The Almoravid conquest of ancient Ghana in 1076 AD is certainly among the most dramatic and controversial single events in the historiography of West Africa. It has been regarded as a crucial turning point, as the battle of Hastings was for England, not only for the existence of Ghana, but also for the destiny of the entire area, opening the gates to a triumphant Islam in sub-Saharan Africa.

Yet the conquest and destruction of Ghana by Almoravid invaders constitute one of the myths which still populate African historiography, like the wonderful voyage of Hanno to the Bight of Biafra, which was carried over from classical Greco-Roman texts into modern European literature as early as 1533. Since then the story of Hanno has been used for various purposes by western Africanists, for instance, to explain the diffusion of iron technology into sub-Saharan Africa. Just the same, no definite evidence has yet been found for any Carthaginian sailings along the West African coast, except the Periplus of Hanno itself, which seems to be a literary composition drawn from earlier classical sources. A reason for the popularity of Hanno, and other such stories in African historiography, has been that many modern writers have been content with using the previous secondary literature, instead of examining carefully all the available primary sources. Consequently, many subjective and hypothetical assumptions created by previous scholars, working on the basis of even less evidence, have been transferred bodily from one corpus of research to the next. Finally, their origin forgotten, stories like the voyage of Hanno have become established historical facts through constant repetition in the authorized literature.

We first conceived of this study several years ago, when our respective lines of research seemed to some extent to converge on the period of the decline of ancient Ghana. It has proved a more complex matter than we imagined, or than the deceptively simple title above might suggest--partly because it is often more difficult to prove that something did not happen than that it did. We have had to divide our analysis into two parts. First we examine the external written sources for the Almoravid conquest of Ghana. Our work in this respect has been immensely facilitated by the appearance in 1975 of Cuoq's Recueil des sources arabes, which in some sense provides the infrastructure for our argument. We have also had the invaluable privilege of seeing in proof Hopkins' and Levtzion's Corpus of early Arabic sources. In the second half of our study, it will be the turn of internal sources, mainly oral traditions. We use the terms external and internal advisedly, since the simple distinction between written sources and oral traditions is too artificial: an oral tradition was an oral tradition, whether written down in Arabic in the Middle Ages or in French in this century, and a great deal of the information in both parts of our article must have started as oral tradition.

Putting it very bluntly, we have discovered no sources, whether external or internal, which unambiguously point to such a conquest. A handful of sources suggest some link between the rise of the Almoravids and the decline of Ghana, but with a puzzling vagueness--a vagueness which decreases as the number of centuries between the alleged event, and the report of it increases.

In Part I of this paper we examined the external written sources and found no unambiguous evidence that an Almoravid conquest of ancient Ghana ever occurred. The local oral evidence reviewed in this part of our study supports our earlier hypothesis, in that we find nothing in the traditions to indicate any conquest of the eleventh-century sahelian state known to Arab geographers as “Ghana.” Instead, the oral traditions emphasize drought as having had much to do with the eventual disintegration of the Soninke state known locally as “Wagadu.”

An immediate problem involved in sifting the oral sources for evidence of an Almoravid conquest is that a positive identification between the Wagadu of oral tradition and the Ghana of written sources has never been established. Early observers like Tautain (1887) entertained no doubts in this regard, and recently Meillassoux seems to have accepted a connection, if not an identification, between Ghana and Wagadu when he notes that “les Wago, dont le nom a donné Wagadu, sont les plus clairement associés à l'histoire du Ghana.” However, much continues to be written on the subject, and the question remains a thorny one. On the lips of griots (traditional bards) and other local informants, Wagadu is a timeless concept, so a reliable temporal connection between people and events in the oral sources on one hand and Ghana at the time of the Almoravids on the other, is particularly elusive. Indeed, any link between the traditions discussed here and a specific date like 1076 must be regarded as very tenuous, as must any association of legendary events with Islamic dates. In western Sudanic tradition influenced by Islam, the hijra (A.D. 622) is both prestigious and convenient, a date with which virtually any event in the remote past can be associated, though such a claim may have nothing to do with any useful time scale.

The changes in lipoprotein metabolism which follow the ingestion of a large fat load have been well described. The hypothesis was tested that similar changes in lipoprotein metabolism would occur after a relatively normal meal. Plasma and lipoprotein triacylglycerol, cholesterol and apolipoprotein concentrations were determined in twenty subjects (ten female) given a mixed meal containing approximately one-third of the daily intake of major nutrients in the typical Western diet. Fasting plasma triacylglycerol concentrations (range 0.38–2.70 mm/l) and the postprandial rise in plasma triacylglycerol varied considerably between subjects and were significantly associated (P < 0.01). The rise in plasma triacylglycerol corresponded to marked increases in the triacylglycerol concentration of the triacylglycerol-rich lipoproteins (TRL; chylomicrons and very-low-density lipoproteins). TRL cholesterol also increased after the meal. An increase in high-density-lipoprotein (HDL)-triacylglycerol following the meal was accompanied by a decrease in HDL-cholesterol concentration, presumably due to the action of the cholesteryl-ester transfer protein. The increases in HDL-triacylglycerol and in TRL- cholesterol were correlated with the postprandial rise in triacylglycerol in the TRL (P < 0.01). We conclude that potentially adverse changes occur in both triacylglycerol-rich and high-density lipoproteins following a typical mixed meal, as they do after large fat loads. The changes are exaggerated in those subjects with greater fasting plasma triacylglycerol concentrations.

The primary purpose of this paper—apart from a personal concern to offer a token of affection and respect to a friend and colleague of many years’ standing—is to consider two analytical models (one drawn from early Christian history, the other from the history of science), and to suggest ways in which these may help us to interpret the data of religious, and in particular Muslim, change in sub-Saharan Africa.

The authors argue that, during the crucial decade of Songhay history which followed the death of Sunni Ali, Askiya Muḥammad pursued, sometimes quickly and sometimes hesitantly, three distinct ‘Islamic’ options, in contrast to the ‘received tradition’ which sharply differentiates between the reign of the last sunni's and the first of the askiyas. Askiya Muḥammad began his reign in alliance with the court clerics of the imperial capital in Gao, who were accustomed to ‘mixing’ Islamic and traditional practices. After his pilgrimage he sought out the advice of the radical Muslim scholar from the Sahara, al-Maghīlī. The strong positions of al-Maghīlī against the Jews and also the Musūfah, a Berber group strongly associated with Timbuktu, led the askiya to his third choice, the urbane and tolerant Islamic practice of the famous center of Muslim scholarship. The authors advance this as a new interpretation of predominantly old available evidence, and they suggest, on the one hand, the complexity and multiplicity of Islamic practices in the Niger Buckle region around 1500 A.D. and, on the other, the necessity of choice among the three options.