Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, AFRICAN HISTORY ( (c) Oxford University Press USA, 2020. All Rights Reserved. Personal use only; commercial use is strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2020

The Empire of Ghana

Summary and Keywords

The Empire of Ghana is one of the earliest known political formations in West Africa. Within the context of a growing trans-Saharan trade, Arabic sources begin to mention “Ghāna,” the name of a ruler as well as of the city or country he ruled, in the 9th century. Repeatedly named in connection with fabulous riches in gold, Ghāna had acquired a preeminent role in the western Sahel and was a leader among a large group of smaller polities. Ghāna’s influence waned, and by the mid-14th century its ruler had become subordinate to the Empire of Mali. Over the course of a complex history of research, the Empire of Ghana became equated with the Soninké people’s legend of Wagadu and the archaeological site of Kumbi Saleh in southern Mauritania was identified as its capital. Yet between historical sources, oral traditions, and archaeological finds, little is known with certainty about the Empire of Ghana. Most questions on this early West African empire remain unanswered, including its location, development, the nature and extent of its rule, and the circumstances of its demise.

Keywords: Empire of Ghana, Wagadu, West African history, Kumbi Saleh, Soninké, West African archaeology, oral traditions, trans-Saharan trade

Ghāna, Wagadu, and the Empire of Ghana

What is commonly referred to as the Empire of Ghana in books and documentaries on Africa’s history, is not as straightforward as it sounds. In fact, several different elements of the past in the borderlands of Mauritania and Mali have become conflated, rightly or wrongly, under this label.1 Firstly, there are mentions of a political entity called Ghāna in written Arabic sources. Secondly, there is the legend of the Soninké people about a polity that ruled in an area called Wagadu. Thirdly, there is a comparatively long history of archaeological research in the wider region that has become attached to the Empire of Ghana. There are few other topics in African history that have been of interest to European scholars for longer than the Empire of Ghana, which has made regular appearances in the literature since the 19th century. In this long engagement, scholarly knowledge about the Empire of Ghana has acquired a colorful history of its own, in which some assumptions have become certainties and some facts have been chosen as real while others were discarded.2 On critical examination, it becomes clear that very little about the Empire of Ghana is known with any degree of certainty.

The Ghāna of Medieval Arabic Sources

The first mentions of the political entity that later became known as the Empire of Ghana are in Arabic texts after the 8th century ce. In the 10th-century work of al-Ma’sūdī, there is a reference to an unknown manuscript by al-Fazārī that appears to be from the late 8th or early 9th century. Al-Fazārī is quoted as mentioning “Ghāna, the land of gold,”3 an association that sticks to Ghāna throughout its mentions in the Arabic source material. Written between 831 and 842, al-Khuwārizmī’s Ṣurat al-arḍ mentions Ghāna and its geographic position, as well as a mountain of the same name, Jabal Ghāna, from which there flows a river into the sea.4 What exactly Ghāna was in this early period, however, is open to debate among the translators: Cuoq translates al-Fazārī as describing “the state of Ghāna”5 while Levtzion and Hopkins translate the same passage as “the province of Ghāna.”6 Some clarity is introduced by Al-Ya’qūbī in his Tārīkh, written in 872–873. In one of the most fundamental quotes for the early political history of West Africa, he states:

Then there is the kingdom of Ghāna, whose king is also very powerful. In his country are the gold mines, and under his authority are a number of kings. Among them are the kingdom of ‘ĀM and the kingdom of Sāma. Gold is found in the whole of this country.7

Together with the preceding section, a list of kings who are subordinate to the nearby king of Kawkaw (Gao), this passage is taken as proof for the existence of a complex political hierarchy in the western Sahel and is the main justification for speaking about an “Empire” of Ghana. We must however note that, by this point, the political structure of Ghāna has undergone three processes of distortion: first, in the way the North African visitors understood what they saw and related it; second, in the way the authors of the Arabic sources interpreted this information; and third, in the way that European scholars interpreted the Arabic sources. To speak of Ghāna as an empire and of its rulers as kings certainly overfamiliarizes a political form whose actual shape we know only in outline and by approximation.

Yet if Al-Ya ̔qūbī states that Ghāna is a “kingdom,” then Ibn al-Faqīh, writing after 903, muddies the waters again by referring to Ghāna as a “land” as well as a “town.”8 Ibn Ḥawqal, in the second half of the 10th century, is of the opinion that Ghāna is the name of the ruler as well as the country he rules.9 The most detailed source, Al-Bakrī in the 11th century, finally states that “Ghāna is a title given to their kings,” while the region has the name Awkar. All the while, however, Al-Bakrī still refers to the “town of Ghāna.” While the Empire of Ghana has become a common term, it is far from clear whether this is in any way a name that this polity used for itself, or whether it rather refers to the ruler, the capital, or the core region it controls. Neither do we know what language the word “Ghāna” is taken from. It is neither Arabic, nor does it appear to be from a Mandé language.10

We know little more about the political structure of Ghāna. Al-Bakrī is able to inform us that the ruler in 1067–1068 was a man named Tunka Manin and that he had taken on this role some five years earlier from his maternal uncle, Basi, who had come to power at a very advanced age. We further learn that rule was hereditary in Ghāna, customarily passing down the maternal line, as it had done with Basi and Tunka Manin.11 The primacy of the rulers of Ghāna over others in the area is often stated in the sources, but the mechanisms by which the empire ruled are only available in glimpses. Part of the rule was certainly based on the use of force, particularly in form of raiding.12 We also hear that the ruler of Ghāna held the sons of subordinate rulers at his court.13 This practice of taking royal hostages is also said to have been employed later by the Empire of Mali and to have been a common practice of rulers in this region.14


Since the contact of North African Arabs with Ghāna was almost exclusively in the arena of trade, this is the one of the main preoccupations of most medieval Arabic sources, besides giving readers entertaining accounts of exotic places. We thus know a fair deal about the trade with Ghāna, albeit mostly from an Arab point of view. From the 9th century onward, there appears to have been a steady exchange between Ghāna and North Africa via the trans-Saharan trade routes. Several sources show that having been to Ghāna was noteworthy yet not unheard of among the North African populations at the time.15 Yet the beginnings of this exchange, and thus also of Ghāna, may lie even further in the past, as Ibn Ḥawqal describes an old route from Egypt to Ghāna that appears to have been abandoned in the 9th century due to climatic and political factors.16

From the sources, it appears that the prime reason for northern travelers to cross the Sahara to Ghāna between the 9th and 14th centuries was gold. Slaves are also mentioned as being imported from Saharan and sub-Saharan West Africa more generally, yet Ghāna is only ever mentioned in connection with the gold trade. This trade became so legendary that myths developed around it, such as that of the silent trade,17 an unlikely scenario that included a barter in which none of the persons involved ever met. This account has been dismissed as most likely fictitious.18

What was brought south in exchange for this gold, however, is more difficult to find out. The sources usually speak about “wares” in general terms, but we can glean that clothes made up a part of these, as did horse furniture. The large majority of Ghāna’s desert-side trade was not trans-Saharan, but involved exporting southern products such as honey and probably other foodstuffs into the Sahara in exchange for vital desert salt.19

Customs of the People of Ghāna

The sources also provide a small amount of detail on the food and habits of the people of Ghāna. They mention sorghum, millet, and cowpeas as food and the use of animal skins and textiles as clothing.20 Al-Bakrī goes into some detail on the ruler’s outfit, which he describes as comprising necklaces and bracelets with a tall, gold-decorated cap and a turban of fine cotton. The ruler and his heir were the only non-Muslims allowed to wear sewn clothes. All others wrapped themselves in unsewn cloth. The ruler’s regalia are described as warlike, with shields and spears.21

More details are available on the capital of Ghāna. Its description is one of the most famous passages from Al-Bakrī, and is worth quoting at length here:

The city of Ghāna consists of two towns situated on a plain. One of these towns, which is inhabited by Muslims, is large and possesses twelve mosques . . . In the environs are wells with sweet water, from which they drink and with which they grow vegetables. The king’s town is six miles distant from this one and bears the name of Al-Ghaba. Between these two towns there are continuous habitations. The houses of the inhabitants are of stone and acacia wood. The king has a palace and a number of domed dwellings all surrounded with an enclosure like a city wall. In the king’s town, and not far from his court of justice, is a mosque where the Muslims who arrive at his court pray. Around the king’s town are domed buildings and groves and thickets where the sorcerers of these people, men in charge of the religious cult, live. In them too are their idols and the tombs of their kings. These woods are guarded and none may enter them and know what is there. In them are also the king’s prisons. If somebody is imprisoned there no news of him is ever heard.22

As well as the geography and building technology, the quote reflects the extent to which the Islamic faith had by this time begun to leave physical traces, though the rulers, and presumably large parts of the rest of the population, remained non-Muslim. Non-Muslim religious practices are also reflected in the burial rites of the rulers. Al-Bakrī describes the building of a huge burial mound with a wooden substructure in which the body of the deceased ruler was placed. He was accompanied by grave goods and human sacrifices, and intoxicating drinks were consumed during the ceremonies.

Overall, it is clear that the sporadic mentions in the Arabic sources, only a few of which contain original insights, do not allow much of a reconstruction of what Ghāna was, beyond basic knowledge of a somewhat expansive and durable political entity with a hereditary male ruler who was not averse to receiving North African Muslims trading for gold. Nor, for that matter, do they clearly state where it was. Much energy has been expended in reconstructing the itineraries described in Al-Bakrī and Ibn Ḥawqal.23 A process of equating places mentioned in the texts with places that still exist or are known indicates roughly the area framed by, and south of, the Mauritanian Dhars Tichitt, Oualata and Néma (see fig. 1). In 20th-century terminology, this area comprises the regions known as Awkar (as al-Bakrī mentions), Hodh, and Bakhunu in the south.

The Empire of Ghana

Figure 1. Map showing the area of the Empire of Ghana and key archaeological sites of the region.

Copyright N. Gestrich.

The Soninké Legend of Wagadu

The Soninké tradition of Wagadu tells of a movement of immigration into and domination of a region called Wagadu, focussing especially on its beginning and its end. While Wagadu is today a relatively small region around the towns of Nara and Gumbu in Mali, it is understood in the tradition to have been a rather large area comprising the border regions of Mali and Mauritania and extending quite far into both countries. The traditions seem to center, however, on the regions of Bakhunu, Awkar, Kingi, and the area of Jajiga.24 As such, it comprises the only reconstructable location of Ghāna from the Arabic sources, the “Awkar” mentioned in Al-Bakrī.

The recorded traditions tend to begin with the peregrinations of an ancestor named Dinga, who is said to have traveled the world far and wide, especially in Yemen and India. In the version recited by Diarra Sylla,25 Dinga marries at a legendary location known as Dalangumbé, which Dieterlen and Sylla give as being close to modern-day Nioro du Sahel, despite the fact that there is at present no known locality with that name. After a prolonged period at Dalangumbé, Sylla has Dinga return east to rule at Aswan in Egypt. His progeny, specifically his son Diabé Cissé, return to their maternal region and apparently set up a polity of some description.

Diabé Cissé is the first to be known under the title Manga, which his successors subsequently also bear. In Sylla’s version, Diabé Cissé is referred to as Kaya Manga, a name that is central to linking this tradition with the Arabic sources. The tradition gives several main characteristics to the polity set up by Diabé Cissé. Great riches in gold, people, cattle, and horses are among the praises sung to him. From the perspective of Diarra Sylla, then, the polity of Wagadu was at once warlike, counting on horses and mounted warriors to increase the population, as well as placing an emphasis on the ownership of cattle, and finally on gold, presumably as the basis for trade.

The emergent Wagadu polity is also linked to several main locations: Kumbi, where Diabé Cissé is said to have ruled, and Kharambéra, Tengume, and Jengédé. The latter three are the territories that Diabé Cissé protected and Jengédé is also the place where an oath was taken to tie the people of Wagadu together. None of these locations can be positively identified, though Dieterlen and Sylla appear to equate Jengédé with the settlement of Yengi, not far from the archaeological sites at Kumbi Saleh.26 Dieterlen states that Kumbi was developed on these three territories.27 The transcription of Diarra Sylla’s account describes these places as gallɛ, which refers to an enclosed or demarcated space. The space could be an orchard, garden, or cattle enclosure, but also extends to the ownership of slaves or to a domestic compound. The rulers are described as gallɛn tangandaana, those who oversee or protect the gallɛ.28 The choice of this word is highly significant and underlines the agropastoral element of Wagadu’s beginnings, as well as having overtones of slavery.

Not much material is available on the period between the Wagadu polity’s establishment by Diabé Cissé and its dissolution. A number of other Manga are usually cited in many varying lists.29 With them are named further territories in the reach of the historic Wagadu region.

While traditions are thin for the period of the Wagadu polity’s rule, the event of its dissolution is one of the best known traditions in West Africa and has been recorded in many versions since the late 19th century.30 It centers on the killing of a monstrous snake and its aftermath. It seems that, at the establishment of Kumbi, Diabé Cissé had made a pact with this snake, called Bida, in order to ensure the prosperity of the country. In return for ensuring steady rains or a steady supply of gold (the traditions vary), a virgin girl was to be sacrificed to the snake each year. This arrangement came to an end when the lover of a designated sacrifice cut off the snake’s head and took flight with the girl. In dying, the snake cursed Wagadu, saying that no rain would fall (or no gold would be found) for seven years. This brought Wagadu to a catastrophic end, leading its people to disperse.

Were Ghāna and Wagadu the Same?

There are a number of points that support the idea that the sources on Ghāna and the legend of Wagadu are referring to the same polity. For one, they both refer to very similar geographical areas and are the only political entities that are known as having ruled these areas. Secondly, the name of the ruler reported by al-Bakrī, Tunka Manin, indicates a Soninké-speaking population, as tunka is a Soninké word for a ruler. It is thus tempting to think that written sources and oral traditions of a powerful Soninké polity in this area must be referring to the same thing. A further indicator might be the persistent association of the rulers with gold, both for Ghana and Wagadu, which is notable since the gold-producing areas of West Africa are usually thought to have been further south.31

There are, however, important caveats to bear in mind, the first being the difficulty of dating oral tradition. It is clear that the Wagadu legend is not recent. The first parts of it enter into written form in the Tarikhs of Timbuktu in the 17th century, by which time it was already considered ancient. Both the Tarikh al-Sudan (TaS) and the Tarikh al-Fattash (TaF) consider it to have preceded the Empire of Mali. They give dates that place the period of its rule in the first millennium ce, relative to the hijra, but these must be taken with extreme caution. Within the known versions of this tradition, there are no temporal markers that would allow it to be fixed to a specific period.32 This is a major obstacle for equating Wagadu with Ghāna. As Bathily states, it is entirely possible that the Soninké legend is inspired by Late Stone Age developments in southern Mauritania in the last three millennia bce, notably the emergence and decline of large settlements on the escarpments near Tichitt and Oualata.33 As the far better researched Homeric epics demonstrate, the preservation of legendary elements in such an extended time frame is entirely possible.

A further difficulty lies in the name Ghāna, whose meaning remains uncertain,34 but it seems unlikely that even earlier variants of the Soninké language would have used it to designate a ruler as Al-Bakrī claims.35 It is clear that Ghāna is not an Arabic term,36 but Masonen thinks it possible that it might be a Wolof expression that became a scholarly term within Arab geographical writing. Finally, Masonen suggests that Ghana might refer to a yard or enclosure, since there is a close equivalent in the northern Soninké dialect of Azer. In this case, which has parallels to names for royal courts in West Africa and beyond,37 we return to a meaning very close to the gallɛ that the rulers of the oral tradition are said to have protected. Nevertheless, the etymology of the term remains speculative and the fact that the word Ghāna appears in no known version of the Wagadu legend is at least puzzling. We also cannot know whether the title tunka really indicates that the rulers or their populations were Soninké at this time, or whether it was simply someone in the chain of al-Bakrī’s informants who gave the ruler this Soninké epithet.

In order to find the first point at which the Ghāna of the written sources and the Wagadu legend are equated, we must return to the TaS, written in Timbuktu in the 1650s: “The first ruler to establish a state there [where the Empire of Mali later ruled] was Qayamagha, the seat of his sovereignty being Ghana a large city in the land of Baghana.”38 Al-Sadi, the author of the TaS, owes most of his historical information to oral accounts of individuals he considered to be trustworthy. But we also know that he consulted works by earlier authors, for instance a work by Ibn Baṭṭūṭa. It thus seems possible that he added knowledge of Ghāna gained from al-Bakrī to the oral traditions of the 17th-century Sahel.

It is interesting to note that the TaF, the other main work of Timbuktu’s brief historiographical tradition, whose original version was completed around a decade later than the TaS,39 does not use the word Ghāna. It similarly speaks of rulers called Kaya Maga, thus of the Wagadu legend, but, as in Sylla’s version, the capital is given as Kumbi, while the last of the rulers resided at Koronga, a known location very close to the site of Kumbi Saleh.

The combination of these two sources, with their uncontrollable mix of 17th-century oral information, influences from earlier authors, and later transformations of the manuscript texts, thus lie at the basis of two problematic claims of 20th-century writing on this topic. Firstly, that the Empire of Ghana and the Wagadu polity are the same,40 and secondly, that the capital of the Empire of Ghana was at Kumbi Saleh, in today’s southern Mauritania. In particular, the addition of the information from the TaF to the TaS left its mark on the work of Maurice Delafosse after 1913 (fig. 2). Delafosse’s work in turn has been highly influential for the historiography of West Africa for decades, and his conviction that Ghāna and Wagadu were the same has remained in the literature to the present day.

The Empire of Ghana

Figure 2. Maurice Delafosse in 1924.

Artist unknown, Wikimedia Commons.

Archaeology and the Development of the Western Sahel in the Late First Millennium

Only from the 1970s onward has modern archaeological research been able to furnish the background of wider developments before which the sparse historical mentions of Ghāna must be understood. Significant advances have been made that shed light on the environmental, technological, and social developments in West Africa in the periods leading up to and between the 8th and 14th centuries.

The wider region of southeastern Mauritania, which is central to the sources about Ghāna, was home to a remarkable agropastoral society between around 2500–1000 bce. The remains of large stone-built settlements on the escarpments of Tichitt and Walata have provided evidence of large communities with domesticated cattle and millet. This way of life developed toward the end of a climatic development that saw a previously verdant Sahara dry out considerably. The sites of this so-called Tichitt tradition went into decline after 1000bce and its populations probably moved south toward the permanent watercourses of the Niger.41 In the period between 300 bce and 300 ce, this drying climate took an extreme turn, with the desert margin moving considerably further south than where it currently lies. After this “Big Dry” event,42 however, the climate became considerably wetter, and the desert margin appears to have moved north rapidly. For the area of the Empire of Ghana, the period after 300 ce thus opened up areas for human exploitation and habitation which had not been inhabitable for the best part of a millennium.43 S. K. McIntosh suggests that we might see the development of the Empire of Ghana as part of this northward expansion of agropastoral populations, which saw new forms of political organization born out of the development of large herds of cattle and horses coupled with millet agriculture, and increasing contact with camel-owning Berber populations expanding southward into the Sahara. From the archaeological evidence of the wider western Sahel, McIntosh is further able to suggest that societies at least in the early periods of this northward expansion had a “significant component of seasonal mobility.”44 A number of items and raw materials such as stone beads and copper circulated widely across the western Sahel at this time, as, presumably, did many more goods that are not as readily accessible to archaeology. These regional trade networks formed the base “onto which the [trans-Saharan] gold trade was later grafted.”45 Especially fascinating are developments of demography and urbanization in the regions on the Niger River that adjoin the supposed area of the Empire of Ghana. The Inland Niger Delta (IND), the Méma, and the Malian Lakes Region all appear to have been the focus of considerable human activity and social and technological development.46 Archaeological excavations and surveys around the cities of Ja and Jenné have shown that settlements developed in the IND during the last millennium bce, which grew in size and gained an idiosyncratic “clustered” urban form during the first millennium ce. The settlements were built from earthen materials, whose collapse and rebuilding became the numerous pottery-covered mounds that dot the landscape of the IND and its adjoining regions to the north, south, and west. Though many of these sites remain unexcavated, they appear to show a demographic boom and urban development that had its apogee in the late 1st and early 2nd millennia ce.

These developments are in no small way linked to iron metallurgy, which was mastered in West Africa during the last millennium bce. Indeed, provisioning a burgeoning agricultural population with iron must have been a key concern during the 1st millennium ce, and archaeological evidence appears to show that iron production changed from small-scale local to intensified operations in several areas around the IND after about 400 ce.47 This intensification, along with the trade networks that must have supported it, has been cited as a key factor that may have supported the political development of this region, including the Empire of Ghana.

It must be noted, however, that the name “Empire of Ghana” or “Empire of Ghana Period” is sometimes used in archaeological literature as shorthand for the time between the 3rd and the 13th centuries and this period’s key developments. This was a period of increasing regional integration, urbanism, a high-volume interregional trade and production network, but these centered on the IND and the areas to its northwest, rather than in the region in which the historical sources describe Ghāna as having been.48 It is interesting that, from this perspective, the latter area is somewhat of a fringe region. Several authors have wondered why such a supposedly powerful polity would not seek to control the rich trading cities of the Niger rather than occupy the grasslands of the desert edge.49

Yet it must be admitted that archaeological knowledge of the region mentioned in the sources is rather slim for the first millennium ce. In contrast to the areas on the Niger, little archaeological work has been carried out here since the 1950s. The main body of work is thus from the time before the nature, dating, and extent of urbanism on the Niger became known, which has significant consequences for which sites were recognized and how they were understood. For many years, political circumstances have made this region inaccessible to archaeological research. As a result, more is known about the archaeology of the Empire of Ghana’s period in regions nearby than in its core area. To learn more about the nature and circumstances of its rule, more archaeological work is needed in southeastern Mauretania and northwestern Mali.

Kumbi Saleh, the Capital of Ghana?

Our archaeological knowledge of the Awkar-Hodh-Wagadu region between 300 and 1400 ce is essentially restricted to two main sites, Tegdaoust (Awdaghust in the Arabic sources)50 and Kumbi Saleh. The latter site is commonly referred to as the site of the capital of the Empire of Ghana. Yet this identification is far from secure and there are several elements of the site that speak against it.

Kumbi Saleh is a large and impressive stone-built site just north of the Mauritanian border with Mali (fig. 3). It was first documented by Albert Bonnel de Mézières as part of a larger survey of the region that he conducted over several years from Walata with the express goal of finding the capital of Ghāna. In particular, he followed al-Bakrī’s text and the directions of informants from Walata. In the course of his surveys, Bonnel de Mézières reported seeing a number of sites on the axis between Walata and Gumbu in modern-day Mali.51 After first favoring other sites, he finally settled on the site of Kumbi Saleh as the capital. This was partly due to Delafosse’s intervention, as he had recently helped translate the TaF. The Kumbi described as the capital of the Kaya Magha in the TaF is very likely the same as the site of Kumbi Saleh, and Delafosse was strongly in favor of identifying Kumbi Saleh as the capital of Ghāna because of the TaS’s association of the Kaya Magha rulers with Ghāna.52

Bonnel de Mézières was looking for a site to fit al-Bakrī’s description of twin towns that were some distance apart with inhabited space between them and a wooded area around. They needed to be built in stone and have a sufficiently royal character. Kumbi Saleh, according to Bonnel de Mézières’s reports, satisfied all these categories, though it is not true that, as Mauny later claimed, the inhabitants of Walata took him straight to this site after having been asked the whereabouts of the capital of the Empire of Ghana.53 Neither was his identification universally accepted at the time, and it is probably the weight of Delafosse’s support that ultimately swung the argument in Kumbi Saleh’s favor.54

The Empire of Ghana

Figure 3. Map of the site ensemble at Kumbi Saleh, with the location of the main urban tell, two necropoleis, and the so-called Columns Tomb.

Map drawn by N. Gestrich based on Capel et al., “The End of a Hundred-Year-Old Archaeological Riddle,” 2015, fig. 1.

Six subsequent archaeological investigations have taken place at the site, the results of which show a stone-built town that was clearly of some importance and had good connections to the trans-Saharan trade. The excavations conducted by Berthier55 were able to produce a dated chronology for the site, which, while beset by problems of inversion and very large standard error ranges, showed a first, very ephemeral, phase of occupation, followed by several phases of stone and earth architecture that characterize the urban tell. The first phase (Niveau 0, Period A) has one usable radiocarbon date whose main probability falls into the range between 610 and 980 cal. ce, alongside two others whose likely dates lie in the 750–1400 cal. ce range, with peaks between 1000 and 1400 cal. ce.56 Some ephemeral occupation of the site thus appears to have taken place before 1000ce, but most of the earlier phase must fall into the 11th century at the earliest. Berthier’s periods B and C (Niveaux I–IV), in which stone-built architecture appears and the site grows to its considerable size of around 60 ha, have dates ranging between 1010 and 1640 cal. ce, with their peaks between the mid-11th and late 15th century. A few later dates also indicate an occupation between the 14th and 17th centuries.57 The main period of Kumbi Saleh’s urban growth therefore seems to fall between the 11th and 14th centuries, and thus at the end and after the supposed period of Ghana’s imperial power. The dates all come from a small area at the center of the tell, which means they do not preclude an earlier occupation elsewhere on the site (fig. 4). From the available data, however, the occupation and growth of Kumbi Saleh as a large urban center appears to have taken place during the period of the Empire of Ghana’s decline. In any case, the date of Al-Bakrī’s writing in 1068, which appears to be based on observations made in the early 1060s, falls at the outer margins of the possible interpretation of these dates.

The Empire of Ghana

Figure 4. View of the central portion of the urban tell at Kumbi Saleh with excavation units by Berthier (1975–1981); Jean Devisse, Serge Robert, and Denise Robert (1972); Raymond Mauny and Paul Thomassey (1949–1951). The locations of excavations by Lazartigues (1939) and Bonnel de Mézières (1914) are omitted here. Map drawn by N. Gestrich based on Berthier, Recherches archéologiques sur la capitale de lempire de Ghana, 1997, p. 4; satellite image underlay © Bing Maps.

Another dissonance between al-Bakrī’s description and the site lies in the twin nature of the described towns, separated by a distance of six miles. Bonnel de Mézières makes mention of the remains of a large site built in earth, called Ghanata, about twelve miles from Kumbi Saleh, as well as more ruins at the village of Kumbi Dioufi not far from the site.58 Curiously, no more substantial mention is made of the remains of earthen buildings by any of the later visitors to the site, even after the large earthen sites of the Middle Niger were well known. It is hard to tell whether these earthen ruins do not exist or were simply ignored by later archaeologists. Particularly in the early days of archaeological research in West Africa, which might be said to have lasted up until the 1970s, the wish to furnish “proof” for historical sources or oral traditions led to situations where the interpretation of archaeological remains was led more by wishful thinking than by the careful observation of the remains themselves.59

The site of Kumbi Saleh is adjoined, especially to its northeast, by areas of monumental tombs. These appear to take the shape of concentric square enclosures with a central tomb monument. The biggest of these is the so-called columns tomb excavated by Bonnel de Mézières in 1914, in which he found the remains of three individuals buried in a chamber supported by columns. The crania of these individuals have been dated to the 11th or 12th century, around the time at which al-Bakrī was writing.60 It is thus interesting to note that the burials around Kumbi Saleh, including those of doubtlessly high status in the impressive columns tomb, do not correspond at all to the description that al-Bakrī gives of the burial rites of the rulers of Ghana.61 Instead of the stone tombs of Kumbi Saleh, whose occupants may have been Muslims, al-Bakrī described a burial rite involving the construction of a large burial mound and various (including human) sacrifices. Burial tumuli such as are implied by al-Bakrī’s description are common site types in Senegal and Mali,62 but none have been found around Kumbi Saleh in particular or in the Awkar or Bakhounou regions more generally.

Kumbi Saleh may well fit with the capital of the Kaya Magha rulers in the TaF. There is little reason to doubt that the area around Kumbi Saleh and Kumbi Dioufi is the one referred to in the TaF and it is by far the most substantial site in the area. Yet what we know of the site of Kumbi Saleh today leaves virtually no possibility of associating it with al-Bakrī’s capital of Ghana, unless his description is wildly inaccurate. Furthermore, there is no particular reason to think that Ghana had only one capital throughout the period of its existence. It is easily conceivable that several capitals existed63 and that the end of the Kaya Magha’s power, which the TaF describes at Kumbi and Karonga, did not occur where the capital was located during the time of Al-Bakrī’s writing. The dating of Kumbi Saleh appears to preclude the existence of a capital there prior to the 11th century. On the basis of the current dating of the site, the mentions of the “city of Ghāna” referred to in the sources before this time must have been referring to a different place. While Kumbi Saleh is a fascinating site with clear links to Saharan Berber populations and the trans-Saharan trade, its identification as the one capital of the Empire of Ghana is too uncertain as to be tenable and the site has thus far allowed no insights into the early development of this polity.

The End of the Empire: Ghana, Islam, and the Almoravids

It has become customary to state that the Empire of Ghana was conquered by the Berber Almoravid movement in the late 11th century (usually given as 1176) and the population was forcibly converted to Islam. The fact that Ghāna continued to play a prominent role in the sources until the early 14th century is credited to the fact that Ghāna was then ruled by the Almoravids. Yet in an incisive review, Conrad and Fisher have shown that the sources, while indicating conflict between Sarahan and sub-Saharan groups,64 do not support the idea of an Almoravid conquest of Ghāna.65 Instead, this conquest appears to be a historical fabrication based on ambiguous statements by al-Zuhri and Ibn Khaldūn, both writing more than two hundred years after the supposed event.66 Masonen and Fisher further follow the hypothesis of an Almoravid conquest through the later historiography, and show how the idea of the conquest became cemented in European scholarship, despite its lack of clear support in the sources.67

What appears to be clear is that parts of Ghāna’s population, or at least its rulers, probably converted to Islam during the 12th century. The mechanism of this conversion, however, is unlikely to have been conquest by the Almoravids, even if Almoravid influence might be reasonably suspected to be behind it.68 Indeed, the 12th- and 13th-century sources reviewed by Conrad and Fisher indicate that there was a significant degree of mercantile and military cooperation between Ghāna and the Almoravids, perhaps even with an element of supremacy of the former. Furthermore, the conversion may have been the culmination of a slow process, but was probably still an important event in local perception. Al-Bakrī’s description of the situation in Ghāna in the mid-11th century shows a ruler who was not Muslim, but a society and royal court in which Islam was welcome and spreading. That the growing influence of Muslim courtiers described by Al-Bakrī could have led to the abandonment of other religious practices within several decades is not beyond the realms of possibility.

The detailed account of al-‘Umarī, written in 1337–1338, first alerts us to the end of Ghāna as an empire. By the time of his writing, Ghāna had become subject to the Empire of Mali, yet still retained a privileged position among the subordinate rulers.69 How exactly this subordination took place is unclear. Ibn Khaldūn states that Ghāna was first conquered by Soso and then, in turn, by Mali. But his is a single source, even if it is echoed by later writers, so that this second conquest of Ghāna, this time by Soso, must be doubted as much as the Almoravid conquest. Less questionable is that the territory of Ghāna eventually came under the rule of the Empire of Mali. After the 14th century, mentions of Ghāna cease entirely in the written sources, and Mali takes its place as the dominant political formation in the Western Sudan.

The Scattering

In stark contrast to the narrative of Ghana’s slow decline and subordination to Mali that emerges from the written sources, oral traditions paint the end of Wagadu as a dramatic event, the reverberations of which are felt among Soninké communities in West Africa to the present day. The killing of the snake Bida and its associated curse led to what is described in oral tradition as sanxiyɛ, the “scattering” of the population of Wagadu.70 This scattering is said to be the reason for the far-flung Soninké communities throughout West Africa, and is fundamental to their identity as communities in an eternal diaspora.

Levtzion has interpreted the Bida myth as a dramatization of Islamic influence and the end of the observance of rituals linked to previous religious practices,71 but this reading has been rejected as too tenuous.72 Conrad and Fisher propose a political and religious dimension: the Bida cult might have served as an important support for Wagadu’s royal power, in that it united the subordinate rulers in a joint ritual activity that tied them to Wagadu. The desecration of the object of this cult thus weakened Wagadu’s central power.73 In any case, it is the drought and famine and the lack of tradeable gold in the aftermath of the snake’s death that are the reasons usually given for the emigration. The problem of dating this event persists, and if it does refer to the end of Ghāna it is surprising that the Arabic sources do not mention such a catastrophe befalling a regular trading partner. Further, it is not certain how immediate this exodus was. Dieterlen for one imagines the emigration from Wagadu as a slow process of several waves over several periods.74 The attractiveness of the story as an origin myth may exaggerate the importance of any underlying historical event.

Ghāna, a Sceptic’s Account

Ghāna provides an early instance of political organization in West Africa on the southern margins of the Sahara. It is known from Arabic sources, between the 8th and the 14th centuries, as having had powerful rulers. This political development took place against a backdrop of increasing social complexity, urbanism, and trade within West Africa, climatic amelioration, and intensifying trans-Saharan contacts. It appears that Ghāna was able to dominate the southern termini of the trans-Saharan gold trade for several centuries.

Over a complex history of research, the Ghāna of the written sources has become equated with the political formation described in the Soninké legend of Wagadu, yet this identification must remain a cautious one, as it requires a great many leaps of faith. Like most sub-Saharan states mentioned in medieval Arabic sources, Ghāna mostly remains beyond the grasp of archaeological research. While the site of Kumbi Saleh has been proposed as its capital in the literature and widely presented as fact in popular accounts, this identification is also problematic. A firm identification of the capital (or capitals) of Ghāna has not yet been possible.

It is a general feature of the historical period in West Africa known mainly from Arabic sources that their descriptions are difficult to reconcile with archaeological evidence. Through most of the 20th century, the primacy of written sources in such cases of conflict was largely unquestioned. It must, however, be remembered that all of the textual sources that relate to the Ghāna consist of secondhand information at the very best and are often derivative and beset with problems in source criticism. In short, they range between the unreliable and outright fantasy. Ghāna appears to have been an important terminus for trans-Saharan traders, and their stories of it are certainly impressive. But an archaeological view might challenge the importance of a desert-edge polity focused on trans-Saharan trade to West African, rather than North African, history if the main demographic concentrations, the development of urbanism, and the hubs of high-volume internal trade were beyond its domains.

Discussion of the Literature

Much of the discussion on the Empire of Ghana is necessarily a discussion of the literature, since sources are thin. The most influential works include volume 2 of Maurice Delafosse’s Haut-Senegal-Niger, the erstwhile classic of a tradition of research that includes Mauny’s publications on the topic75 and culminates in Nehemia Levtzion’s Ancient Ghana and Mali.76 Of no less importance are two works that critically assess the history of research by Triaud77 and, in far greater length, Masonen.78

The question of the Almoravid conquest of Ghana gave rise to a number of works that are exemplary in their attention to detail of the Arabic sources, in their consideration of the effects of historiographical misunderstandings and inventions, and of their considered treatment of oral sources.79 It is somewhat regrettable that such treatment has not yet extended to the “Question of Ghana” as a whole.

Oral traditions on Ghana are found in many places and discussed by numerous authors. Especially among West African specialists, the literature is constantly growing, though it is of varying quality.80 For the moment, it is hard to surpass Dieterlen and Sylla’s publication,81 both for the striking quality of Sylla’s tradition and the range of Dieterlen’s commentary, though some of its contents are somewhat too esoteric for the tastes of modern research.

Primary Sources

Strictly speaking, there are no primary sources for the Empire of Ghana. All sources that exist are based on oral evidence and the quality of any source is a question of the degree of its removal from the event and the clarity of its transmission. Nevertheless, three types of source materials should be considered: Arabic sources, oral traditions, and archaeology.

The easiest access to Arabic sources is afforded by the corpora of Cuoq and of Levtzion and Hopkins. Both of these, however, provide only French and English translations, respectively. A particular benefit of both is the careful and extensive notes that accompany the texts, as well as the biographical notes on the authors. The latter are more extensive in Levtzion and Hopkins’s edition. Readers who wish to consult the Arabic texts are directed to Yusuf Kamal’s Monumenta Geographica Africae et Aegyptii, which contains facsimiles of most of the manuscripts contained in these corpora.

The 17th-century tarikhs of Timbuktu record, in Arabic, are versions of the oral tradition relating to Ghana at this time. The Tarikh al-Sudan has a French translation and Arabic edition by Octave Houdas,82 as well as a partial English translation by John Hunwick, which includes the section relevant to the Empire of Ghana.83

The Wagadu legend has been recorded numerous times since the 18th century, but transcriptions of the actual texts are rare. Readers are directed to Germaine Dieterlen’s edition of Diarra Sylla’s text, which usefully includes earlier publications on the topics in the annex.84

The equivalent of an edited primary source in archaeology might be a site report. However, the number of these reports required to give a reader a good overview is somewhat too large to be included here. For the outstanding sites of the period and region, it is suggested that readers look toward a series of monographs on the site of Tegdaoust,85 the excavations at Jenné-jeno86 and Dia87 in the Inland Niger Delta, the survey of archaeological sites in Mali edited by Raimbault and Sanogo,88 the work of Togola in the Méma,89 and a number of separate works on the site of Kumbi Saleh.90 An excellent overview of the state of archaeological work concerning the Empire of Ghana is provided in S. K. McIntosh’s Reconsidering Ancient Ghana and Mali.91

Further Reading

Bathily, Abdoulaye. “A Discussion of the Traditions of Wagadu with Some Reference to Ancient Ghana: Including a Review of Oral Accounts, Arabic Sources and Archaeological Evidence.” Bulletin IFAN 37 (1975): 1–94.Find this resource:

Berthier, Sophie. Recherches archéologiques sur la capitale de lempire de Ghana: Étude dun secteur dhabitat à Koumbi Saleh, Mauritanie; campagnes II, III, IV, V; (1975–1976)–(1980–1981). Cambridge Monographs in African Archaeology 41. Oxford: Archaeopress, 1997.Find this resource:

Bonnel de Mézières, Albert. “Recherches sur l’emplacement de Ghana et sur le site de Tekrour.” Mémoires présentés par divers savants à lAcadémie des inscriptions et belles-lettres de lInstitut de France. Première série, Sujets divers dérudition 13, no. 1 (1923): 227–273.Find this resource:

Capel, Chloé, Antoine Zazzo, Jean-François Saliège, and Jean Polet. “The End of a Hundred-Year-Old Archaeological Riddle: First Dating of the Columns Tomb of Kumbi Saleh (Mauritania).” Radiocarbon 57, no. 1 (2015): 65–75.Find this resource:

Conrad, David, and Humphrey Fisher. “The Conquest That Never Was: Ghana and the Almoravids, 1076. I. The External Arabic Sources.” History in Africa 9 (1982): 21–59.Find this resource:

Conrad, David C., and Humphrey J. Fisher. “The Conquest That Never Was: Ghana and the Almoravids, 1076. II. The Local Oral Sources” History in Africa 10 (1983): 53–78.Find this resource:

De Moraes Farias, P. F. “Great States Revisited.” Journal of African History 15, no. 3 (1974): 479–488.Find this resource:

Dieterlen, Germaine, and Diarra Sylla. Lempire de Ghana: Le Wagadou et les traditions de Yéréré. Paris: Éditions Karthala; Association Arsan, 1992.Find this resource:

Levtzion, Nehemia. Ancient Ghana and Mali. London: Methuen, 1973.Find this resource:

Levtzion, Nehemia, and John F. P. Hopkins, eds. Corpus of Early Arabic Sources for West African History. Princeton, NJ: Markus Wiener, 2000.Find this resource:

Masonen, Pekka. The Negroland Revisited: Discovery and invention of the Sudanese Middle Ages. Annales Academiae Scientiarum Fennicae Sarja Humaniora 309. Helsinki, Finland: Suomalainen Tiedeakatemia, 2000.Find this resource:

Mauny, Raymond. “The Question of Ghana.” Africa: Journal of the International African Institute 24, no. 3 (1954): 200–213.Find this resource:

McIntosh, Susan Keech. “Reconceptualizing Early Ghana.” Canadian Journal of African Studies/Revue Canadienne des études Africaines 42, 2/3 (2008): 347–373.Find this resource:

Thomassey, Paul, and Raymond Mauny. “Campagne de fouilles à Koumbi Saleh.” Bulletin IFAN XIII, no. 2 (1951): 438–462.Find this resource:


(1.) In this article, the different entities are kept apart: Ghāna refers to the polity mentioned in the Arabic sources and Wagadu to that mentioned in the Soninké legends, while the Empire of Ghana is the historiographical conglomerate of the former two, with added archaeological dimensions. The state of Ghana, formed out of the former British colony of the Gold Coast in the late 1950s, is not further discussed here. The name, however, is no coincidence: the newly independent state was named after Ghāna. This was fueled by traditions among some Akan peoples that claim descent from the north. These traditions were well known to some leaders of the independence movement, for instance J. B. Danquah, and the choice of the name underscores the wish to associate the new state with what was perceived as being a glorious example of indigenous African political tradition.

(4.) Levtzion and Hopkins, Corpus of Early Arabic Sources for West African History, 5–6, 8–9.

(5.) Joseph M. Cuoq, Recueil des sources arabes concernant l’Afrique occidentale du VIIIe au XVIe siècle: (Bilad al-Sudan), Réimpr. par procédé photomécanique de l’éd. publ. en 1975, corr. et complétée, Sources d’histoire médiévale (Paris: Éd. du Centre National de la Recherche Scientifique, 1985), 42.

(6.) Levtzion and Hopkins, Corpus of Early Arabic Sources for West African History, 32.

(7.) Levtzion and Hopkins, 21.

(8.) Levtzion and Hopkins, 28.

(9.) Levtzion and Hopkins, 52, 79.

(10.) Masonen, Negroland Revisited, 516–519.

(11.) Levtzion and Hopkins, Corpus of Early Arabic Sources for West African History, 79.

(12.) Levtzion and Hopkins, 98.

(13.) Levtzion and Hopkins, 80.

(14.) John O. Hunwick, Timbuktu and the Songhay Empire: Al-Saʿdī’s Taʾrīkh al-sūdān Down to 1613 and Other Contemporary Documents, Islamic History and Civilization 27 (Leiden: Brill, 1999), 7.

(15.) See for instance the “Ibadi Extracts” in Levtzion and Hopkins, Corpus of Early Arabic Sources for West African History, 88–91.

(16.) Levtzion and Hopkins, 45.

(17.) Levtzion and Hopkins, 31–32, 36.

(18.) P. F. de Moraes Farias, “Silent Trade: Myth and Historical Evidence,” History in Africa 1 (1974): 9–24.

(19.) See for instance “al-Bakri,” in Levtzion and Hopkins, Corpus of Early Arabic Sources for West African History, 49, 68, 76, 95.

(20.) Levtzion and Hopkins, 27–28, 80.

(21.) Levtzion and Hopkins, 80.

(22.) Levtzion and Hopkins, 79–80.

(23.) See, for instance, Maurice Delafosse, Haut-Senegal-Niger: II: L’Histoire (Paris: Larose, 1912), 13–19.

(24.) Jajiga is more a name of a people than a geographic space, but centered on what is today known as the region of Kaarta.

(26.) Dieterlen and Diarra Sylla, L’empire de Ghana, 126.

(27.) Dieterlen and Diarra Sylla, 57.

(28.) I thank Mamadou Diawara for his comments and translation of the idea of the gallɛ.

(29.) See, for instance, the accounts of M. Adam, “Légendes Historiques du Pays de Nioro (Sahel),” in Revue Coloniale (Paris: Challamel, 1903–1904); R. Arnaud, “La Singulière Légende des Soninkés,” in L’Islam et la politique musulmane française en A.O.F. (Paris: Publication du Comité de l’Afrique française, 1912), 144–159); or C. Monteil, “La Légende du Wagadu,” in Melanges ethnologiques (Dakar: IFAN, 1953).

(30.) Masonen, Negroland Revisited, 450, table 7.

(31.) The Bambuk region has usually been assumed to have fed Ghāna’s trans-Saharan gold trade, see Nehemia Levtzion, Ancient Ghana and Mali (London: Methuen, 1973), 27–28.

(32.) Abdoulaye Bathily, intervention recorded in Actes du Colloque: Histoire et tradition orale; Projet Boucle du Niger, 2e.

(33.) Bathily, Actes du Colloque, 113–117.

(34.) Masonen, Negroland Revisited, 516–519.

(36.) Levtzion and Hopkins, Corpus of Early Arabic Sources for West African History, 172.

(37.) Masonen, Negroland Revisited, 516–519.

(38.) Hunwick, Timbuktu and the Songhay Empire, 13.

(39.) Paulo Fernando de Moraes Farias, “Intellectual Innovation and Reinvention of the Sahel: The Seventeenth-Century Timbuktu Chronicles,” in The Meanings of Timbuktu, ed. Shamil Jeppie (Cape Town, South Africa: HSRC Press, 2008), 95.

(40.) This matter is thus not as unproblematic as is claimed by Jean Devisse and Boubacar Diallo, “Le seuil du Wagadu,” in Vallées du Niger (Paris: Editions de la Réunion des Musées Nationaux, 1993), 103.

(41.) Sylvie Amblard-Pison, Tichitt-Walata (R.I. Mauritanie): Civilisation et industrie lithique, Recherche sur les civilisations, Mémoire 35 (Paris: Ed. Recherche sur les Civilisations, 1984); Augustin F. C. Holl, Économie et société néolithique du Dhar Tichitt (Mauritanie), Recherche sur les civilisations, Mémoire 69 (Paris: A.D.P.F, 1986); Patrick J. Munson, The Tichitt Tradition: A Late Prehistoric Occupation of the Southwestern Sahara, Sahel MU 035 (Urbana, IL: n.p., 1971); Kevin C. MacDonald et al., “Dhar Néma: From Early Agriculture to Metallurgy in Southeastern Mauritania,” Azania: Archaeological Research in Africa 44, no. 1 (2009): 3–48; K. C. MacDonald, “Before the Empire of Ghana: Pastoralism and the Origins of Cultural Complexity in the Sahel,” in Transformations in Africa: Essays on Africa’s later past, ed. Graham Connah (London: Leicester University Press, 1998); and Peter R. Coutros, “The Malian Lakes Region Redefined: Archaeological Survey of the Gorbi Valley,” Antiquity 91, no. 356 (2017): 474–489.

(42.) Roderick J. McIntosh, The Peoples of the Middle Niger: The Island of Gold (Malden, MA: Blackwell Publishers, 1998), 72.

(44.) McIntosh, “Reconceptualizing Early Ghana,” 360.

(45.) McIntosh, 360.

(46.) S. K. McIntosh, ed., Excavations at Jenné-Jeno, Hambarketolo, and Kaniana (Inland Niger Delta, Mali), the 1981 Season (Berkeley: University of California Press, 1995); R. M. A. Bedaux et al., eds., Recherches archéologiques a Dia dans le Delta intérieur du Niger (Mali): Bilan des saisons de fouilles 1998–2003 (Leiden: CNWS, 2005); T. Togola, ed., Archaeological Investigations of Iron Age Sites in the Mema Region, Mali (West Africa), Cambridge Monographs in African Archaeology 73, BAR International Series 1736 (Oxford: Archeopress, 2008); and M. Raimbault and K. Sanogo, eds., Recherches Archéologiques au Mali: Prospections et inventaire, fouilles et études analytiques en Zone lacustre (Paris: Karthala, 1991).

(47.) McIntosh, Excavations at Jenné-Jeno, Hambarketolo, and Kaniana; Randi Håland, “Man’s Role in the Changing Habitat of Méma During the Old Kingdom Of Ghana,” Norwegian Archaeological Review 13, no. 1 (1980); Nikolas Gestrich and Kevin C. MacDonald, “On the Margins of Ghana and Kawkaw: Four Seasons of Excavation at Tongo Maaré Diabal (AD 500–1150), Mali,” Journal of African Archaeology 20, no. 290 (2018); and Caroline Robion-Brunner, Forgerons et sidérurgie en pays dogon: Vers une histoire de la production du fer sur le plateau de Bandiagara (Mali) durant les empires précoloniaux, Journal of African Archaeology Monograph Series, vol. 3 (Frankfurt: Africa Magna, 2010).

(48.) For example, Gestrich and MacDonald, “On the Margins of Ghana and Kawkaw.”

(49.) For one explanation, see Devisse and Diallo, “Le seuil du Wagadu,” 113–114.

(50.) Abdallah O. Babacar, Jean Polet, and Denise Robert-Chaleix, Tegdaoust, Éditions recherche sur les civilisations, Mémoire 25 (Paris: Éd. Recherche sur les Civilisations, 1983); Jean Polet, Tegdaoust IV: Fouille d’un Quartier de Tegdaoust (Mauritanie Occidentale) : Urbanisation, architecture, utilisation de l’espace construit, Éditions recherche sur les civilisations, Mémoire 54 (Paris: Éd. Recherche sur les Civilisations, 1985); Denise Robert-Chaleix, Une concession médiévale à Tegdaoust: Implantation, évolution d’une unité d’habitation, Recherche sur les civilisations, Mémoire 82 (Paris: Ed. Recherche sur les Civilisations, 1989); and Claudette Vanacker, Tegdaoust II: Fouille d’un Quartier Artisanal, Tegdaoust: Recherches sur Awdaghost 2 (Paris: Institut Mauritanien de la Recherche Scientifique, 1979).

(51.) Albert Bonnel de Mézières, “Note sur ses récentes découvertes, d’après un télégramme adressé par lui, le 23 mars 1914, à M. le gouverneur Clozel,” Comptes-rendus des séances de l année—Académie des inscriptions et belles-lettres 58, no. 3 (1914); Albert Bonnel de Mézières, “Recherches sur l’emplacement de Ghana et sur le site de Tekrour,” Mémoires présentés par divers savants à l’Académie des inscriptions et belles-lettres de l’Institut de France. Première série, Sujets divers d’érudition 13, no. 1 (1923); and Henri Cordier, “Nouvelles de la mission Bonnel de Mézières en Afrique,” Comptes-rendus des séances de l’année—Académie des inscriptions et belles-lettres 58, no. 5 (1914).

(52.) Bonnel de Mézières, “Note sur ses récentes découvertes,” 255.

(54.) Masonen, Negroland Revisited, 520–521.

(56.) Berthier, Recherches archéologiques sur la capitale de l’empire de Ghana, 30; and Doosselaere, Le Roi et le Potier, 31–32.

(57.) Berthier, Recherches archéologiques, 31–35.

(58.) Mézières, “Note sur ses récentes découvertes.”

(59.) cf. François-Xavier Fauvelle, African Archaeology and the “Chalk Line Effect”: A Consideration of Māli City and Siğilmasa,” in Landscapes, Sources and Intellectual Projects of the West African Past, eds. Toby Green and Benedetta Rossi (Leiden: Brill, 2018).

(61.) Capel et al., “The End of a Hundred-Year-Old Archaeological Riddle.”

(62.) Susan K. McIntosh and Roderick J. McIntosh, “Field Survey in the Tumulus Zone of Senegal,” African Archaeological Review 11, no. 1 (1993); Sonja Magnavita, “First Geophysical Exploration in the Tumuli Zone of Central Senegal: A Multidimensional Approach,” Azania: Archaeological Research in Africa 52, no. 1 (2017).

(63.) See, for example, Bathily, “A Discussion of the Traditions of Wagadu,” 19.

(64.) P. F. de Moraes Farias, “Great States Revisited,” Journal of African History 15, no. 3 (1974).

(66.) Sheryl L. Burkhalter, “Listening for Silences in Almoravid History: Another Reading of ‘The Conquest That Never Was’” History in Africa 19 (1992) criticizes Conrad and Fisher’s reading of al-Zuhri on several points, yet the overall argument appears to hold: There is nothing approaching clear evidence for an Almoravid conquest of Ghana.

(67.) Pekka Masonen and Humphrey J. Fisher, “Not Quite Venus from the Waves: The Almoravid Conquest of Ghana in the Modern Historiography of Western Africa,” History in Africa 23 (1996).

(68.) Burkhalter, “Listening for Silences in Almoravid History,” 115.

(69.) Levtzion and Hopkins, Corpus of Early Arabic Sources for West African History, 262.

(70.) Mamadou Diawara, personal communication, August 2018

(71.) Levtzion, Ancient Ghana and Mali.

(72.) Farias, “Great States Revisited,” 484.

(73.) Conrad and Fisher, “The Conquest That Never Was.”

(74.) Germaine Dieterlen, “Premier aperçu sur les cultes des Soninké émigrés au Mande,” Systèmes de pensée en Afrique noire 1 (1975): 5.

(75.) Raymond Mauny, “The Question of Ghana”; and Mauny, “État actuel de la question de Ghana,” Bulletin IFAN XIII, no. 2 (1951).

(76.) Levtzion, Ancient Ghana and Mali.

(77.) Jean-Louis Triaud, “Le nom du Ghana: Mémoire en exil, mémoire importée, mémoire appropriée,” in Histoire d’Afrique. Les enjeux de mémoire, ed. Jean-Pierre Chrétien and Jean-Louis Triaud (Paris: Ed. Karthala, 1999).

(78.) Masonen, Negroland Revisited.

(79.) Masonen and Fisher, “Not Quite Venus from the Waves”; Conrad and Fisher, “Conquest That Never Was”; Humphrey J. Fisher, “Early Arabic Sources and the Almoravid Conquest of Ghana,” Journal of African History 23, no. 4 (1982); and Burkhalter, “Listening for Silences in Almoravid History.”

(80.) For a recent example, see Youba Bathily, Les racines de l’empire de Ghana (Saarbrücken: Éditions universitaires européennes, 2017).

(81.) Dieterlen and Sylla, L’empire de Ghana.

(82.) Octave Houdas, ed., Tarikh es-Soudan: par Abderrahman b. Abdallah b. ‘Imran b. ‘Amir es-Sa’di (Paris: Adrien-Maisonneuve, 1964).

(83.) Hunwick, Timbuktu and the Songhay Empire.

(84.) Dieterlen and Sylla, L’empire de Ghana.

(85.) Babacar, Polet, and Robert-Chaleix, Tegdaoust, 25; Polet, Tegdaoust IV, 54; Robert-Chaleix, Une concession médiévale à Tegdaoust, 82; and Vanacker, Tegdaoust II, 2.

(86.) McIntosh, Excavations at Jenné-Jeno, Hambarketolo, and Kaniana.

(87.) Bedaux et al., Recherches archéologiques à Dia dans le Delta intérieur du Niger (Mali).

(88.) Raimbault and Sanogo, Recherches archéologiques au Mali.

(89.) Tereba Togola, Archaeological Investigations of Iron Age Sites in the Mema Region.

(90.) Berthier, Recherches archéologiques sur la capitale de l’empire de Ghana; Barbara van Doosselaere, Le Roi et le Potier; Thomassey and Mauny, “Campagne de fouilles à Koumbi Saleh,”; Bonnel de Mézières, “Note sur ses récentes découvertes,”; Bonnel de Mézières, “Recherches sur l’emplacement de Ghana et sur le site de Tekrour”; Capel et al., “The End of a Hundred-Year-Old Archaeological Riddle.”

(91.) McIntosh, “Reconceptualizing Early Ghana.”