Exploring long distance trade in Somaliland (1000-1900 AD): Preliminary results of the 2015-2016 field seasons Alfredo González-Ruibal. Incipit-CSIC, alfredo.gonzalez-ruibal@incipit-csic.es Jorge de Torres. African Rock Art Image Project, British Museum, JDeTorres@britishmuseum.org Manuel Antonio Franco. Incipit-CSIC, antonio.franco.fernandez@gmail.com Mohammed Abdi Ali. Department of Tourism, Ministry of Sports, Youth and Tourism of Somaliland. Abdisallam Mohamed Shabelle. Department of Tourism, Somaliland, shabeletour@hotmail.com Candela Martínez Barrio. Incipit-CSIC, candelamb@gmail.com Khader Ahmed Aideed. Department of Tourism, Somaliland. Abstract Somaliland was a key region in the trade routes connecting the Mediterranean, the Middle East and the Indian Ocean from antiquity to the nineteenth century. However, little archaeological work on this topic has been carried out in the region to date. A new project by the Spanish National Research Council (CSIC) intends not only to assess the role of Somaliland in the Indian Ocean network, but also to understand indigenous societies, their participation in long distance trade and the impact this had on their sociopolitical organization and culture. In this article, we present data from two field seasons (2015, 2016). We conducted a preliminary survey of the ports of Zeila and Berbera and the latter’s hinterland, and gathered important information on the relationship between Somaliland and the wider world, between coast and interior, and between merchants and indigenous communities during the second millennium AD. Keywords Indian Ocean trade, long-distance trade, Islamic Middle Ages, Ottoman Empire, nomads. Introduction The participation of the Eastern African coast in the Indian Ocean network has attracted increasing attention from archaeologists in recent years, working in prehistoric (Boivin et al. 2013), ancient (Tomber 2008; Seland 2014) medieval (Hawkes and Wynne-Jones 2015; Zhao 2015) and modern periods (Croucher 2014). This can be considered as part of a wider trend for revaluing Africans’ participation in global networks in the past (Mitchell 2005). The impact of long distance connections in African and Asian contexts has been re-assessed, along with the history of indigenous communities and endogenous historical processes, both on the coast and inland (Kusimba et al. 2013; Fleisher et al. 2015). This has helped to balance a picture that was biased towards the foreign element. The study of long-distance trade in antiquity has been particularly intense in the Red Sea basin, between Egypt and Eritrea, whereas the medieval and modern periods enjoy from a well-established research tradition in Tanzania and Kenya. Lying between these two zones, Somalia has been largely forgotten, in large part due to the conflicts that have ravaged the region for the last four decades. The situation has started to change in the de facto state of Somaliland, where several projects in recent years have revealed the enormous archaeological potential of the region (Gutherz 2002; Hirsch and FauvelleAymar 2011; Mire 2015). The cultural dynamism and long-distance connections of Somaliland during the Middle Ages had already been glimpsed in the 1930s. Between 1930 and 1934, A.T. Curle, a colonial administrator, employed his free time surveying the so-called ruined towns of the area east of Hargeisa and the eastern Ethiopian borderland, of which he inventoried 18 (Curle 1937). He did some survey and limited test pitting, which yielded a variety of import materials, including celadons1 of the Song and Ming dynasties (twelfth -fifteenth centuries AD), blue and white porcelain (sixteenth-seventeenth centuries AD), Egyptian and Ottoman coins, Persian pottery and Egyptian glass, including perfume bottles and bracelets (Curle 1937: 320-322). However, no further archaeological work on the medieval towns of Somaliland was carried out before the investigations of François-Xavier Fauvelle-Aymar and his team in 2007 (Fauvelle-Amar et al. 2011), except for very brief visits by Neville Chittick (1976) and Huntingford (1978). Fauvelle and his team worked mainly in the town of Zeila, one of the main ports before the nineteenth century along the Somali coast. They conducted sondages in the 1 Chinese green-glazed wares have not been analyzed yet. We refer to celadons generically, although we are aware of the inadequacy of the term. Qiblatayin mosque, reputedly the oldest in sub-Saharan Africa, although the excavations could not find evidence antedating the thirteenth century. Test pits were also dug in-between the Qiblatayin mosque and another medieval religious building, the Shahari mosque. The excavations and some surface collections in the refuse heaps along the northern shore of Zeila, already identified by Curle, yielded a variety of materials from the eighteenth to the early twentieth century, although some earlier pottery was also found. The town of Ferdusa, in the Berbera hinterland, and surrounding area were also preliminarily explored. Unfortunately, the project was discontinued shortly after it had begun. Drawing upon these previous experiences, the Institute of Heritage Sciences of the Spanish National Research Council (Incipit-CSIC) started a new project in 2015 to explore long-distance trade in Somaliland. The idea is to study both the role of the region in the Indian Ocean networks and the local cultural processes, including the sedentary communities and the nomads, of whom virtually nothing is known, apart from the existence of numerous tombs and cairns (Chittick 1992; Mire 2015). Our project takes a long-term perspective: we intend to examine the period comprised between Late Antiquity and the Ottoman era, which is when long-distance connections seem to be more intense. For this, we planned a series of surveys and excavations in different sectors of the coast and the hinterland. So far, we have conducted two field seasons (2015 and 2016). Some survey work has already been carried out in the country (Curle 1937; Chittick 1976; Fauvelle-Aymar et al. 2011; and particularly Mire 2015), but the published data for each of the inventoried sites is, with a few exceptions (FauvelleAymar et al. 2011), very limited. In fact, under labels such as megalithic sites or towns there might be, as we will try to show in this article, sites of very different nature and chronology. Therefore, instead of finding new places, we decided that it could be more productive to visit those that were already known through the published literature or through the work of the Ministry of Culture and Tourism of Somaliland, which was undertaken in large part by one of us (Mohamed Abdi Ali). We selected the sites that looked more promising a priori for the objectives of the project. The data gathered in those sites cover a period comprised roughly between 1000 and 1900 AD. No clear data related to ancient trade have been located to date, but those for later periods are rich and diverse. Relevant evidence of the involvement of Somaliland in long-distance trading networks during the second millennium was collected, as well as of the indigenous participation in it. Data were gathered in four kinds of sites: ports, inland towns, caravan stations and nomadic spaces (Figure 1). Figure 1 Area of study in NW Somaliland. 1. Bulahar; 2. Bender Abbas; 3. Qorgab; 4. Bagan ; 5. Qalcadda ; 6. Qubuuraale; 7. Iskudar; 8. Ferdusa; 9. Gidheys. Ports Our project targeted the area comprised between the two most important historic ports of Somaliland: Zeila and Berbera, as well as their immediate hinterland. After a first visit to the region, we decided to focus on the Berbera area and its hinterland for the second field season. Here, we surveyed three coastal towns, which are complementary in terms of their main material remnants: Berbera is a typical example of nineteenth century Ottoman architecture; in Bulahar we were able to document traces of a long history spanning from the thirteenth century to around 1900, whereas Bender Abbas seems to have had a shorter occupation, ca. 1000-1400 AD. In turn, Zeila yielded materials covering the second millennium AD, with a peak around the fourteenthfifteenth centuries AD. Zeila Zeila was one of the most important ports of the Somali coast until its decline at the beginning of the twentieth century. The old city has virtually disappeared due to the ravages of time and war. However, for at least a millennium Zeila was the main emporium of Somaliland, from which the products of the Horn were shipped to the wider world and a variety of imports entered the continent. Some consider that this was the ancient port of the Avalites mentioned in the first-century AD Peryplus of the Erytraean Sea, although no archaeological evidence has come forward to support this theory. From the tenth century AD onwards, however, it is mentioned in several Arab sources (Fauvelle-Aymar et al. 2011). At that time, it was a small non-Muslim port from which hides, incense, ambergris and tortoise shell were exported (Fauvelle-Aymar et al. 2011: 54). The town disappears from Arab sources for a couple of centuries and reappears at the end of the twelfth , which marks the beginning of the Islamization of the region (ibid.: 60-61). Its heyday seems to span from that moment until the sixteenth century, when it is sacked by the Portuguese in 1517 and again in 1528. However, the town continued to be an important multicultural trade center until well into the nineteenth century, at times depending on Moccha (Yemen). Thus, Zeila is described as “the only town with permanent trading facilities and government... frequently visited by Arab, Afar and Banyan ships” (Abir 1968: 15). During our work, we recorded some of the ruined historical buildings in the town using digital photogrammetry—including the Shahari Mosque and several Ottoman arcades (Figure 2)—and revisited the dumps near the Governor’s house where surface materials had previously been collected by Curle and Fauvelle. We found a large amount of green glass (perfume flasks, phials and some beakers). There are many containers with pushed-up, dome-shaped bases (fig. 3, nºs. 9 and 11) and loop-like (nº. 8) or thickened bulging rims (nº7), all of which have good parallels in Egyptian toilet bottles of the ninth to twelfth centuries (Shinnie and Shinnie 1978: 84-85; Lester 1996: 202-207; Rougeulle 2015: 327-328, figs. 234-235). There are also some examples of bottles with ring bases (fig. 3, nº 10) with parallels in thirteenth-fourteenth century levels at Shanga (Horton 1996: 316, fig. 25) and phials (fig. 3, nºs 2 and 12), again with parallels in the Kenyan site, where they are dated between the twelfth and fifteenth centuries AD (Horton 1996: 316, fig. 241) (Figure 3). Figure 2 Orthogonal image of Ottoman-era arcades in Zeila. All pottery recovered from the dumps is imported. The materials that we collected seem to be older than those detected by Fauvelle-Aymar et al. (2011) and predate almost in their entirety the British occupation. Finds include many Chinese and Japanese porcelains, most of them dated to the eighteenth-nineteenth centuries (witness to the relevance of trade at that time), but also many older materials, including a fragment of an early Ming bowl (fourteenth - fifteenth centuries) and a fragment of a late fifteenth century Jingdezhen bowl (Figure 4), many fragments of celadon, monochrome turquoise-glazed pottery and blue and white frit –the latter two from southern Iran and dated between the fourteenth and seventeenth centuries (Priestman 2013: 612-615; 635-636). In addition, we recovered a large amount of ribbed bangles in green and dark blue glass (a minimum number of 50), for which a thirteenth to fifteenth century date is most likely. They were used as a sort of currency and are frequent in other Red Sea settlements of the period (Insoll 1997: 386). No materials predating the early second millennium AD were found during our survey, but the items documented show the importance of Zeila during the fourteenth to sixteenth centuries, which is consistent with its political role during this period, as a main city of the Adal Sultanate. The artifact scatters can be interpreted as dumps associated to a harbor area, as they lie along the northern beach of the town. To document older periods it would be necessary to conduct test pits, which has not been possible for the moment. Figure 3. Glass beaker, bottles and phials from Zeila. Figure 4. Chinese and Japanese porcelain from the harbor dumps of Zeila, all eighteenthnineteenth centuries, except B4 (Chinese, fourteenth-fifteenth century) and B5 (Jingdezhen, China, late fifteenth-early sixteenth century). Berbera A port is supposed to have existed in Berbera since Antiquity, but the town itself is not mentioned before the thirteenth century. After it was sacked by the Portuguese in 1518, little is known of its fate before the nineteenth century (Lewis 1960: 217). When Richard Burton visited the place in 1855 he found only the ruins of four towers and temporary huts erected by the people coming to Berbera to trade (Burton 1856: 408), so the town had doubtless declined after the sixteenth century, despite being the center of a multicultural fair that is described as “one of the most important commercial events of the east coast of Africa” at the beginning of the nineteenth century (Burton 1856: 408410; Abir 1968: 16). As it stands, the town is basically the product of the Egyptian occupation after 1870 (Turton 1970: 358). The present architectural ensemble is very coherent in terms of style and includes merchant mansions, commoner houses and mosques, which were probably constructed in the short period between 1870 and the British occupation of the town in 1884, which explain their homogeneous character. Domestic architecture bears important similarities with the Red Sea style, whose roots go back to the sixteenth century (Um 2012), and has good parallels in other coastal towns like Suakin (Philips 2014). Larger houses systematically occupy street intersections, have the greater spatial control and enjoy a high degree of spatial permeability and visibility (Figure 5). They are two or three stories high with commercial storerooms at ground level and have plastered facades, topped by crenellations, and often lavish woodwork (jalousies and verandahs). Commoner houses have only one story and are not usually plastered. Both commoner and elite structures are made of coral and have semicircular doors and windows. The urbanism diverges from that typical of the Red Sea in that it is of an artificial character, typical of a colonial undertaking: the town follows a rigid orthogonal plan. Two Turco-Egyptian forts exist in the surroundings of Berbera: a large one with barracks and administrative buildings controlling the harbor, which is partially destroyed, and another one in Dubar, overlooking a natural spring that provided water to the city. Figure 5. Elite Turco-Egyptian house in Berbera. Commoner houses can be seen in the background. The latter was mapped using digital photogrammetry. Materials found at the site include some Chinese porcelain, transferware, Turkish ware and spongeware all dated to the late nineteenth century. The place was apparently bombed during the Second World War, as shown by craters with remains of bomb fragments. Shell casings, tin cans and a trench dug around the wall indicate its use as a stronghold at that time. However, traces of an older occupation also exist and include a wheel-turned jar with wavy decoration, typical of Yadhgat (Yemen) between the 10th and 11th centuries AD (Rougeulle 2015: fig. 158), and a polychrome bangle, which has close parallels in a later level from the site of Debeira West, Sudan (Shinnie and Shinnie 1978: 94-95), which can be dated around the twelfth century AD and in the Dahlak Islands (Insoll 1997). The bangle is similar to bracelet type D1 of Spaer (1992: table 3), which has a Mamluk (1250-1517) and later chronology. Ancient glass and pottery had been observed here already in 1848 (Cruttenden 1849: 55-56), but much certainly disappeared with the construction of the fort three decades later. In any case, the medieval materials associated to a fortified site controlling the town’s water supply can be taken as evidence for the existence of an ancient occupation in Berbera. Bender Abbas Ten kilometers along the coast to the east of Berbera there is a place called Bender Abbas which was perhaps an emporium of similar importance, if not Berbera’s forebear. The site is today mostly covered by sand dunes, as it lies immediately behind the beach. However, a portion of about three hectares has been exposed by wind and rains. Although they have destroyed most of this sector of the site, they have allowed us to take a glimpse at its history. The site occupies a flat consolidated dune which is almost completely covered in pottery sherds, glass, bones, shells and other remains. There is only one substantial building, a large rectangular structure (13 x 6.5 meters), which had brick walls over stone foundations. The use of bricks is unknown in Somaliland, but is well attested in the Persian outposts of Yemen (Rougeulle 2015). We excavated a test pit in one corner of the building, but the structure turned out to be eroded underneath the pavement, so we could only document the foundations (Figure 6). The other structures visible on the surface are Muslim burials that probably postdate the occupation of the settlement. Human bones could be seen protruding from some of the tombs, including those of an infant. Figure 6. Location of the artifact scatter of Bender Abbas (dashed area) in relation to the coast; remnants of a large building with stone foundations and brick walls; plan and cut of the test pit in the building. The vast majority of the materials documented in the site are imports from a variety of places. Pottery is mostly wheel-turned, while a minority of hand-made vessels evince exchanges with the indigenous communities. Most of the domestic pottery has parallels in Yemen, although some cooking pots with edged rims and complex profiles are clearly Indian: they are abundantly represented in Yemen, though (Rougeulle 2015: figs. 172, 173, 176, 301) (Figure 7). A Yemeni origin can be assigned to several fragments of soapstone vessels, too (Rougeulle 2015: 369-371). Fine wares comprise Iranian sgraffiato pottery (4 MNI of monochrome mustard, 3 polychrome and 1 hatched) and many celadons, including what seem to be early productions. There are also many glass fragments, predominately pertaining to beakers and bowls (cf. Horton 1996: 312-315) and less frequently to flasks and phials (MNI=3 out of 15), a situation that is the reverse of that documented at Zeila. Several personal ornaments were found, mostly of foreign origin: five glass beads (2 green, 1 yellow, 1 turquoise, 1 deep red), of which two are of Indo-Pacific type (Shay and Poissonnier 2016: 67-68) and a carnelian cabochon, probably from Gujarat (Hawkes and Wynne-Jones 2015). IndoPacific beads become common from the 11th century onwards, whereas carnelian has been attested in East African sites as early as the seventh century AD. A limestone pendant, a pendant on an ammonite and several perforated cowries (Cypraea annulus) were also documented. Figure 7. Pottery from Bender Abbas: A. Mustard Ware (Yemeni Yellow); B. Iranian sgraffiato; C. and Indian cooking pots; D. Yemeni pots and storing jars. Subsistence activities documented at the site included fishing (fish bones, net sinks) and animal husbandry. Bones of sheep/goat and cattle were plentiful all over the site. In some cases, they appear in dense concentrations, mixed with some ashes and charcoal, indicating dumps of organic refuse. Goats and sheep could have been raised locally, but it is highly unlikely that that was the case with cattle, since the site is located in a desert environment. Its presence might be indicative of exchanges with highland pastoralists. Several rotary quern stones in volcanic rock were found reused in the tombs. In general, the pottery assemblage is almost identical to that of Sharma in Yemen, which is a Persian outpost dated to the tenth-twelfth centuries AD (Rougeulle 2015). However, Bender Abbas might have had a longer occupation, as suggested by the presence of some Yemeni Yellow (Mustard Ware) (MNI=3), which can be dated to the thirteenth-fourteenth centuries (Kennet 2004: 53). The population of the “Port of Abbas” was probably foreign as well, as both domestic pottery and the brick structure indicate. Sirafi migrations are known between the tenth to twelfth centuries, which led to the foundation of several emporia along the Arabic coast of the Red Sea (Rougeulle 2015: 472-473). During the tenth and 11th centuries, Siraf was the main trading center of the Persian Gulf, with extensive connections with Arabia, China and India (Ricks 1970: 345-348). The variety of imports documented in Sirafi colonies shows that the merchants brought these connections with them in their diaspora. Bender Abbas could thus be a Persian establishment, or one with a Persian element to it, on the African side of the Red Sea. It is possible that it was a secondary foundation dependent on any of the coastal towns that emerged along the northern shores of the Gulf of Aden. Bulahar The last port that we were able to visit was Bulahar (also transcribed as Bulhar, Bulloxaar, etc.). This is considered one of the historic coastal towns of Somaliland, with Berbera and Zeila, although much less is known about it. Despite its vicinity to Berbera, its privileged location in a main route leading to the highlands and next to a large wadi probably explains the town’s foundation. Its impressive ruins cover around 15 hectares, with a central core of dense construction of about 6 hectares (Figure 8). The buildings are mostly Ottoman, although there are also several Turco-Egyptian structures: with Berbera, the place was occupied by the Egyptians in 1870 (Turton 1970: 358). At that time, it is described as “the greatest marketplace of the country” by a German traveler (Haggenmacher 1876: 36). If we are to believe a British colonial officer, it still maintained its status in 1890: “The ports of Berbera, Zeila, and Bulhar have considerable trade, which is increasing every year. The principal exports are sheep and goats, gum, hides, and coffee (from the interior), and the chief imports are dates, rice, and other food-grains, piece goods (mostly American), salt, &c.” (Nurse 1891: 663). The architecture of arcades in coral stone resemble closely those of Zeila and can be probably dated between the sixteenth and early nineteenth centuries. However, there is a wall with triangular niches that might be of older date: similar features have appeared in the town of Amud (Curle 1937: 318, pl. II; Huntingford 1978: 184), which is dated to the sixteenth AD and before. Niches for displaying china are characteristic of Swahili architecture (Horton 1996: 41-44). More recent structures are a fortlet, military barracks and a two-story building that can be dated to the Turco-Egyptian occupation of the 1870s. Figure 8. Map of the center of Bulahar. 1. Wall with triangular niches; 2. Ottoman arcades; 3. American school from the mid-twentieth century; 4. Turco-Egyptian building; 5. Sheikh Seberein mausoleum. Three places yielded artifacts: one of them is the tomb of Sheikh Seberein, situated in the eastern part of the town. An entire incense burner and fragments from several others turned up in front of the mosque—with no relation to the building, which seems to be of nineteenth century construction (Figure 9). They appeared mixed with bones of sheep or goat. Near this site appeared some green-glazed sherds as well. The assemblage can be dated roughly during the early second millennium AD. A second place with findings is an area of ruined buildings adjacent to the beach. Most of the finds are dated to the late nineteenth century: Turkish painted ware with crescent moons and stars, Scottish spongeware, and Japanese and Chinese porcelain, which would corroborate Haggemancher’s and Nurse’s description of Bulahar as an important trade center in the late nineteenth century. The Chinese porcelain is identical to that recorded by Fauvelle et al. (2011: fig. 2.4) in Zeila. We also recorded two beads (one polychrome, one monochrome), a polychrome glass bead and several celadon sherds that might be of an earlier date. Figure 9 Incense burner found with fragments of others and faunal remains in front of Sheikh Seberein mausoleum (early second millennium AD). The place that provided more finds is an area of ruined buildings that had been recently exposed by erosion. The great accumulation of pottery sherds and bones (goat/sheep and fish bones) suggests that it is a dump area. Three dumps were found, with materials spanning a wide period, from the Middle Ages to the nineteenth century (Figure 10). The medieval assemblage is composed of local hand-made pottery (including two fairly complete incense burners, similar to those of Sheikh Seberein), wheel-made domestic pottery with wavy decoration of Yemeni origin, Mustard Ware also from Yemen, and a glass bracelet with wavy decoration identical to examples from the Dahlak Islands (Insoll 1997). The Yemeni pottery can be dated during the thirteenth to fifteenth centuries, thus offering a tentative date for the older layers of Bulahar. Later material (sixteenth - early nineteenth centuries) comprises Ottoman wheel-turned pottery, polychrome glass bracelets, monochrome cobalt-blue bracelets with pressed decoration, and shisha pipes. The late nineteenth century is represented by overglaze painted cups, bowls and dishes from Turkey and the Netherlands, Scottish spongeware, and Chinese porcelain, a set that is ubiquitous elsewhere in Africa and the Persian Gulf (Malan and Klose 2003; Grey 2011). Bulahar offers an enormous challenge to excavators, as it is largely covered by thick deposits of rubble, dunes as tall as six meters above the original ground, and mangrove. However, the materials documented on the surface of the site demonstrate both the long history of the town and its diverse long-distance connections. Inland towns The establishment of coastal towns from the late first millennium AD onwards has to be understood in connection to the emergence of sedentary settlements in the interior. The relationship between the two phenomena is one of the issues that we intend to elucidate with this project. A priori, it seems that the arrival of merchants was a boost for the creation of permanent settlements. The label of towns that has been used to describe these centers, however, is misleading. During our surveys, we identified sites of different size, monumentality and occupation span, which suggest a rather complex landscape. During the 2015 and 2016 field seasons we visited four “towns” in the hinterland of Berbera: Qubuuraale, Bagan, Ferdusa and Gidheys. We decided to focus on this area, because the towns along the route from Zeila to the interior are slightly better know (Curle 1937; Fauvelle-Aymar and Hirsch 2011; Huntingford 1978): these include Maduna, Qorgab, Abasa and Amud, all in the surroundings of the modern town of Borama. As we mentioned at the beginning, surveys and limited excavation by Curle identified ruins of mosques and documented imported materials. However, we did pay a visit to Qorgab for the sake of comparison. This is a small settlement of two hectares Figure 10 Finds from the dumps of Bulahar: A. Ottoman shishas; B. Wheel-turned jars with wavy decoration (fourteenth-fifteenth c.); C. Local incense burners (early second millennium AD); D. Chinese porcelain bowl (late nineteenth century). composed of several rectangular stone structures, most of them with two rooms. A lot pottery was seen on the surface, all of it indigenous and hand-made. It is characterized by rough modelling and thick temper. The majority of the vessels are bowls and open pots with everted, thickened rims. Storing jars with inverted rims and an incense burner were also documented. The indigenous ware is quite different from the one we have seen in the hinterland of Berbera, except for the horizontal handle, which appears in both areas. We only found one imported item on the surface: a rim from a celadon bowl. The location of most medieval towns, around Borama and elsewhere, is very similar: they are situated near the limit of the escarpment that divides the highlands of Somaliland and the coastal plain and along the main routes connecting coast and interior. There is, however, a notable internal diversity within the sites labelled as towns. Qorgab, Bagan and Gidheys are similar. They are large villages, more than towns proper. They have a surface of less than six hectares and are composed of less than a hundred stone structures, either rectangular or square in shape. Most have only one or two rooms, but a few display more complex layouts. The cell module is of around 20 square meters. Houses typically have between 20 and 50 square meters in plan (although the smaller size is more prevalent) and are elevated over the surrounding ground. The walls are made of two rows of stones with well-dressed faces to the outside and inside. All the sites are located beside a seasonal course of water which allows for some cultivation at least during part of the year. In turn, places like Amud, Abasa or Ferdusa are settlements of 20 or more hectares (Figure 11). Although the location, organization of the space and architecture are similar, the larger settlements usually have monumental public structures (mosques) that are absent in smaller sites. Figure 11 Comparison of four “towns” in Somaliland: A. Gidheys; B. Bagan; C. Amud; D. Ferdusa. A and B from maps made on the ground with GPS; C. from satellite images; D. from satellite images and GPS data on the ground. Bagan With 6 hectares, Bagan is the largest settlement of the three. We draw the plan of Bagan with two handheld GPS, thus making it the first medieval settlement in Somaliland to be fully mapped (Figure 12). Fortunately, in this and other sites all or virtually all stone structures are visible on the surface, due to high erosion and very limited sedimentation, so we can assert with some certainty that if not all, probably over 90% of the structures of Bagan were mapped. A total of 85 were recorded, many of which are to be identified with corrals or other annexes, rather than houses, due to their small size and poor construction. Apart from this, a burial area with two small cairns was recorded 200 m to the NE of the settlement. Two other cairns flank the road that gives access to the town next to a ford of the river that flows to the SW of Bagan, around 800 meters in a straight line from the site. Figure 12 Map of Bagan and pottery from surface collection: 1. Incense burner; 2. Bottom from a celadon bowl; 3. Wheel-turned cooking pot; 4-7. Indigenous pottery. The map shows that there is no obvious geometric arrangement of the houses within the settlement in the form of streets, squares or buildings that articulate the space around. This does not mean that there is no order whatsoever, but that the order responds to a sociocultural logic that is not translated into a clear geometric layout. In fact, the arrangement of the buildings in Bagan is redolent of the organization of nomadic campsites, with wide spaces between buildings and structures clustered in small groups that may correspond with compounds of extended families. Furthermore, the idea of the campsite is further reinforced for what seems to be a ring of structures in the center of the site. The interior of the ring is occupied by an empty, flat area covered with gravel, probably a public space, and several multi-roomed buildings whose size is considerably bigger than the rest. Among these buildings, a square structure surrounded by a spacious courtyard stands out, which has parallels in the town of Amud (Huntingford 1978: 184). At present it is difficult to know whether these structures belonged to some sort of elite group, were ritual/social spaces or all of them. The organization of some of the large structures clearly suggests a domestic use: they started as single-celled or two-celled buildings and grew by addition of other rooms. We did not find many artifacts on the surface. Elements related to trade include the bottom of a celadon cup, a sherd of celadon of a different fabric, two bangles in blue glass, and some cowries (Cypraea annulus). Several fragments of indigenous pottery were documented, among which several rims and handles. All pots are hand-made, with thickened, flat rims, decorated on the edge by impression and/or incision. Only the rim is decorated. There seems to be a very limited array of forms, a feature shared by the pottery assemblages of all the local medieval settlements. A ubiquitous element, here and elsewhere, is the horizontal handle. A decorated incense burner similar to those documented in Bulahar and Iskudar (see below) was also found. Apart from pottery, a dozen iron slags were recovered. There is not enough evidence to date Bagan with any precision, but we would suggest a rough chronology of 1000-1500. Gidheys Gidheys is similar to Bagan, although of smaller size (around 2 hectares, of which only have are densely occupied). There are many tombs in the site, some of which probably post-date the use of the settlement. Among these tombs, the oldest are probably groups of adjacent cists made of quartzite slabs, with two stelae, one to each end. Although the structure is similar to Islamic tombs, they may not be, as the orientation is E-W, instead of towards Mecca. The architecture of the houses is similar to the one seen in Qorgab and Bagan and documented in other places (Curle 1937; Huntinford 1978): one- or tworoomed buildings in stone of rather small size, with quadrangular or rectangular layout. The level of destruction here and the amount of rubble prevented us from identifying individual buildings as we did in Bagan. In any case, the number is obviously much smaller, although they seem to be more densely packed. We were able to find very little material on the surface: this is limited to some indigenous hand-made pottery, a fragment of which with typical incised decoration, a sherd from an imported wheelturned cooking pot, and a blue glass bracelet. The chronology is probably similar to that of Bagan and Ferdusa. Gidheys is located in a natural route leading to Ferdusa (38 km away), which is dotted with cairns and tombs. Those that are closer to Gidheys are two circular graves made of large slabs and filled with small stones (Figure 13). Without excavation, it is difficult to ascertain whether they are contemporary to the site. Figure 13 Circular tomb with orthostats near the “town” of Gidheys. Ferdusa If there is a place that can be properly defined as a town this is Ferdusa (modern Sheikh), located 60 km to the south of Berbera. The ruined town, also known as Fedowsa or Fardowsa, lies in a plateau immediately after the mountain pass that connects Berbera with the interior, a strategic position which undoubtedly was fundamental for its development and growth. With a surface of around 35 hectares, Ferdusa is the biggest town in the area. The site is located to the north of the modern village, whose expansion has severely affected its southern part and destroyed many of the structures, but in the central area buildings are still surprisingly well preserved, in some cases with walls over a meter and a half tall. As happens with the rest of settlements of this type, there is no evidence of urban planning, the buildings being scattered throughout the area, although they are more concentrated to the east of the site. No remains of public buildings—like mosques—were documented during the survey of the place, but they surely existed, given the size of the town. Rubble and bush prevent from adequately identifying each of the structures that make up the site. Information about Ferdusa was first published by Fauvelle-Aymar et al. (2011: 41-42) who visited it briefly and documented the general characteristics of the site. During his survey they collected an Arab coin, glass bracelets, cowries and a significant amount of pottery sherds, either glazed or unglazed and including some Chinese pieces of porcelain, which provide a chronology of the fourteenth-fifteenth centuries. The author suggests the height occupation for the site between the thirteenth to the eighteenth centuries. During the 2016 field season, the site was surveyed again and the entire perimeter was recorded, increasing the area previously estimated for the settlement. In addition, two sondages (FER01 and FER02) were conducted to the northeast and the southwest of the site. To the northeast, a square structure with at least two rooms was selected following the evidence of several medium sized pots broken but still in their original position, in what seemed the entrance to the building. A bead of translucent crystal and a piece of mother of pearl (probably a pendant) decorated with incisions and punctuations were found around near the structure. A test pit of 2 x 1.5 meters was excavated, showing the remains of a poor quality pavement only preserved to the south. Under the pavement and the wall foundations, a series of layers alternating ashes and soil were documented (SU 1003, SU 1005-1007), reaching to a depth of 0.40 meters and yielding a significant amount of animal bones, including a vertebral spine still in anatomical connection. Pottery was abundant too, consisting mainly in coarse, undecorated sherds, in some cases burnt. Unfortunately, due to some problems with the owners of the land, the excavation could not be finished and had to be refilled without reaching the end of the archaeological deposits. The second sondage provided far more interesting results (Figure 14). It was carried out in the southwest of the site, an area affected by the construction of modern houses, but with remains of older square structures that can be seen scattered among them. One of these buildings, a two-room structure, was selected, with dimensions of 7 x 6 meters and a partition wall which divided the space in two; and a test pit was excavated at the southwest corner, between the external and partition walls. The pit had 3 x 2.60 meters and was in-filled by 0.2-0.3 m of an undisturbed collapse level (SU 1006) under which existed a thinner layer of soil (0.15 – 0.2 m) probably corresponding to the abandonment of the building. In this layer a number of significant pieces were documented, including several glass fragments of good quality, part of a glass bracelet and a number of sherds belonging to the same piece of coarse ware. In addition, two pieces of rotary quern stones were found, with evidence of a prolonged use, lying on the floor of the house (SU 1009), made of packed earth. The doorstep which communicated the two rooms has been located at the northwest corner of the test pit, providing evidence to trace the hypothetical layout of the house. Figure 14 Plans and stratigraphic cut of the excavation of FER02. The excavation of FER02 has provided relevant information about the architectural features of a common house in medieval Ferdusa. Following the local tradition the house appeared isolated, was of an average size (42 m2), and was built with middle sized stones bound with mud. The walls were 55 cm wide and the partition wall (size) was not attached to the external ones. All these features are very similar to those described by Curle for the towns located in other areas (Curle 1937: 318). The presence of quernstones in situ suggests that the room that we excavated was used for food-processing activities and therefore the other could have been the sleeping area. The materials gathered at FER01 and FER02 did not help us to define the chronology of Ferdusa, being too generic to date the occupation of the sites. Probably the most interesting parallel comes from FER02, where some shards of local wares show strong decorative similarities with those collected at Bagan. This lack of accuracy is compensated, however, by the relevance of the artifacts collected during our survey of the site (FIGURE 15 and 16). Imported materials were very abundant and included celadons, blue and white Chinese porcelain and several types of glazed potteries including turquoise, green, yellow and whitish examples, as well as two examples of unglazed sgraffiato very similar to the one found in Qalcadda. Regarding the local pottery, there is a predominance of middle sized pieces with globular bodies, the decoration usually consisting in punctuations or incisions on the upper part of the rim. The smaller pieces have similar decorations, but placed on the body. As mentioned above, the closer parallels for this indigenous pottery are those found in Bagan, in a similar if more modest context. Figure 15 Imported wares from Ferdusa: 1-4. Green/Turquoise glazed pottery; 6-7. Possible Manganese Purple underglazed-painted ware; 8-9. Green glazed-painted ware (Ayyubid?). 12-14 Celadon. 15 Yellow Yemenite ware. 16. White and Blue glazed pottery; 5, 10-11, 17. Blue and white Chinese Porcelain Figure 16 Local pottery from Ferdusa. Many of the imported materials suggest a chronology which would range between the twelfth and the sixteenth centuries, slightly older than the one proposed by Fauvelle-Aymar et al. (2011: 42). This is the case, for example, of several fragments of a white, iridescent glazed pottery which could be related to the manganese purple underglazed-painted ware found in Ras-al-Khaimah (Kennet 2004: 51, color plate 10), with a chronology of the 11th to thirteenth centuries. Some glazed pieces painted in green could correspond to Ayyubid productions (twelfth -thirteenth centuries). Finally, some yellow Yemenite (Mustard Ware) shards point to a slightly later chronology (thirteenth-fourteenth centuries). The more modern materials can be dated in the sixteenth century, probably marking the final moments of the town which could be related to the collapse of the Sultanate of Adal, the Ottoman intervention in the coast and the Oromo invasions that affected the region. If the excavation carried out during 2016 is of any indication, the abandonment happened peacefully and progressively, due to the disturbance and reallocation of the trade networks rather than a traumatic, violent episode. Ferdusa is whithout any doubt, one of the key places for the study of the medieval trade networks in central Somaliland. Its excellent geographic position, the entity of the preserved remains and the sample of materials collected point confirm the potential of the site. The slow but continuous destruction of the site caused by the growth of Ferdusa makes its study the more urgent. Qubuuraale Whereas we can discuss whether the label “town” is more or less appropriate for the settlements mentioned so far, in the present case we are not even sure if the concept “settlement” is adequate. Qubuuraale is certainly not a town, but it does not resemble the settlements describe to date. The site is also located on the escarpment area (in the Golis Mountains), but in this case it lies on its very edge, with the land falling down rather abruptly immediately to the north. The ruins occupy a hill surrounded by higher elevations. This in itself is a peculiar feature, because most “towns” lie either in plateau areas or on gentle slopes. The hill is limited by a creek to the south and southeast that has some water all year round. The area has cultivable soils and in fact the valley bottom is today covered in orchards that grow guava, banana, papaya, vegetables and sunflowers. Beyond the river, the land is covered by a dense forest of acacia. The site occupies the top of the elongated hill and has axes of 285 x 50 meters. The structures of Qubuuraale (Figure 17) have little to do with those of Bagan, Qorgab or Gidheys. First, there is a wide diversity of buildings, with very different layouts. Second, most of these buildings are multi-celled. We mapped four of these structures with total station and GPS. Figure 17 Map of Qubuuraale with the buildings that can be identified on the surface and detailed map of one of the buildings. Dashed areas indicate accumulations of rubble. QUB01 is built on a massive plinth with huge stones, cursorily worked. The original structure was a two-roomed rectangular structure. Each room is 5 x 5 meters, therefore the structure had a surface of 50 square meters. A single rectangular room, 15 x 5 meters, with a similar fabric, was annexed at some point, clearly using the same 5 x 5 module. Later, two semicircular structures were added, although these are mere rows of stone that may be recent (perhaps sheep folds?). The structure is oriented to the SW. The monumentality of the structure and its isolated location made us hypothesize that it was a ritual structure. However, no materials were found on the surface that could have helped the interpretation. QUB02 is a long structure made by the addition of rectangular rooms. The compound is oriented N-S and the rooms E-W. We conducted two test pits. The first (S1) yielded many bones of sheep/goat; the second (S2) only an undiagnostic sherd. No pavements were found in either case. The walls are very roughly made: in some cases they are a mere alignment of stones. It seems that only some of the rooms were roofed and the others were courtyards. The sizes and shapes of the rooms are very different, in stark contrast to what we saw in QUB01 and, as we will now see, QUB03. QUB03 has some resemblance to QUB02, but there are also important differences. The walls are crudely made, with no worked faces. A particular kind of wall was seen here, that is present in other structures at the site: two rows of slabs stuck in the ground leaving an empty space inside. This suggests that the walls were made of brush or some other perishable material, but not stone. It seems that in this case the structure originated from two two-celled buildings that were annexed at some point. The maximum axes of the building are 18 x 14 m (NNW and SSW). Interestingly, the module is the same as in QUB01 (5 x 5 m). The overall aspect is that of a large domestic compound. QUB04 is a very spacious unpartitioned structure (10 x 16 meters), with the longest axe oriented to the SW. A very similar building was documented in the ritual site of Iskudar, that we will describe later. They have the same size and orientation. In the case of Qubuurale, the orientation differs from the other buildings and is reasonably consistent with an alignment to Mecca. It can then be hypothesized that it is a mosque. Very little material turned up on the surface (or in the test pits, for that matter). Finds comprise some hand-made pottery, with the characteristic horizontal handles (six elements have been found) and a globular pot decorated by simple impression. Imports include the rim of a wheel-made pot, a bottom of Yemenite Yellow ware, two undiagnostic wheel-turned sherds and a small fragment of blue glass. Other findings include a fragment of rotary quern stone in volcanic rock, four iron slags and four cowries (C. annulus as all the cowries documented in our work). The natural access to Qubuuraale (from the south) is flanked by several big cairns. The funerary function of cairns has been demonstrated in other cases, both in Somalia and elsewhere (Chittick 1992, Davies 2013, Fauvelle and Poissonnier 2016: 66-67). No other site has so many large tumuli in the immediate vicinity and those of Qubuuraale are clearly associated to the site. Before wider excavations are carried out, it is difficult to ascertain the nature of this site. In the meantime, a religious function for some of the buildings cannot be ruled out. Unfortunately, the paucity of materials makes it difficult to date. The bottom of Yemenite Yellow ware is the only firm dating evidence for the site and places it at the thirteenth-fourteenth centuries AD. The strategic location of Qubuuraale, in one of the entry points to the highlands from the coastal plain, might explain its peculiar nature. Caravan stations So far, we have discussed two central elements in the long-distance trade networks of Somaliland: the ports from where the imports arrived and raw materials (incense, ivory, tortoise shell, ambergris) were exported, and the towns that demanded the imports and channeled the raw materials towards the coast. However, another element participated in this network: caravan stations. Many of these probably existed that have left little or no trace, as they were probably just thorn or brush fences. Others might have had a stone enclosure and some substantial building and in fact, in the route from Zeila to the highlands some of these can be seen, although their precise function and date remain unknown. A very different case is the caravenserai of Qalcadda. The name comes from the Arabic word (of Persian origin) Qalāt, meaning “fortified place”, and refers to a fort that stands along the route connecting coast and interior through Bagan and Qubuuraale. The qalāt is a rectangular enclosure (55 x 90 m) with thick walls around 1 meter high made of dressed stone. The corners are defended by round bastions. Inside, very few surface materials turned up: we can mention two tiny fragments of white and blue Chinese porcelain and celadon. The fort, however, is just one of the many structures and activity areas that exist in the place (Figure 18). To the south of the fort there is a flat space where many traces of walls can be glimpsed. Those that can be more easily discerned form a rectangular, partitioned nave parallel to the road. All walls are made of dressed stone. From what can be assessed on the surface, it seems that we have a large building with many rooms surrounding an open space, which is the typical plan of a caravanserai. The caravensarai or khan is the typical caravan station of the Middle East. They are particularly popular between the twelfth and sixteenth centuries (Onge 2007; Tavernari 2009), although examples exist at least from Antiquity (Thareani-Sussely 2007). They were state or privately owned, urban or rural, and offered their services to merchants, who used the premises to pass the night and store their merchandises safely. Caravanserais have continued to exist until recently and their structure has remained largely unchanged. The presence of a fort next to a caravan station was quite common, as garrisons were stationed near the them to secure the roads (Thareani-Sussely 2007: 126-127). Qalcadda is 85 km away from Berbera through a main road, which means a journey of probably three days from the coast for an average caravan. It is the first place the caravan would find with cool weather, pasture and reliable sources of water, after the hot coastal desert. Figure 18 General map of Qalcadda with the fort, the caravan station and the artifact scatter that can be indicative of a market area or a place for loading and unloading the caravan. We conducted a sondage of 2.5 x 4 meters in the main nave of the building, next to what seems to be an access from the road (figure 19). The structure was sedimented by 0.6-0.7 m of deposits, covering a lime floor (SU 1006). This was just the last of four floors, each with its own occupation layer, separated by about 5 cm each. The penultimate deposit (SU 1007), whose pavement (SU 1008) showed traces of fire with many fragments of charcoal and ashes, was radiocarbon dated to 318 ± 28 BP (cal. AD 1486-1646)2, which dates the use of the room most likely at some point during the sixteenth century. The deposit (SU 1004) covering the last pavement (SU 1006) yielded some materials of interest, including an unglazed sgraffiato-decorated cup and a piece 2 D-AMS 015989 (CAIS 24947). of indigenous pottery. Over the pavement itself, we recovered two small fragments of thin green glass, two cowries and a hand-made incense burner or perhaps rather a lamp. Figure 19 Left: Plan of the main nave of the caravanserai visible on the surface with location of the sondage. Right: Materials retrieved in Qalcadda during the excavation of the caravanserai (A): from top to bottom, sgraffiato cup, lamp and indigenous pot; and the survey of the open area (B). To the other site of the road, there is an artifact scatter covering around 8 hectares. Materials comprise hand-made indigenous pottery, glass, turquoise- and greenglazed pottery, celadons, blue and white Chinese porcelain, tortoise shell and many cowries. Tortoise is mentioned as an export of Somalia from the early Middle Ages (Ricks 1970: 343). Indigenous pottery is different from those retrieved in the towns: decorated items are very rare and instead of flat thickened rims, it is inverted rims that predominate, apparently belonging to jars and containers. The artifact scatter extends to the south of the courtyard structure. There are no visible structures, however, associated to these materials. We interpret the activity area as a possible trading zone, perhaps where caravans were unloaded, animals kept and merchants and locals met to exchange goods. According to Thareani-Sussely (2007: 126), South Arabian ethnographers report the existence of an open space that is located next to the merchants’ quarters and the market that enables caravans to leave their camels. The chronology of the site coincides with the heyday of the Adal Sultanate (1415-1559). This could explain the construction of the fort, which is in strike contrast with the ruined settlements, which lack fortifications. Qalcadda attests to the changing sociopolitical context during the fifteenth and sixteenth century, which sees the general decline of towns in a climate of violence and war with the Ethiopian kingdom, which made frequent forays into the country, and the expanding Oromo (Lewis 1960: 222224). The fort can also be seen as a manifestation of the control of long-distance trade by the state, a panorama that again tallies well with the period of the Adal Sultanate. We know that forts and castles are erected at that time by other powers, such as the sultanates of Ajuran and Warsangali. The location of the fort in a strategic position, between coast and mountain and along a key route, could plausibly be related to the exaction of tribute. However, the relationship of the building that we identify as a caravanserai and the fort is not clear. At present, we cannot ascertain that they are synchronous, although the surface materials look quite similar. Yet the caravan station seems to have had a complex architectural history, judging from the walls with different orientations that suggest additions, reconstructions or modifications. Only further open area excavations will help us interpret the site. Archaeological parallels for waystations and caravanserais in Sub-Saharan Africa before modern times do not abound—for a nineteenth-century example see Chami et al. (2004). We can mention the site of Tabot (Magid 2004), which is a unique example of an ancient inland cavanserai, the medieval site of Yendi Dabarai in West Africa (Mitchell 2005: 165), and, much closer to Qalcadda’s chronology, that of the port of Suakin (Insoll 2003: 99), both in Sudan. The nomadic landscape The landscape that we have described up to this point lacks a crucial element: the nomads. Towns, ports and trading stations were just a part, perhaps not the most important, of the cultural landscape of Somaliland between 1000 and 1900 AD. The countryside was then, as still is today, inhabited by nomadic groups. Of these, very little is known. The main source of evidence is burial monuments (cairns and other kinds of tombs) that dot large parts of the landscape (Chittick 1992). During our work, we mapped hundreds of burial sites, including isolated tombs, clusters and extensive necropolises. They are not located at random: they tend to cluster along the main routes connecting coast and interior, at important natural crossroads and near fords in wadis (Mire 2015: 127). There is a variety of tombs that likely has a chronological reading, not incompatible with others (sociopolitical and geographical). To date, only simple cairns have been excavated on the coast (Chittick 1992) and they have variously furnished early Roman (Ballat 1996), Late Antique (Chittick 1992) and even nineteenth-century materials (Lewis1961). The largest cairns, such as those recorded near Qubuuraale, seem to be among the oldest structures, if not the oldest. Other structures include the circular tombs with slabs from Gidheys, large rectangular stone platforms (as documented in this same site), small cairns, rectangular cairns, cairns with a perimeter ring (either circular or rectangular) and cruciform tombs (see also Davies 2013). Many cairns use different kinds of stones (such as basalt and sandstone) to produce chromatic effects that make the structures visible from afar. Before a variety of tombs is excavated, we will be unable to offer a chronology and even less to understand their place in the social process of Somaliland and their relationship to long distance trade. During our fieldwork, we had the chance to study a site that is particularly relevant for understanding the nomadic landscape of Somaliland. The site, known as Iskudar, lies in the Golis Mountains, at the edge that separates the coastal plain and the highlands, a location that is redolent of Qubuuraale. The name Iskudar means “aggregation”, “mixing” or “combination” in Somali. This is what this site seems to do. Iskudar lies in a natural crossroads: it is a small elevation located in the confluence of three wadis. The site is surrounded by hills from all sides, but is easily accessible through the water courses. It has an oval shape that covers around 3 hectares and is surrounded by a stone fence of almost 700 meters (Figure 20). The fence is made of two rows of stones and has roughly-hewn standing stones or stelae stuck in between the two lines every couple of meters. Inside the perimeter there are large parts covered in rubble, where it is virtually impossible to make out the outline of any feature. The NW part of the enclosure, however, has many funerary structures, that are relatively well preserved: these are stone rings and clusters of cists. There is also a cruciform tomb, flanked by two massive stelae and surrounded by a wall, which is located at some distance from the main group of structures, and another one outside the perimeter and next to the river, both of them looted. The walls of four rectangular structures were also identified, including a large structure that has almost the same shape and dimensions as the building QUB04 in Qubuuraale, that we identified as a possible mosque: 140 square meters. In addition to the burials within the enclosure, the accesses through the valleys are flanked along several kilometres by large cairns and stone circles, which form a sort of ritual avenues. Interestingly, no similar cairns were found inside the site itself. Figure 20 A. General map of Iskudar.1: Cruciform tombs; 2. Areas covered in rubble; 3. Area with remains of feasting activities; 4. Cists cluster; 5. Structures; 6. Cists cluster; 7. Spaces delimited by partition walls. Triangles indicate original entrances to the site. The asterisk indicates the finding of the marvered glass. B. Incense burner found amid faunal remains in Locus 5. C. Detail map of the area with faunal remains (nº3 in general map). The date of the four types of funerary monuments (cairns, stone rings, cists and cruciform tombs) is difficult to ascertain (Figure 21). We did not have the time to excavate any of the large cairns, but, as noted above, these are probably the oldest monumental funerary structures in Somalia, judging from the results of previous excavations, which have yielded materials as early as the first century AD (Ballet 1996: 821, 825). The cists, in turn, are probably the most recent. They are often marked with two stelae and strongly resemble Islamic burials. However, they are usually not oriented in the direction of Mecca (definitely not in the case of Iskudar) and they are not identified as Muslim by any of the local communities with which we had the opportunity to broach the issue. In fact, several of the cists had been looted by people from the local community—the same fate met by some of the large cairns and the two cruciform tombs, thus confirming the disconnection between the present people and the ancient burials in the local imagination. The similarities with Islamic tombs may indicate some kind of influence from them, though, and therefore a relatively recent chronology. More important, perhaps, is a stratigraphic argument: a cluster of cists appeared annexed from the outside to the perimeter wall (nº 6 in the map), probably because the space inside the enclosure was already occupied. After being given permission by the elders of Iskudar, we excavated one of the cists that looked intact. In fact, there was a layer of small white stones covering the space delimited by slabs, which often seals this and other tombs. However, we could only find three small fragments of human bones and not a trace of funerary offerings. The bones are a fragment of right rib, a possible fragment of sacrum or coxal and a fragment of the first right metacarpian, all belonging to an adult or several adults. There is no physicochemical reason for the bad preservation of the bones, as well-preserved human bones have turned up in the surface3 and faunal remains during excavation (see below), so we can only hypothesize that the burial was disturbed and for some reason the layer of white stones restored at some later point. 3 Including a femur and fibula that appeared together in the area of the perimeter wall disturbed by bulldozers. They belong to an adult of indeterminate sex but with a height of over 1.70 m. Figure 21 Typology of tombs: Cist from Iskudar that was excavated in cluster nº4 in the map; cists cluster from Matagiso (Geri) similar to thouse found in Iskudar; stone ring from Xabaalo Tumaalod, similar to those found in Iskudar but with a cist burial in the middle and a secondary cairn (instead of a main cairn); C. Cruciform tomb from Xabaalo Tumalod, similar to those of Iskudar. The dark grey stone indicates a stela. Cruciform tombs were first reported in Somaliland by Mire (2015). They have the layout of a segmented Greek cross made with large slabs stuck into the ground. It is usually marked with huge, roughly hewn stelae, usually two, which can be over two meters high. As we have pointed out, the two cruciform tombs had been looted recently. We were told by the diggers that seven complete decorated pots were found along with many bones. We cleaned the burial and were able to retrieve several fragments of incense burners decorated by impression, similar to those documented in Bulahar and Bagan, which would situate the tomb around the first half of the second millennium AD. In addition, we visited a necropolis with two cruciform tombs near Berbera called Xabaalo Tumaalod (“Cemetery of the Blacksmiths”) (cf. Mire 2015: 124). The burials here were untouched, but around them we found cowries, fragments of hand-made and wheel-made pottery and a bottom of monochrome green-glazed ware, all of which can be roughly dated around the same period as Bagan and the early levels of Bulahar. Due to their cruciform shape, the tombs are considered Christian by the local population (also Mire 2015: 127-128). The existence of rich funerary offerings inside the tombs would run against this interpretation, but the perfect East-West orientation marked by the position of the stelae and the Greek-cross shape are, at least, intriguing. Cruciform graves seem to cluster in the area south of Berbera and Bulahar (see Mire 2015: map 4). Stone rings are circular rubble walls with either a cist or a cairn in the middle. We might have some chronological arguments. The northeastern part of the site had been damaged by the excavation of an irrigation channel by bulldozers. The excavation destroyed a large tract of the perimeter wall and adjacent area. The disappearance of the wall, in turn, facilitated the erosion of the area (nº 3 in the map) which has exposed an archaeological layer full of animal bones, ashes, charcoal, pottery, cowries and a single tiny sea shell perforated to be used as a bead. The materials seemed to come from a flat area surrounded by stone rings at least in part delimited by a rubble wall or terrace. We conducted five shovel tests of 0.5 to 2 square meters, codenamed Loci 1-6. The larger density of finds appear closer to the front of erosion to the north and they become sparser and finally disappear towards the south. The largest amount of faunal remains comes from Locus 3, the northern half of Locus 2 and Locus 5. More faunal remains and pottery were found in burrows dug within the perimeter of another stone ring further to the north (Locus 6). The majority of the finds are faunal remains. These are still under study, but a first evaluation indicates the presence of sheep/goat (MNI=5), cattle (MNI=2), camel, large birds, and perhaps some wild ungulates. Some of the sheep/goat bones belong to immature individuals. Two fragments of wheel-turned pottery were also found, and indigenous hand-made pottery decorated with incised designs. In Locus 5, we found an incense burner, as we saw, mixed with the bones. Many of the bones show cuts and fleshing marks made with iron implements. This is clear evidence of feasting, probably related to the burials: the layer of bones appears at only 20 cm underneath the surface. We dated charcoal samples from Locus 3 and 5. Locus 3 furnished a date of 827 ± 24 BP (cal. AD 1166-1260). The charcoal from Locus 5 was dated to 663 ± 20 BP (cal. AD 1276-1393)4. The lack of overlap between both dates can be tentatively interpreted as recurring feasting activities taking place in the burial area. 4 Locus 3: D-AMS 015990; Locus 5: D-AMS 015991. Apart from this eroded area, few artifacts turned up on the surface of the site: a dozen hand-made pottery sherds, some decorated by incision, fragments from two different hand-made incense burners (which again point towards the ritual use of the site), and half a dozen undiagnostic wheel-turned sherds. Local pottery is crude, compared to that documented in the towns, although the shapes (thickened inverted rims and horizontal handles) are similar. The most significant finds are a complex rim from an Indian wheel-turned cooking pot, which has a long chronology from the ninth to the sixteenth centuries AD (Rougeulle 2015: 458, fig. 301) and a fragment of marvered glass from a perfume bottle. Marvered glass is typical of the Ayyubid period in Egypt, with a heyday in the twelfth -thirteenth centuries AD (Whitcomb 1983: 102-103; Lester 1996: 212; Rougeulle 2015: 328-329, fig. 236), a chronology that is consistent with our radiocarbon dates. The bottle was probably part of a funerary offering. Before more excavations are undertaken we cannot know whether the site has a more ancient origin, as is probably the case, or a later existence. In any case, if the Islamization of Somaliland starts from the coastal towns by the end of the twelfth or the beginning of the thirteenth century, as Fauvelle-Aymar et al. (2011) surmise, then it makes sense that a large pagan sanctuary existed in the mountains during the early second millennium AD. Its decline can perhaps be associated to the rise of the Adal Sultanate at the beginning of the fifteenth century, when Islam is already well established in Somaliland. To understand the landscape of Iskudar, we have to take into account not only the sanctuary and the funerary avenues leading to the site, but also a large oval enclosure, known as Gamada, situated on top of a hill overlooking the ritual center. Nothing can be seen on the surface, except for the rubble wall that surrounds a space of around 7.5 hectares. The wall is 950 meters long and between 1 and 2 meters thick, which implies an enormous investment of labor. A similar, and equally empty, enclosure was documented by Chittick (1976: 130-131) in Wargaale. We suggest that the enclosure could be the space of a large campsite where the nomads coming to perform the rituals at Iskudar could reside temporarily in mat and brush huts. The lack of stone structures is consistent with the nomadic lifestyle and the short occupations, which would have left little trace in terms of residues, as most activities would have taken place in the ritual enclosure at the valley bottom. The complexity of the site, the landscape and the activities that were carried out in it clearly indicate that this was more than a cemetery. We can infer that nomads from different areas came to Iskudar through pre-established routes—a sort of pilgrimage ways— which were marked with large cairns and tombs. In Iskudar, they buried and honored their dead and performed rituals that included the collective consumption of animals and the use of incense. The slaughter of valuable animals, such as young specimens and cattle, are suggestive of the important investment made in these feasts. The wheel-turned pottery and the unguentarium evince the participation of nomads in the long-distance trade network. A cursory GIS analysis (Figure 22) shows that Iskudar lies outside any of the optimal routes leading from the coast to the highlands, unlike the towns and caravanserai discussed above, which lie directly along them. At the same time, Iskudar as a natural crossroads, was a place of confluence, as the modern name indicates. It was probably the sacred nature of the site, not its strategic location, that attracted people: it responds to a logic that has nothing to do with the trade and a lot with nomadic patterns of movement and indigenous cultural values. Figure 22 To the left, map of the ritual site of Iskudar and the stone enclosure of Gamada. The symbols indicate cairns. To the right, GIS analysis of optimal routes in the vicinity of Iskudar. It is necessary to make a detour from any main route to go to Iskudar. The place, however, lies in a confluence of minor routes. Conclusions In this article, we have presented evidence collected in two archaeological field seasons in NW Somaliland (2015 and 2016). Although we have only started to scratch the surface of the country’s rich archaeological heritage, we have been able to collect significant data for the study of long-distance connections in Somaliland and the indigenous communities that participated in it. This has been largely possible thanks to previous research work in the area undertaken during the last decade both by the Somaliland authorities and by archaeologists (Fauvelle-Aymar and Hirsch 2011; Mire 2015). We have drawn upon this pioneer work and tried to cast more light on the articulation of landscape during the second millennium AD by mapping significant sites, doing intensive surveys of those sites, collecting representative materials and conducting sondages to obtain datable materials. Recent archaeological work in the Horn of Africa is offering a more complex picture of the late first and second millennia AD—a period that has traditionally received little attention in contrast to earlier ages. The world that is emerging is one of interactions with the wider world and cosmopolitan communities (Insoll 1997; Hirsch and Fauvelle-Aymar 2011; Insoll et al. 2014; Mire 2015), but also of complex and diverse indigenous societies of which little or nothing was known before (Hirsch and Fauvelle-Aymar 2004, 2008; Fauvelle and Poissonnier 2016). Our work in Somaliland adds to this increasingly rich panorama. Thus, sites like the nomadic sanctuary of Iskudar and its surroundings reveal the existence of a virtually unknown indigenous society, which produced impressive monuments and performed rituals likely involving a diversity of clans coming from a large territory. Places like the coastal site of Bender Abbas are witness to the presence of diasporic communities living in their own settlements along the Somali coast but interacting in different ways with indigenous groups. Coastal sites, however, where not just foreign outposts: the documentation of indigenous materials and religious practices in the port of Bulahar offers a glimpse at the role of local groups in the creation of trading centers. The wealth of imports documented in this and other ports, such as Bender Abbas or Zeila, speak of a millennium of connections with distant and not so distant places: China, Japan, India, Egypt, Yemen and Iran. The products that were imported include fine wares, perfume and personal ornaments (beads, cowries and bangles). These categories seem to change little during at least a thousand years, although their consumption undoubtedly changed through time and in different contexts. From written sources we know that incense was one of the main exports of Somaliland; yet the appearance of incense burners in each and every site that we surveyed suggests that this resin was also a crucial substance in the local culture. Finally, a comparative study of ruined “towns” is showing the internal diversity of the sedentary settlements of Somaliland. Overall, the picture that emerges is one of complex articulations between nomads, town-dwellers, diasporic communities and cosmopolitan merchants. References Abir, M. 1968. Ethiopia: The Era of the Princes. New York: Praeguer. Ballet, P. 1996. De la Méditerranée à l'océan indien. L'Égypte et le commerce de longue distance à l'époque romaine: les données céramiques. Topoi 6(2): 809-840. Boivin, N., Crowther, A., Helm, R., & Fuller, D. Q. 2013. East Africa and Madagascar in the Indian Ocean world. Journal of World Prehistory 26(3): 213-281. Burton. R. 1856. First footsteps in Africa or an Exploration of Harar. London: Longman, Browns, Green and Longmans. Croucher, S. K. 2014. Capitalism and Cloves: An Archaeology of Plantation Life on Nineteenth-Century Zanzibar. New York: Springer. Curle, A. T. 1937. The ruined towns of Somaliland. Antiquity 11(43): 315-327. Chami, F., Odunga, S., Kessy, J. and Maro, E. 2004. Historical archaeology of Bagamoyo: excavations at the caravan-Serai. Dar es Salaam: Dar es Salaam University. Chittick, N. 1976. An archaeological reconnaissance in the Horn: the British-Somali Expedition, 1975. Azania: Journal of the British Institute in Eastern Africa, 11(1): 117-133. Chittick, H.N. 1992. Cairns and other drystone monuments in Somali regions. In H.M. Adam and C.L. Geschekter (Eds.): Proceedings of the First International Congress of Somali Studies, pp. 35-42. Atlanta: Scholars Press. Cruttenden, C.J. 1849. Memoir on the Western or Edoor Tribes, Inhabiting the Somali Coast of N.-E. Africa, with the Southern Branches of the Family of Darrood, Resident on the Banks of the Webbe Shebeyli, Commonly Called the River Webbe. Journal of the Royal Geographical Society of London 19: 49-76. Davies, M. I. 2013. Stone cairns across eastern Africa: a critical review. Azania: Archaeological Research in Africa 48(2): 218-240. Fauvelle-Aymar, F.-X., and Hirsch, B. (eds.) 2011. Espaces musulmans de la Corne de l'Afrique au Moyen Age. Études d'archéologie et d'histoire. De Boccard/Centre Français d'Études Éthiopiennes. Fauvelle-Aymar, Hirsch, B., Bernard, R. and Champagne. F. 2011. Le port de Zeyla et son arrière-pays au Moyen Age. Investigations archéologies et retour aux sources écrites. In Fauvelle-Aymar, F.-X., and Hirsch, B. (eds.) 2011. Espaces musulmans de la Corne de l'Afrique au Moyen Age. Études d'archéologie et d'histoire, 26-74. De Boccard/Centre Français d'Études Éthiopiennes. Fauvelle, F. X., & Poissonnier, B. 2016. The Shay Culture of Ethiopia (10th to 14th Century AD): “Pagans” in the Time of Christians and Muslims. African Archaeological Review 33(1), 61-74. Fleisher, J., Lane, P., LaViolette, A., Horton, M., Pollard, E., Quintana Morales, E., Vernet, T., Christie, A. & Wynne‐Jones, S. 2015. When did the Swahili become maritime?. American Anthropologist 117(1): 100-115. Grey, T. 2011. Late trade wares on Arabian shores: eighteenth- to twentieth-century imported fineware ceramics from excavated sites on the southern Persian (Arabian) Gulf coast. Post-Medieval Archaeology 45(2): 350-373. Haggenmacher, G.A. 1876. Haggenmacher's Reise im Somali-Lande, 1874. J. Perthes. Hawkes, J. D., & Wynne-Jones, S. 2015. India in Africa: Trade goods and connections of the late first millennium. Afriques. Débats, Méthodes et Terrains d’Histoire 6. Hirsch, B., & Fauvelle-Aymar, F. X. 2004. Cités oubliées. Réflexions sur l’histoire urbaine de l’Éthiopie médiévale (xie-xvie siècles). Journal des Africanistes, 74(12): 299-314. Hirsch, B., & Fauvelle-Aymar, F.-X. 2008. Établissements et formations politiques musulmans d'Éthiopie et de la corne de l'Afrique au Moyen Âge: vers une reconstruction. In Annales Islamologiques 42: 339-372. Horton, M. 1996. Shanga: The archaeology of a Muslim trading community on the coast of East Africa. London: The British Institute in East Africa. Huntingford, G. W. 1978. The Town of Amud, Somalia. Azania: Journal of the British Institute in Eastern Africa, 13(1), 181-186. Insoll, T. 1997. An archaeological reconnaissance made to Dahlak Kebir, the Dahlak Islands, Eritrea: preliminary observations. In thirteenth International Conference of Ethiopian Studies, 382-388. Kyoto: Shokado. Insoll, T. 2003. The Archaeology of Islam in Sub-Saharan Africa. Cambridge: Cambridge University Press. Insoll, T., Habtamu Tesfaye and Malik Saako Mahmoud. 2014. Archaeological Survey and Test Excavations, Harari Regional State, Ethiopia, July-August 2014. A preliminary fieldwork report. Nyame Akuma 82. Kennet, D. 2004. Sasanian and Islamic Pottery from Ras al-Khaimah: Classification, Chronology and Analysis of Trade in the Western Indian Ocean. Society for Arabian Studies Monographs 1. Oxford: Archaeopress. Lester, A. 1996. The Glass from Yoqne’am: The Early Islamic, Crusader, and Mamluk Periods. Qedem Reports 3, Yoqne’am I: The Late Periods: 202-217. Institute of Archaeology, Hebrew University of Jerusalem. One of the most peculiar kind of tombs is the cruciform type. This tomb has the Lewis, I.M. 1960. The Somali Conquest of the Horn of Africa. The Journal of African History 1(2): 213-230. Lewis, I.M. 1961. The So-Called'Galla Graves' of Northern Somaliland. Man 61: 103106. Magid, A.A. 2004. The Site of Tabot: An Old Waystation in the Southern Red Sea Hills, Sudan. In Neueste Feldforschungen Im Sudan und in Eritrea: Akten Des Symposiums Vom 13. Bis 14. Oktober 1999 in Berlin , 155-172. Wiesbaden: Otto Harrassowitz. Malan, A., and Klose, J. 2003. Nineteenth-century ceramics in Cape Town, South Africa. In S. Lawrence (Ed.): Archaeologies of the British: Explorations of Identity in Great Britain and Its Colonies, 1600-1945, 191-210. One World Archaeology 46. London: Routledge. Mire, S. 2015. Mapping the Archaeology of Somaliland: Religion, Art, Script, Time, Urbanism, Trade and Empire. African Archaeological Review, 32(1), 111-136. Mitchell, P. 2005. African Connections: Archaeological Perspectives on Africa and the Wider World. Lanham. MD: AltaMira. Onge, M. 2007. Caravanserais as Symbols of Power in Seljuk Anatolia. In J. Osmond and A. Cimdiņa (Eds.): Power and Culture: Identity, Ideology, Representation, 49-69. Pisa: Pisa University Press. Phillips, J. 2014. Architecture in Suakin. Encyclopaedia of the History of Science, Technology, and Medicine in Non-Western Cultures, 1-8. New York: Springer. Priestman, S.N. 2013. A Quantitative Archaeological Analysis of Ceramic Exchange in the Persian Gulf and Western Indian Ocean, AD c.400 – 1275. PhD Thesis, University of Southampton, Centre for Maritime Archaeology, Department of Archaeology. Rick, T.M. 1970. Persian Gulf Seafaring and East Africa: Ninth-Twelfth Centuries. African Historical Studies 3(2): 339-357. Rougeulle, A. (Ed.) 2015. Sharma. Un entrepôt de commerce médiéval sur la côte du Hadramawt (Yémen, ca. 980 – 1180). Oxford : Archaeopress. Seland, E. H. 2014. Archaeology of trade in the western Indian Ocean, 300 BC–AD 700. Journal of Archaeological Research 22(4): 367-402. Shinnie, P. L., & Shinnie, M. 1978. Debeira West, a Mediaeval Nubian Town. Warminster: Aris & Phillips. Spaer, M. 1992. The Islamic glass bracelets of Palestine: Preliminary findings. Journal of Glass Studies 34: 44-62. Tavernari, C. 2009. Medieval road caravanserais in Syria: An archaeological approach. In P. Matthiae, F. Pinnock, L. Nigro and N. Marchetti: Proceedings of the 6th International Congress on the Archaeology of the Ancient Near East, May, 5thtenth 2009, “Sapienza” - Università di Roma, 191-205. Wiesbaden: Harrassowitz. Thareani-Sussely, Y. 2007. Ancient Caravanserais: An Archaeological View from 'Aroer. Levant 39(1): 123-141. Tomber, R. 2012. Indo-Roman trade: from pots to pepper. London: Bloomsbury. Turton, E.R. 1970. Kirk and the Egyptian Invasion of East Africa in 1875: A Reassessment. The Journal of African History 11(3): 355-370. Um, N, 2012. Reflections on the Red Sea Style: Beyond the Surface of Coastal Architecture. Northeast African Studies 12(1): 243-271. Whitcomb, D.S. 1983. Islamic glass from Al-Qadim, Egypt. Journal of Glass Studies 25: 101-108. Zhao, B. 2015. Chinese-style ceramics in East Africa from the 9th to sixteenth century: A case of changing value and symbols in the multi-partner global trade. Afriques. Débats, Méthodes et Terrains d’Histoire 6.