Looking east to the Jarigole hills.

Evidence for Early Trade Between the Coast and Interior of East Africa

Prepared for the WAC Mombassa Intercongress Conference Volume

©1993 by
Charles M. Nelson
Department of History
University of Nairobi

Figure 2. Strigatella paupercula from Jarigole.

Beads made from the shells of Strigatella paupercula have been recovered from a Pastoral Neolithic cemetery on the shore of Lake Turkana, 840 kilometers from the Indian Ocean. Comparative ceramic analysis suggests that the site dates from about 4,000 BP. Deeply penetrating coastal/interior trade was established well before the rise of trading ports on the coast of Eastern Africa. Data associated with the later Pastoral Neolithic period, from 2,300 to 1,600 BP, suggest long continuity in coast interior exchange networks. Pastoral peoples, with their extended ranges and mobility, were well positioned to play an important role in facilitating this trade as it developed.

KEY WORDS: East Africa, Trade, Pastoral Neolithic

The Jarigole Burial Complex consists of a series sites which ring Lake Turkana (Fig. 1), including Jarigole, the Kalakol Pillar Site (Ng'amoritung'a II) (Lynch and Robbins 1978; Soper 1982), the Lothagam Pillar Site, and the Suguta Pillar Site (Robbins and Lynch 1975:8). The essential associations at these sites are low mounds of gravel into which bundles of bones and grave goods were buried, natural columns of basalt brought to the sites and erected adjacent to the mounds, and ceramics of Nderit Ware as one of the principle grave goods. The complex also includes smaller mound ossuaries which do not contain pillars, for example at Il Lokeridede, near Koobi Fora (Koch: Personal Communication), and very small sites which may contain interments or pits used to prepare the dead before their bones were moved to an ossuary (GbJj4 at Jarigole)(Kamau 1991). These smaller sites also contain ceramics of Nderit Ware.

Figure 1. Distribution of Nderit and Ileret Ware Sites. Nderit Ware (black plaques): 1, Jarigole, GbJj4; 2, Il Lokeridede, Dongodian, GaJi2, GaJi24, GaJj9; 3, Kalakol (Ng'amoritung'a II), Kalakol Road Site; 4, Kangantotha 5, Lothagam; 6, Apia, Bb-9, Bb-14, Site N, Site O, Site Q, GdJi2; 7, Suguta; 8, Stables Drift, Nderit Drift, Makalia Burial; 9 Lukenya (GvJm44); 10, Seronera. Ileret Ware (white plaques): 1, Ileret Stone Bowl Site; 2, GaJi13, GaJj13.

Only two pillar burial sites have been tested to date. The evidence for early trade reported in this paper comes from the pillar site at Jarigole (GbJj1) and is based on a ca. five percent sample of the central mound (Table 1). The stratigraphic context is simple, but chaotic. After construction, the central mound, which is 15 meters in diameter and one meter high, was used as a repository for bundles containing bones and grave goods which were brought to the ossuary for interment. The contents of these bundles were already fragmentary at the time they were buried. Repeated excavation of small pits to deposit these bundles ensured that earlier bundles were intruded and their contents scattered in the tailings and pit fillings of more recent interments, a process which further fragmented the artefacts and exposed bones and artefacts on the surface for long enough periods to be destroyed by weathering. Sherds with evidence of prior weathering at the surface are common. The weathering histories of some individual sherds reveal as many as two-and-one-half cycles of exposure and reburial (original burial, exposure, reburial, re-exposure, reburial). As a result, much of the original content of the mound has been lost and what remains has been badly fragmented. Nevertheless, it still contains the remains of some 200,000 artifacts (estimate of number of objects before fragmentation). A brief summary of these objects is presented in Table 1.

Figurines12200mostly local
Ceramic vessels of Nderit Ware>650>13,000about 40% made outside volcanic area
Ostrich Egg Shell Vessels6120Local
Ostrich Egg Shell Beads6,200120,000Local
Ivory Beads & Ornament Fragments19380Local
Ivory-like Mystery Beads7140Possibly tortoise toes
Bone Beads & Ornaments11220Local
Beads of Strigatella paupercula13260from coast of Indian Ocean
Other Shell Beads & Ornaments10200Local
Amazonite Beads & Pendants30600not local, possibly southern Ethiopia
Basement Complex Beads11220from outside volcanic area of basin
Carnelian Beads240Local
Agate Beads360Local
Calcite Beads19380Local
Soft Stone Beads (includes Fluorite)41820Origin uncertain
Phallic Ornaments51003 also included in bead categories
Small Stone Spheres360Local?
Polished Pebbles of Quartz11220from outside volcanic area of the basin
Porphyry Pebbles240Origin uncertain
Pestle Rubbers (for ochre?)240from outside volcanic area of the basin
Large Palette120Local
Small Whet Stones6120Local
Flaked Basalt72514,000Local
Flaked Silica1703,400Local
Flaked Vein Quartz7140from outside volcanic area of the basin
Flaked Obsidian2,24044,800nearest source about 80 km.
Table 1. Contents of the mound at the Jarigole Pillar Site (GbJj1). Bold = transported more than 60 kilometers.

The people who created the Jarigole Ossuary Complex were pastoralists. We know this for two reasons. First, the Jarigole Pillar Site has yielded four cattle figurines and one figurine which may be that of a sheep. Second, the site of Dongodian, which contains an Nderit ceramic assemblage like that from Jarigole, also contains the bones of cattle and ovicaprids (Barthleme 1984). Sites containing Nderit Ware are distributed widely in the Turkana Basin at elevations which are more than 60 meters above that of the present lake (Robbins and Lynch 1975:14; Barthleme 1981:548). They are conspicuously absent from lower localities, such as Eliye Springs, which have long and diverse records of occupation. These facts suggest occupation of the basin at a time when the lake was consistently much higher than today, but well below the overflow level.

Most Nderit sites are relatively small and larger sites, such as Dongodian, are quite probably a palimpsest of sporadic seasonal occupations, quite possibly over long periods. No structural remains have been found and both stone tool and ceramic densities are modest when compared with many later Pastoral Neolithic occurrences from the Central Highlands. Nderit Ware develops into Ileret Ware in the northeast quadrant of Turkana basin. Sites containing Ileret Ware reveal the same basic site patterns as their earlier counterparts, but developed forms of the ware, at GaJj13 and GaJi13, are found at lower elevations, suggesting that the lake fell significantly during the Ileret Period.

The picture which emerges is that of a pastoral society which occupied a very large area and may have been fully nomadic, but which made what is arguably the most complex of all Pastoral Neolithic ceramic wares and buried many of its dead in central cemetery sites that required a great amount of labor to construct.

Direct dates for the Jarigole Ossuary Complex have yet to be obtained. The Jarigole Pillar Site, though made of well-drained gravel and in an arid environment, is none-the-less quite active geochemically, making dating a complicated issue in the absence of well associated charcoal samples. For the moment, we must fall back on ceramic correlation with assemblages which are reasonably well dated. Fortunately, there is a clear evolution of Ileret Ware out of Nderit Ware which involves changes in motif frequencies, basic decorative techniques, the specific techniques of design element production, design structure within motifs, rim forms, rim decorations, and vessel shapes. For example, one of the major motif clusters in Nderit Ware involves preparing the surface of the vessel with a fine repetitive design which is used as a ground over which broad, curvilinear channels are inscribed to form a bold design. This motif cluster survives as a minor element in the earliest stage of Ileret Ware, but repetitive ground design is made using a typical Ileret decorative technique and decorative implement rather than one of the ground techniques typical of Nderit Ware. Several lines of evidence such as this all indicate that the Ileret Stone Bowl Site (Barthleme 1981:260-301, 1984, 1985) represents an early stage in the development of Ileret Ware out of Nderit Ware, and that Dongodian (Barthleme 1981:204-259, 1984, 1985), GaJi2 (Barthleme 1981:182-203, 1985), Apia (Robbins and Lynch 1975:13), Kangatotha and Site Bb-14 (Robbins 1972) are all mainstream Nderit sites. The C-14 dates from these sites are presented in Table 2.

Site SASES Date Lab No Material and Comments
ApiaGdJi25,420 80UCLA-2124MCharcoal
Kangatotha4,800 100Lake Sediments w/Nderit Occurrence
Bb-145,020 220N-814Charcoal
Karari RidgeGaJi24,160 110SUA-634Charcoal from Lower Horizon
Karari RidgeGaJi23,970 60P-2609Charcoal from Lower Horizon
DongodianGaJi43,890 60P-2610Charcoal
DongodianGaJi43,945 135SUA-637Charcoal
DongodianGaJi44,100 125SUA-637BHumic acid from charcoal
DongodianGaJi43,405 130GX-4642-IAApatite Lev 3 Hor B
DongodianGaJi4noneGX-4642-IGGelatin fraction not preserved
DongodianGaJi44,580 170GX-4642-IIAApatite (fish bone) Lev 3 Hor B
DongodianGaJi4noneGX-4642-IIGGelatin fraction not preserved
Ileret Stone Bowl SiteFwJj54,000 140GX-4643AApatite Spits 2-6
Ileret Stone Bowl SiteFwJj5noneGX-4643GGelatin fraction not preserved
Six Mile Airstrip, Bur 2GaJj93,125 210GX-6400AHuman bone, apatite fraction
Six Mile Airstrip, Bur 2GaJj9noneGX-6400GHuman bone (gelatin poorly preserved)
Six Mile Airstrip, Bur 1GaJj9ModernGX-4641AHuman bone (altered by exposure?)
Six Mile Airstrip, Bur 1GaJj9noneGX-4641GHuman bone (gelatin poorly preserved)

Table 2. Radiocarbon dates from sites in the Turkana Basin with Nderit Ware and Ileret Ware.

The Ileret Stone Bowl Site has a C-13 corrected bone apatite date of 4,000 140 BP (3,660 140 BP, uncorrected), while Dongodian, which has the most recent dates associated with Nderit Ware in the Turkana basin, has a wide spread of dates on a variety of materials. The most reliable of these are most likely the two uncorrected charcoal dates of 3,890 60 BP and 3,945 135 BP. These agree closely with the uncorrected charcoal dates of 4,160 110 and 3,970 60 from a second site on Karari Ridge, GaJi2. Given our lack of experience with the C-13 correction of charcoal dates in the active geochemical environments of the Turkana basin, it is probably not wise to rely on an arithmetic calibration of these dates. C-13 correction would certainly have increased their age, but by exactly how much is uncertain. This suite of dates suggests that the changeover from Nderit Ware to Ileret Ware occurred about 4,000 BP, while the dates from Apia, Bb-14 and Kangatotha suggest that the Nderit occupation of the Turkana basin could approach a thousand years in duration, ample time for the creation and use of centrally located cemetaries.

There is, however, one clearly anomalous date in the series: 3,125 210 BP from the Six Mile Airstrip burial site (Barthleme 1981:357-366). Two partially in tact burials were excavated here about 1.5 meters apart. Burial 1 contains two vessels which are comfortably accommodated within Nderit Ware, but could possibly come from an early phase of Ileret Ware. Fragments of human bone from this burial produced a modern date based on apatite and contained gelatin which was too poorly preserved to date. The bones from this burial are described as partly fossilized by Barthleme and as being an advanced state of weathering by the lab. The apatite produced a C-13 value of -10.3, considered anomalous for uncontaminated human bone. Burial 2 contained an ostrich egg shell necklace but no ceramics. It produced the date of 3,125 BP on apatite, but gelatin was again too poorly preserved to date. Barthleme describes the bone from this burial as highly weathered but not as being partially fossilized. The poor condition of the bone raises questions about the accuracy of the date while the differences in preservation between the two burials raises a further question of association since the date is not on the burial containing the probable Nderit pots. For these reasons, the date should be set aside until corroborated.

Strigatella paupercula has been identified by Dr. Pieter W. Kat (National Museums of Kenya) on the basis of 13 specimens from the central mound at the Jarigole Pillar Site. It is a small snail common on the beaches of the Indian Ocean from South Africa to the Horn. Fresh, the shell is typically white and brownish red. Some of the specimens from Jarigole preserve this reddish pigment in narrow bands that spiral up the body of the shell, following its whorls. The color is similar to that of many of the ceramic vessels, which were finished with mineral slips and paints that fired to various hues of red and orange. Other specimens are weathered until chalky at the surface. These do not preserve the pigmentation. Variation in the condition of the beads is probably due to exposure and reburial in the same fashion as that described for the ceramics.

Figure 3. Strigatella beads with surviving pigmentation. White bar is 1 cm.

Strigatella beads were manufactured by grinding away the surface on the face opposite the mouth of the shell in the same manner as the money cowry is prepared for sewing onto clothing (Figs. 2 & 3). Usually the ground opening is formed by a single facet parallel to the ventral face of the shell, but sometimes there are two facets (Fig. 3, right) forming the opening as if the bead were designed to be laid in the angle between two intersecting surfaces or doubled side-to-side with a second bead. Fishermen on the coast of Kenya commonly use Strigatella as bait and prepare them by breaking through the back of the shell to expose the flesh within. The specimens from Jarigole are carefully ground and were probably sewn on articles of clothing, but they could have begun their cultural careers as some one's bait before being turned into ornaments.

Strigatella paupercula is a very common shell on the East African coast. Its supply is virtually infinite when considered in economic terms. Why, then, is it important when there are about 260 such beads in a 4,000 year old cemetary on the shores of Lake Turkana, a cemetery that contains some 200,000 objects? This question needs to be placed into a broader context and given the perspective of scale by contrasting the transport of Strigatella shells, which were brought some 840 kilometers from the cost, with the transport of more local items of trade or portage that occur at Jarigole.

Of particular interest among these are beads and pendants of Amazonite, a gem quality form of feldspar. These come from a single source which must be very small in extent and located someplace in the basement complex, most probably to the east or northeast. No known Kenyan source matches the specimens from Jarigole, but there is an Ethiopian source which might. In any event, what we know of the regional geology suggests a transport distance between 150 and 300 kilometers. Amazonite is about twice as abundant as Strigatella shells at Jarigole. Local raw materials of hard stone, such as agate, are abundant and widespread in alluvial sources, but beads of these materials are less common than those of Amazonite. They are also not as well finished and usually have larger perforations, which might suggest that the Amazonite is being traded mostly in the form of finished beads rather than raw material.

On the other hand, ceramics which contain no obvious volcanic temper but of quartz derived from the basement complex, are common. These must have been made at least 80 to 120 kilometers distant. Likewise, artefacts of obsidian, most of which are likely to come from a mere 60 to 80 km. to the north, number in the thousands.

Taken together, these facts suggest that the obsidian and ceramics are coming from within the territory of the kin group(s) using the cemetary and are subject to conditions of internal trade and portage, while the Amazonite and Strigatella come from outside this territory and are subject to conditions of external trade and portage. In these circumstances, the low frequencies of externally procured items is suggestive of two interpretations. First, direct long-distance trade between the coast and the Turkana Basin is highly unlikely. There should be a much higher frequency of exotic items among the artefacts from the Jarigole Pillar Site if such trade existed. Second, a trickle-in model of exchange without professional traders in the interior hinterland is sufficient to explain the material at Jarigole.

Nderit Ware and the Jarigole Burial Complex are associated with the earliest well-known, fully pastoral society in eastern Africa. The design structure and production technology of its ceramics are arguably the most diverse and complex of any Pastoral Neolithic society in the region. The cemetaries include the largest and most labor intensive cemeteries associated with pastoral societies in eastern Africa.

Nderit Ware spread southward along the rift and in the adjoining highlands as far as Tanzania at an early date (Leakey 1931; Bower 1973; Bower et al. 1977:140-1; Nelson 1976) and is probably the base-line pastoral occupation in most parts of this central corridor. Sites with early stylistic expressions of Nderit Ware are shown in Figure 1. At present, however, the only well associated radiocarbon date from south of the Turkana Basin comes from GvJm44 at Lukenya Hill. Here the lowest occurrence contains Nderit Ware and is associated with a date of 3,290 145 (GX-5348, charcoal, C-13 corrected; the uncorrected age is 3,285 145). There is also a limiting date from Seronera of 2,020 BP at SE2 (Bower 1973:85). What follows the spread of Nderit Ware in this entire region, including the Turkana Basin, is the simplification of the material expression of the ceramic and funerary traditions. It is fair to ask to what extent this apparent simplification reflects a broader simplification of social and economic organization, including participation in long distance exchange networks.

The problem of sample structures makes it almost impossible to answer this question in any straight-forward or definitive way. If the makers of Nderit Ware in the Turkana basin did not have central cemetaries which they packed full of material goods, we would not have found the early evidence for trade with the coast. Changing burial practices, which include the evident proscription first of ostrich egg shell and ceramics, and later of all ornaments as appropriate grave goods, means that we must rely almost exclusively on open sites for the recovery of traded shells. In this context, both locally made and traded ornaments are very uncommon.

Nevertheless, some interesting points can be made. The one complex cemetery in which grave goods are abundant is Njoro River Cave (Leakey and Leakey 1950), dating to 3,000 BP (Merrick and Monaghan 1984). Although stone beads were relatively common here, ostrich egg shell beads and shell beads of coastal origin are not present. It is interesting, however, that the stone beads include one Amazonite specimen which is visually identical to the specimens from Jarigole, indicating a southern extension of the exchange network in operation in the Jarigole region some centuries earlier.

Although no direct evidence of coastal trade in the interior is reported in the literature, there are a few fragments of pendants made from marine shells from the central ash deposit at Vaave Makongo (GvJm48) on Lukenya Hill, just east of Nairobi (Nelson 1976). This is a Pastoral Neolithic occurrence with a date on the central ash heap of 1,810 135 BP (GX-5347G) on bone gelatin. In addition, obsidian was transported widely and in great quantities during the Pastoral Neolithic period. Raw material from the Naivasha basin in the Central Rift Valley is found as far east as Kulungu (Merrick and Brown 1984), south of Machacos, and the foothills of Mt. Kilimanjaro (Mturi 1986). These localities are about 400 kilometers from the coast and range from 140 to 250 kilometers from the obsidian sources, and well east of Lukenya Hill, where the pendent fragments of marine shell are associated with abundant obsidian debris and tools. On the east side of the Rift, overlap in the distribution of coastal shells and obsidian from the Central Rift shows that in exchange network extended from the Rift to the coast. On the west side of the Rift, obsidian from the same sources was transported nearly to the shores of Lake Victoria, a distance of 170 kilometers (Robertshaw 1990). Thus, it is likely that the exchange network extended all the way from the coast to Lake Victoria.

Although the evidence is scanty, it is clear that, once established, pastoral exchange networks endured and were available for development as trading ports emerged along the coast of eastern Africa. However, a great deal more work needs to be done to determine to what extent, and how early, such pastoral networks were actually used.

The trade of shells from the coast of East Africa to far off Jarigole, however modest, is nevertheless quite significant. It demonstrates that there was an extensive trading network in existence before the rise of coastal trading ports or the involvement of interior-derived commodities in an international system of trade between states. This pre-existing trading network is likely to have been the basis for the early stages of larger-scale coastal/interior trade.

More interesting still is the fact that pastoralists were involved in extensive trading networks from their earliest well-documented penetration into Eastern Africa. Pastoral peoples maintain large territories within which they are highly mobile and they maintain far-flung contacts to facilitate the movement and pasturage of stock during times of draught. This also places them in a position to facilitate trade in ways which hunters and gatherers or hill-bound agriculturists can't. Among these are armed escorts, locally supplied beasts of burden, and safe passage for considerable distances. By 2,000 years ago, pastoral communities dominated the landscape from Kilimanjaro to Lake Victoria and from Lake Eyasi to the Ethiopian Highlands, a vast potential market already accessible through established exchange networks.

I would like to thank the Koobi Fora Field School of Palaeoanthropology and Archaeology for supporting the excavation and survey work on which this paper is based, and the Fulbright Exchange Program for supporting my presentation to the conference and the comparative research necessary for the preparation of the manuscript.

Barthleme, J.W.
1981Late Pleistocene-Holocene Prehistory to the Northeast of Lake Turkana, Kenya. Ph.D. Dissertation, University of California at Berkeley. Anthropology.
1984Early Evidence for Animal Domestication in Eastern Africa. Hunters to Farmers: The Causes and Consequences of Food Production in Africa (J.D. Clark and S.A. Brandt, eds.), 200-203. University of California Press.
1985Fisher-Hunters and Neolithic Pastoralists in East Turkana, Kenya. BAR International Series, 254. Oxford.
Bower, J.R.F.
1973Seronera: Excavations at a Stone Bowl Site in the Serengeti National Park, Tanzania. Azania, 8:71-104.
Bower, J.R.F., C.M. Nelson, A.F. Waibel and S. Wandibba
1977The University of Massachusetts' Later Stone Age/Pastoral 'Neolithic" Comparative Study in Central Kenya: an Overview. Azania: 12, 119-146.
Kamau, John
1991Report on a "Stone Pillar" Site Northeast of Jarigole Burial Mound, East Turkana. Nyama Akuma, 36:6-8.
Leakey, L.S.B.
1931The Stone Age Cultures of Kenya Colony. Cambridge University Press.
Leakey, M.D., and L.S.B. Leakey
1950Excavations at the Njoro River Cave. Clarendon Press, Oxford.
Lynch, B.M., and L. Robbins
1978Namoratunga: the First Archaeoastronomical Evidence in Sub-Saharan Africa. Science, 200:766-8.
Mturi, A.A.
1986The Pastoral Neolithic of West Kilimanjaro. Azania, 21:53-64.
Nelson, C.M.
1976Results of Salvage Archaeology at Lukenya Hill, Near Nairobi, Kenya, unpublished report on behalf of the National Museums of Kenya, for the Department of Tourism and Wildlife, Government of Kenya, Nairobi. 30 pp. + 28 figs. and tabs. On file, National Museums of Kenya.
1995Work of the Koobi Fora Field School at the Jarigole Pillar Site. Kenya Past and Present, 27:49-63. Kenya Museum Society, Nairobi.
Merrick, H.V.M., and F. Brown
1984Obsidian Sources and Patterns of Source Utilization in Kenya and Northern Tanzania: Some Initial Findings. The African Archaeological Review, 2:129-152.
Merrick, H.V.M., and M. L. Monaghan
1984The Date of the Cremated Burials in Njoro River Cave. Azania, 19:7-12.
Robbins, L.H.
1972Archaeology in the Turkana District, Kenya. Science, 176:359-66.
Robbins, L.H., and B.M. Lynch
1975Prehistory and Paleoenvironment in The Lake Turkana Basin: Final Report on Archaeological Fieldwork Carried Out in Turkana District, Kenya. National Museums of Kenya and The National Science Foundation.
Robertshaw, P.T.
1990The Early Pastoralists of Southwestern Kenya. British Institute of Eastern Africa. Nairobi.
Soper, Robert
1982Archaeo-astrnomical Cushites: Some Comments. Azania, 17:145-162.