The regular reader of Expedition and, almost certainly, every professional archaeologist has at one time or another experienced the spell cast over the imagination by the ancient settlement mounds of early floodplain civilizations. Once teeming with thousands, they now lie dry and bare under a relentless sun. The Asian alluvium, of course, comes foremost to mind as the mother of ancient civilizations built on tells—the Sumerians, Babylonians and Harappans. But forgotten in the center of another continent entirely is yet another floodplain with hundreds of ancient tells rivalling those of Asia in area and in clues to the emergence of city life.
French officers in the vanguard of colonial forces penetrating the interior of West Africa filled pages of official dispatches with descriptions of enormous abandoned mounds littering the Niger River’s broad interior delta. To late 19th century Europeans, this was eloquent testimony to centuries of depopulation, political chaos, and suffering that a strong colonial hand could bring under control. For al-Sa’di, a chronicler with strong ties to the Middle Niger commercial centers of Timbuktu and Jenne, who wrote in 1655, the tells were artifacts of a past more prosperous than his own very troubled times.
The 1960s were years of national consolidation as the French Soudan became the Al-Sa’di records local traditions of a time when the plains around Jenne were so densely populated that pronouncements and summons from the town’s sultan were delivered hundreds of kilometers into the interior simply by shouts repeated from one village to the next (al-Sa’di 1964:14-25). This was a fabled time when peace reigned and both Timbuktu and Jenne grew wealthy on commerce in gold from the forests to the south. But by al-Sa`di’s day even the locals had forgotten the age of the tells along the Niger River. Even less was known about the communities which had occupied those mounds.
During the decades of stupendous discoveries at Mesopotamian mounds, the French colonies of sub-Saharan Africa attracted few archaeologists. A few trained prehistorians travelled from the Institut Français d’Afrique Noire in Dakar, Senegal to make surface collections at a handful of tells. In 1907 a military administrator, Lt. Desplagnes, conducted the only complete excavation of a mound along the Niger (el-Oualadji excavations, Desplagnes 1951). That mound turned out to be a huge funerary tumulus, not an occupation site. In the absence of systematic research, the colonial period passed with little more knowledge than in al-Sa`di’s day, of the date of the ancient settlements. independent Republic of Mali. The meager national budget could support little archaeology. The plains of the Niger, as part of the great semi-arid belt south of the Sahara, suffered an enormous reduction in human and livestock population during the Sahel Drought from 1967 to 1974. Archaeology was further postponed. The Sahel Drought dramatically illustrates the unpredictability in the midst of plenty which today characterizes life on the Niger floodplain. ‘Plenty’ is an ancient aspect of this floodplain, which in the distant past supported a population many times greater than today; ‘unpredictability’ may be a more recent aspect.
In the middle of the Sahel, the Niger flows across a great alluvial depression, called the Inland Niger Delta, dominated to the south by Jenne and to the north by Timbuktu, where the river turns from its apparent suicidal rush into the Sahara to flow southeast towards the Gulf of Guinea. The Niger River annually floods some 80,000 square kilometers during the winter inundation, which nourishes a large crop of native African rice and sends millions of fish into the hundreds of Inland Delta streams to be caught by the local Bozo fisherfolk. (A succulent fish called “capitane” [Lutes niloticus] from the region is flown fresh to the most elegant tables of Paris.) The area produces an abundance of cereals, but the scheduling of crops is tenuous; early rains which enable farmers to plow the sun-baked alluvium may be insufficient or tardy, or the flood may arrive too late to nurture the seedlings. Early or high floodwaters invite schools of fish to feast, decimating the rice fields (Galleis 1967). As the flood abates in the months of January through April, the pastoral Fulani drive thousands of cattle, sheep, and goats from southern Saharan pastures into the grassy meadows of Inland Delta back-swamps. This is the beginning of the summer dry season in the Sahara. Without the reserves of floodplain pastures the Fulani’s livestock would perish—as they did by the thousands during the Sahel Drought. In an ironic turn, many cattle driven into the region in recent years have drowned in swamps and river beds still containing deep floodwaters. But when the rains and the floods arrive as expected, the land is bountiful. The floodplains around Jenne produce an ample harvest which has provisioned southern Saharan towns, Timbuktu in particular, in a riverine commerce in fish, rice, and cattle flesh since at least al-Sa`di’s day. And, as we shall see, this trade must have been flourishing Iong before his time.
This is the context within which we must consider the tells that dot the Jenne region by the hundreds. The number between Jenne and Timbuktu (never more than cursorily inventoried) must run to the thousands. We chose to concentrate on Jenne, sister city to Timbuktu and its principal partner in the medieval trade exchanging West African gold for Saharan salt and Mediterranean manufactures (Bovill 1968). As we pored over aerial photographs of the Jenne region before our first field season in 1977, an initial possibility dominated our thoughts. If all these tells had been occupied contemporaneously, the Inland Delta’s economic potential had at one time been superbly exploited. The mounds displayed great differences in size and a dis tinct clustering tendency. What clues did these clustered tells hold about the beginnings of urban life here?
Our 1977 field season began with a two-part inquiry. Jenne was already a city at the moment of its first historical mention in late medieval times. How old was this city, still thriving today with 10,000 inhabitants? No city emerges in isolation: what role did Jenne’s hundreds of neighboring tells play in its evolution and why do they now lie abandoned? Asking these questions in tandem, we believe, is the greatest single improvement in contemporary urban archaeology. This assertion deserves brief explanation as it lies at the heart of our research strategy.
In the earlier days of tell archaeology—the ‘Heroic Age’ of Layard, Botta, Woolley and other pioneers—excavations focussed on the principal mounds, generally concentrating on central public structures such as ziggurats or palaces. This resulted in a very lopsided view of the sites themselves, and even more so the processes of early urbanization. For cities did not grow in isolation, but emerged from agricultural village antecedents which can be understood only through a regional study. Principal settlements flourished because of a supporting network of smaller towns, villages and hamlets within their sphere of influence. Regional connections supported the city with staple crops, materials for its industries, enormous labor reserves, and markets for the diverse products of its specialized population. Accepting this concept of the regional roots of urbanism, it becomes not just appropriate, but imperative to ask “what role did the hundreds of neighboring tells play in Jenne’s evolution?”
Jenne’s regional roots extend throughout the complex Niger floodplain. The landscape over which Jenne and nearby tells are distributed is what geomorphologists call an anastomosing floodplain. The Niger and its major tributary, the Bani River, converge not far away. In the extremely flat, lowlying floodplain between them a myriad of small streams, or distributaries, form a fine capillary system between the main riverine arteries (R. McIntosh n.d.). Today, farmers in the interior transport their produce along these distributaries to Jenne for warehousing and eventual shipment down river to Timbuktu. In the past, these streams played an even greater role, to judge from the number of tells along channels now choked by dune sands or silt barriers. Jenne and a particularly high density of large tells lie on the banks of the Souman-Bani, the principal distributary linking the Bani and the Niger. Smaller, but still substantial mounds loom over more interior streams. The preference for settlements on these different landforms—dune and channel levee permanently above flooding, floodplain soils suited for rice, or deep basins useful only for fishing and dry-season pasturing—provided the initial clue to occupational specialization contemporaneous with the emergence of Jenne-jeno.
Survey of the region proceeded in two stages, in 1977 and 1981. Aerial photographs and satellite imagery simplified mapping of landforms and soil units. Once in the field, the discovery that tells on the floodplain were one hundred percent visible on these photographs allowed a randomly-selected sample of mounds to be visited and surface-collected. Sites on dunes or levees—that is, above the floodplain—were less easily detectable on air photos, so these were divided into transacts, a sample of which we then walked in search of ancient remains. During the first season in 1977, cataloguing over 400 mounds during survey of 1,100 square kilometers and excavation at the ancient site of Jenne-jeno provided the first of several unexpected lessons of Inland Niger Delta settlement, Stone Age peoples apparently avoided the Inland Delta. More intensive survey of a smaller region in 1981, a search of over 20 kilometers of rivercuts, and excavation at several other mounds confirmed the absence of Late Stone Age occupation. This completely contradicted our expectations. The Sahara, once a pastoral haven for Late Stone Age fisherfolk and herders, had experienced pro. gressive desiccation since the 6th millennium B.C. (S. McIntosh and R. McIntosh 1981:602-608]. It seemed reasonable to presume that the Inland Delta had been a natural magnet for these people at least from the 2nd millennium B.C., when aridity became pronounced. So far, there is no trace of them. It may simply be that remains of Stone Age camps are buried far beneath alluvium or that wildly meandering streams erased all traces of their presence. However, it is equally possible that the Inland Delta until the last millennia B.C. was inaccessible except by boat or uninhabitable due to the presence of water-borne or water-associated diseases affecting cattle and humans. The range of the invidious parasites which cause bilharzia, river blindness and sleeping sickness may have extended into the West African Sahel, well to the north of their present home in the savannas (Smith 1979:359-360]. Some geomorphologists have even claimed that the Inland Delta was for millennia a vast shal– tow lake (Tricart 1965:36-37, 43]. Whatever the reason, the evidence is quite firm. The first peoples to penetrate the region of Jenne were iron-using farmers, fishermen, and herders. The settlement which began during the last half of the last millennium B.C. was no tentative trickling of pioneers. The Inland Delta was the scene of a prehistoric land rush.
Imagine yourself an early colonist looking for the first time upon an uninhabited plain of rich grasses for your herds and lush rice paddy lands. What are your basic requirements for permanent settlement? Deep in the interior are the clay basins supporting the dry season pastures and the only ponds available year-round for fishing, drinking, bathing, etc. Ringing these basins are the finest rice-growing soils, those inundated to about 3 meters each year. Near this idyllic scene are elevated features capable of maintaining villages permanently above flood high-stage. However, for reasons not yet understood, higher land such as dunes and levees was almost without exception avoided until the beginning of the present millennium. Nevertheless, the first colonists persisted on the floodplain. Their final solution to the dilemma was to found villages on mounds—tells built up of centuries of domestic debris and collapse of mud houses constructed of the nearby alluvium. Perhaps the high-flood months were spent away from the new villages until sufficient height was reached, or per haps the floods of that period were generally lower (evidence discussed in R. McIntosh n.d.). Whatever the details of early occupation, it was highly successful. Hundreds of tells were established at the clay basin/rice lands boundary, apparently within a few centuries of the opening to settlement of the Inland Niger Delta. Of all sites catalogued within the 1,100 square kilometer survey region, 34% were located near clay basins and paddies. But fully 55% of all sites were within channels of distributaries. These sites certainly had year-round access to water and to fish-runs, but why would the inhabitants have chosen to construct settlement mounds in the very place where floods would be most destructive?
Again, imagine yourself an early colonist of the Inland Delta. Many localities in the floodplain have excellent potential for fishing, farming, or pastoralism but, more importantly, all are absolutely lacking in certain resources critical to the Iron Age way of life. Imported items found during excavation demonstrate those resources which the floodplain could not provide. Evidence from the very lowest levels of several sites shows that, as early as the last few centuries B.C., the inhabitants conducted trade with places beyond the Delta for a variety of utilitarian and decorative items. Traffic was in iron ore (a medium-carbon steel was smelted and refined, apparently on industrial scale, as early as 200 B.C. at the site of ancient Jenne; R. F. Tylecote 1982:personal communication), sandstone grinders (which were imported in a number of standardized shapes and sizes), and exotic volcanic stone required for production of beads and other ornaments. The nearest source for any of these materials was at least 50-75 kilometers away. These were heavy, bulky materials, and the amounts discovered at Inland Delta sites imply an organized, voluminous transport and exchange (S. McIntosh and R. McIntosh 1980: part ii, 444-461). By about A.D. 400 copper from Saharan sources over 1,000 kilometers away is found at the site. Salt, another distant Saharan material for which there has always been a thriving market in the Sahel, presumably accompanied caravan shipments of copper, although archaeological evidence of this necessary mineral is notoriously elusive. And by A.D. 600-800 we find ornaments made of gold from mines far to the south in West Africa.
In exchange for these items the inhabitants of the channel tells undoubtedly exported the same things that ancient peoples of Mesopotamia or the Indus exchanged with their distant suppliers: vast quantities of food and the specialized products of their manufacture. In the case of the Inland Delta trade this is still conjecture because no modern investigations have been conducted at appropriate exterior sites (see Szumowski 1956). Nevertheless, the first historical references and local traditions emphatically state that settlements such as Jenne became important in the first place because they controlled an ancient trade with the fringes of the Inland Delta, and only later with the other towns farther down the Niger River. Thus trade necessarily accompanied the regional land rush. In this pursuit of commerce lay the seed of the next settlement transformation—the rapid emergence of the true city.
A legacy of the mass circulation, high-excitement reporting of the archaeological discoveries during the ‘Heroic Age’ is the romantic image of the abandoned city mound. Looming over a parched and barren plain, it astounds as much by its massive size as by the tumbled palaces or eroded ziggurats still highly visible on its surface. Rapid site expansion also characterized the urban transformation in the Niger floodplain. The accompanying photographs illustrate the evidence for this: excavation at the principal tell in the region, Jenne-jeno, reveals a reasonably complete picture of the city’s evolution. Jenne-jeno, or “ancient Jenne” in the Songhai language, is the ancestor of modern Jenne and occupies the same distributary channel merely three kilometers away. Founded by the 3rd century B.C., Jenne-jeno expanded to at minimum twelve hectares by about A.D. 50 and was a settlement of thirty-three hectares by sometime before A.D. 600-800. By this last date Jenne-jeno was enclosed by a 3.3 meter wide mud-brick wall of about 2.0 kilometer circumference—the settlement was of true urban dimensions.
The dramatic expansion of the principal site of Jenne-jeno is just one aspect of the urban transformation, however. During the early period of urbanism here, starting around A.D. 400, cities were clustered settlements. As Jenne-jeno expanded, smaller communities sprang up all around it. Jenne-jeno was one site in a cluster of twelve sites, each within a half kilometer of the next. This whole cluster is still recalled in the oral traditions as the “Jenne” of the past. Within the ancient Jenne cluster is another site test-excavated in 1981, Hambarketolo. Founded by the 2nd century A.D., Hambarketolo grew apace with Jenne-jeno and its maximum expansion (9 hectares) was reached at the same time as Jenne-jeno’s. At the climax of these two sites they had begun to merge. Evidence is growing that the other sites of the ancient Jenne cluster shared parallel histories.
Early urban clustering was not confined to Jenne-jeno. Only four kilometers down the same channel is another massive clustered unit of seventeen sites, ranging in size from 12.4 to less than 0.05 hectares. Eventually, clustering became the order of the day on all floodplain soil units and land-forms within the region. Clusters deep in the interior were limited to two or three, or. infrequently, up to six sites, But in major and minor distributary channels three to ten settlements, clustered around a principal site of 10 or more hectares are a common pattern, But how can we know the clusters were a true unit, and not just a series of sequentially-occupied communities? And if a true, contemporaneous unit, what was the purpose of this striking settlement pattern?
The answer to the first question makes the second all the more intriguing. During the 1981 field season we again used probability sampling techniques to select half the individual sites in each of seven clusters within a four-kilometer radius of Jenne for intensive surface collection. Seven other isolated sites were studied. We estimated the date of site abandonment by comparing surface ceramics with the 1,500 year stratified pottery sequence established at excavated sites nearby (at the type site of Jenne-jeno, for example, an internally consistent series of 26 radiocarbon dates anchors that sequence in time). Without exception, all sites within each cluster had been abandoned within a short span. Contemporaneous abandonment implies contemporaneous rather than sequential occupation. Of thirty sites dated by this method, twenty-five were abandoned within two or three centuries of the desertion of Jennejeno at about A.D. 1300. Of course, we will not know without excavation just when these satellite mounds were founded or, for that matter, whether occupation was con tinuous, as it appears to have been at Jennejeno. Test excavations at Hambarketolo, and the height to which debris had accumulated at all sites, suggest a massive population build-up in the region contemporaneous with Jenne-jeno’s expansion. If high settlement density was an integral feature of the first colonization of the Inland Niger Delta, why did it take the form of multiple satellite sites?
The 1981 season gave us some clues to the clustering question. Not all sites in each cluster are uniform in all artifacts present, although the uniformity of ceramics—the dating markers—parallels precisely what archaeologists find repeatedly in integrated urban regions. Patterns of differences in other artifacts can be deciphered fully only when a large enough sample of sites can be excavated. Nevertheless, some gross trends are apparent even now: these trends point clearly to differences of function among sites within clusters.
Iron smelting is an example of a specialized activity found at a restricted number of sites. Evidence of on-site smelting occurs in various forms: surface occurrences of scattered slag in significant den sity, discrete slag piles, furnace and tuyère parts, and piles of ore. In light of the fact that all ore had to be imported from outside the Inland Delta, it is noteworthy that all clusters surveyed in 1981 have at least one site with two or more pieces of primary evidence of smelting. Large clusters may have several. The presence of smelting appears to be correlated with site size. Sites 0.05 hectares (500 m²) or less never have smelting evidence. It is also rare at sites measuring 1.0-5.0 hectares. By contrast, smelting debris is abundant at all sites 0.5-1.0 hectares in size and at every site larger than 5.0 hectares. Crude as these observations may be, they clearly point to systematic differences in tasks carried out—and presumably to ‘communities’ responsible for such tasks at these sites. Systematic patterns allow prediction: in this case, sites with in situ evidence of smelting can be predicted by the number of sites in a cluster and by their size.
Are we perhaps seeing in these clustered tells the earliest expression of the region’s remarkable heterogeneity? From the earliest historical documentation to the present, Jenne has been a town of multiple ethnic groups engaged in different subsistence occupations (Bozo fishermen, Marka rice farmers, Fulani pastoralists) supporting a clearly defined hierarchy of fiercely independent cross-ethnic craft corporations (S. McIntosh and R. McIntosh 1980: part ii, 338-344: Galleis 1967]. The craft corporations are particularly important in the Jenne economic and social hierarchy today. Preeminent and renowned throughout West Africa are the Sorobana—the masons who, incidentally, are avid followers of the archaeology at Jenne-jeno. Next in prestige are the male smiths and the female potters (married to smiths, but in a separate corporation) who, because of the magic associated with their skills, live subtly ostracized in a distinct part of town. Successively lower in status are the weavers, leather craftsmen, and even griots (reciters of oral traditions and family histories—see Monteil 1903).
We propose the following hypothesis for the clustering principle observed during the early period of Inland Delta urbanism. A paradox arose soon after the Inland Delta land rush began. As population exploded on the floodplain, specialized production evolved in order to create surplus foodstuffs and goods for shipment to exterior trade partners. Specialist fishermen or rice farmers may have become divided along ethnic lines; other groups were skilled in manufacture. These latter evolved into the formal craft corporations of today and, from the debris left by their activities, they can be recognized at archaeological sites as distinct ‘communities.’ As production matured and commerce increased, different specialists found it advantageous to come together to reduce the cost of transporting services and goods between interdependent ‘communities’ and to facilitate bulk warehousing for efficient shipment. Specialist ‘communities’ gravitated together but were initially unwilling to surrender completely their identities to a monolithic urban presence. The resolution to the paradox was made through clustering. The whole functioned as a city of many heterogeneous communities, each nevertheless able to maintain a separate physical identity.
By the end of the 1st millennium A.D., some urban clusters had become very large indeed. The ancient Jenne-jeno cluster of twelve sites could boast a combined area of 51.2 hectares. Settlement density on the floodplain was several times greater than today, and one cannot help but recall al’Sa`di’s wistful half-mythical recollections of a period of long-past prosperity in the Inland Delta. This settlement climax ended precipitously. By A.D. 900-1,000 the inhabited part of Jenne-jeno began to shrink rapidly. Within three centuries the site was abandoned, along with the vast majority of sites within the ancient Jenne cluster and throughout the surveyed vicinity. Some inhabitants presumably moved to the present mound of Jenne. Occupation in the interior shifted to a far reduced number of single-component settlements. The hitherto avoided dunes and levees, permanently above the floods, became the preferred locations for villages. Evidence of population decline and settlement reorganization during the 2nd millennium A.D. is dramatic and indisputable, but the reasons for these changes are unclear.
Climatic changes anticipating modern environmental unpredictability may have played a contributory role, as may the migration into the area of new ethnic groups (the martial Bambara, in particular). The beginning of settlement contraction at Jenne-jeno coincides with evidence from several other West African localities of a climatic deterioration substantially more devastating than the 1968-1974 Sahel Drought. At about A.D. 1,000 both Lake Bosumtwi (Ghana] and Lake Chad dropped rapidly, and lake levels rose again only centuries later. The pan-West African precipitation drop implied by this evidence would have severely affected the rice and fish harvests of the Inland Delta and may have intensified the seasonal competition for land exploitation rights between pastoralists and farmers. Significant population decrease in the Jenne region may be linked to environmental causes. But can these alone explain the fundamental transformation of urban settlement distribution from dense to sparse, from clusters of variable sizes to isolated sites with a narrower range of size? We believe that other, social, agents were involved.
By the early 2nd millennium the sphere of West African trade, with Jenne-jeno as a principal participant in that trade, expanded beyond the great desert to embrace North Africa. By the mid-13th century the region was brought under the hegemony of the Islamic Empire of Mali. New currents of political centralization swept the Delta and a new unifying ideology, Islam, spread rapidly. Islam had begun to penetrate parts of West Africa centuries before as a rather exclusive religion of kings, courts, and worldly merchants. According to local traditions, the King of Jenne converted to Islam around A.D. 1250. By the 15th century, when the city first appears mentioned on a written page, Jenne and its sister city, Timbuktu, were centers of Islamic learning. Arabic manuscripts rivalled gold as the items of greatest value transported by camel caravan in the trans-Saharan trade between these cities and North Africa. Here, at last, were the centralizing forces missing earlier. Illustrating the power of the new political loyalty, one local ethnic group, the Nono, began to call themselves Marka—the “Men of Mali.” So powerful was the appeal of the new religion that, throughout northern West Africa, many communities abandoned venerable villages inhabited by their ancestors for new localities unpolluted by pagan practices (R. Mauny 1978: pers. comm.; R. McIntosh and S. McIntosh 1981:19-21). Oral traditions of present-day Jenne record a similar transformation of identity with the coming of Islam and a similar disdain for the beliefs of the ancestors at Jenne-jeno. Nevertheless, like al-Sa`di, modern Jennéens long for the commercial Golden Age to which the abandoned tells give eloquent witness.
Memories such as these underscore the special value of the Inland Delta mounds to the urban archaeologist. Jenne emerged because an environment of extreme plenty lacked certain critical necessities—an interesting parallel with early urban Mesopotamia (Redman 1978:225-6, 231-4), These mounds also provide general lessons, possibly of wide applicability, for understanding urban evolution. For example, archaeologists working in Mesopotamia recognize a period of settlement ‘implosion’ beginning during Uruk times (3600-3100 B.C.] and culminating by 2700 B.C. (beginning of the Early Dynastic). Small dispersed settlements expanded in number, then rapidly concentrated, or imploded, around one locality (Redman 1978:263-8). The center of implosion eventually fused to become a true, integrated city-state. For anyone familiar with that process, the analogy with the clustering principle in the Inland Niger Delta is remarkable. Likewise, Chinese archaeologists writing about Shang and Chou capitals of the Bronze Age remark on the concentration about the royal residences of small, discrete artisan hamlets or communities (Chang 1974). Can we see in this pattern a period when clustering was tolerated, but political centralization was fast crystalizing? Evidence of an indigenous urbanization which evolved close to the frontier between history and prehistory and for which oral traditions penetrate deep into prehistory should be invaluable for the comparative study of urban civilizations. As the forgotten tells of Mali become better known, their contribution to an understanding of indigenous urban beginnings will be considerable.