Looking northwest across Jarigole to Mount Sibiloi.

The Work of the Koobi Fora Field School at the Jarigole Pillar Site
© 1995 by Charles M. Nelson
Published in Kenya Past and Present, 27:49-63. Nairobi.

The people who made Nderit pottery were among the first pastoralists to herd their cattle, sheep and goats in Eastern Africa. Charles Nelson describes one of their chief cemeteries overlooking Lake Turkana and the complex ceramics that are the hallmark of their culture.

The Koobi Fora Field School of Archaeology and Palaeoanthropology
The Koobi Fora Field School (KFFS) is a programme run jointly by the National Museums of Kenya and Harvard University. Its purpose is to introduce students to the areas of knowledge they must master if they are to work in the field as palaeoanthropologists. Each year about 45 students participate in the Field School, including a number of Kenyans who are supported by international scholarships made available through the National Museums.

One of the sites which KFFS uses to train students is the Jarigole Pillar Site (GbJj1), a Pastoral Neolithic cemetery situated at the foot of the Jarigole Hills near Alia Bay and Mt. Sibilot. This site is used to teach students excavation techniques, how to take excavation notes, field cataloging procedures, piece plotting of individual artefacts, instrument leveling for site mapping and excavation, and how to recognize and recover small fragments of human bone. Because artefacts and bone fragments are very common, three days of work at Jarigole is equivalent to three weeks of work at a stone age site. Jarigole is also used to show students how research questions are formulated and answered in the context of site excavation.

Figure 1. Location of Jarigole and
selected Pastoral Neolithic sites in
the Turkana Basin.

The Jarigole Pillar Site is one of five known archaeological sites in the Turkana Basin that combine large, low mounds, pillars, platforms and cairns (Figure 1). Robbins and Lynch measured pillar alignments at the Kalakol Pillar Site and argued that it was an astronomical observatory made by Cushitic speakers and dating to the first millennium B.C. Re-measurement of the pillar alignments and other observations by Soper(1) and Stiles(2) show that this interpretation is not likely to be correct.

When Frank Brown and Craig Feibel discovered the Jarigole Pillar Site, the Field School was provided with a marvelous opportunity to test the likelihood of the interpretation offered by Lynch and Robbins for pillar sites. The obvious alternative interpretation is, of course, that these pillar sites are cemeteries since all of their defining characteristics are also found associated individually with burials dating to a number of different periods within the basin. So it was decided to excavate the site to determine if its primary function was that of a cemetery. Not only did this prove to be the case, but the material recovered from the first season of work, in 1986, showed that the Jarigole Pillar Site would provide the first definitive look at Nderit Ware, the most ancient pottery known to be associated with pastoralists in East Africa.

This article has two purposes. The first is to report the preliminary results of the Field School's work at Jarigole, particularly with respect to the structure and contents of the site. The second is to give the readers of Kenya Past and Present an idea of how archaeologists learn about the prehistoric past.

Site Environment
The Jarigole Pillar Site is situated on an ancient beach which was formed when Lake Turkana receded from its last major high stand around 6,000 years ago. This high stand, which occurred when the lake was at overflow, is about 80 metres above the 1973 lake level and the Pillar Site is about 70 metres above. The four well-documented pillar sites in the basin were each constructed on this recessional beach and at about this elevation. The pottery from these four sites all belongs to a variety known as Nderit Ware. Assemblages of Nderit Ware have also been recovered from a number of sites where early pastoral peoples had villages or encampments some of which, such as those from Dongodian, were at or very close to the lake shore. These are all found well above the modern level of the lake, while low-elevation sites in prime localities, such as Eliyie Springs and Koobi Fora Spit, have no trace of Nderit Ware despite their repeated occupations. For these reasons, we know that the lake was probably 30 to 40 metres above its 1973 level at the time the Jarigole Pillar Site was constructed.

Although the lake has a long history of a fluctuating water level, the overall trend has been one of decline during the last 6,000 years. Brief declines to modern or even lower levels have been followed by rebounds to high, relatively stable stands, but each rebound has brought the lake back to a lower elevation than pervious high stands. It is likely that the Jarigole Pillar Site was used during a period in which the lake was consistently high, but substantially below its overflow level. At the time the site was in use, it is probable that the shore was somewhere near the toe of the high alluvial apron which slopes gently away from the base of the Jarigole Hills and has been planed flat by the ancient advance and retreat of Lake Turkana. Much of the broad, shallow valley which separates the Jarigole Hills from Mt. Sibilot, would have been inundated, forming an extensive embayment to the east of the site.

Today, Jarigole Laga runs from the heart of the Jarigole Hills past the spring known as the Jarigole Water Hole, and then northward across the alluvial apron into Alia Bay. It has deeply dissected the apron and the underlying sediments of the Koobi Fora Formation, forming what is now a broad, stepped valley which has been repeatedly eroded and refilled with alluvial debris washed down from the hills. A preliminary survey of the laga suggests that this process began before Lake Turkana last rose to its overflow level. It appears that Jarigole Laga was well established before this event, though its valley was not as deep or nearly so wide. When the lake receded, this valley became a long, shallow lagoon with its head perhaps as little as one kilometre northwest the site.

Thus, the Jarigole Pillar Site was situated at the base of a peninsula which ran northward into the lake, with broad prospects of water to the east and north, easy access to the lake via a sheltered lagoon, and a permanent source of fresh water close by just within the hills. The basin-wide environment was certainly somewhat moister than today, with extended and wetter rainy seasons, and the more extensive pasturage which this supported. Water tables were higher and supported more luxuriant gallery forests along seasonal water courses.

This environment was conducive to pastoralism. Food remains from sites with Nderit pottery at Koobi Fora and ceramic figurines recovered from the Jarigole Pillar Site indicate an economy based mainly on cattle, sheep and goats, and supplemented by fishing.

Site Structure
Unlike older stone age sites, which owe the bulk of their structure to geological processes such as the accumulation of slope wash or floodplain sediments, Jarigole is wholly an artificial construction made by a community of people about 4,000 years ago and most likely used by them for several generations as shown below in Figure 2.

Figure 2. Map of the Jarigole Pillar Site Showing the Main Structural Features (Detailed Version Allow time to load).

The first step in constructing the site was to make a large, oval platform contained by a low curb of large cobbles. This platform is only about 15 cm to 20 cm high, but it covers an area of more than 1,000 square metres and contains at least 150 cubic metres of sediment which was dug or scraped from surrounding areas on the beach. The platform became the site of two construction activities. A circular mound was built slightly to one side of the centre of the platform. It is about 1 metre high, 26 metres in diameter, and contains about 100 cubic metres of the same material of which the platform is constructed. In the open area of the platform, to one side of the mound, more than 28 pillars of columnar basalt were erected. These pillars each weigh hundreds of kilos and were transported about two kilometers from the top of a ridge midway between the site and the Jarigole Water Hole on the west side of Jarigole Laga.

The labour involved in the initial construction of the site was substantial, probably something in the neighborhood of 175 to 225 man-days, not counting ritual activities or the construction of perishable surface structures of wood and fibre.

The surfaces of the mound and platform host a number of secondary features which intrude, damage and cap the primary construction features at the site. These may have been added during the use of the site or at some time after the site had been abandoned by its makers. These include a small cairn in the centre of the mound, small piles of stone accumulated around the bases of pillars and around fallen pillar fragments possibly from piling stones on the tops of pillars, and scatters of calcite pebbles which encompass some of the small pillar cairns. A few of the pillars and two rocks on the surface of the mound are decorated with simple designs which have been pecked into the stone. These appear much fresher than the pillar surfaces and could have been added long after the site had been abandoned. Similar petroglyphs may be found widely scattered on rocks near the top of the ridge which overlooks the site. Finally, some pillar fragments were carried from the site to make two small enclosures along the margin of a large pastoralist encampment which probably dates from sometime during the past 50 years.

Excavation Strategy
The structure of stone age sites is usually dominated by the geological processes that laid down their sediments. The people who created such sites adapted their activities to the evolving landscape and did little to alter its structure. For this reason, excavation trenches and grids at stone age sites are set out so that one axis of the excavation runs in the same direction as the energy which deposited the sediment, while the other axis is oriented at right angles to the direction of this geological process. For example, in a floodplain, one axis of an excavation would be set out parallel with the river and the others at right angles to the river. This procedure allows the structure of a site to be visualized clearly and makes it easier to record and interpret variation in the sediments within the site. It also makes it easier to recognize subtle anomalies in the micro-structure of sediments that are the product of human activity.

The surface structure of the Jarigole Pillar Site presents a very different problem. This is a site which is almost entirely the product of human labour, imposed on a stable land form on which there has been very little deposition while the site was in use or over the millennia since its abandonment. At localities where cultural activities control site structure, archaeologists hope to use an excavation strategy which aligns trenches and grids to the shape of the spaces being used and to the intensity of the activity within each space. For example, in a settlement of rectangular houses with standing walls which delineate rooms, each room would be excavated as a separate space with its own grid usually set out parallel with the long axis of the room. The Jarigole Pillar Site presents a set of concentric circular spaces: curb, platform and mound. There is little to hint at the details of subsurface structures. Since a simple circle has no preferred orientation, there is no obvious basis for selecting a particular orientation. That is why the grid for mapping and excavation is simply set out to the cardinal points, with North/South and East/West axes (Figure 2). It is impractical to use a polar coordinate grid because it would require that excavation pits approximate pieces of pie, so a normal rectangular grid is employed. This can be manipulated so that two grid lines will intersect at right angles in the centre of the mound, quartering it. This is why one edge of our trench has been set out on the East 15 metre grid line. In addition, you will notice two test pits on the platform to confirm its structure and two probes off site to document its geological context. One of these produced a harpoon fragment associated with the beach horizon.

Site Function
The central part of the mound, in a circular area between 12 and 15 metres in diameter, was used to dig burial pits (Figure 2). So many pits were dug that there is no part of the excavated portion of this central space which does not contain an overlapping series of pits. To fill the central area, which covers between 115 and 160 square meters, with overlapping pits each perhaps a metre in diameter, must have required the excavation of several hundred individual pits in many separate interment episodes. We can tell this because no virgin construction fill from the building of the platform and mound survives in the excavated portion of the central burial area, which is 6% of the surface area of the mound. Throughout the excavation, the construction fill contains high densities of fragmentary bones and artefacts which have been mixed into the fill as new burials intruded into and scattered the contents of the old.

Our test trench is only just penetrating the lower part of the fill between 65 and 90 cm below the surface. So far, only one intact burial has been found, with the base only 60 to 65 cm below the surface of the mound. It contains an elderly individual, tightly flexed, who was covered by a single layer of large rocks. The rest of the burial pit was filled with the debris removed during its excavation. Although not yet fully excavated, this burial is already affecting the way we think about the site. Four points are of particular interest.

First, if large rocks were customarily used to cover a body in the bottom of the burial pit, it would help explain the amount of mixing that has gone on in the mound. This seems likely since we are beginning to uncover other clusters of large rocks which appear to outline pits. Since large rocks would be difficult to dig out of the fabric of the mound when working in a small pit at considerable depth, it is likely that most would have to be collected from the beach nearby. The volume of the body together with the volume of the rocks would mean that a substantial fraction of the debris excavated from the burial pit would remain on the surface after it was completely filled. This material would tend to spread laterally before some of it was reincorporated more deeply into the mound in the process of filling other burial pits. This interpretation is supported by the fact we find numerous pot sherds within the mound which have been weathered by exposure on the surface and then reburied. Some have been weathered on opposite ends or surfaces at two different times indicating a cycle of burial, exposure, reburial, exposure and reburial yet again.

Second, the addition of rocks in the process of burial would add substantially to the volume of the mound, itself. Rocks from the first excavated burial occupy between 0.1 and 0.2 of a cubic metre. The most conservative possible interpretation of the extent and internal structure of the inner pit area yields a minimum of 100 burial pits, that is between 10 and 20 cubic metres of large rocks added in the process of interring bodies. This is between 10% and 20% of the entire volume of the mound. The number of burial pits likely to me in the mound is two to five times this minimum estimate, so the mound is likely to have grown substantially simply through use.

Third, although the first intact burial is only partly uncovered, no associated grave goods have yet been found. Given the average size of vessels at the site, it seems likely that whole vessels would be visible by this point. There should also be some evidence for ear spools or necklaces if these were being worn. So it is likely that this first burial comes to us with no grave goods or very few grave goods, in stark contrast to what must be the norm for this site. What could this mean? It could reflect differences in age, sex, personal wealth or status, resulting in some individuals being buried with nothing while others are buried with substantial amounts of material. It could indicate a radical change in custom involving the proscription of grave goods shortly before the cemetery was abandoned. Or it could mean that the mound was used much more recently by an entirely different group of people; that the burial represents an opportunistic use of a preexisting structure. This last possibility seems least likely given the condition of the skeleton, the lack of a surface cairn and the presence of other subsurface rock features with a similar structure.

Fourth, in the early stages of work at Jarigole, the extremely fragmentary nature of the human bone and artifacts suggested that we might be dealing not with the interment of bodies but rather the burial of bundles of bones and artefacts which had already been exposed or broken. The proposition is that a pastoral people spread over hundreds of kilometers would have a difficult time in moving bodies to a central location for burial. Prepared bundles would be far easier to move and the schedule for their movement and burial more easily managed. This might still be correct for some proportion of the burials, but the evidence from the first burial we have encountered, a flexed inhumation, makes such an interpretation less likely. However, if inhumation is the norm, then the site may have been used longer or more intensively than first suspected.

The platform has been tested in one place immediately adjacent to the mound at the base of the main trench and in two places away from the mound adjacent to the pillar field. These tests reveal no intrusive pits. However, the area immediately adjacent to the mound contains a substantial amount of debris, such as pot sherds and beads, which have broadcast from the mound due to the digging of pits there. Ms. Veronika Waweru, a Field School student from the University of Nairobi, has made a detailed comparison of the contents of the mound and the platform, focusing particularly on flaked stone artefacts of basalt. Her study shows that there was a much higher rate of stone tool preparation on the platform than on the mound, adding yet another dimension to the functional separation of these spaces within the site.

The function of the pillars erected in the platform remains unknown. No work on pillar alignments has been done because all of the pillars have been broken and only a few fully reconstructed. Many are also badly tilted and are probably shifted from their original position. Therefore, it will not be possible to determine if there are significant astronomical alignments among the pillars. However, since we have demonstrated that the most obvious function of the Jarigole Pillar Site was as a cemetery, the simplest explanation would be that the pillars are memorial markers erected in honor of the dead.

Site Age
The Jarigole Pillar Site has not yet been dated using radiocarbon or any other absolute dating method. We are waiting to recover suitable samples from in situ burial remnants so that we can date events accurately in the history of the construction and use of the site. However, there are numerous dates associated with assemblages of Nderit Ware from the Turkana Basin. The most reliable of these dates come from charcoal and are summarized in Table 1.

SITE Ref. SASES   Date BP     Lab. No.
Apia 3 GdJi2 5,420   80 UCLA-2124M
Bb-14 4 5,020 220 N-814
Dongodien Upper 5 GaJi2 4,160 110 SUA-634
Dongodien Upper 5 GaJi2 3,970   60 P-2609
Dongodien Lower 5 GaJi4 3,945 135 USA-637
Dongodien Lower 5 GaJi4 3,890   60 P-2610
Table 1. Radiocarbon dates from charcoal associated with Nderit Ware in the Turkana Basin.

These dates provide the most probable range of use of the Jarigole Pillar Site. In addition, there is a third date on bone apatite, which is considered less reliable, of 4,000 140 BP associated with an Ileret Ware assemblage. Ileret Ware appears to develop out of Nderit Ware. Hence, any Ileret Ware assemblage should date from a more recent period. Given the potential error ranges for each of these dates, which are about twice the one standard deviation (plus/minus) given after each date, the changeover from Nderit Ware to Ileret Ware is likely to be about 3,900 to 4,000 BP. The Jarigole Pillar Site is likely to have been abandoned about this time. By the by, BP means 'Before Present' and 'Present' is defined as 1950 AD by international agreement. This is the date at which radiocarbon age determinations began to accumulate and also the date by which atmospheric radiocarbon pollution from atomic testing had altered its balance in the atmosphere.

Site Contents
About 40,000 artefacts have been recovered from the Jarigole Pillar Site. The vast majority of these come from the eight cubic metres of mound and platform fill removed from the main trench (Figure 2). Pottery and ostrich egg shell beads dominate the inventory, but there is a wide variety of other material objects

Shell Beads. A dozen beads have been recovered which are made from the shell of Strigatella paupercula by grinding away the back of the main whorl in the same way that cowries are ground to facilitate sewing unto clothing. This species is found on the shores of the Indian Ocean and demonstrates that the people who created the Jarigole Pillar Site had access to an extended trade network.

Amazonite Ornaments. More than 20 beads and pendants of blue Amazonite, a form of microcline feldspar, have been recovered. This material is not available in the Turkana Basin and the nearest source that might match the specimens from Jarigole is in southern Ethiopia, many kilometers to the east and north. A single specimen of the same material was found by Mary and Louis Leakey at Njoro River Cave.

Carnelian and Agate Ornaments. There are three large carnelian and three agate beads and pendants made from material found along the eastern margin of the Turkana Basin.

Fluorite (?) Beads. There are about a dozen beads which are violet, pale yellow or clear, that appear to be tube drilled. The color range and hardness is consistent with that of fluorite. The origin of these beads is unknown.

Calcite Ornaments. The assemblage includes more than 30 beads and pendants of calcite, many fashioned from layered plates of crystals that grow in soil cracks in the sediments of the Koobi Fora Formation.

Phallic Ornaments. There are eight objects carved from a variety of stones which appear to be phallic ornaments. Some (Figure 3) are quite graphic representations. These intergrade with more representational forms which might have other meanings.
Figure 3. Bead, possibly of calcite, in the shape of a phallus. Scale bar of 1 cm.

Other Stone Ornaments. There are about a dozen beads and pendants of other, unidentified rocks. The raw materials are all relatively soft and have weathered considerably since their manufacture.

Tubular Bone Beads. There are about eight short tubular beads made from bones, probably of rodents.

Tortoise Toe (?) Beads. There are a dozen beads that are flat on one surface and dome-shaped on the other, made from a very dense, grainless bone with rods of enamel-like material running between the surfaces. These have been tentatively identified as the terminal phalanges of tortoise or turtle with their bluntly pointed tips ground into a dome shape.

Ivory Disk Beads. There are a small number of ivory disk beads made in the same shape and size as ostrich egg shell beads. The exact number is difficult to determine because weathered ivory of certain kinds and cuts appears very similar to weathered ostrich egg shell. Both of these materials tend to become translucent and loose structural detail as they weather.

Ivory ring and bracelet fragments. There are small fragments of several finger rings and narrow bracelets of ivory, all probably undecorated.

Other Ivory Objects. There are several items which may be large ivory ear spools, each about 6 cm long and 2 cm in diameter, flanged and pointed at one end and cup-shaped at the other. There is also a fragmentary, handle-like object decorated with fields of incised dots and two perforated pieces which may be fragments of elongated pendants.

Ostrich Egg Shell Beads. There are nearly 7,000 disc beads of ostrich egg shell. Ninety-two of these were found in a small cluster as though they had been enclosed in a pouch. The others are scattered throughout the mound fill and, to a lesser extent, in the platform fill at the very edge of the mound. Mr. Veruku Mandu, a Field School student from the University of Nairobi, has done a detailed attribute analysis of these beads. It reveals quite a range of techniques in their manufacture suggesting that the beads may come from a variety of sources. There is also a bimodal distribution in diameter which argues for a range in decorative function.

Ground Stone Artefacts
Ground stone artefacts are relatively uncommon. They include oval whetstones 7 to 10 cm long; flat oval stones with faceted, polished edges, which may be pot burnishers; pestle rubbers and fragments of pallets that may be for grinding ochre; and spherical hammer stones with small ground facets or dimples.

The assemblage includes over a dozen well rounded, white quartz pebbles which are circular or oval in outline. They range from 2 to 4 cm in length and some have been slightly polished. There are also a small number of pebbles of porphoritic lava, flatter and slightly longer than the specimens of quartz.

Stone Bowls and Axes
These are conspicuously absent.

Flaked Stone Artefacts
Context. The fill in the mound and the platform is derived from beach and alluvial deposits which contain a low-density, off-site scatter, mostly of basalt but with small quantities of obsidian and silica as well. Therefore, the question arises to what extent the flaked stone artefacts in the fill were made by the people who built and used the cemetery and to what extent they are derived from the fill used to build the site. In addition, the entire surface of the old beach and site have been used by pastoralists since the abandonment of the cemetery, creating a thin scatter of flaked stone debris on its surface. Ms Veronika Waweru has solved this problem using flaked artefacts of basalt, which is the most abundant raw material both on and off the site. Specimens from the mound and platform fill which were waterworn from the process of beach formation or had been weathered on the surface before burial were eliminated from the comparison. Her analysis shows that the remaining specimens in the mound and platform, which are fresh, occur in much higher densities than they are found in geological probes near the site or on the old beach surface surrounding the site. This relationship is even stronger for flaked artefacts made from obsidian and silica. These facts demonstrate that the bulk of the flaked stone artefacts found in the mound and platform must be associated with cemetery construction and maintenance, and with activities carried on in the course of using the site.

Obsidian. Obsidian occurs in the northeast corner of the Turkana Basin, about 60 km to the north-north-east of Jarigole, in the form of small pebbles. It is likely that almost all of the obsidian at the site comes from this source. Not surprisingly, then, the size of the obsidian artefacts is very small, always less than 3 cm in length. Several hundred pieces of discarded waste, and smaller numbers of tools and cores have been recovered. The most common items which may have been used as tools are small, parallel-sided flakes between 1 and 2 cm long with thin sharp edges which would have been suitable for cutting soft materials. In addition, there are a small number of flakes with an edge reinforced by removing numerous small flakes which rise vertically from the dorsal and/or ventral surface of the flake. These are used to remove the thin, brittle portion of an edge and moves it into the centre of the flake where it is thickest. This creates a thick blunted (backed) edge which is both difficult to fracture and blunt enough so that it can be pressed very hard without fear of cutting the pressing finger. The edge opposite the blunted edge is left unmodified and therefore very sharp. The blunted edge allows this sharp edge to be manipulated at will.

A few of the specimens in this group may be barbs which were inserted into the edges of the tips of arrows. There are also a few thicker flakes which have convex scraping edges and a variety of small cores from which flakes were struck for subsequent use as tools.

Silica. Silica, or chalcedony, is common in the Turkana basin where it occurs as cavity fillings, in the form of agates, in volcanic rocks as a result of hydrothermal action, or as a precipitate in lake sediments, called chert, which accumulated in broad seams and nodules deposited when the ground water conditions beneath the lake were appropriate. There is also agatized wood from Mt. Sibilot, less than 10 km from the site.

Since silica is locally available in pieces the size of small cobbles, it is possible to make larger tools form it than one finds made from the local obsidian. It is also less brittle, providing a tougher working edge. Nevertheless, silica is the least common raw material from the site, with only about 100 fresh pieces which are likely to be associated with the use of the site. These include a few barbs for the tips of arrows, some borers of the sort which might have been used to manufacture beads, and a few small scrapers.

The relative abundance of obsidian, which must be carried a considerable distance, and the relative paucity of silica, which can be picked up in quantity not far from the site, suggests that obsidian was the preferred raw material at the cemetery. This might be either because it is exotic and therefore more valuable, and its discard at the site a sign of how the living value the dead, or because short razor-sharp edges are required for preparation of grave offerings or other activities.

There are many small fragments of tooth enamel which come from bovids which would be the same size as goat or sheep, and a small number of fragments from the teeth of cow-sized animals, which may be the remnants of funerary offerings or meals prepared for the living. There are also about two dozen very small unfossilized fish vertebrae such as those which might come from the tail of a fish where fillets might be left joined if fish were being dried on poles. Sharp obsidian flakes would be ideal for apportioning small quantities of meat and fish as offerings or servings.

Basalt. Basalt is the type of lava which composes the fault block which has been thrust up to form the Jarigole Hills. The Jarigole Pillar Site is situated only a few hundred metres from this fault and the base of the hills, where a number of different lava flows are exposed. Some of these are composed of very fine-grained basalt which is readily flaked. These fine-grained basalts are also more resistant to mechanical and chemical weathering. As a consequence they survive geological transport longer and are commonplace in the beach gravels which form the gently-sloping plain on which the site stands. So basalt was the most readily available of the raw materials used at the site. Accordingly, it is also used more than the other raw materials, forming an assemblage of more than 2,000 specimens.

Figure 4. Side scraper of basalt made by trimming one edge onto the release surface, which is created as the flake breaks away from the core. Modification of the release surface prepared a working edge. Its plano-clinal structure defines it as a scraping edge and the location as a side scraper.

Large basalt cores and waste from the manufacture of tools are particularly abundant. Tools consist mostly of thick, heavy-duty scrapers and smaller flake scrapers between 5 and 8 cm long. There are also occasional long flakes with blunted edges which were most likely used as hand-held backed knives.

Ethnoarchaeological Tools. Between sessions of the Field School, the trench is lined with pieces of plastic and back filled with sterile sand and gravel. Each year the Gabra frequent the area with their herds. When they discovered that we had buried plastic, they dug the plastic up because they found it useful. To do this, they manufactured digging implements out of large cobbles taken from the surface of the site. These are bifacially flaked at each end and closely resemble Middle Stone Age core axes of the kind common in the Sangoan technology. We have found six of these digging tools after their plastic raids. Archaeologically, we have tools of a similar shape and weight, but they are flaked in a different way and do not show the edge battering that the Gabra digging tools show.

Pottery Figurines
Fragments of several pottery figurines (Fig. 5) have been recovered from the Jarigole Pillar Site. These include three of cattle, one which appears to be a sheep, an equid (horses and related animals), an elephant, a giraffe, a leopard, a hippo and a bird with a throat which is painted white, as well as numerous limbs, horns and torso fragments. A small satellite site (GbJj4), located a few hundred metres to the north, has also produced figurine fragments on the surface. These include one figurine of a person. These are the oldest known figurines from East Africa and provide evidence that the people who built and used the Jarigole Pillar Site were pastoralists. They are also quite interesting anatomically because the artists have carefully reproduced certain anatomical details which are characteristic of a species while allowing others, such as limbs, to be more schematically represented. For example, the characteristic notch behind the horn of a cow, where the horn enters the skull, is very exactly reproduced. This is also true of the wattle of the sheep, raised spine of the elephant, the shoulders and neck of the equid, the spinal profile of the leopard, and the anatomy of the mouth of the hippo. The figurines vary widely in size from about 4 cm to 40 cm in length.

Figure 5. Cow (left) seen from the top, with head and neck turned to the right. The right horn is broken off but the angle suggests a different orientation than the left horn, indicating that horn-bending may have been practiced. Sheep (?)(right) with upper neck rough and wattle. Scale bars of 1 cm.

Importance of Ceramic Analysis
Utility. In the American Southwest, Pueblo bowls and ladles become larger and smaller as serving sizes change in response to the availability of food. In ancient Greece, the selection of a particular ceremonial serving ware was a public statement of allegiance to a particular religious and political faction. In Sub-Saharan Africa, the development and spread of certain ceramic wares is the most obvious evidence for the spread of Bantu speaking peoples and Iron Age technology. In Mesopotamia, ceramic tablets are the vehicle for the oldest known complex system of writing, cuneiform. Because ceramic technology is plastic, it can be molded to any number of needs, especially those in which cultural expression is also essential.

Pastoral Neolithic Ceramics. Pastoral Neolithic adaptations emerged first in northern Africa, becoming widespread by 7,000 BP and persist at least until the local introduction of iron technology or the development of ceramic craft specialization. In some regions they survive these innovations. The early ceramic hallmarks of these cultures are complex designs which use both texture and form to cover large areas on the outer surfaces of vessels. The classic textural devices are burnishing, impressions made by a stylus not unlike that used to produce cuneiform writing, combs with three to nine blunt teeth (Figure 7), and grooves, incisions and cuts made with stylus, comb or knife. Over time and locally, other textural devices are added, including the roulette and fibre impression. Commonly color is controlled through the firing process and slips of fine clay added to the surface of the vessel. Wares having these characteristics are widely associated with pastoral peoples over much of north Africa and the Sahel. They also define the Pastoral Neolithic wares of eastern Africa and set them apart from most subsequent Pastoral Iron Age wares. The ceramics found at the Jarigole Pillar Site belong to Nderit Ware, a Pastoral Neolithic ceramic which is distributed in and around the Gregory Rift from northern Kenya to central Tanzania.

Significance of Nderit Ware. Although the manner and date of the earliest spread of domestic animals and pastoral adaptations into East Africa is still an open question, the earliest well-dated sites and the first reasonably well-documented spread of a pastoral adaptation are associated with Nderit Ware. This ware is known particularly for a series of characteristic decorations which appear on the inside of vessels and is widely distributed in the Turkana Basin, where it is associated with radiocarbon dates ranging from 5,400 to 3,640years ago (Table 1). At Koobi Fora, John Barthelme(6) found it in association with domestic cattle and ovicaprids (goats and sheep; ovicaprid is used when species identification is uncertain), most importantly at the sites at Dongodian (GaJi2, GaJi4).

Nderit Ware is the most complex of all the Pastoral Neolithic ceramic wares. This complexity is reflected in an impressive range of decorative techniques, design motifs and vessel forms which are combined to create almost endless variation. Nderit Ware spread southward from the Turkana basin and became the basis for the development of a number of local Pastoral Neolithic wares along the Gregory Rift. This historical process was accompanied by the simplification of decorative technique, design principles and vessel forms in each of the daughter wares. How and why this occurred will occupy archaeologists for decades as they untangle the complex web of relationships which exist among the Pastoral Neolithic wares of East Africa.

The Jarigole Assemblage of Nderit Ware. The Jarigole Pillar Site was in use at a time before Nderit Ware had undergone appreciable simplification so the assemblage from this site provides a reasonably good picture of the pool of traditional elements which were drawn upon selectively as daughter wares emerged. The assemblage of Nderit Ware from Jarigole is also larger than all previous assemblages of this ware combined, documenting the range and interaction of these elements in much greater detail than was possible in the past. There are over 30,000 sherds, and this summary is based on a sample of 659 vessels which have been defined on the basis of rims alone. Vessels represented by body sherds only are not tabulated in this study. Not surprisingly, this sample reveals aspects of the ware which were previously unknown or only suspected.

Indeed, the ceramic assemblage is so richly diverse that it is difficult to know what to focus on and how much detail to present. One way to visualize the complex diversity of the assemblage, is to summarize the choices which confronted the potter at each stage of making a pot.

Creating an Nderit Vessel
Paste. The paste (clay body) used to form a pot is composed of clay, a temper to give the clay greater body so that it can be worked and fired properly, and water to produce the necessary plasticity for forming and decorating the vessel. Our Nderit potter had two basic kinds of paste from which to choose. The first uses a clay which produces a relatively dense, moderately porous, hard ceramic when fired. It is mixed with a sand temper and a moderate amount of water and can support the weight of vessels up to 40 cm in diameter without deforming prior to firing. This paste burnishes and slips well and accepts all decorative techniques. However, extensive decoration may require that the clay be remoistened at intervals while the decoration is being applied. Decorative techniques which require an especially plastic paste will be difficult to employ. In firing, this paste may also require a somewhat higher temperature then the second type.

The second kind of paste uses a clay which produces a very porous, friable ceramic when fired but may have the advantage that it can be fired at somewhat lower temperatures. It contains very little visible temper and was usually mixed with a lot of water so that it was quite wet and plastic when worked. As a result, it was only used for smaller vessels, was difficult to burnish and unsuitable for interior decoration. However, it is ideally suited to one particular decorative technique, stylus evulsion, in which a stylus is used to lift and evulse the clay body to produce a complex, raised, textured surface. After firing, it may also have accepted paint and red ochre most readily.

Thus, the choice of paste depended on the size and form of the vessel intended by the potter and by the decorative techniques which were to be employed.

Choosing the Form of the Vessel. More than 20 vessel forms and numerous subvarieties were made by Nderit potters. These include carinated, in-turned, hemispherical and semi-spherical bowls; broad shallow bowls with straight sides, small bowls with broad constricted necks and everted rims, and oval or boat-shaped bowls; pointed and flat-bottomed beakers; large saculated and spherical jars without necks; large jars with broad necks and straight, inturned or everted lips; saculated vessels with narrowly constricted necks and small head-like tops; jars with conical tops and small openings which are girdled with binding channels; small jars with long, tapering cylindrical necks in-turned to form small openings at the top; platters; internally divided bowls and a number of other unusual forms such as footed ring stands. Many of these forms may come with and without rim folds and tab-like epaulettes; tab, strap and ring handles; spouts; pre-firing holes for handles or suspension; and feet.

The choice of vessel form constrained later choices, particularly with respect to the form of decoration. For example, vessel forms which were not open could not be decorated on the interior surface. Vessels with necks or binding channels have exterior surfaces which are divided into spaces by their structure. These spaces present certain design possibilities. But over and above such structural constraints, specific vessel forms came to have certain design motifs associated with them as part of the cultural tradition associated with Nderit Ware. For example, there is a large, thick-walled saculated vessel form with a small round opening at the top. The surfaces of these vessels are left quite rough without burnishing, slip or paint. Unlike other forms, in which decoration begins at the rim, here it begins between five and ten cm below the rim.

Selecting Suitable Decorative Motifs. For our purposes, a motif is a design structure which is used to organize the space on the surface of a vessel. A single motif may be used to cover the entire exterior surface, interior surface or vessel rim. Vessels whose exterior surfaces are divided into separate spaces by structural features, such as necks and binding channels, often exhibit different motifs in these spaces. To date, we have recognized five motif clusters commonly used in the decoration of exterior spaces on vessels and five motif clusters commonly used to decorate the interior of open vessels. There are four motif clusters commonly used in decorating rims. Each motif cluster consists of a family of from three to eight closely related, but distinct sub-clusters, and each sub-cluster is composed of between one and six specific motifs. Each motif may employ a series of as many as ten interchangeable design elements which can be executed using six different decorative techniques. The potential for artistic expression and variability is therefore enormous. In fact, there are more than 200,000 potential design outcomes not counting variation introduced by the distinctive shape of comb or stylus impressions; the distinctive techniques employed when applying combs and styluses to the clay body; and the use of slip, burnish, paint and ochre to give different characters to otherwise similar designs.

In reality Nderit potters developed only a small range of this vast potential into a series of traditional ceramic designs. In the case of the Jarigole Pillar Site, representing only a small selection of Nderit potters, this may mean scores of minor variations on each of 50 to 80 basic design combinations. Nonetheless, this amount of variation is still significantly larger than our excavated sample. Each excavation season reveals new, sometimes strikingly different realizations of the design principles being employed by Nderit potters.

In the space available here, it is only possible to touch on some of the common expressions that characterize decoration in the Nderit Ware assemblage from Jarigole. The most common motif clusters and sub-clusters, which define the organization of design on vessel surfaces, are as follows:

    External Surfaces
      Burnished Ripple Cluster (Fig. 7)
        Netted Ripple Sub-cluster
        Butted Ripple Sub-cluster (Fig. 7)
        Offset Ripple Sub-cluster
      Channeled Ground Cluster (Figs. 7-8)
        Reticular Net Sub-cluster
        Serpentine Sub-cluster
        Branching Sub-cluster
      Banded Cluster (Figs. 9-10, 15)
        Herring Bone Sub-cluster (Figs 9-10)
        Horizontal Line Sub-cluster (Fig. 15)
        Diagonal Line Sub-cluster
        Compound Band Sub-cluster
      Vertical Field Cluster
        Sub-clusters not yet defined
      Panel Cluster (Figs. 11-13)
        Butted Panel Sub-cluster (Fig. 11)
        Separated Panel Sub-cluster (Figs. 12, 13a)
        Floating Panel Sub-cluster (Fig. 13b)
        Bossed Pane Sub-cluster
    Interior Surfaces (Subclusters not yet defined)
      Evulsed Pocket Cluster (Fig. 9)
      Evulsed Net Cluster
      Radial Net Cluster
      Band and Pendent Cluster
      Banded Chevron Cluster (Fig. 10)
      Vertical Brush Cluster (Figs. 11a, 12a)
    Rims (Sub-clusters not yet defined)
      Milled Rim Cluster (Figs. 6, 9, 11, 12a, 12c)
      Segmented Rim Cluster (Fig. 13b)
      Body Design Extension Cluster
      Banded Verge Cluster (Figs. 8, 10, 12b)

For the purposes of the discussion which follows, only the external vessel surface will be considered; decoration of the vessel rim and internal surface will not be discussed.

The Burnished Ripple Cluster (14% of all vessels) is named after the chief decorative technique, ripple burnishing, employed to create its design motifs. In ripple burnishing, the gently rounded edge of a polished pebble is used to compact and polish the surface of the vessel, creating a shallow, polished trough. When these are aligned edge to edge, the margins of the troughs intersect creating low, sharp crests. The alternating crests and troughs create a pattern which resembles ripples formed in the sand by wind or water, common natural motifs along the shores of Lake Turkana (Fig. 7).

Figure 6. Rim sherd from spherical vessel showing the long continuous throughs typical of the Butted Ripple Sub-cluster. Sherd is tilted to show the use of stylus impressions on the rim. Scale bar of 2 cm.

The ripples may be made of various lengths, bent in a number of ways and overlapped or closely butted at their ends, but they are always oriented diagonal or subvertical to the rim, and most often slope downward from right to left, and cover the vessel completely. Motifs using long ripples, common in the Butted Ripple Sub-cluster, are occasionally made using grooves which are then over-burnished. Vessels in the ripple cluster are frequently covered with a pigmented slip, fired paint or unfired paint. Only about 1% have internal decoration, though the vessel form and paste are compatible in more than 85% of vessels. This demonstrates that the motif complexes are not simply the result of the random recombination of design elements, but are the outcome of decorative choices which follow definite rules.

With the exception of a single vessel from an early Ileret Ware assemblage at the Ileret Stone Bowl Site, the technique of ripple burnishing is known only from assemblages of Nderit Ware.

In the Channeled Ground Cluster (13% of vessels) the exterior of the vessel is usually covered completely with a repetitive pattern of fine impressions, but sometimes the vessel base or a broad band below the rim is left undecorated. At the Jarigole Pillar Site all but one of these vessels has a ground design created by comb stamping (99%); the other has a girded stylus background. But as Ileret Ware develops out of Nderit Ware on the east side of Lake Turkana, the Channeled Ground Cluster declines somewhat in frequency and stylus impressions are most commonly used to produce the background design, a trend seen at sites in the Koobi Fora area.

After the ground design has been laid down, a pebble or fingertip is used to inscribe broad, moderately deep channels which obliterate the impressed ground design wherever they pass. In come cases channel pathways have been outlined prior to channeling by leaving space for them within the patterned ground. The channels were used to create large-sized design elements, usually from 5 to 12 cm across, so secure identification of sub-clusters requires large sections of the vessel surface to be reconstructed. Since we have not yet undertaken vessel restoration, the identification of sub-clusters is still provisional.

Figure 7. Comb of Ostrich egg shell with five teeth together with the impressions made by it in modeling clay. Different effects are produced by spacing, dragging and impression depth. Scale bar is 1 cm.

In Nderit Ware, the Channeled Ground Cluster is closely associated with the use of comb stamps. One such comb, the first found in East Africa, has been recovered from Jarigole (Fig. 7). It is made of ostrich egg shell and has five small, rounded "teeth" which are capable of making a variety of patterns depending on how they are impressed in the clay.

Vessels of the Channeled Ground Cluster are usually decorated over the entire surface. They are most often open bowls and commonly exhibit rim elaborations such as vertical tabs, horizontal tab handles, hollow pole handles, eared tabs which might be ladle rests, special marks on the interior or exterior, and rim folds with decorative appliqués at the pinches between folds (Fig. 8).

Figure 8. Rim sherd of the Channel Ground Motif Cluster with a comb-stamped ground. The vessel is a hemispherical bowl whose rim is folded inward. Where folds meet, the top of the rim is decorated with bosoms (inset). Scale bar of 2 cm.

Although the Banded Motif Cluster (12% of vessels) has a very simple spatial structure in which the vessel is simply divided into a series of relatively narrow horizontal bands, these bands can be filled with a variety of decorations generated by any of the decorative techniques found in the assemblage. The most common sub-cluster uses a horizontal herring bone pattern most commonly made of grooves, incisions, stylus impressions or comb stamps. This sub-cluster also has a particularly high frequency of internal decoration. The result is an assemblage of vessels with a wide range of decorative characteristics which demonstrate that even the simplest motif can be elaborated in dozens of very distinctive ways.

Figure 9 shows the rim of a vessel in which the herring bone design has been made by pressing the curved edge of an implement deeply into the clay to form impressed incisions about one centimetre long. These are overlapped to produce a roughly textured surface. When viewed from a distance that looks very much lake coarse basketry, a motif which is carried onto the rim where the herring bone wraps around the top and looks like plaiting. The interior is also decorated with rows of evulsed pockets made by pushing the clay forward and outward to form a raised cowl not unlike a vegetable grater.

Figure 9. Rim sherd of the Banded Motif Cluster with impressed incisions in the herring bone motif and evulsed pockets on the interior surface. Scale bar is 1 cm. Select figure for color picture.

In marked contrast, the vessel shown in Figure 10 has the herring bone motif executed meticulously using a small rectangular rod or stylus. This stylus would look much like an unsharpened pencil if you shaved the end to be rectangular in cross section. By impressing the end, edges, corners and sides of the stylus at different angles you can make impressions of different shapes and sizes. By butting or overlapping the impressions in different ways, designs formed by parallel lines of impressions can be given a variety of visual textures as light strikes the overlapping impressions from different angles. This is illustrated in the left and centre pains of Figure 10 which show the same section of vessel lit from two different directions. The impressions are overlapped on all edges, forming a completely imbricated surface. Each step in each imbricated row is the remains of a single stylus impression. On this vessel there are an average of 12 impressions per square centimetre and it required about 7,500 impressions to complete the design. The interior of the vessel is decorated with the herring bone motif, executed by dragging the corner of a stylus to produce deep V-shaped grooves.

Figure 10. Rim sherd of the Banded Motif Cluster with herring bone motif made of overlapped stylus impressions and a grooved herring bone design on the interior surface. Exterior surface (left and center) shown under two lighting conditions. Scale bar is 1 cm.

The most abundant motif cluster is the Panel Cluster which includes 46% of the vessels from the Jarigole Pillar Site. In terms of design structure, this is perhaps the most diverse and least understood of the motif clusters, and will undoubtedly be broken into a larger number of sub-clusters as larger vessel sections are reconstructed.

The Panel Cluster is defined by the use of interlocking panels and fields which are filled with design elements such as stylus impressions, comb stamped impressions, grooves, incisions and burnished ripples. A panel is an area which is enclosed by a border design, whereas a field is a filled space which lacks a defined border. Instead, the space is defined by the decorative element which it contains. Panels are often composed of from two to five interlocking fields (Fig. 12, b-c).

Typically, vessels are covered in a complex structure of interlocking panels or fields which may be continuously butted (Fig. 11), surrounded and separated by bands which may be decorated or undecorated (Fig. 12), or as "emblems" floating in a larger design structure or on the otherwise undecorated surface of a vessel. Discussion of the two most common sub-clusters illustrates the complexity within each of these groups.

Figure 11. Examples of the Butted Panel Motif Sub-cluster. Rim sherds: a, c. Body sherd: b.
a: vertically intersecting fields of diagonally oriented stylus impressions on exterior mirrored by vertical and diagonal incisions on the interior.
b: fields of stylus impressions and groves illustrating curvilinear design elements and field boundaries.
c: grooved panels with burnished and slipped surface. the slip contains small pieces of shell that generate carbon dioxide gas during the firing process. This gas pops the surface of the vessel away as it expands, leaving the shell exposed in shallow pits and producing a red surface with small white speckles.

The Butted Panel Sub-cluster (74% of the Paneled Cluster) is characterized by continuously butted panels and fields which are used to cover the entire vessel surface or, less often, the upper half of the vessel (Fig. 11). Panels are usually geometric polygons, sometimes with curved edges, which are distorted to accommodate the curving spaces on the vessel exterior. Fields within panels are often less regular in shape and may introduce considerable asymmetry into the design structure. The design elements which fill fields and panels are usually arranged in straight lines, but may on occasion be curved (Fig. 11, b). Grooves and stylus impressions are the most common decorative elements and are often mixed in the same vessel. About 12% of the vessels also have internal decoration.

The second sub-cluster is the Separated Panel Sub-cluster (18% of the Paneled Cluster). Individual panels or concatenations of panels are separated from one another by avenues which form an angular, geometric design on the outside of the pot. These avenues are usually about 1 to 3 cm wide. They are usually left undecorated or filled with alternating decorated and undecorated bands (Fig. 12, c), but in some cases the entire band is filled with grooves or rows of stylus impressions running parallel to the margins of the adjacent panels. Panels are commonly filled with stylus impression, grooves or comb stamping, and rarely with burnished ripples. 28% of vessels in this sub-cluster have internal decoration.

Figure 12. Rim sherds of the Separated Panel Motif Sub-cluster. Notice the very narrow separating band in a and the broader and increasingly more complex separating bands in b and c.
a: stylus impressions articulated in rows but not fully imbricated; with vertically brushed interior.
b: comb-stamped fields and panel borders.
c: mixture of comb stamps and stylus impressions. Scale bars of 1 cm.

Decorative Complications. Two additional decorative complications should be mentioned simply because they are very common. The first has to do with the variety of styluses which are commonly employed. Although most are of the rectangular variety described above, styluses with circular and oval ends were sometimes used, and natural objects, such as shells and reeds, are quite common (Fig. 13, b). These are usually used in grids of impressions and only occasionally articulated or imbricated.

Figure 13.
a: Separated Panel Motif Sub-cluster showing evulsed and normal stylus impressions. Note the complex field boundary in the upper panel.
b: Rim sherd of the Floating Panel Motif Sub-cluster with a tightly spaced grid of crescentic impressions most likely made with the base of a shell. (3/4 view to show top of tightly curled rim.) Scale bars of 1 cm.

The second complication involves the technique of stylus evulsion (Fig. 13, a). In this technique, the stylus is used to push and lift the clay, making it roll forward. This creates a rounded lobe which stands above the surface of the vessel. This technique is usually used in conjunction with a special soft paste which moves easily when worked. Evulsed impressions are frequently filled with red ochre after firing, producing a striking decorative effect. Vessels of the soft paste usually used when this technique is employed are not decorated on the interior, possibly to ensure stability of the clay when it is fired.

Technique, Design Element and Motif
Traditionally, analysis of Pastoral Neolithic ceramics in East Africa has used the technique of decoration (e.g., comb, stylus, groove) and specific design elements (e.g., pendent triangles) to describe and classify assemblages of pot sherds. This approach is useful when an assemblage is composed primarily of small fragments and there are very few sherds from any individual vessel.

However, to understand ceramic wares at the level of cultural and artistic traditions, it is necessary to study them in broader terms. This second approach attempts to determine and use concepts that were significant to the people whose cultural products are being analyzed. To understand the expressive language of the artist or study ethnicity and cultural identity, the analysis should reveal categories that had significance to the artists who made Nderit pots and to Nderit people in general. The material presented in this article is a first approximation in the application of this approach to the assemblage of Nderit Ware from Jarigole.

Motif Cluster
and Subcluster
Banded Cluster Vertical Field
Panel Cluster Bands or Panels
Number of Vessels * 1 91 3 83 8 20 2 3 2 9 - 7 - 3 - 2 - 1 18 127 10 26 - 12 - 4 4 6 16 106 9 51 32 659
Stylus Impressions       * For each category, left column = no. w/internal decoration; right column = no. w/external decoration only.
Evulsed               1   7                   2   3                 13
Imbricated         2 1 1 1 1 2                 7 6 2 7       1 1 2 1 2 36
Lapped in rows         2 3                   2     7 42 3 11   7     2 1 12 27 119
Crescentic               1                       10   1   1   1     2 7 23
Angular         1 1                         3                       5
Slits       1                                                     1
Ovals                                       1 1 2                 4
Comb Stamped
Impressions     3 82   4       2                     3 1   3   2 1       101
Grooved and Insised
Incisions or Cuts         2 5           5   3       1   1                   4 21
Simple Grooves         4 12 1 1       5             1 107 1 9   3   2   3 1 66 216
Compound Grooves                       1                                     1
Ripple Burnishing
Burnished Ripples   86       2 1                             2                 91
Burnished Grooves 1 5                                                         6
Table 2. Methods of Decoration Tabulated by Motif Cluster for 659 Vessels from Jarigole Pillar Site (GbJj1).


Table 2 is a quantitative representation of the progress in this analysis to date. It shows the choice of decorative technique on the exterior of the vessel in relationship to the major motif clusters and some of their sub-clusters. There is a total of 567 vessels with exterior decoration out of an assemblage of 659. However, there are a total of 640 representations of specific design techniques because 65 of the vessels each exhibit two or three of the techniques (see Fig. 13). Thus, Table 2 shows the relationship between a potter's choice of design and the decorative techniques selected to execute that design. These choices are further tabulated against the use of interior decoration. Clusters of numbers represent popular patterns of choice and were possibly of consequence to the Nderit. Low frequencies may represent very specialised categories or design experiments by innovative potters; the absence of numbers represents design combinations considered inappropriate. Hence, Table 2 represents a first step in documenting the aesthetic system of Nderit potters.

Ceramics as Representational Art
As the figurines and the representation of bosoms suggest, some of the design elements and decorative motifs on Nderit pots almost certainly had representational significance to the people who made and used them. For example, two of the Ripple motifs call to mind the ripples made by wind and water in sand. Some are quite tantalizing such as the theme which involves the use of roughly conical and breast-shaped appliqués on rims and around the necks of vessels. Most of these are not realistic representations of breasts, but one vessel has a collar of appliqués which appear to have well formed breasts with proper asymmetrical cross section from top to bottom, but lacking nipples. Another vessel has what appear to be pairs of breasts which decorate the intersection of rim folds not unlike breasts on the prow of a ship (Fig. 8). Do these really represent breasts? We may never know for certain. But within the stone ornaments, there are two very well carved phalluses and a number of beads which range from realistic phallic representations to elongated beads in which the phallic symbolism is strictly a matter of opinion. Fertility is certainly a common theme in decorative art from many periods and places, so it would not be surprising to find it reflected in ceramic decoration.

Another intriguing example is the sherd illustrated in Figure 14. This is one of five sherds from a large vessel. The other sherds all repeat the same basic design which appears on this one, viz., rows of well-spaced comb stamps. This sherd is different, however, because it shows the end of a row of these comb stamps and the unusual way in which that portion of the vessel was treated. The sequence of events is as follows:

1) Before firing, the end of a stylus was used to make a series of impressions which combine to form what look like stalks of grass. Long lines of these were made over most of the surface of the vessel. Three of these "stalks" survive in tact on the lower part of the sherd.

2) A comb stamp with compound teeth was used to over-stamp the "stalks." This is evident from the fact that the tops and bottoms of "stalks" can be seen protruding beyond the ends of the comb stamp on all of the sherds recovered from this vessel. For some reason the last three "stalks" in the row were not over-stamped.

3) The same technique used in ripple burnishing was used to add three channels to the top of the three comb-stamped "stalks" to the left of the "stalks" which were not stamped. This partially obliterated the "tassels" protruding from the top of stamped impressions.

4) The pot was fired.

5) After firing, an area just above the top row of stamped "stalks" on the sherd was heavily abraded, removing the tassels which protrude from beyond the comb stamped impressions. Their original presence is highly likely because we can see the "stems" of the stalks protruding from below the comb stamped impressions.

Figure 14. Vessel of the Banded Motif Cluster with a complex history of decoration. Scale bar of 2 cm.

Select image for a color enlargement.

What does this sherd mean? Is it evidence for agriculture in the form of grain production? Not necessarily. No carbonized domestic plant remains have been found in Nderit sites. No grain impressions are present in the 30,000 sherds from Jarigole or from the sherds at other Nderit sites. Grindstones of the sort used to process hulled grains are not known from Nderit sites. Even if the "stalks" represent grasses, these grasses need not have been domestic crops. They could be wild grasses depicted to symbolize pasturage for domestic animals, which we know Nderit folk kept in numbers. And, of course, they may not be grass stems at all, but plumes or staffs with feathers tied to them or other less likely objects.

Explaining the over-stamping of the "stalks" is even more challenging. Perhaps the "stalks" merely represent a design idea that did not prove pleasing to the eye? Not likely. First, that should have been obvious after the first dozen or so. Second, self slipping with water would have erased such an experiment cleanly and quickly. And third, a comb stamp was used which did not obliterate the "stalks" completely, while some stalks were never removed. Of course, we might have a child or a lazy potter who is doing a very sloppy job, unconcerned by the fact that the pot may be destined to become a grave offering.

The alternative is that the over-stamping was deliberate and has some specific significance. And what could this significance be? The possibilities are limited only by the fertility of your imagination. A tally to record access to pasturage; a mnemonic device used in story-telling; a commemoration of the wealth or deeds of the person for whom it was buried; the product of a ritual using sympathetic magic to ensure abundant pasturage; a commemoration of the number of living who paid homage to the deceased; a mnemonic device used to commemorate and hand down those in each generation of a lineage. Shall we go on?

This is, of course, the difficulty with art which is representational without being obviously pictorial. It is difficult to isolate representational patterns and virtually impossible to interpret them with any degree of confidence. Nevertheless, the problem and analysis of representational ceramic art is worth while considering seriously in the present context for two reasons. First of all, just demonstrating that our ceramics have substantial representational content would add a new dimension to the meaning of Nderit ceramics and culture. Second, since continued excavation of the Jarigole Pillar Site could accumulate upwards of ten thousand vessels, there is some real hope that representational elements can be isolated and their frequency quantified. More interesting still, we might be able to determine how the more obvious representational elements enter into design motif construction. If this follows any particular rule, then it would be possible to work backwards from that rule to identify other design elements or motifs which follow the same rule and might therefore stand a good chance of being representational in character. An intriguing prospect, but a daunting task.

The Jarigole Pillar Site and Nderit Ware in Archaeological Perspective
There is no record of the prior development of Nderit Ware, the cemetery form represented by the Jarigole Pillar Site, or the domestication of animals within the Turkana Basin. This cultural package is almost certain to have been introduced from the north, most probably the southernmost extension of the Ethiopian Highlands. Over 1,500 pillar sites have been recorded in Ethiopia. Most of these are attributed to the medieval period, but since systematic excavations have not been conducted, it is likely that the erection of pillars is a far more ancient custom. A survey of the southernmost of these sites would be a good starting point from which to trace the origins of the culture embodied in the Jarigole Pillar Site.

Not long after occupying the Turkana Basin, this culture spread southward along the Rift Valley, leaving centers of pastoralism in the wake of this expansion, for example in the Nakuru Basin, the Lukenya region, and in the Serengeti. Meanwhile, in the course of the continuing development in the eastern half of the Turkana Basin, both funerary practices and the design structure of the ceramic ware were simplified. There appears to be ceramic continuity right down into the Iron Age, implying a stable, gradually changing population.

The fate of the Nderit folk in the western half of the Turkana Basin is not yet known. They either developed along different lines or were absorbed or pushed out by a new influx of pastoralists from the north. Perhaps both of these things happened. Certainly around 2,000 years ago, the appearance of Turkwell Ware(7) in the southernmost part of the Sudan, along the lower Omo River, throughout Karamoja in northern Uganda, and in the western half of the Turkana Basin, suggests a very different cultural history than that seen in the eastern half of the basin.

The author would like to acknowledge the support of Harvard University and the National Museums of Kenya, who jointly sponsor the Koobi Fora Field School and whose support has made the work at Jarigole and this report possible. This is the first comprehensive description of the Jarigole Pillar Site and it is fitting that it appear in a journal published by the Kenya Museum Society.

I would also like to thank the many individuals who have had a hand in teaching and working at Jarigole, including Harry Merrick, Craig and Marsha Feibel, John Kimengich, Lora and Alan Morton, Tam Smith, John Bower, Peter Robertshaw, and Cosmos Kiwia, and finally the many students, Kenyan and other, who have contributed simply by their participation and learning.

Illustrations and Photographs by the author


  1. R. Soper. Archaeo-astronomical Cushites. Some comments. Azania 17 (1982) pp. 145-165.
  2. D. Stiles. he Azanian civilization and megalithic Cushites revisited. Kenya Past and Present, 16 (1984), pp. 38-41.
  3. L. Robbins. Archaeology in the Turkana District. Science, 176 (1972) pp. 359-366.
  4. Ibid.
  5. J. Barthelme. Fisher-Hunters and Neolithic Pastoralists in East Africa. Oxford, B.A.R. International Series, No. 254.
  6. Ibid.
  7. Robbins and M. Lynch. Lopoy: A Late Stone Age Fishing and pastoralist settlement in the Lake Turkana Basin, Kenya. Michigan State University Museum, Anthropology Monographs, Vol. 1, No. 3.