Pre-Colonial Origins of Urban Spaces in the West African Sahel: Street Networks,Trade,and Spatial Plurality

Trade and Coastal Regions

Trade on the edge of the Sahara was of paramount importance in the pre-colonial era in terms of politics of social structure and economy (Figure 1), and very likely, it played a role in where new towns were established and how they developed in size and prosperity. ¹¹ The West African Sahel represents a liminal zone on the edge of the Sahara, the word Sahel derived from the Arabic sāḥil, which means “coast.” The region was a zone of great importance for trans-Saharan trade related to Islam from the ninth century CE, ¹² but probably represented a hub of trade between the various environmental zones of West Africa already in the first millennium BCE. ¹³ At the time of the great West African empires of Ghana, Mali, and Songhay, which were involved and controlled the trade between the ninth and sixteenth century CE, the main articles were gold and slaves from the south and salt coming from the Sahara in the north. ¹⁴ The north-south flow of exchange across the Sahara broke up further into local networks in the Sahel zone, making use of the east-west transport on the river Niger in the Inland Niger Delta and the Niger Bend, in what is today Mali and Niger.

OPEN IN VIEWER

Figure 1.

Map of Africa showing the location of all sites mentioned in the text.

Trading towns elsewhere in Africa characterized by spatial plurality also developed on or were tightly linked to coastal zones. For example, trade on the coast of the Indian Ocean in East Africa saw its height between the thirteenth and sixteenth century with the Swahili-managed trade networks. ¹⁵ Similarly to the West African Sahel, the Swahili towns were ports on a border zone, where goods from around the Indian Ocean were exchanged for goods from the African interior, while in parallel, the Swahili managed the trade through the network of interlinked Swahili towns along the coast. ¹⁶ During the Atlantic era, the West African coast flourished, building their prosperity on trade with the Europeans, especially between the sixteenth and eighteenth centuries, until which time sourcing the goods for commerce from the African interior was exclusively in hands of the local African kings that dealt with the Europeans.

Coastal cities may be understood as frontier settlements on the edge of major environmental zones, at the start and end points of caravan routes, but they are also central hubs and market places of exchange networks represented by social ties. Although such urban societies lived in permanent settlements, the lifestyle of its inhabitants was seasonally that of nomadic or seminomadic people (sailors, porters, and traveling merchants) or at least tightly intertwined with these. The West African Sahel and East African coast were predominantly Islamic at the height of the pre-colonial trade period, and emphasis on community matters and formalization of human interaction was made possible by urban life. ¹⁷ In the coastal cities, goods and people arriving from rather clearly defined directions such as from the ocean or from the desert were brought together, stored, and redistributed. These activities might have been happening in spatial context of particular buildings, on open spaces such as markets, caravanserai, or in the houses, but while on the move in and out the city, objects of trade and people needed to engage with the city’s streets physically or through representatives.

Cities as Street Networks

An urban settlement as a collection of particular buildings or open spaces and streets can also be seen as a network of nodes and edges. The streets represent edges that somewhat differ from a simple schematic network, because especially in organically developed towns, they are usually not straight but mostly bent on junctions and open spaces, and these turns do not always represent nodes or end points with a given function in their own right. The bends and junctions are important nevertheless. As experiments in environmental psychology have confirmed, these places are where decisions must be made, and people moving in the network perceive their surroundings with more attention, and they become prominent in human memory. ¹⁸ Street as a network’s edge has an important role, because in movement on the main (usually long and wide) streets, people regardless of cultural background acquire knowledge of the city as a whole, first remembering landmarks along the route, then routes to places of their interest, and only then configurational knowledge of other places that may lead them to explore the city further. ¹⁹ The stages in this process are even more pronounced when maps are not available. ²⁰

Streets represent abstract heritage of the past, yet they share with buildings the ways in which they structure social action through creating obligatory passage points and stabilizing social action by increasing cost of innovative use, ²¹ which might be, for example, changing the points or set of places they interlink in the network. In analyzing streets as edges in the network and the way they facilitated particular social structuring and interaction, we cannot provide insights into individual experience of the town but rather look at what collective use was promoted by the spatial structure the urban society organically produced.

The understanding and perception of streets may of course differ from one culture to another. For example, in Islamic cities with which this article is most concerned, streets are known to be understood as potentially polluting or liminal spaces. ²² Apart from connecting places in the city and facilitating traffic of people, goods, and information, streets may also represent barriers. In Islamic cities with private blocks of houses and residential units with semiprivate courtyards, street may represent borders of a territory as a certain zone of protective power of the houses. While some researchers have clearly distinguished between territorial and spatial material representations, ²³ others engage with this classification less rigorously. ²⁴ Yet Islamic towns with their informal structure ²⁵ can be taken as an example of territorial behavior, which is an expression of power and degree of access to places, ²⁶ where streets delimit residential units and vice versa in a stereotypical way across the town to provide a feeling of security, privacy, and identity. ²⁷

Cultural factors such as these must have had an impact on the nature and distribution of interactions that took place on the streets. Specific insights are, however, difficult to discern from material evidence and hence our analysis is based solely on universal principles of bodily movement. ²⁸ The size of pre-industrial cities in Africa mostly allowed for convenient movement on foot, ²⁹ with associated sensory perception of sights, haptic stimuli (including wind, sand storms etc.), sounds, and smells that must have affected and tempered all social interactions taking place. People irrespective of their cultural background move and visually perceive their environment through their bodies and senses. Hence, we first set out to define each individual street on the basis of these principles, where the start and end of a street is defined by its course running approximately straight until it is interrupted with a building or an open space, or it bends at an angle greater than forty-five degrees, which obscures visual apprehension of the street’s continuation.

In towns that remain in use over generations and centuries, and are built using durable material, streets represent a fixed-feature element that changes only slowly and rarely. Especially the main streets—in terms of length, width, or centrality—represent arterial routes playing a major role in giving visitors an image and memory of the entire settlement. ³⁰ A historic town then never simply becomes old but is made such ³¹ through long-term preservation of visual sceneries and concepts inherent to its structure. That does not mean that settlements do not change, or that we must assume that social shifts occur while spatial structure remains unchanged. Construction of buildings, and, by extension, of urban streets, using certain building material is also a social choice, and so is the repair and maintenance of the built environment. Every spatial structure, hence, creates divisions and caters for making connections, and analyzing these, we can ask how society in the spatial setting chooses to regenerate or alter itself. The presented analysis of street networks aims to establish how the location of trade routes might have affected the evolving street networks within the towns of Timbuktu and Djenne, and whether there were any differences between the towns in this respect.

The Trade Towns of Timbuktu and Djenne

In this article, we analyze the street network and public spaces of two pre-modern towns in present-day Mali, Timbuktu, and Djenne. What is known about their environmental, political, and economic history suggests that they had a lot in common. Both towns have been predominantly Islamic: Timbuktu since its foundation ³² and an earlier settlement of Djenne becoming Islamic in the thirteenth century. ³³ They are located in the Sahel zone on the edge of the Sahara, Timbuktu to the north close to the Niger Bend, and Djenne in the Inland Niger Delta.

In terms of urban development, it is likely that both towns or their old parts at the core of modern cities gradually grew together out of smaller clusters or quarters that were set up near the trade routes and on the most suitable landscape features. The towns were linked by trans-Saharan trade in salt, gold, slaves, as well as book trade connected to Sankore University in Timbuktu. ³⁴ For this trade, travel on the river Niger played a crucial part according to Ibn Battuta’s accounts, ³⁵ especially since mid-fourteenth century. This is confirmed by later accounts, which suggest that the towns were as “two halves of the same city”; Timbuktu served as a trading outpost on the Niger directly sponsored by Djenne merchants. ³⁶ The salt arriving to Timbuktu from the Sahara was probably taken on canoes further south to Djenne, and from there, at least part of it was distributed on land to the gold mines—in case of Djenne probably those in modern Ghana and Ivory Coast. ³⁷

Timbuktu is closer to the desert and enjoyed less favorable environmental conditions. Trade routes led up to the town from the north and the Sahara, and the south in the direction of the Niger. It is likely that wadis once allowed a direct access to the Niger by boat, ³⁸ but, for example, Ibn Battuta who visited Timbuktu in the fourteenth century already described a six kilometer stretch of land between the river and Timbuktu town, ³⁹ a distance that today is nearly twenty kilometers. ⁴⁰ According to the written records, Timbuktu was founded in the early twelfth century CE as a Tuareg camp, ⁴¹ with previous (possibly early urban) occupation attested archaeologically in the wider hinterlands between the first and seventh centuries CE. ⁴² Timbuktu then fell under the political influence of the Empire of Mali in the thirteenth century. ⁴³ The visual landscape of Timbuktu was dominated by three “towers”, or minarets of mosques, the two most monumental standing on the limits of the city. ⁴⁴ The building of the great mosque on the south edge of the city, Djinguereber, was visited and sponsored by Mansa Musa, a fourteenth century Malian emperor on his pilgrimage to Mecca. ⁴⁵ Sankore mosque at the north end represents the well-known university of Timbuktu and is somewhat later than Djinguereber, and the third mosque, centrally- positioned Sidi Yahya, is the youngest and smallest dated to the fifteenth century, ⁴⁶ when the Empire of Songhay controlled the town. Archaeological and historical studies agree that the joining of Timbuktu’s quarters together was only completed in the sixteenth century. ⁴⁷ Thanks to its direct links with Djenne, Timbuktu replaced Walata as the center of trade. ⁴⁸

Djenne (also spelled Jenne) was founded possibly in the thirteenth century CE, ⁴⁹ in close vicinity of an even older settlement known as Jenne-jeno, which can be classified as urban at least from the fifth century CE. ⁵⁰ Djenne was originally a settlement of Bozo fishermen and politically subject to Soninke. ⁵¹ It is known from the earliest historical accounts of the fifteenth century ⁵² and was perhaps less isolated for European travelers from the late eighteenth century onward, when Mungo Park visited Djenne but did not reach Timbuktu, and Felix Dubois visited both towns. Djenne is located directly by the river, being surrounded on most sides by water and becoming an island during floods. According to the historical sources, this characteristic made it different to Timbuktu and more difficult to conquer; in the thirteenth and fourteenth centuries, both towns were part of the Empire of Mali, but after its decline, Djenne flourished as an independent town. ⁵³ Trade routes reached Djenne from west and east, traversing the river or making use of bridges. ⁵⁴ As opposed to Timbuktu, Djenne’s Great mosque is located at the heart of the town, removed from the point where trade routes enter the city. The present-day mosque is a complete redevelopment undertaken by Sheikh Amadou in the early twentieth century, when also a significantly smaller religious school was established in the center of the town to the east of the market. The original mosque was founded in the thirteenth century.

The two towns probably developed to their highest economic prosperity and importance in trade at about the same time between the fifteenth to sixteenth centuries. Situated on the convergence of major trade routes that were channeling the flow of goods and people, the spatial layout of these settlements might be expected to have revolved around the placements and distribution of trade-related practices. We might ask whether trade was the defining force in affecting the inner structure of the towns or whether the material definition of street networks and interconnectedness of spaces within the towns display evidence of being influenced by other social factors, such as that Timbuktu was a center of learning or that Djenne functioned as an independent political center. By analyzing the spatial layout of Timbuktu and Djenne, we can learn which of these aspects might have gained an upper hand in defining the resulting street layout of these towns, as we know them today.

Analysis of Street Networks, Open Spaces, and Movement

Both Timbuktu and Djenne are today much larger towns than in the past, having outgrown their earlier boundaries, especially in the case of Timbuktu. The built environment of Timbuktu has for centuries been subject to not only desertification and covering by wind-blown sands ⁵⁵ but also continual upkeep of buildings and house blocks. ⁵⁶ The extent of the historic town is apparent from the sharp contrast between the irregular street grid of the historic core and more right-angle regular layout of the modern town. In our analysis, we focused on the historic core of Timbuktu. Historical and pictorial evidence further confirms the fact that the extent of the historical core matches the former extent of the town. The three mosques of Timbuktu have been known to dominate its visual landscape, and their positioning, with the largest mosques on the edges of the town have approximately delimited its boundaries. ⁵⁷

The clay architecture of Djenne is particularly durable, and although the layout of individual houses or their blocks might change, the distribution and course of streets that had been formed by continual line of buildings can be expected to have changed, only little and with great difficulty once established. The town of Old Djenne is located virtually on an island, nearly completely enclosed by branches of the Niger. The natural environment, hence, approximately delimits its extent. The development of the built environment in both towns was affected during the period of French colonialism, when certain buildings were subject to remodeling following contemporary political agenda or virtually built anew (such as the Great Mosque of Djenne).

The analysis presented in the following paragraphs contributes to an understanding of the movement of people in relation to specific structuration of urban space in Sub-Saharan Africa. ⁵⁸ The Table 1 presents the quantified data we observed analyzing both towns. It shows that apart from sharing common history of development and trade, they are of approximately the same size, interestingly with virtually the same perimeter length (difference of approximately 24 m). The survey data on the street network in these towns conducted by the Zamani project (University of Cape Town) were mapped and further edited in geographic information systems (ESRI ArcGIS 10) and combined with information from satellite images. The streets were plotted as linear features defined by their unbroken course, that is, a single street is supposed to continue until it ends in an open space, edge of settlement, is blocked by a building, or turns at an angle that would prevent a moving person to anticipate its further direction (approximately forty-five degrees or more). They are referred to as axial streets, because they represent visual and movement axes of the town. In their definition, they somewhat differ in our study from the category of axial lines as used in standard space syntax method. ⁵⁹ Here, we wanted to understand the role of open spaces, so we include them among the features that end the axial streets.

OPEN IN VIEWER

Table 1.

The characteristics of the street networks of Timbuktu and Djenne.

	Timbuktu	Djenne
period	13th-15th -17th century	13th-14/15th century
polity	Mali, Songhay	Mali, independent
Total area analysed =size of the Old Town	520 491m2	453 741 m2
Perimeter of the Old Town	2 761 m	2 737 m
Total number of axial streets	276	382
No. of N-S streets	126	194
No. of W-E streets	150	188
No. of open spaces	21	21
Town area relative to number of streets	52 ha : 276 =5.3 street/ha	45.3 ha : 382 =8.4 street/ha
Average street length	71.1m	56.4 m
Average length of N-S streets: W-E streets	66 m : 76 m	54 m : 59m

Trade

We can also compare the impact movement related to trade had on defining the communication network within the towns. We may argue that trade would be represented by the distribution of markets and also by following the continued course of major trade routes as they enter the town and become avenues that would naturally bear the most traffic of arriving and departing merchants and goods. In the case of Djenne, two major trade routes are known to have led to the city from the west, and one left the town toward the river on the east, all of these following the edge of the town forming an avenue before ending at the core of the town on the largest open space by the great mosque. ⁶⁰ These are prominent on the analysis of the street network, where longest axial streets are highlighted by increased relative thickness (Figure 2). However, there are other streets within the town that display similar qualities, unrelated to the direction of trade routes. These are somewhat related to the central area but leading up to it indirectly. They might have serviced the inner-city traffic, and interestingly correlate with the edges that existed between the quarters of the town before they were built up with development. ⁶¹ These quarters were characterized by the occupation of the residents—merchants, artisans, fishermen, and farmers/herders clockwise from the north. ⁶²

OPEN IN VIEWER

Figure 2.

Analysis of Djenne street network in geographic information systems (GIS). The thickness of the axial streets represents their relative length. The black dashed line represents approximate limits of the original urban quarters, along which they subsequently grew together.

Less detailed information is available for Timbuktu, but one major trade route must have been reaching the town from the north or north-west servicing caravans coming from across the Sahara. To the south, traffic would move in the direction of the Niger and to Kabara. ⁶³ The north-south focus on trade-related movement is discernible from the distribution of longest axial streets (Figure 3). These divide the town into approximately four parallel vertical parts. However, Timbuktu was originally built on two parallel longitudinal dunes, and the residents of both were using water standing in the interdunal depression known as Badjinde or The Stream of Hippos in Songhai. ⁶⁴ As opposed to Djenne, this way, the main axial streets cut across and through the original residential quarters, instead of passing along their edge. A number of west-east streets running from Timbuktu’s main avenues then serviced the intra-town traffic, but interestingly the most central avenue, which passes by the two mosques and through the main markets, has the least number of such off-shooting west-east streets. As a result, those parts of the city adjacent to its core are the least connected with the rest of the town. This pattern is further accentuated with the lack of open spaces in this area.

OPEN IN VIEWER

Figure 3.

Analysis of Timbuktu street network in GIS. The thickness of the axial streets represents their relative length. The white dashed line represents approximate limits of the original urban quarters, along which they subsequently grew together.

While the concentric distribution of city quarters that gradually grew together towards the center of the town is reflected in the street network of Djenne, in Timbuktu, the distribution of major axial streets is not in accordance with the edges of the original city quarters. To consider whether the position of trade routes, which was mostly west-east in Djenne and north-south in Timbuktu, affected the intra-town direction of streets, we calculated the number of streets running north-south and east-west (Figures 4 and 5). The results show that Djenne has approximately similar number of north-south (194) and west-east (188) streets, with somewhat greater difference in the case of Timbuktu (126 and 150, respectively). The same pattern is present when we look at their average length, which is for both directions very similar in case of Djenne (54 and 59 m) and slightly larger for Timbuktu (66 and 76 m). Djenne’s street network in terms of directional distribution of streets is even and reflects neither the course and placement of major trade routes. In Timbuktu, the west-east streets, which can be related to intra-town and intra-quarter movement, generally predominate in number and average length. On the other hand, the three north-south axial avenues connect the major public buildings and spaces, and most likely bore the load of trade-related traffic. These main arterials crossed the natural depression between the two dunes on which the town was established, which underlines their special standing in the entire street network.

OPEN IN VIEWER

Figure 4.

Analysis of Djenne street network in GIS highlighting the distribution of North-South and West-East running streets.

OPEN IN VIEWER

Figure 5.

Analysis of Timbuktu street network in GIS highlighting the distribution of North-South and West-East running streets.

Duality, Plurality, and Urban Quarters

Because both towns became spatially continual settlements only with the joining of the individual town quarters, which took several centuries to occur, it is important to consider how the quarters came to be spatially interconnected with one another and, also, whether they differ in comparison of their individual street networks. For this end, we consider the relative length of streets in individual quarters and the placement of longer axial streets that cross the approximate boundaries between quarters reaching deeper into the quarters’ cores.

The eastern part of Djenne was described as consisting of two quarters occupied by merchants, who were Muslim, ⁶⁵ that is, Dubois’ “Moorish quarter,” ⁶⁶ and a south-east quarter occupied by artisans (Figure 2). On the easternmost edge of the city was the house of the town’s leader. ⁶⁷ These, however, appear spatially grown together seamlessly without an edge between them apparent, and the street network does not bear evidence of spatial inequality in terms of separation. Together, they are distinct from the rest of the city in featuring shorter streets and only six squares/open spaces positioned away from the center of the city with mosques. The center of Djenne has a less dense street network but longer avenues, as well as seven or eight larger open spaces approximately. In terms of movement, it facilitates communication between the western and eastern halves of the city and connects to the major trade routes. The western part of Djenne known as Yoboukayna and smaller Djoboro ⁶⁸ features three long avenues running in west-east direction, which cut through Yoboukayna. It is hence more comprehensive for orientation, especially for people unfamiliar with the environment. One of the main axial streets serves as an edge between the quarter of farmers/herders (north-west) and fishers (south-west).

Timbuktu was divided by the natural relief in two quarters, the northern part featuring the Sankore mosque as the scholarly center. The core part of the town in the interdunal depression is the youngest in the town, but apart from being traversed by the three north-south main axial avenues, it features comparatively short streets (Figure 3). It forms an integrated part of the urban spatial fabric and yet it does not provide the radiating communication channels to the rest of the city. This results in a pattern where the mosques and markets are evenly spread out along the main communication channels attracting traffic, which can easily reach them or pass through. The street network from the west and east is spread out to draw movement toward the main north-south axes, and it does so using a much denser web of relatively long and straight west-east axial streets. The quarters do not seem to be distinguished in terms of differential street networks; in fact, the two halves of the town are similar if considered separately, although in between them lies the less interconnected Bajinde quarter. The west and east part of the city, if we consider them divided by the central avenue (regardless of natural relief), have approximately the same number of open spaces (eleven and ten). The north part of the town with Sankore mosque, which has been so often noted for by foreign travelers, ⁶⁹ does not display differential street pattern either, but it is not positioned in direction of any of the major trade thoroughfares, and it is not connected to the main market space by a direct axial line; instead, the approach route is set off course. The north part of the town also features the least open spaces (three in total, located north of the main market).

A number of towns in West Africa have been ascribed a dual nature in spatial terms. Al-Bakri in 1068 described the towns of Sila and Ghana as consisting of two towns, characterized by Muslim and non-Muslim pagan population with the king’s enclosure with the latter and nevertheless featuring a mosque for the king and visitors, while between both parts of the town, there was continuous habitation. ⁷⁰ On the basis of archaeological research, the same characteristic is claimed for a number of other towns, some of which are contemporary with Old Djenne and Timbuktu, such as Gao. ⁷¹ The tradition of dual centers, which co-exist as complementary sharing a sphere of influence over a single polity, is also known from elsewhere in West Africa. ⁷²

The analyses presented in this article point out that the concept of plurality represented by urban quarters within a single town should be more frequently brought into the debate on urbanism in the West African Sahel. Parallels can be drawn with another “frontier” zone of the East coast of Africa, where trading Islamic towns of the Swahili were in the same period tied into far-reaching trade networks. The spatial arrangements of the towns, place of residence, and distribution of quarters have been established as deeply meaningful for enacting political power and social relationships of the Swahili urban communities. ⁷³ Studying the spatial development of the towns in the West African Sahel over time and the patterns preserved in the layout, we argue that the towns evolved around a certain degree of spatial plurality represented by individual quarters as distinctive entities. In case of Timbuktu, the spatial arrangement of quarters and the street network suggest that the town served and facilitated the flow of trade-related traffic. In Djenne, the quarters of the town emerged in almost-island conditions delimited by the Niger, and they were in more competitive or complementarily specialized spatial relationships respective to trade-related traffic. This fits in with the historical argument that Djenne was the producer and investor, while Timbuktu represented the depot and intermediary in the relationship between both towns ⁷⁴ as well as later architectural historical and ethnographic analysis. ⁷⁵

Conclusion

This article aimed to establish how two major social defining forces, involvement in trade and urban life represented by place of residence in individual quarters, might have affected and led a dialogue on the level of spatial fabric and urban layout of Timbuktu and Djenne.

The street network of Timbuktu was probably more affected by the placement of trade routes than that of Djenne. In Timbuktu, major avenues that represented continuation of trade routes were channeling trade-related traffic to cross through the individual town quarters, but in Djenne, such routes were following their limits and passing through the borders between them.

As for the social role of individual urban quarters as it is discernible and promoted by the spatial layout, at Timbuktu, the quarters brought together by trade were, on the other hand, clearly distinguished with landmarks (the three minarets of major mosques) evenly distributed on the edges of the town with the smallest mosque in the middle. At Djenne, trade was channeled to pass where all quarters had access to it, but it did not facilitate movement between quarters.

Although both towns had approximately four to five main quarters, they grew together in a way, which resulted in the towns seemingly having an eastern and western half. In Timbuktu, the two halves have a very similar layout within. Although the quarters might have differed in terms of resident majority, such as the learned families residing in the vicinity of Sankore mosque, ⁷⁶ the individual quarters do not display a differential spatial pattern. The street network of the town brings traffic to the main routes, and there is a pattern in preference for open spaces on the main routes also. In the case of Djenne, the eastern half incorporating the original Muslim quarters of traders and artisans developed to feature comparatively short winding streets, while the western half of Djenne became more open and, hence, potentially more comprehensible for foreigners entering the town.

For both analyzed towns, the concept of spatial plurality known in West Africa mostly in the dual form, hence, appears to be present and inherent to their nature, given their development in the pre-colonial period. By analyzing the spatial organization attained by the towns involved in the trans-Saharan trade, we can more reliably disentangle the trajectories of urban social development, which characterized the edge of the Sahara in this period. More intensive and more systematic analyses of the urban fabric in this part of the world would provide information for further testing of the hypotheses presented here. It has been shown that the preserved street networks in the historic core of cities hold considerable potential for teasing out more information on the spatial logic that provided an ever-present context to the life of the urbanites in the recent past. Current research can make use of an increasing volume of available data and analytical tools such as satellite imagery, terrestrial and airborne 3D scanning, spatially indexed databases, and geographic information systems. Future studies should aim to expand this type of enquiry geographically, making cross-cultural comparisons. Spatial plurality being enacted in the form of urban quarters makes it an all the more promising theme for a wider research program of comparative analyses.