This essay interprets a classification of Africa's Bantu languages which used statistical tools guided by assumptions about farming and its chronology to analyze fresh vocabulary evidence. It shows a peeling movement from Cameroon's grassfields, into southern Cameroon, then along a savanna corridor through West Central Africa's rainforests, into the Savannahs, then to Southern Africa, the Great Lakes, and Indian Ocean coast. The clear sequence of movement masks methodological and historical factors. Language death, multilingualism, and the limits of vocabulary evidence restrain the classification's authority. ‘Transformations’ from food collecting to food producing or from no metals to full engagement with metals were mutable, unfolded at different speeds, and involved interactions with firstcomers. In Central Africa, Bantu speakers were often the first farmers and metal-users in the region but elsewhere they were commonly neither. Their arrivals did not immediately displace firstcomers. Computational methods can accommodate many of these issues.

Interest in the question of Bantu expansion rose dramatically in the 1950s as historians, archeologists, and anthropologists all joined in the fray. This reflected both the rise of Africa in world affairs and the expansion of research in general. The scholars involved were typically a new breed of professionals, and as such more dependent than their predecessors on universities or research institutions. The School of Oriental and African Studies in London achieved overwhelming dominance from about 1950 until the late 1960s, so that opinions held by its staff found the widest audience. The new scholars also were, for the most part, anti-racist, sympathetic to African nationalisms, and of liberal or socialist persuasion. They tended to reject the notion of “conquest,” believing in gradual change rather than abrupt cataclysmic mutation, perhaps because they were repelled by their recent experiences during the war. As had happened earlier, these extraneous circumstances left a deep imprint on the speculations that were now proposed. Early in this period a new paradigm almost achieved consensus, but after 1968 this fell apart and during the last decade two new trends have appeared: the single-minded quest for a new paradigm and the search for better understanding through the study of analogous processes, coupled with a more radical skepticism.

In 1754 the newly-arrived governor of Angola, António Alvares da Cunha, announced to the court in Lisbon that, inspired by plans his uncle had made in 1725, he had decided to revive the old attempt to find a way overland to link Angola and Mozambique. He was initiating this enterprise by sending, at his own expense, an expedition to the fair of Cassange, the farthest known point eastwards of Luanda, to gather information about the lands between Angola and Mozambique. The governor's envoy was Manoel Correia Leitão— an experienced trader, but one who had never been to Cassange— accompanied by a pilot, António Roiz Grizante, whose task was to fix the latitude of Cassange.

A suspicious reader could conclude with the subsequent governor António de Vasconcelos, that this venture was a slave trading expedition by Leitão to recover some debts. But it was financed by the governor and one suspects that the whole enterprise was in fact a joint business venture, disguised as an exploration, because governors were forbidden to trade. Be that as it may, the reaction from Lisbon to this news was unexpected. The governor was ordered to abandon his plans and to recall his envoys because this enterprise was too “delicate.” Knowledge about it was to be shrouded in absolute secrecy and a report about this was to be treated as top-secret. Whereupon the governor recalled Leitão.

In the initial paragraph of a set of book reviews dealing with a sixteenth-century relation and a map by Ptolemy, Michel Doortmont informs us casually that:

Quite a statement! How arrogant and how gratuitously insulting to all those who do study “ancient and early modern history.” If Doortmont were alone in his views, one might just as well ignore this as an example of regretable idiosyncrasy. But the cavalier way in which he delivers his opinion suggests that he merely voices a truism, i.e., an opinion that he thinks is shared by most of his colleagues. Moreover, such a statement will surely frighten budding scholars away, which in itself is a good reason not to let such a sweeping condemnation of the study of earlier African history pass without comment.

Four main claims are made here: (a) a dichotomy and a contrast exists between those who who study the more remote past and those who study the recent past; (b) the first group is small and out of the mainstream; (c) its research yields disappointing results; (d) its scholarship lacks seriousness and method—with (b) and (c) of course in implicit contrast to the second group.

A few years ago I was reading a doctoral dissertation concerned with the history of São Tomé, when one particular word fairly leapt off the page. It came at the end of an episode telling how a particular ethnic group on the island today had arrived there. The story told how a number of Angolan slaves, later called Angolares, were shipwrecked just off the coast of São Tomé in 1544 (or 1554), “though (of course) no accurate date or account of the circumstances exists.” The survivors swam to shore and moved into the bush “building there a series of small villages called Quilombos.” It was that word, “Quilombos,” that made me jump to attention, for it is a very important word in Angolan history and a written record of it from either 1544 or 1554 would be sixty to seventy years earlier than the hitherto first-known mention of the word. If it were true, profound revisions to the accepted political history in Angola during the sixteenth century would be called for, and even the chronology of the onset of the Lunda (Ruund) kingdom far inland would have to be revised.

From time to time one encounters a situation where a wisp of a source yields evidence of momentous consequence for the understanding of a whole society in the past. Such a situation raises the question of how much evidentiary weight a slender source can credibly bear, for when the evidence is so slender or tenuous, the slightest misunderstanding or misinterpretation can lead to hugely magnified errors in historical reconstruction. At the same time, the more slender the evidence, the more tempted researchers will be to interpret it according to their own a priori vision of the past in question. This is a genuine problem which can only be overcome by finding more relevant evidence.

Such situations naturally crop up wherever sources (written or otherwise) are very rare, so that every scrap of evidence must be scrutinized and exploited with the greatest care and to its fullest extent, a common circumstance in African history in relation to written documents before ca. A.D. 1450 and documents at all times about topics such as, for instance, women's history or specific techniques of agriculture that were normally not recorded. The situation then, while unusual, is frequent enough to warrant a more detailed discussion of one such instance to illustrate what the problem is and how it can be tackled. This illustration will be the case of oratoi, a single word in the Periplus Maris Erythraei.

The Periplus is justly famous in African historiography. Known from a single manuscript now in Heidelberg and held to date from the early tenth century and a much later copy from that now in the British Museum, written in a mixture of classical and vulgar Greek by either a Greek living in Egypt or an Egyptian, and firmly dated to between AD 40-55, the Periplus contains the first, albeit succinct, information about the inhabitants of the East African coast, more than half a millennium before other comparable written mentions occur.

The later precolonial history of a vast area in west central Africa between the Kwango and the Lubilash rivers starts with—and is dated by—the tradition of exodus of Kinguri and his companions from the heartland of the Lunda commonwealth. For the last two decades, however, several scholars have claimed that this tradition is merely a later addition to the older body of the traditions told by a dozen or so different peoples in west central Africa. Yet so far no one has examined where and when and how the Kinguri exodus tradition could have grafted itself onto the traditions of so many peoples over such a vast area. If true, this claim also requires a radical revision of the accepted history of western Lunda expansion. To examine the claim and its consequences is the aim of this article, which begins with the earliest written report of the Kinguri's exodus story.

Historical tales are the most abundant and the sources most used to reconstruct the history of the kingdoms of the area between the Great Lakes. This is especially true for the history of the Nyiginya kingdom in Rwanda, where such tales, preserved at the court as well as by local people on the hills, are even more abundant than anywhere else. It is not surprising then that they form the bedrock on which authors have built their reconstructions on the history of that kingdom. Yet little attention has been paid to a general critical examination of these tales. Here, and elsewhere in the region, their contents have generally been accepted as credible, after the arbitrary erasure of all references to passages judged to refer to miracles, after the arbitrary dismissal of the bits and the variants that do not conform to one's preferred version, and after either the exclusion of local “provincial” narratives, or as happened after 1960, the exclusion of all those that stemmed from the court. Such practices will simply not do.

Because these tales form the bedrock of the history of the kingdom the Institute for Scientific Research in Central Africa (IRSAC) under my direction instituted a large collection of such sources between 1958 and 1962, and made them available in the original and in translation first in depots at Butare and Tervuren and later in 1973 on microfilm. Surprisingly enough, this collection as well as even major editions of other sources, have been completely ignored by most scholars working after 1960.

Despite their uninviting aspect, dictionaries make for engrossing historical reading. Seemingly dull compilations conceal an ample description of a society and its culture at a given moment in time. Each entry in the original language constitutes a bit of information. For African languages the amount of information varies between one thousand or fifteen hundred entries in the shorter vocabularies to as many as eighty thousand for the fullest dictionaries. Anything less than a thousand words may be considered to be only a wordlist, the longest of which rarely exceeds five hundred items. Most dictionaries seem to contain between six thousand and thirty thousand entries. The extent of information is astonishing because the form is usually so condensed that it consists only of a gloss as translation of the term given. The best dictionaries contain explanations of the gloss, references to relevant publications, and examples taken from everyday speech as well as from literary usage, and they begin with a short grammatical introduction. Unfortunately, this kind of dictionary has been and remains all too rare; usually the gloss is unnecessarily laconic. Even so, the most exciting as well as the most trivial information is to be found in these tomes. For example, we learn from the 1652 Kongo dictionary (7200 entries) collected or recopied by van Geel, both that the Kongo were matrilineal–a fact stated nowhere else for the old kingdom–and that playing cards was a popular pastime.

Unesco's General History of Africa (henceforth GHA) is now nearing completion—seven out of the eight volumes having been published in English, at least—and the time is ripe to draw attention to its role in the historiography of African history, a role very different from that played by the Cambridge History of Africa with which it is often compared. In certain ways the Unesco project has been a unique venture, and not just in African history but, in general, because it broke with old established practice. The work was not guided by one or two editors but by a large committee. History-writing by committee seemed not only distasteful but impossible to achieve to many, both because of the practical difficulties involved and because it seemed incredible that so many editors could agree on a common text, without falling into sheer banality. Now that the volumes are out, readers can judge for themselves. Many among them actually wonder how exactly these volumes and chapters were created. Because I have been an active member of the committee since its creation in 1971 and of its bureau since 1983, I can provide a general answer to this question. But the time has also come to draw attention to the records generated by this project. Future researchers will find a huge mass of papers involving hundreds of historians of Africa that touch on practically all aspects of Africa's historiography between ca. 1965 and today.

The aim of this note is to attract attention to a new venture in scholarly publication—Beatrix Heintze's Alfred Schachtzabel's Reise nach Angola, 1913-1914. Und seine Sammlungen für das Museum für Völkerkunde in Berlin. Rekonstruktion einer Ethnographischen Quelle (Köln, 1995), the first issue of Afrika Archiv, a new series to be devoted to text editions. That this is not to be confused with an ordinary text edition is already intimated by the work's title, which translates as Alfred Schachtzabel's Voyage to Angola, 1913-1914, and his Collections for the Museum für Völkerkunde in Berlin. Reconstruction of an Ethnographic Source. In other words, it speaks of a “reconstruction” and “a source edition,” not “text edition.” As well, Heintze is described as the author of this book rather than the editor. Yet she presents the work as a new kind of edition—the “integrated edition of sources.”

Heintze is an experienced text editor, having published both a voluminous conventional collection of documents and a set of ethnographic drawings. Yet when she examined the surviving documentation pertaining to Schachtzabel's expedition in Angola, she became convinced that a standard edition of all his work would be unwise. The unpublished materials at hand are too disparate and slender.

The historian of pre-nineteenth century Africa…cannot get far without the aid of archaeology.

Nevertheless, historians have good reason to be cautious about historical generalisations by archaeologists and about their own use of archaeological material…: it would be a rash historian who totally accepted the conclusions of Garlake and Huffman with the same simple-minded trust as I myself accepted the conclusions of Summers and Robinson.

In the beginning, historians of Africa put great store by archeology. Was its great time depth not one of the distinctive features of the history of Africa, a condition that cannot be put aside without seriously distorting the flavor of all its history? Did not the relative scarcity and the foreign authorship of most precolonial written records render archeological sources all the more precious? Did not history and archeology both deal with the reconstruction of human societies in the past? Was the difference between them not merely the result of a division of labor based on sources, so that historical reconstruction follows in time and flows from archeological reconstruction? Such considerations explain why the Journal of African History has regularly published regional archeological surveys in order to keep historians up to date.

More than one-third of Africa is occupied by people who speak related languages belonging to a single family called Bantu. This has been recognized for more than a century. As early as 1886 Harry Johnston argued that this situation was the result of differentiation from a real single ancestral language, later called UrBantu or Proto-Bantu. The inevitable question arises: How could one language or a group of closely related dialects diffuse over such a vast area? The fact of Bantu expansion remains a major puzzle in the history of Africa. Many have risen to the bait of solving it.

My main goal here is to recount the salient features of this century-long inquiry and in doing so to lead to an assessment of the present situation. Given the nature and the paucity of the available data, much of proposed reconstruction has been conjectural, so that the study of Bantu expansion also has been an exercise in conjectural history and in speculation. The available data are disparate and drawn from different disciplines, and the results tell us something about what can and what cannot be done in interdisciplinary research. In the telling I hope to demonstrate how much different considerations of the question have been moulded by the major themes in European and American intellectual history of the last century and how much scholarly tradition, once established, has directed and limited the solutions proposed.

Despite its very visible career in anthropology and folklore, structuralism has been little used by historians of precolonial Africa. Only Ronald Atkinson has applied the method of Lévi-Strauss in the Edmund Leach variant, although a number of historians have attempted to elucidate symbolic meanings by other means. Rather surprisingly as well, given the two decades or so that have elapsed since Lévi-Strauss developed its axioms and analysis, no historian of Africa has ventured to discuss the validity of structuralism for coping with the interpretation of myths of origin or other oral traditions, except in passing. The topic has surfaced only here and there in the never-ending debate about traditions as expressions of the present, or of the past, or of both. Given the influence of structuralism elsewhere, though, it is due time that the approach be discussed for its own sake.

The reticence to do so became especially incongruous when a senior structuralist, Luc de Heusch, began to cover ground that historians had recently trod. In his Le roi ivre he discussed at length myths in the kingdoms of southeastern Zaire and adjacent areas. This did prompt publication of two articles about his Luba and Lunda interpretations, but no general assessment of this work in toto. Jeffrey Hoover faulted de Heusch's Lunda material but still praised his “provocative ideas” and the method, “which bore some good fruit,” while Thomas Reefe prefaced his critique of Luba material by calling the book “stimulating” and sidestepped the issue by noting that “no matter what the final assessment of this book will be by historians…” Others were equally bland in their references to this work, while still refuting de Heusch on specifics. Everyone felt, it seems, that a general assessment was beyond or outside their competence. Yet a general critique would have been of use for de Heusch is one of the oldest and most experienced structuralists in anthropology, perhaps the first disciple of Lévi-Strauss. Trained in Paris, he imbibed the approaches of the Griaule school, the protostructuralism of Georges Dumézil, and the early teaching of the master himself. Of all structuralists he remains the most faithful to the method of Lévi-Strauss.

Any mention in a text to something demonstrably more recent than its attributed date of composition is an anachronism. The presence of an anachronism is absolute proof that the purported date given for the redaction of the final the text is wrong and that the real date occurred later than the date implied by the anachronism. In rare cases an anachronism can also occur in reverse, namely, when it can be shown that the anachronism implies the presence of something (an archaism, a practice) which had been replaced or had disappeared by the purported date for the manuscript. Anachronisms can be very useful both to date a document and in some cases to show us something about the dynamics involved in the creation of a text. The detection of anachronisms is common with regard to written documents, but they can also be used for the study of oral documents which have been memorized word by word, especially when only one version of such documents has survived. That is the case for the Rwandan esoteric code of kingship, which yields an excellent example of how a systematic search for anachronisms throws light on such a document and allows a historian to use its contents with much greater confidence than was the case otherwise.

Ubwiiru is the name given in Rwanda to a set of eighteen pieces in prose, called “roads” or “ways,” which vary in length between 74 and 1252 lines and were learned by heart since “immemorial times” by specialists called abiiru.

One wonders what Fernand Braudel and the school of the Annales have done to become a kind of Trojan Horse for the wholesale condemnation of the historical value of oral tradition. Yet they are the banner raised by W.G. Clarence-Smith in a recent article in his journal to preach jihad against its historical value. Clarence-Smith claims that the historiographical revolution effected by Annales has resulted in the definitive exclusion of oral traditions from the halls of Clio. Oral traditions are at best ambiguous “signs” about the past and are very much of the present. They lack absolute chronology and they are selective, so away with them. If they be worthy of attention at all, let anthropologists and sociologists be concerned, save in a few rare instances where a historian wants to check on some European printed source. And even then, caveat emptor. Significantly, the article is not just the expression of the views of one person; rather it is symptomatic of much of the criticism which has been leveled at oral tradition, mostly by fasionable anthropologists. And it brings this criticism to its logical conclusion.

But first a word about Braudel, the Annales, and oral tradition in general. The Annales School was founded by Lucien Febvre and Marc Bloch before World War II. Fernand Braudel is its most distinguished exponent. His major theoretical pronouncements can be found in his Ecrits sur l'histoire, a collection of articles reprinted and published in 1969. This and his two major historical works should be read by those who want to know more about his views and ways of dealing with history. The basic tenets that members of the Annales School hold is that the history of events is but the spray of past developments; other time depths tell us more about the waves of the past. There is the time of the conjoncture, the trend, and the even longer time periods -- sometimes many centuries long -- the longue durée or long term. Successful history writing does not liminate the study of events, but analyzes them against the movement of these longer and deeper-running trends.

Around 1850 the peoples of central Africa from Duala to the Kunene River and from the Atlantic to the Great Lakes shared a common view of the universe and a common political ideology. This included assumptions about roles, statuses, symbols, values, and indeed the very notion of legitimate authority. Among the plethora of symbols connected with these views were the leopard or the lion, the sun, the anvil, and the drum, symbolizing respectively the leader as predator, protector, forger of society, and the voice of all. Obviously, in each case the common political ideology was expressed in slightly different views, reflecting the impact of differential historical processes on different peoples. But the common core persisted. The gigantic extent of this phenomenon, encompassing an area equal to two-thirds of the continental United States, baffles the mind. How did it come about? Such a common tradition certainly did not arise independently in each of the hundreds of political communities that existed then. However absorbent and stable this mental political constellation was, it must have taken shape over a profound time depth. How and as a result of what did this happen? Is it even possible to answer such queries in a part of the world that did not generate written records until a few centuries ago or less?

This paper addresses this question: how can one trace the social construction of such a common constellation over great time depths and over great regional scale? All the peoples involved are agriculturalists and the political repertory with which we are concerned could not easily exist in its known form outside sedentary societies.

Not so long ago this journal published an excellent overview of the issues related to the reliability of translations. But neither this, nor apparently any other study, considered the case of a text translated by its own author, hence a note on this topic may have some use. As I just finished translating of my own Le Rwanda ancien (Paris, 2001), I have been prompted to do so. The issues still are: how reliable is this sort of translation to the original? Is it really equivalent to the original?

Translations of modern works can be grouped in three categories according to the knowledge an author has about the language in which his or her work is being translated: the author does not know the language of the translation at all; the author knows that language well enough to check on the reliability of a translation; or the author knows that language to the point that he or she can translate the work him-or herself. In the first case, even the author has no idea how faithful the translation is to the original. In the second case the overall reliability of the translation can be guaranteed by the author, but his or her limited knowledge of the language in question means that he or she may be unaware that some parts of the translation may not be faithful in the sense that readers might understand them in ways the author did not intend. Moreover, the great majority of translations by others will still strike native readers of the language into which it is translated as stilted because its rhetorical devices, and perhaps its syntax, do not wholly conform to the usual practice in that language. Indeed some passages may seem quaint or even especially striking merely because standard idioms or metaphors have been too literally translated. But surely self-translation should be completely reliable, especially when the translated text has been also copy-edited by a native speaker of the language in which it is translated?

Alfons Vermeulen published a novel about his life in a Congo town as a trader in 1933 and also left drafts about the same subject for his memoirs when he died in 1965. This essay systematically compares both documents using the rules of evidence and concludes that although it does not appear to be so on the surface the novel distorts the evidence he adduces considerably more than the memoirs. At the same time though both documents are equally deeply affected by the spirit of the time of their composition. Hence the combined evidence of both documents together is not only richer but also more reliable than their separate use.