supported by horizontal beams, as suggested by Curle (1937:1209), and two photographs taken by him (British Museum col. Number Af,B53.17) and Chittick (1976:plate XIV, a) point to the existence of square capitals on top of the columns, under which the remains of wood beams could still be seen. Another of the British Museum photographs (Af,B53.25) shows a small rectangular window with a blind arch over it, the only documented evidence of this type of architectural feature in the medieval archaeology of Somaliland.

The 2018 intervention included the excavation of a test pit inside the mosque, aiming to document its floor and foundations, and the existence of previous, older mosques underneath, as is common in many cases in Muslim settlements. The 2 × 2 m test pit was placed in an area relatively clear of rubble and vegetation, partially over one of the cruciform pillars of the central row of columns, to better document the relationship between these structures and the floor of the mosque. The stratigraphy documented during the excavation was extremely simple. After the rubble and vegetation were cleared, three stratigraphic units were documented: a surface level with a thickness of 10–15 cm corresponding to the surface layer; a thicker 25 cm stratum of medium- or small-sized stones, corresponding to the collapse of the roof or the upper part of the pillars. Below this collapse layer a thin (4–8 cm) layer was found, identified as the soil deposited immediately after the abandonment of the site. This level yielded the only archaeological piece found during the excavation: a bronze fragment of a bangle, 3 cm long, with a triangular section and decorated with impressed circles.

Below this post-abandonment layer the limestone base was found, with evidence of leveling work to provide a flat platform upon which the mosque could be built (Fig. 5). The cruciform pillar which was part of the test pit had a squared base of about 15 cm high, with stones placed directly on the rock floor. Once this 15 cm high square was set, the cruciform shape was raised, keeping the original square as a reference for the total dimensions of the pillar. No foundations for the columns have been found, and the square platforms of the pillars were most probably visible, although the rock floor could have been covered by mats.

Fig. 5
figure 5

Test at Abasa showing the leveled stone floor and the base of the pillar

Full size image

The work conducted in the main mosque of Abasa has fully documented this important building, amending minor mistakes in previous descriptions and adding information about other structures such as the surrounding wall. It has also raised some questions about its chronology and role within the story of the site. Given its size, this building has been interpreted as the main mosque of Abasa, but its peripheral position somehow contradicts this role, as mosques are usually in a more central position. The lack of previous structures underneath—a common feature in main mosques—also points to a later period for the construction of this building in the history of the town. With the available data, our more plausible hypothesis is that this mosque was built at a time when the settlement was being remodeled, either because the town became Islamized (if we accept the existence of a pre-Islamic Abasa, as the story recorded by Burton suggests) or because it went through an expansive period that demanded a bigger prayer building. Regardless of the reason, at that time the central area of the site would already have been occupied by other buildings, and the requirements of this large mosque would have forced its emplacement in a peripheral position, although there was obviously a significant investment of work in the erection of the building.

The Secondary Mosque

During the 2018 survey, the Incipit team documented a second mosque located at the easternmost point of the site. This mosque was much smaller, set on a platform of large, flat stones, and with architectural differences from the main mosque. These differences are especially evident at the gate, which was built with two slabs at each side acting as doorjambs, and a well-preserved threshold made of carefully placed flat stones. This use of slabs differs from the squared gates made with flat stones common elsewhere in Abasa and other medieval sites. The interior of the mosque is small—4.1 × 3.8 m—and it has no pillars inside, as they were not needed to support the roof of the building. As is the case with the rest of the medieval mosques in Somaliland, the mihrab protrudes from the qibla with a square shape, although in this case no arches were documented in the interior of the structure. Leaving aside the attention paid to the entrance, the mosque presents in general a less planned outline, with walls slightly curved and no proper right angles at the corners. An enclosing wall, better preserved than the one of the main mosque, was also documented (Fig. 6).

Fig. 6
figure 6

Main (top) and secondary (bottom) mosques at Abasa

Full size image

The existence of more than a mosque is a rare feature in the medieval sites of western Somaliland, where so far it has only been documented in Dameraqad, a religious center (Cornax-Gómez 2023); Torres et al. 2018:34. They are common in other medieval sites in the region, where they have been interpreted as neighborhood mosques or as the prayer building of a specific community, such as foreign traders or followers of a specific Muslim school (Pradines 2017:16–18). We cannot rule out two different chronologies for the two mosques found at Abasa, although the small size of the second one and its outright peripheral position make it difficult to believe that it was at some point the only mosque in the town. As with the main mosque, only a proper excavation of the interior and the surrounding area can shed light on the role of this building.

The Monumental Building

As mentioned above, Abasa’s western limit seems to be defined by a 5 m-deep ravine running south–north, which acts as a natural barrier in this direction. This barrier is not fixed, as some buildings can be found on the western side, among them a large building (Fig. 7) whose function is undetermined, and which was never mentioned by previous visitors. The building is located on the northwest side of the settlement and consists of a long room 31 × 8 m (248 m2) in size, oriented northwest–southeast, and with the walls preserved to about 1 m in height and 1 m in width. The entrance was located to the southeast and consisted of a long, slightly curved corridor running north–south and built with big flat stones that lend a certain monumentality to the entrance (Fig. 8). Two roughly square rooms protrude from the building to the east, one at each side of the main room and resting on an outcrop. The northern room was smaller (12 m2), while the southern one measured about 37 m2; its southern wall was part of the access to the main room. The area occupied by the building was approximately 300 m2.

Fig. 7
figure 7

Orthophoto of the monumental building at Abasa

Full size image
Fig. 8
figure 8

Monumental entrance to the monumental building

Full size image

It is difficult to determine the functionality of this building because no excavation was conducted in this area and no materials were found during the survey. The size, the construction technique, the big boulders used for the walls and gate, and the distribution of the rooms rule out its being a common house. The monumentality of the building could point to a military use as a stronghold, but this interpretation is inconsistent with its isolated location, separated from the settlement by the ravine. Another interpretation could be that of a monumental building related to the civil administration, either the house of a local sheikh, a trade control post, or a prison, or maybe a mix of all these different functions. Regardless of its function, it was most likely a public building and as such it could be one of the few pieces of evidence of urban management over the household level found so far in the medieval settlements of Somaliland.

The Graveyards

Two cemeteries have been documented in Abasa, one to the north and the second to the southwest of the site (Fig. 9). Both are relatively small, covering an area of 0.26 ha and 0.20 ha respectively. Significantly, each of the graveyards presents differentiated types of tombs: the cemetery to the north is composed exclusively of square or rectangular graves delimited by straight slabs—in some cases two parallel ones—which delimit a cist where the body was placed. In some cases, two or three slabs run perpendicular to the tomb and divide the interior into two or three parts. In other cases, the upper part of the slab has been lowered to present a small step to the interior, maybe to place horizontal slabs to cover the tomb. No stelae have been found marking the tombs, but they are unmistakably Muslim, oriented east–west so the deceased can be placed facing north and toward Mecca. Tombs appear scattered throughout the cemetery, sometimes following irregular rows. Some sections of small walls were found in the area, but it was unclear whether they surrounded the whole graveyard or just some of the tombs.

Fig. 9
figure 9

Views of the north (top) and southwest (bottom) cemeteries of Abasa. The upper one shows the canonic cist-like graves common in most medieval sites of western Somaliland, and the lower one shows the small slabs type, with a Muslim cairn behind it

Full size image

The type of tomb described above seems to have been the usual kind of burial for Muslims in medieval Somaliland, and they have been documented in many other archaeological sites of this period such as Fardowsa, Aroqolab, Dameraqad, and Kabab (Cornax-Gómez 2023:64; Torres et al. 2018:32–33). At the southwestern graveyard, this type of tomb is also found, but it represents a minority, the most common tomb consisting of a series of three or four plain slabs oriented east–west, placed one along the other in a single line, likely marking one of the longer sides of the tomb. The variety of graves at this second graveyard is complemented by the existence of several cairns with two stelae the top. These stelae seem to mark the head and feet of the deceased, following Muslim funerary practices.

As is the case with the two mosques, the presence of two different graveyards is difficult to explain without a comprehensive study of the site and the excavation of tombs. The two most common explanations could be either the diachrony of both sites (one being older than the other) or the synchronic use of both cemeteries by different groups. The first explanation would be coherent with the expansion of the site and the construction of a new mosque, while the existence of two different graveyards used by different groups could also make sense with the existence of different mosques at the settlement. It would also incorporate quite well the presence of nomads in the region—cairns being the most widespread type of burial for nomadic groups in Somaliland.

The Material Culture

The earliest references to the material culture of the medieval towns of western Somaliland were provided by A. T. Curle, who collected a significant sample of materials in about 20 sites in the region and donated them to the British Museum. With the help of specialists, he studied most of the imported materials and published brief notes about their origin and chronology (Curle 1937). His publication focuses mostly on the imported materials, stating that celadon and a piece of Chinese Ming ware were found at Abasa as well as coarse, possibly wheel-thrown pottery (Curle 1937:320). The sample from Abasa was completed with four pebbles used as hammer stones and a barbed arrow tip made of iron (Curle 1937:322). Later surveys confirmed the types and chronologies suggested by Curle: Chittick mentions Islamic and Chinese pottery dated in the fifteenth to sixteenth centuries (although a ribbed celadon could date back to the fourteenth century) (Chittick 1976:129). Fauvelle-Aymar et al. (2011:40) did not find any Chinese or glazed pottery during their survey, although they documented several pierced cowries and a type of white pottery different from the local, handmade wares.

Although all the publications remark on the abundance of artifacts scattered throughout Abasa, the descriptions of the materials point to a low presence of imported wares in the town. That was indeed the case in the 2018 survey, when the number of imports was astonishingly low, considering the size and importance of Abasa and its position near a major route. After several days of work at the site, the findings of imported pieces consisted of just five pieces of pottery, three fragments of softstone objects, and several cowries. One of the pottery sherds was a small fragment of celadon of unknown chronology (Fig. 10, 4). The second one was a fragment of speckled ware (see Fig. 10, 3), the most common type of glazed pottery in late medieval sites in Somaliland, to the point that it can be considered a kind of “fossil type” for the Adalite period (González-Ruibal et al. 2021:11) and dating to the fifteenth to sixteenth centuries. The third fragment of glazed pottery was a blue and white fritware of good quality (Fig. 10, 2). Although fritware started to be produced around the fourteenth century, its heyday seems to have been the fifteenth to sixteenth centuries (González-Ruibal et al. 2021:11). Finally, two fragments of Yemenite White Cream ware with incised and punctuated decoration were also documented during the survey (see Fig. 10, 5). This type of pottery is also very common in fifteenth-to-sixteenth-century settlements in Somaliland. The sample of imports included a group of eight pierced cowries and a small sample of softstone objects, including a bead and a large fragment of an unidentified vessel perforated and decorated with molding (see Fig. 10, Nos. 9 and 11). Although scarce, the sample offers a consistent chronology of the fifteenth to sixteenth centuries and provides a good summary of the most common imported wares of the Adalite period.

Fig. 10
figure 10

Imports found at Abasa (1–5, 9, 11), Hasadinle (6–8, 10). 1. Bronze bangle 2. Fritware 3. Speckled ware 4. Celadon 5. Yemenite White Cream ware 6–8. Softstone tokens 9. Bead 10–11. Fragments of softstone vessels

Full size image

Aside from these imports, the vast majority of the archaeological materials gathered correspond to local, handmade pottery of good quality, but a reduced number of types (see Fig. 11). Most of the vessels can be classified into three main types: open bowls with the rim slightly enlarged, globular vessels with a short straight neck, and spherical-like pieces with an almost horizontal, flat rim. Handles are abundant, either horizontal, curved handles with oval sections or smaller vertical handles with circular sections. Bases are scarce, but those found are ringed. Decoration is usually scarce and limited to the neck or the upper part of the rim. It usually consists of simple incised designs (a series of parallel horizontal or oblique lines). Only in very few cases are other types of decorations present, such as nail incisions or clay appliqués. Millstones, either saddle querns or rotatory, could be found throughout the site, evidencing the importance of agriculture in an area at the time.

Fig. 11
figure 11

Local pottery of Abasa

Full size image

The Village of Hasadinle

Hasadinle is another of the medieval sites located along the border with Ethiopia. Hasadinle was regarded as a minor site (Curle 1937:316) with no relevant architectural or archaeological features, therefore no further information was provided beyond the height of the walls (Curle 1937:316) and the presence of celadon (Curle 1937:320). No photographs or archaeological materials associated with Hasadinle have been found in the collections of the British Museum, and unlike Abasa, the site seems to have been ignored by the archaeological teams that visited Abasa. In 2018, the Incipit team located the site and documented its main features.

Hasandile is about 20 km northwest of the town of Borama and 9 km to the southeast of the modern village of Quljeed, right by the border with Ethiopia. The site is located in a very similar position to that of Abasa, in the proximity of a big wadi (the Okholale) but slightly set back, at the base of a group of low hills that give the site its name, and with a deep ravine delimiting the site to the east. According to a 1939 map (GSGS 1939), two wells with permanent water exist in the area, although they were not located during our survey. The area is suitable for agriculture, and many small farms can be seen throughout the landscape. In a straight line, the site is close to Abasa (about 6 km), although both settlements are separated by the Wenole mountains which surround Abasa. A pass among these mountains can be seen from Hasandile, which probably was the fastest path between both sites when they were inhabited.

Contrary to what was expected from Curle’s report, when the Incipit team visited Hasandile it found a large village of about 60 houses of good size, occupying an area of about 11 ha (Fig. 12). The area was densely covered with acacia trees, shrubs, and cacti, making the identification and documentation of the archaeological remains difficult, especially at the upper part of the site (Fig. 13). The concentration of houses is higher close to the hill, while those on the plain are more scattered, distributed in rough lines following the level curves. In general, buildings are in a worse state of preservation than those of Abasa, standing just 40–50 cm above the ground. However, the interiors of the houses are more visible, and it has been possible to determine the internal partitions of many of them. They are usually rectangular houses oriented east–west, divided into two (and, more exceptionally, three) rooms, with the doors aligned in the middle of the room. The dimensions of the houses vary, but they are on average 11 × 7 m, with walls measuring about 60 cm. Some houses are significantly bigger: one of them measured 14 × 9 m for an area of 125 m2. 

Fig. 12
figure 12

Plan of Hasadinle

Full size image
Fig. 13
figure 13

View of Hasadinle from the north. The mosque and the main cluster of buildings are at the bottom

Full size image

Interspersed throughout the site are several circular structures defined either with flat slabs, each around 4–5 m in diameter (Fig. 14, top). The function of these circular structures is unknown, although according to local informants, stone circular structures are used in nearby areas as a base for haystacks, preventing the construction of termite mounds. These haystacks—without the stone platform—could be seen in the nearby houses of farmers (Fig. 14, bottom), reinforcing this preliminary interpretation. The agricultural vocation of the site is supported by the wide alluvial plain in front of the houses, where cultivated parcels are still visible, and by the number of querns found throughout the site.

Fig. 14
figure 14

Circular structure at Hasadinle (top) and contemporary sorghum haystack (bottom)

Full size image

The most important building in Hasandile is the mosque, which Curle did not document. It has a remarkable size—11 × 13 m—and in general is in a very poor state of preservation. The mosque was built on the most densely occupied part of the hill, on an artificial platform prepared to counter its slope. As was the case in Abasa, it was surrounded by a wall, and the mihrab had an external square shape protruding from the qibla wall. At the interior of the mosque, the bases of four square pillars (65–70 cm in width) were documented. The dimensions, architectural features, and building technique of the mosque at Hasandile relate it directly to the two mosques of Abasa, and point to an interesting standardization of this type of building in the Sultanate of Barr Saʿd al-Dīn. As for the cemetery, it could not be located during the survey.

The material culture collected in Hasadinle is almost completely of local origin, and no glazed pottery, celadon, or porcelain were found at the site. Only three pieces of pottery can be considered imports: a wheel-thrown body sherd of unknown provenance, and two pieces of polished pottery, one of them a small knob, which very likely have a northern Ethiopian origin. The medieval and modern wares of the Ethiopian highlands have a characteristic polish which makes them easily identifiable (Torres 2017), and the knob found in Hasadinle is identical to those used for beer glasses identified in sites in northern Ethiopia. The presence of pottery from the north of Ethiopia is not surprising considering the proximity of the two areas and the long (and often conflictive) history of diplomatic, political, and economic relationships between the Christian kingdom of Ethiopia and the Muslim sultanates of the Horn of Africa. However, the lack of proper studies about Ethiopian wares and a preference for international trade imports have made these pieces generally go unnoticed. In small numbers, they are starting to be documented in medieval sites in Somaliland such as Fardowsa (Torres et al. 2020), and the collection Curle donated to the British Museum included two fragments of this ware from unknown provenance (registration numbers Af1935,0709.82 and Af1935,0709.83). The sample of imported materials is completed with nine cropped cowries and four pieces of softstone, including a bead and two circular small stones, which could be interpreted as game tokens.

Regarding the local materials (Fig. 15), the handmade pottery found at Hasadinle was abundant and very similar to that of Abasa in terms of technique, shape, wall treatment, and decorative patterns. The main difference with Abasa’s sample is the variety, which includes several types of bowls with outward or beveled rims, pieces with knobs, and containers of larger size than those of Abasa. A remarkable amount of lithic tools was also gathered, including an arrowhead and several well-made scrapers. As with Ethiopian wares from the highlands, not much attention has been paid to lithic tools in medieval Somaliland, although they appear recurrently in the sites and probably were commonly used in daily activities. Along with the flint implements, many fragments of hand and rotatory querns and hand stones were found throughout the site, reinforcing the idea of Hasadinle as a rural settlement with few connections to the main international trade routes.

Fig. 15
figure 15

Local pottery of Hasadinle

Full size image

The Hamlet of Iskudarka Dayeergalka Kifiile (Abasa 2)

The site of Iskudarka Dayeergalka Kifiile does not appear on Curle’s maps, but its existence was well known both to the officers of the Department of Archaeology of the Somaliland government and to the local authorities in Boon. The site was generically named Abasa 2, but a local name of the area was selected during our visit to avoid problems of identification with the main site of Abasa. That name was “Iskudarka Dayeergalka Kifiile,” or, literally, “the mountain where monkeys meet,” referring to a rocky, prominent hill which is in front of the site. The settlement is just about 8 km to the north of Abasa, following the river Abasa to the northeast and around 5 km to the east of the village of Boon. As with Abasa and Hasandile, the site is located slightly back from the main wadi, on the upper part of a small hill, and delimited to the south by a small but deep ravine created by a tributary of the Abasa. It also has a good cultivation area in the proximity, just 1 km away.

Iskudarka Dayeergalka Kifiile is a small village of just about 20 houses covering 1 ha, located at the top of a hill and its immediate slope, with the buildings at the top concentrated while those on the slopes are more scattered (Fig. 16). The structures are similar to those of the previous sites, with rectangular buildings with internal partitions defining two or three rooms. The construction technique is, however, much less elaborate than in Abasa, with medium-sized stones of different shapes arranged more irregularly instead of the carefully laid flat stones used in the larger settlements. The size of the houses is also remarkably smaller, with average dimensions of about 6–7 m in length and 4–6 m in width. The state of preservation of the buildings is poor and similar to that of Hasadinle, and, as in the other sites, the ground is covered by shrubs and cacti, making the survey and the documentation of structures challenging. Sometimes the houses are attached to each other in pairs. The orientation of the buildings is more irregular than in the previous sites too, although many of them have an entrance facing down the slope. A cemetery was found during the survey.

Fig. 16
figure 16

Plan of Iskudarka Dayeergalka Kifiile

Full size image

The most significant building in Iskudarka Dayeergalka Kifiile is the mosque, located on the upper part of the hill, with a rectangular plan and dimensions of 9 × 6 m. The mosque has almost completely disappeared but the mihrab remains, enabling us to identify it. Unlike the mosques of Abasa and Hasadinle, it is not surrounded by an external wall, and no pillars could be found inside, although given the dimensions of the building, it is likely that they existed. The mihrab, as in the other mosques, has a rectangular shape and protrudes from the qibla. In this case, the mosque was not isolated, but was part of a series of houses that delimited an L-like street whose main axis runs east–west (Fig. 17). This street is one of the few pieces of evidence of urban planning found in a medieval site in Somaliland so far, although it looks to be more the result of the progressive occupation of the space rather than a well-designed street (the mihrab of the mosque protrudes into the street). The presence of this relatively large mosque (bigger than the secondary mosque of Abasa) highlights the importance of religion in the Sultanate of Barr Saʿd al-Dīn and an advanced process of Islamization in the region, even in small hamlets in secluded rural areas.

Fig. 17
figure 17

Street at Iskudarka Dayeergalka Kifiile

Full size image

One of the most important features of Iskudarka Dayeergalka Kifiile is its closeness to a former meander of the Abasa river, now dry as the wadi has changed its course, but still close to the current wadi, and therefore retaining excellent soil quality for growing crops and other vegetables.

No archaeological materials were found, but the Incipit team confirmed the use of these parcels even today. We also documented a well more than 10 m deep that explicitly shows the capacity of local communities to develop infrastructure to improve the irrigation of the area and to water their animals.

Archaeological materials were relatively abundant throughout the site, but again imports were extremely scarce: only a fragment of speckled ware was found during the survey, suggesting a chronology of the fifteenth to sixteenth centuries (González-Ruibal et al. 2021:11). Part of a softstone bangle and five cropped cowries were also collected. However, the most interesting artifact was half of a bored stone with a diameter of 15 cm, almost certainly the weight of a digging stick (Fig. 18). This tool is the first one discovered in an archaeological context in Somaliland, although there are reports that they were traditionally used to tend gardens in the border between Somaliland and Ethiopia and in the region of Harar (Clark 1944:31; Simoons 1959:303). The rest of the sample consisted of handmade, local pottery of medium quality, identical to the specimen found at the other sites (Fig. 19).

Fig. 18
figure 18

Bored stone found at Iskudarka Dayeergalka Kifiile

Full size image
Fig. 19
figure 19

Local handmade pottery found at Iskudarka Dayeergalka Kifiile

Full size image

Conclusions: Shared Identities in a Complex World

Despite the differences in size, the three sites presented in this paper show a remarkable similarity that speaks strongly of a shared identity. This shared materiality is easily perceived in basic aspects of daily life, such as the local pottery found throughout the sites, or the construction techniques of the buildings, but it can also be appreciated in more complex elements such as the design of the mosques, which follows very specific architectural criteria. Even the emplacement of the settlements, recurrently located near a wadi but slightly set back and protected by ravines, speak of well-established practices of territory occupation and use. The replication of all these almost identical features all along the Ogo Mountains and as far as central Somaliland (Torres 2020) supports the idea of a common identity among the inhabitants of this region during at least the fifteenth and sixteenth centuries, the period traditionally identified with the so-called Sultanate of Barr Saʿd al-Dīn.

Obviously, shared materiality does not imply uniformity, and the surveys and excavations conducted by the Incipit archaeological team in recent years are unraveling a much more complex world than expected. The list of settlements only differing in size, as described by Curle (1937), now includes religious settlements and fortresses, caravan stations, trading posts, and nomadic aggregation centers, which raise, for the first time, the multi-layered and complex society which occupied Somaliland during the medieval period (González-Ruibal et al. 2017; González-Ruibal and Torres 2018; Torres 2020). The hierarchization of the sites presented in this paper is a good example of this complexity, although at this stage it is difficult to discern if these differences in size had political implications. The written sources of the fifteenth to sixteenth centuries show a hierarchized, complex society that included several administrative and political titles, as well as the existence of elites (Chekroun 2013:207–211), although the terms used are often confusing. For our area of study, written evidence is non-existent.

The data gathered in Abasa and other medieval sites excavated in recent years can to some extent fill these gaps in our knowledge of the political and social structures of medieval Somaliland. The documentation of what seems to be a public building or stronghold suggests the existence of some kind of political control, in a way similar to that suggested for the caravan station of Qalcadda (Torres 2020:177). The presence of public buildings other than mosques could be indirect evidence of political decisions taken at a higher level than the household. Combined with other archaeological features, such as the existence of two mosques and two cemeteries with differentiated types of burials, they could suggest the existence of a relatively complex society with economic, political, social, and maybe ethnic differences. Given the lack of archaeological excavations, the existence of political and social elites is difficult to track, but the recent discovery of two large houses at the site of Fardowsa (Sheikh), which were full of imported materials, suggests there could be substantial variations between households in the medieval settlements of Somaliland (Torres et al. 2020).

It is also unclear who were, ethnically speaking, the inhabitants of these permanent sites, as the Somali cited in Arab texts are considered nomadic groups and differentiated from the inhabitants of the Barr Saʿd al-Dīn Sultanate, even if in some cases they shared the same territory (Chekroun 2013:191, 194). Interestingly, there is no strong evidence of nomadic presence around the settlements: the biggest cairn field identified so far in the region is located more than 50 km to the north at the important pass of Jidhi. Cairns—the traditional nomadic type of burial in the region (González-Ruibal et al. 2017:161–162)—and nomadic mosques (consisting of a simple line of stones outlining the shape of the building) are present in the region but not in great numbers, although in Abasa some small cairns with stelae on the top were located at the southwestern cemetery. Although a more systematic survey has yet to be made in the region, satellite surveys—which have detected hundreds of structures in neighboring areas—have not found significant concentrations of cairns in this region, somewhat reinforcing the idea of a mostly agricultural area.

From a material point of view, although many elements of the material culture of the medieval communities of western Somaliland are similar to those of other areas belonging to the Sultanate of Barr Saʿd al-Dīn, there are also remarkable differences. The most obvious is the local pottery tradition, which is very different from those of the medieval sites of Shewa (Fauvelle-Aymar et al. 2006a, 2006b) or the Harar regions (Insoll et al. 2016, 2017; Joussaume and Joussaume 1972; Tait and Insoll 2021Joussaume and Joussaume ), which are stylistically and technically speaking more similar to those of the Ethiopian highlands (Torres 2020:173). Other interesting differences can be discerned in the plans of the mosques. The medieval mosques of western Somaliland present a very well-established design with the mihrab protruding from the qibla wall, without a minaret or minbar, and a wall around the main building defining a courtyard. The only differences are the size (the larger mosques have stone pillars) and the quality of the construction technique. From the published plans and photographs (Fauvelle-Aymar et al. 2006b; Pradines 2017:33), most of the mosques in northern Shoa do not adhere to this plan: several have minarets, several of the mihrabs are inserted within the qibla wall, or in some cases the courtyard is located before the roofed area. Regarding the Harar region, the only medieval mosque in published research so far (Insoll et al. 2016:26) shares some features with those of the Shoa region, such as the mihrab inserted in the qibla. These differences have been recently studied and suggest the existence of regional styles based on specific political, economic, and social influences (Cornax-Gómez and Torres Rodríguez 2023).

Finally, some indirect hints about who the inhabitants of Abasa and the other medieval settlements were come from the legend of Queen Kola written down by Burton in 1852. The existence of powerful individuals—usually magicians—named generically “Galla,” who ruled the territory before their defeat by Muslim holy men, has strong roots in Somali oral traditions (Lewis 1998). The best known of these holy men is Sharif Yusuf Aw Barkhadle, a Somali saint who battled against pagan sorcerers or chiefs (Abderahman 1977:127–128), and a similar story is recorded by Richard Burton (1894:92–93) for the coastal trading post of Siyaara. Although legends, they probably reflect the complex process of the Islamization of the Somalis, which took place later than expected and had a deep impact on the social and political history of the Horn. Discussions about the area of origin and the expansion of the Somalis throughout what is now Somaliland fall well beyond the scope of this paper (Lewis 1959, 1966). However, they all point to an east–west movement in which they occupied the territory belonging to the so-called Galla (a term which is now considered derogatory and which could have two different meanings for the Somali) (Lewis 1959:23). This movement was slow: the first time Somalis appear in Ethiopian texts is the early fifteenth century (Chekroun 2013:189), and “The Conquest of Abyssinia” (sixteenth century) still refers to many of them as scarcely Islamized and living outside the territory of the Barr Saʿd al-Dīn Sultanate (Chekroun 2013:194).

By the sixteenth century, the impact of Islamization can be felt in the archaeological record of the region, with at least three medieval sites bearing the name of Muslim saints preceded by “Aw” (meaning “holy men,” in Somali). The most important is Aw Barkhadle, close to Hargeisa, but at least two other sites (Aw Bare and Aw Boba) bear the name of a holy man. Aw Boba is, indeed, the town described as the bitterest enemy of Queen Kola, Auboba being a revered holy man whose descendants fought alongside Ahmed Gragn in the first half of the sixteenth century (Stenhouse and Pankhurst 2003:281). Therefore, the story written down by Burton could actually have merged two different processes: the expansion of the Somalis to the west and the arrival of Islam to the region, which seems to have occurred around the thirteenth century (Fauvelle-Aymar and Hirsch 2011:42).

Although these events cannot currently be tracked in archaeological records, there is some evidence that the settlements went through an expansive period at some point before their abandonment. The plan of Hasadinle shows what looks like an expansion from an original, compact cluster of houses around the mosque, and the outward position of the mosques and the stronghold at Abasa leads us to think that they were incorporated into a preexistent urban design. This expansive episode must have taken place during the fifteenth to sixteenth centuries, before the dismantling of permanent settlements following the collapse of the Barr Saʿd al-Dīn Sultanate, and would be coherent with the heyday of this state and its subjugation of wide areas of the Horn of Africa. The disturbance of trade caused by the blockage of the Red Sea by the Portuguese and the defeats against Ethiopians and Oromo led to a widespread collapse of the sultanate in the last decades of the sixteenth century (Trimingham 1965: 97). None of the sites studied so far have shown evidence of destruction, and the good state of preservation of many of the buildings even in the 1930s points to a peaceful, progressive abandonment once the political and economic structures that sustained them disappeared. At that time, the nomadic lifestyle proved far more efficient than ever-hazardous agriculture in a semidesert region such as Somaliland (Torres 2020:182). After three centuries, what remained of these settlements was, in Burton’s (1894:146) words, a scene “still and dreary as the grave; for a mile and a half in length all was ruins—ruins—ruins.”