Changing Representations in Central African History

The Journal of African History http://journals.cambridge.org/AFH Additional services for The Journal of African History: Email alerts: Click here Subscriptions: Click here Commercial reprints: Click here Terms of use : Click here CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY WYATT MACGAFFEY The Journal of African History / Volume 46 / Issue 02 / July 2005, pp 189 - 207 DOI: 10.1017/S002185370400043X, Published online: 01 July 2005 Link to this article: http://journals.cambridge.org/abstract_S002185370400043X How to cite this article: WYATT MACGAFFEY (2005). CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY. The Journal of African History, 46, pp 189-207 doi:10.1017/S002185370400043X Request Permissions : Click here Downloaded from http://journals.cambridge.org/AFH, IP address: 128.178.131.113 on 22 Mar 2015 Journal of African History, 46 (2005), pp. 189–207. f 2005 Cambridge University Press doi:10.1017/S002185370400043X Printed in the United Kingdom 189 CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY BY WYATT MACGAFFEY Haverford College ABSTRACT: This article examines how historiography makes its objects and includes critical reflections on the epistemological frames that have shaped historical representations of Central African states and social structures. The article examines the seductive quality of migration narratives ; mythical features of some classical models, creating order from reduced totalities ; historiographic burdens imposed by questionable anthropological models of kinship and matrilineal descent ; and asks if the prevalence of dual regimes of priest and king is a product more of ideology than history. The article argues for increasing recognition of the value in political studies of data relating to religion and art. KEY WORDS: Central Africa, historiography, kinship. All history as reconstruction of the past is of course myth. Jan Vansina T H E historian or anthropologist pursuing empirical data can hardly be aware of contemporary collective representations that shape both the analysis and the reception of it, that confer apparent significance on some facts rather than others, and assure some stories of ready acceptance. (It was once thought valuable to measure the shape of people’s heads.) These contemporary collective representations help scholars to identify reasonably stable objects to work with ; even changing and uncertain situations must be pinned down somehow. Those who study other people, other continents, are obliged, whether they are aware of it or not, to answer at least implicitly two questions : is the order we claim to have observed like what we think of as our own or very different from it ? Are we describing the alien or the universal ? The delights of exoticism compete with the impulse to remake others in our own image, or at least to assume the universality of our current concepts. Africa, for Europe and its offshoots, has long been the proving ground for solutions to these problems, which are not so much epistemological as political. The quotation from Vansina that serves as epigraph to this article encapsulates the ambiguity by suggesting that we, too, might be ‘natives ’, that our current concepts are in some way strange. The passage of time makes it easier to see, with benefit of hindsight, that this might be so. Once upon a time (c. 1880–1957), the diversity of African cultures was explained by the effect on an indigenous Negro population of successive waves of Hamitic invaders from the northeast. This story is now recognized and dismissed as a myth, although remnants of it – the ideological jetsam of imperialism – are still to be found in encyclopedias and on the World Wide Web. Scholars who thought of the Hamites as a real factor in history were never able to say exactly who they were or how they could be recognized. 190 WYATT MACGAFFEY Unanimously admitting that the situation was vague, contradictory and in need of further research, they relied on a combination of linguistic and physical features, their arguments slipping conveniently from one to the other criterion.1 To explain the hold of this myth on the imagination of scholars fully equipped with academic credentials we should look at what myth is. Myths have special properties, as C. Lévi-Strauss showed in a famous essay; mythical thought, as he puts it, ‘works from the awareness of oppositions towards their progressive mediation ’. The mediating term between the polar opposites retains something of their duality, giving it ‘an ambiguous and equivocal character ’.2 The polar opposites that underlie the Hamitic myth are the Civilized Caucasian and the Primitive Negro; the mediators are many, depending on which version of the myth we are reading, but the principal ones are ‘the Bantu’ and ‘the Nilo-Hamite ’. The trickster in the whole structure is the Hamite, neither black nor quite white, uncivilized yet a civilizer, an African who comes to the continent from elsewhere. A myth’s explanatory value consists in the story’s reduction to a simple, orderly, commonsensical form of a situation that is not only uncertain as to the facts but politically charged. It acquires operative value in a given context because the pattern it locates in the past is deemed to explain and legitimate the present. In its day, the Hamitic myth served these purposes admirably. It implied that the European conquest of Africa merely continued an ancient and progressive history, but it ceased to satisfy us when the political context of its telling changed in the late 1950s. Yet the underlying political problem remained that of not only discerning but explaining and legitimating social order. THE FALL OF EMPIRES In the European imagination, political order is understood as a function of an administrative hierarchy, a monarchy. This image lends itself to historiographic shorthand, because we think we understand what a monarchy is. It is difficult to understand what the absence of monarchy might be, if not anarchy. That was the ground of Hugh Trevor-Roper’s notorious remark to the effect that Africa had no history because unrewarding gyrations do not lend themselves to historiography. Post-Hamitic myths, still preoccupied with order, included the idea that, in Sudanic Africa at least, ‘the idea of ruling ’, as one scholar put it, was introduced, in the form of divine kingship, by migrants who brought with them an Egyptian, or perhaps Meroitic or Ethiopian ideology to the Sudanic populations among whom they settled; ex oriente rex. Images of movement and crossing are particularly seductive as explanations of change, so much so that their metaphorical character disappears and they seem to be historical facts. For a while, historical maps of Africa were embroidered with migratory arrows leading in all directions. The assumption that kingship was the hereditary endowment of a racial 1 W. MacGaffey, ‘ Concepts of race in the historiography of Northeast Africa ’, Journal of African History, 7 (1966), 1–17. 2 C. Lévi-Strauss, ‘ The structural study of myth ’, in C. Lévi-Strauss, Structural Anthropology (New York, 1963). CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 191 group was discarded, but notions of ethnic essentialism lingered. Nowadays they are still to be found on the outer fringes of art history. Bantu civilization was accounted for by one great migration, or perhaps two, western and eastern. Roland Oliver noted that early concepts of the Bantu expansion relied on migration and conquest; he preferred the idea that expansion was facilitated by agricultural and metallurgical skills, but even he continued to speak of ‘ an unending sequence of migration, conquest and absorption ’; the story was still one of heroes on the move.3 All this is now suspect, but for a while, as historians short of archival data turned to oral history, it seemed to be supported by Africans’ own histories of migration and conquest. Within the Bantu-speaking area, indigenous stories of the founding of the Luba, Lunda, Kongo and other ‘empires ’ constituted amenable objects for historians. Thus, writing on the origin of the Kongo kingdom in perhaps the thirteenth century, J. Cuvelier, in his L’ancien royaume du Congo, based himself on a story reported by Cavazzi in the seventeenth century and on indigenous materials that he collected himself, to which he added colorful and heroic details of his own invention.4 He told how Ntinu Wene killed his aunt in the course of a dispute and fled south across the Congo River, where he made himself king by force. Ntinu Wene then convened a grand celebration, at which he commissioned his clan chiefs to go out and govern the provinces of his new kingdom. He was not consecrated as king, however, until he made peace with the local earth-priest, Na Vunda, who cured him of a possession fit. A succession of historians has repeated this story without seriously questioning Cuvelier’s anachronistic synthesis of heterogeneous sources. Migration and conquest sounded reasonable, and everyone assumed that matrilineal clans, a ‘primitive ’ form of social organization, had existed since early times, in Kongo as elsewhere. Relegating the politically inept, conquered population to a religious role seemed ‘natural ’. John Thornton has re-examined Cuvelier’s procedure and his sources, in the process radically, albeit speculatively, rewriting the story of the origin of the kingdom.5 Gone are the domestic dispute and the crossing of the river ; Ntinu Wene, also called Lukeni, now figures as a military entrepreneur and the kingdom as a loose federation, the product of a mix of conquest, alliance and voluntary affiliation, a rickety arrangement much less rational-legal than Thornton’s earlier book suggested.6 Cuvelier’s source for the picture of primitive Kongo as an organization of clans was a Catholic convert in Vungu named Petelo Boka, writing in 1912, who was trying to make history out of the clan traditions of his day. Thornton believes that traditions in which clans are supposed to be the basic units of social organization date only from the mid-nineteenth century, when the kingdom was in decline.7 His new 3 R. Oliver, ‘ The problem of the Bantu expansion ’, in J. D. Fage and R. Oliver (eds.), Papers in African Prehistory (Cambridge, 1970). 4 J. Cuvelier, L’ancien royaume de Congo (Bruges, 1946). 5 J. K. Thornton, ‘ The origins and early history of the kingdom of Kongo, c. 1350–1550 ’, International Journal of African Historical Studies, 34 (2001), 89–120. 6 J. K. Thornton, The Kingdom of Kongo : Civl War and Transition, 1641–1718 (Madison, 1983). 7 Broadhead argues, however, that the decline of the kingdom has been greatly exaggerated, mostly because its organization, based largely on ritual and magic, no longer 192 WYATT MACGAFFEY story is based on the older traditions in the historical record, which are dynastic, implying a different political structure not based on clans. Recent studies elsewhere in Central Africa present a number of kingdoms as similarly messy political constructions, made up of elements of migration, assimilation, imitation, commercial competition and local ambition.8 Vansina now prefers the term ‘commonwealth’ to describe the former Lunda ‘empire’, because its unity was no more than the acceptance of the Mwaant Yaav as a ruler superior to others. He dismisses the heroic story of its founder as a nineteenth-century invention related to the development of the trade network between Luanda (on the coast) and the Lunda heartland. The expansion of the Lunda entity in northeastern Angola in the seventeenth century was as much a matter of influence as conquest. Responding to new thinking in anthropology as well as history, Igor Kopytoff wrote a masterly new myth, synthesizing precolonial history into a story of ceaseless flux on a turbulent internal frontier.9 The official histories of African polities, he notes, are ‘remarkably repetitive’ in attributing the foundation to migration and conquest, but local histories more modestly tell of small groups splitting, drifting, reforming in various ways. In Kopytoff’s synthesis there are still migrations, conquests and kingdoms, but the central story is about the unfolding of a tradition. Frontiersmen, he says, do not arrive empty handed or empty headed ; they bring with them pre-existing conceptions of social order, and the society that they construct cannot be explained without reference to this model. The principal traits of the tradition include the right of the first settler; the despotic cast of rulership ; the assumption of hierarchy in all relationships; the use of kinship as a metaphor for political relations ; and the importance of the corporate kin group. The idea of an evolving tradition now seems inescapable, but we might ask, concerning the allegedly Pan-African or at least Sub-Saharan tradition that Kopytoff outlines, and which he derives from a single source in ancient times, whether it is convincing as an historical object. The traits he dwells on are all political; there is little reference to cosmology, religion, technology or environments, and the story has neither beginning nor end. Jan Vansina likewise employs the idea of a tradition, but puts more flesh on the bones, confines it more or less to the Congo basin, and locates events in space and time.10 THE KING AND THE PRIEST Kopytoff discusses at length the ‘problem ’ of legitimating authority on the frontier. One ideological solution, that of beginning history anew, explains the frequency of stories in which migrants enter a supposedly empty land. seemed to correspond to the expected rational-legal model. S. H. Broadhead, ‘ Beyond decline: the kingdom of Kongo in the eighteenth and nineteenth centuries ’, International Journal of African Historical Studies, 12 (1979), 615–50. 8 The Kanyok kingdom was never more than a work in progress, and intrinsically so. R. Ceyssens, Le roi Kanyok au milieu de quatre coins (Fribourg, 2003), 183. 9 I. Kopytoff, ‘ The internal African frontier ’, in I. Kopytoff (ed.), The African Frontier: The Reproduction of Traditional African Societies (Bloomington, 1987). 10 J. Vansina, Paths in the Rainforests : Toward a History of Political Tradition in Equatorial Africa (Madison, 1990). CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 193 Alternatively, if the presence of predecessors is acknowledged, the newcomers can incorporate the indigenous inhabitants in a variety of ways and co-opt their mystical powers in relation to the land. The story of Ntinu Wene and Na Vunda, in which Wene becomes king and Na Vunda becomes an earth priest responsible for consecrating the king, is an example of the second type. The frequency with which this arrangement is said to have occurred in Central Africa is suspect, because history is rarely so neat.11 The pairing of priest and king occurs widely, but often without reference to conquest. Consider some examples. In modern Kongo foundation narratives, the hero and his followers, displaced by some incident of violence, leave the capital, Mbanza Kongo (or, if the story is told on the north side of the Congo River, Mwembe Nsundi), and arrive at the river (not necessarily the Congo). There they are obliged to separate into groups because they cannot all fit in one canoe, or for some other reason; that is why we are now divided into nine clans in our new country, which we occupied peacefully because it was empty. Nevertheless, the inauguration of a chief, like that of Ntinu Wene, requires the participation of the priest of an nkisi nsi, a ‘ nature spirit’ such as Bunzi or Mbenza. The priest clan may provide the new chief with his ritual wife, the mpemba nkazi ; there is no suggestion of ethnic difference. In eastern Kongo, the priestly function is performed by the smith, also associated with nature spirits.12 In the land of Kazembe, in the lower Luapula valley on the border between Zambia and Congo, the same relationship is enacted in somewhat subtler form, with many linguistic and other resemblances to the Kongo ritual. At the foundation of a new village, a magician, with the headman and his wife, make a charm called nshipa which they bury and which is never seen again unless the village moves, when it is destroyed. This charm is similar in its composition and function to an nkisi nsi, and distinct from the calabash which is the headman’s personal ritual object and dies when he does.13 The rituals of the northern Yaka are somewhat more complex. The chief’s installation celebrates the creation of the state by Lunda immigrants who subordinated Kongo and Yaka groups as well as others regarded as autochthonous. The principal ritual officer, the Tsakala, is ‘linked to the autochthonous landowners ’, although he is not one of them but a matrilateral relative of the chief (chiefship is inherited patrilineally). Towards the end of the ritual, the representative of the landowners gives to the chief the symbolic anvils that are part of his regalia. Although the chief’s personal life is said to display strong Kongo traditions, his chiefly insignia are said to be Lunda.14 In fact, however, all of the insignia can be found all the way to the Atlantic, well beyond the reach of any Lunda migration. Whatever 11 As Kopytoff notes, it occurs widely elsewhere, but the foundational narratives and associated rituals of Central Africa are not at all like those one finds in, say, northern Ghana or Burkina Faso. M. Izard, Gens du pouvoir, gens de la terre (London, 1985). 12 W. MacGaffey, Kongo Political Culture : The Conceptual Challenge of the Particular (Bloomington, 2000), 65, 148. 13 I. Cunnison, ‘ Headmanship and the ritual of Luapula villages ’, Africa, 26 (1956), 2–16. 14 R. Devisch, ‘ From equal to better : investing the chief among the northern Yaka of Zaire ’, Africa, 58 (1988), 261–89. 194 WYATT MACGAFFEY migrations occurred, they are not the origin of the opposition between chief and priest. The Mbundu tell two stories, simultaneously, about their origins. One says that the hero Ngola Inene arrived from the northeast, married, and left descendants as founders of various subdivisions. The other, that the ancestors emerged from a body of water called Kalunga ; their descendants keep wooden figures called malunga, which are associated with bodies of water, govern the use of land and are responsible for rainfall and agricultural success. J. C. Miller argued that because the personae in both myths are human beings with no remarkable attributes, the myths must be in some way ‘historical’. Seeking to write the beginnings of Mbundu political history by ‘identifying and placing in the proper chronological order the most important innovative techniques of social organization’, Miller then solved the problem of the simultaneity of the stories by making the malunga an innovation adopted by the earlier, lineage-based Mbundu society. ‘The lunga brought a form of territorially based authority into the lives of the Mbundu’, and thus made possible a first step towards greater organizational flexibility and, eventually, state-like formations.15 In this instance, the chief came first, the priest second. In Luc de Heusch’s version of early Kongo history, the autochthons are already earth-priests but conquest does not generate a cult of its own. Whereas Thornton dismisses the story of Ntinu Wene as recent and doubts that there was a priest of the earth (kitomi), called Mani Kabunga, later Na Vunda, who represented the conquered inhabitants and took on the role of sacralizer of the kings, De Heusch needs both of them. Ntinu Wene (Lukeni) no doubt belongs to the Central African corpus of myths about founding kings from elsewhere, but the story nevertheless tells De Heusch that the original Kongo kings, unlike their counterparts in Loango, Kakongo and Ngoyo, were mere political leaders with no magical powers. Taking up a suggestion by Anne Hilton, De Heusch then explains the sudden conversion to Christianity of King Nzinga Nkuwu in 1491 : the king, being dependent for his moral authority on the ritual action of the Mani Vunda, seized the opportunity provided by the newly arrived Portuguese to equip himself with a cult of his own.16 If the Kongo king were ever deficient in magical powers, he would, I believe, be unique in Central Africa, where the distinction between political and ritual roles, so important to those for whom the separation of church and state seems obvious, cannot be made.17 De Heusch rightly dwells on the ubiquity of what he calls ‘dual systems ’, but evidently the pairing of earth-related and dynastic rituals is independent 15 J. C. Miller, ‘ Lineages, ideology and the history of slavery in western central Africa ’, in P. E. Lovejoy (ed.) The Ideology of Slavery in Africa (Beverly Hills, 1981). 16 L. de Heusch, Le roi de Kongo et les monstres sacrés (Paris, 2000), 75–6. De Heusch synthesizes ethnography from different periods into one body of data, and relies without critique on derivative writers, including G. Balandier, K. Ekholm and A. Custodio Gonçalves. 17 Ibid. 33. In another example of this ethnocentrism, Ehret questions Vansina’s gloss of – kúmù as ‘ big man ’, because the etymology of the term connotes ‘ social influence and ritual importance rather than material authority ’ and this is therefore ‘ a ritual, rather than a political role ’. C. Ehret, An African Classical Age (Charlottesville, 1998), 147. CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 195 18 of the narratives that purport to account for them. That is the conclusion of J. Ceyssens, writing on the peoples of Mbuji-Mayi. He notes a complex of oppositions between ‘invaders’, thought of as superior, ‘above ’ and associated with fire, who are believed to have introduced cannibalism ; and the ‘autochthons ’, who are of the below and associated with water. Though local conquests and subsequent colonial policies may have given political substance to these oppositions, they are more fundamentally ‘ intellectual’, a way of defining oneself vis-à-vis an Other. ‘Dyadic opposition must therefore correspond to a need that goes beyond historical contingencies’.19 The pairing can be understood sociologically, in that every community exists both in space and time, which are the necessary dimensions of production and social reproduction and will be ritualized to some extent in every agricultural society. Space is the earth itself and the forces of nature on which all depend in common. Time, on the other hand, is the source of authority and the measure of social differentiation ; reference to the past purports to distinguish older from younger, first-comer from late-comer, aristocrat from commoner. These two dimensions are what Victor Turner called communitas and societas, although he thought of communitas as occurring only in marginal situations outside the reach of societas.20 As Michael Jackson put it, ‘The complementary principles of social organization which are variously called lineage/locality, kinship/residence, ancestors/earth, descent/territoriality, can be abstractly and heuristically polarized as a distinction between temporal and spatial modes of structuring ’.21 Not the origin of this polarity but how it works out in practice is a contingent, historical question. THE MYTH OF REAL KINSHIP Underlying both Miller’s reconstruction of Mbundu social development and Kopytoff’s synthesis are the remains of an older myth that I will call the myth of real kinship. The myth is that originally, or fundamentally, kinship terms denote what Europeans think of as family, but that they can be extended to cover other relations. Kopytoff is well aware that kinship terms can be manipulated, but nevertheless uses a contrast between the early stages of a frontier polity, when communities were organized by kinship, and later stages, when kinship terms were used ‘metaphorically’ to express relations of dependence between the founding, dominant group and its client groups : In a growing frontier settlement, the kinship metaphor … provided an almost imperceptible transition and a bridge between two systems; the earlier one, in which real kin and quasi-kin relations held together the founding group and its 18 L. de Heusch, ‘ The symbolic mechanisms of sacred kingship : rediscovering Frazer’, Journal of the Royal Anthropological Institute, 3 (1987), 213–32. 19 ‘ La structuration dyadique doit donc répondre à un besoin qui va au-delà des contingences historiques ’. J. H. C. Ceyssens, ‘ Pouvoir et parenté chez les Kongo-Dinga du Zaire ’ (Ph.D. thesis, Catholic University of Nijmegen, 1984), 72. 20 V. W. Turner, Revelation and Divination in Ndembu Ritual (Ithaca NY, 1975), 15–29. 21 M. D. Jackson, ‘ Structure and event : witchcraft confessions among the Kuranko ’, Man, 10 (1975), 387–403. 196 WYATT MACGAFFEY close adherents, and the later system in which political relations between rulers and subjects (though still often expressed in kinship metaphor) were more contractual, more formal, more distant and more instrumental.22 Although both Kopytoff and Miller disavow any idea of a necessary evolutionary sequence, the echoes of the movement ‘from status to contract ’ and from ‘kinship to territory ’ in Henry Maine’s Ancient Law are strong.23 Of course, children have parents and grandparents, but as Kopytoff says elsewhere, ‘to modern Westerners the kinship metaphor suggests nurture and closeness; in Africa, and elsewhere, it conveys authority and subordination ’. Moreover, ‘the kin group may have the right to sell or kill its ‘‘ free ’’ members’.24 That being so, the distinction between ‘real’ and ‘ metaphorical’ kinship is misleading and unnecessary, except as a concession to Western habits of mind. Miller wrote his own myth of the making of Central African political structures. As he put it, the gaps in the historical record can be filled in from ‘the theoretical literature on lineages, ideology and slavery ’.25 Basing himself on anthropological speculations about the conditions in which descent groups arise, he believed that lineage structures were basic to western central African societies from the first millennium AD, when land was abundant. Segmentary lineage systems arose as groups increased in size and then divided. Matrilineal descent prevailed in the savanna because ‘it happened ’ that the people chose to aggregate mother’s rather than father’s relatives. Later, increased production for exchange created a demand for labor. Slavery was introduced as a means of acquiring additional labor, although it violated the fundamental precept of the lineage because slaves were not kin. The ideology of the lineage persisted, however, long after the disappearance of ‘relatively pure descent-based societies ’, because both the old men, ‘ clinging to the reins of social control into their dotage’, and their cadets, ‘coveting with ill-concealed impatience their uncles’ wives ’ and eager to be elders themselves, both found it useful. The soap-opera prose here distracts attention from the lack of any historical foundation for the story. Much of the theoretical literature on which Miller based this reconstruction is itself now recognized as ideology, ours rather than theirs. It includes the naturalistic fallacy that kin groups, descent groups in particular, arise because communities notice what they have been in the habit of doing and give it a name. The functions usually imagined are co-residence and inheritance: people usually live with their mothers, or inherit from a mother’s brother, hence matrilineal descent. When the group grows too large for its terrain it ‘segments’, but retains some sense of its original unity. In the 1950s this sort of assumption was common in the materialist anthropology of, for example, Julian Steward. ‘Lineage ideology ’ is therefore supposed to be an ‘idealized version ’ of a reality, though the reality may be somewhat different in practice. With respect to matrilineal systems in particular, later discussion 22 Kopytoff (ed.), Frontier, 59. H. S. Maine, Ancient Law (London, 1861). The parallel movement ‘ from religion to politics ’, whose ideological legacy I discussed in the previous section, is asserted by foundational thinkers such as Auguste Comte and James George Frazer. 24 I. Kopytoff, ‘ Slavery ’, Annual Review of Anthropology, 11 (1982), 207–30. 25 Miller, ‘Lineages ’, 42. 23 CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 197 concerning their alleged fragility in ‘ modern’ times confused political issues with those of family, gender and patriarchy. The corporate character of descent groups cannot be attributed to such diffuse factors. LINEAGE THEORY AS MYTH As an alternative I want to argue that there is not and never was any such thing in Africa as a matrilineal society.26 There are societies with matrilineal descent groups, but such groups are not what they are usually thought to be ; in any case they are only one of several bases for social organization in a given society, and relatively superficial.27 Matrilineal descent groups are units in political competition, and slavery is what they are about. To pursue this theme, we have to review some aspects of ‘lineage theory ’. Lineage theory, in its British version, arose from the same intellectual milieu as the Hamitic myth. In the nineteenth century, anthropologists were fascinated by matrilineal descent, which they confounded with matriarchy as a supposedly earlier stage of social evolution than patriarchy. Matriliny thus became a discrete object of exaggerated importance. In 1935, rejecting ‘conjectural history ’, A. R. Radcliffe-Brown argued that to maintain order a primitive society was necessarily patrilineal or matrilineal because some corporate body had to be responsible for children ; it was apparently a matter of happenstance whether mother’s or father’s group were chosen.28 Elaborating on this theme, Meyer Fortes and E. E. Evans-Pritchard declared that most areas of Africa lacking monarchs maintained order by segmentary lineage systems.29 These two models, respectively hierarchical and egalitarian, reiterated an opposition with roots in British political thought that reach back to Hobbes and Locke. Both models were construed primarily as administrative orders; that is, they left out the politics, too suggestive of anarchy. This bias is clear in Fortes’s account of matrilineal descent in Ashanti and subsequent critiques by Thomas McCaskie and others.30 When Fortes noticed challenges to official genealogies he dismissed them as ‘ without justification ’. In fact, to this day, arguments about pedigree are central to Ashanti politics. The classification into centralized states and descent-based systems broke down not long after it was put forward. Critics pointed to states with lineages at their core and to societies that did not fit either model. If segmentation was supposed to be a demographic process, how did it manage to generate structures of similar scope and form, all providing the balanced opposition 26 ‘ Patrilineal society ’ is equally mythical. A. Southall, ‘ The illusion of Nath agnation ’, Ethnology, 25 (1986), 1–20. 27 In highly intermarried communities, everyone is linked to others in multiple ways, of which one will take precedence only in a particular situation. ‘ There is no such thing as general primacy with regard to any form ’. E. Peters, ‘ Some structural aspects of the feud among the camel-herding Bedouin of Cyrenaica ’, Africa, 37 (1967), 261–82. 28 A. R. Radcliffe-Brown, Structure and Function in Primitive Society (London, 1952). 29 M. Fortes and E. E. Evans-Pritchard, African Political Systems (London, 1940). 30 M. Fortes, Kinship and the Social Order (Chicago, 1969); A. N. Klein, ‘ The two Asantes : competing interpretations in Akan-Asante culture and society ’, in Lovejoy (ed.), The Ideology of Slavery ; T. C. McCaskie, State and Society in Pre-colonial Asante (Cambridge, 1995). 198 WYATT MACGAFFEY that supposedly guaranteed order ? In fact, segmentary opposition is characteristic of all political systems, and does not guarantee anything. As African countries became independent, anthropologists discovered that Africans were political after all; they lost interest in descent groups and the question of order.31 As a result, historians were left with a radically flawed model of a kind of organization they thought they needed to write about. With benefit of hindsight, Adam Kuper declared, My view is that the lineage model, its predecessors and its analogs, have no value for anthropological analysis. Two reasons above all support this conclusion. First, the model does not represent folk models which actors anywhere have of their own societies. Secondly, there do not appear to be any societies in which vital political or economic activities are organized by a repetitive series of descent groups.32 Depending on just what he means by ‘ the lineage model ’, Kuper is almost certainly wrong. Kongo, for example, have a clear idea of a corporate matrilineal clan subdivided into matrilineages, and think of their society as organized by a repetitive series of them. This model is not, however, a true description of what exists in real life now or at any time in the past. Nor is it an idealized or approximate description, except perhaps in the sense that if, in the view of any given elder, there were any justice in the world, then society would be so ordered, and to his advantage. In short, the model is an agreed formula for making political claims; such claims may be temporarily successful, at the expense of similar claims advanced by others, and to that extent a set of supposedly perpetual descent groups may be said to exist, albeit temporarily. MATRILINEAL DESCENT AS POLITICAL PROCESS Claims to what? A matrilineal descent group need not be genealogically organized, but it must claim as its estate the reproductive capacities of its female members from generation to generation ; hence the need for a presumptively perpetual corporate identity.33 Both inheritance practices and residence patterns can change without affecting the collective interests of such a corporation, but if marriage contracts transfer children, the exchanging groups become patrilineal.34 In practice, there are many examples of communities that switch contracts depending on what seems advantageous, and most supposedly ‘unilineal ’ descent groups are in fact cognatic, meaning 31 ‘ Matriliny as a topic in anthropology is as dead as a dodo, one would think ’. P. Peters, ‘ Revisiting the puzzle of matriliny in south-central Africa ’, Critique of Anthropology, 17 (1997), 125. 32 A. Kuper, ‘ Lineage theory : a critical retrospect ’, Annual Review of Anthropology, 11 (1982), 71–95. 33 Fortes, Kinship, 184. A group’s estate may include other long-term interests. Vansina relates corporate matrilineal groups of southwestern Angola to the adoption of cattle-herding. On the other hand, he clings to the idea that matrilineal descent, once developed, might spread elsewhere without any such motive. J. Vansina, How Societies are Born: Governance in West Central Africa before 1600 (Charlottesville, 2004), 88–97. 34 Residence is only critical in the case of the Lele, where ‘ matrilineal ’ clan sections are corporate residential associations whose members are not related genealogically. This arrangement occurs amid chronic shortage of male labor. M. Douglas, The Lele of the Kasai (London, 1963). CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 199 that many members can trace their descent through both father and mother.35 The historical question must be, in what circumstances can groups hope to expand by retaining reproductive capacity rather than by exchanging it? As a reproductive strategy, matrilineal descent only makes sense when a supplementary source of wives for the male members of the group is available. The economics and demographics of the slave trade since the seventeenth century made it possible for groups in both West and Central Africa to acquire women without giving women in exchange.36 As Kinkela Ngoma wrote from Vungu in 1915, ‘Great chiefs stood out as buyers and sellers of slaves, so that their villages should prosper and increase … but though a chief may have had wealth and followers and been invested, if his followers and his women die off, then he has no more power and respect ’.37 In the societies of the ‘matrilineal belt ’, from the Atlantic at least as far as Zambia, descent groups include lineages deemed to be descended from strangers, ‘ slaves ’ whose women are available to the ‘free’ members as wives, with the result that the group can recruit the offspring of both its male and its female members, and ambitious individuals can hope to advance their fortunes more rapidly than the simple reproductive activity of their sisters would allow.38 This is only the beginning of the possible complexities. The politics of it all center on the eminently disputable question, who is free, who is the slave? The outcome of the politics is often that losers, whatever their actual ancestry, can become slaves, and that an entire group can lose its corporate integrity, its claim to autonomous control over an estate in women.39 This internal generation of slaves modifies both Meillassoux’s assumption that slaves always become such by violence and Kopytoff’s assumption that they are strangers who have been to some degree assimilated.40 In her survey of changes in Kongo social structure from the sixteenth to the nineteenth century, Anne Hilton made the mistake of reading too much modern ethnography into the sixteenth-century data. On the other hand, she recognized the flexibility of a bilateral system in which corporate forms, patrilineal as well as matrilineal, could emerge to serve new long-term interests. She believed that the ‘original ’ system of matrilineal descent was weakened during the seventeenth century by the accumulation of slaves, and that Kongo ‘ reverted’ to it in the eighteenth century (precisely the period in 35 Kopytoff (ed.), Frontier, 44–5. W. MacGaffey, ‘ Lineage structure, marriage and the family among the Central Bantu, Journal of African History, 24 (1983), 173–87. There may well be countervailing considerations, of course ; for example, that in given circumstances men are relatively more useful as laborers or warriors. Note that giving bridewealth for a wife is only one step away from purchase on the continuum of property rights. Kopytoff, ‘ Slavery ’ ; W. MacGaffey, ‘ Economic and social dimensions of Kongo slavery ’, in S. Miers and I. Kopytoff (eds.), Slavery in Africa (Madison, 1977), 242–3. 37 W. MacGaffey, Religion and Society in Central Africa (Chicago, 1986), 86. 38 J. Van Velsen gives a detailed example of such a tethered matrilineage among the Lakeside Tonga, notes the relationship between slave status and nonpayment of bridewealth and rightly complains that not enough students have inquired into the actual practice of cross-cousin marriage, as opposed to local statements of preference. J. Van Velsen, The Politics of Kinship (Manchester, 1964), 133–7. 39 MacGaffey, Political Culture, 71–2. 40 One source of this difference is that whereas Meillassoux’s field experience was in Mali, Kopytoff’s was primarily in Congo. 36 200 WYATT MACGAFFEY which slaving was at its height !). If we drop the widely accepted but unsupported assumption of original matrilineality, the rest of Hilton’s account is persuasive.41 Matrilineal descent emerged in the eighteenth century in response to new opportunities for competitive accumulation of women. Thornton agrees that clan traditions (and, presumably, the clans) were related to trade routes but dates them to no earlier than ‘the trade revolution ’, c. 1850, when the commerce in peanuts and wild rubber required large numbers of porters.42 Traditions recorded in the twentieth century do not permit historical inferences any older than 1850, but the trade routes are much older and were busy long before that. Persistent dynastic traditions related to the politics of the declining kingdom co-existed for some time with emergent clan traditions. I am arguing that matrilineal descent is an unstable and relatively superficial phenomenon whose supposed importance is mostly a product of certain European preoccupations. The implications for ancient history are disturbing. Most reconstructions assume a relatively definite and stable object, an orderly system of social organization whose units give ‘meaning ’ to the words that denote them. A cluster of such assumptions underlies parts of Christopher Ehret’s reconstruction of proto-Savanna Bantu.43 Discussion centers on the term *-gàndá, whose denotations are deemed to have drifted from descent groups to residential units and back again. Its derivatives in easterly Bantu areas mostly refer to a residential unit, but today, according to Ehret, it appears in a relict distribution among a block of matrilineal peoples in the Lower Congo in the form kanda, which he says is ‘a kin term ’ denoting ‘clan ’. He concludes, ‘it seems probable that in the proto-Savannah-Bantu period, society was composed of matriclans divided into lineages ’. However, in Kongo at least, kanda means ‘group or category’, as in makanda ma nza, ‘the peoples of the earth’, or minkisi myena makanda matatu, ‘there are three kinds of minkisi’. It is not a kinship term, though it does also denote ‘ clan’. The assumption that clan is what matters leads to an overemphasis on one particular significance of a word.44 The same preconception led anthropologists to misrepresent kinship terms. Most of the anthropological literature on kinship suffers from the ethnocentric assumptions that each relative gets one label, and that on a genealogical diagram the labels form a pattern that corresponds to the social structure ; at about the time anthropologists began to notice that role labels were situational and therefore multiple, they lost interest in the whole subject. The Kongo are supposed to have a Crow type of terminology, which is generally associated with matrilineal descent. In fact, their terminology is much less determinate. In given situations, Kongo usage, when projected on to an anthropologist’s diagram, generates Crow terminology, but in other situations it generates Hawaiian terminology. The Kongo can therefore 41 A. Hilton, ‘ Family and kinship among the Kongo south of the Zaire River from the sixteenth to the nineteenth centuries ’, Journal of African History, 24 (1983), 189–206. 42 43 Thornton, ‘ Origins ’, 97. Ehret, Classical Age, 150–1. 44 The range of meanings of important Kongo words is a constant problem for the translator. Words for technical objects and processes are much more definite, and less open to ethnocentric interpretation, than words for social units and processes. MacGaffey, Political Culture, 59. CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 201 45 be said to have two systems, an anthropological paradox. The difference between the two is most apparent in the alternative names that can be given to cross-cousins ; ambivalence of this kind is a common feature of kinship terminologies from the Congo basin to southern Africa, but has not moved anthropologists to much rethinking. There is more. In Kikongo, the term mpangi, which according to the dictionary means ‘sibling ’, in fact applies reciprocally to any two people who stand in the same relationship to a third person or group. Among the results of this application, the term ngudi a nkazi, ‘mother’s brother’, can refer to a man who is not a member of the speaker’s matrilineal clan and may have, in the narrow sense of the term, no genealogical relationship to the speaker at all.46 This consideration and others too detailed for this context undermine Vansina’s reconstruction of ‘the invention of matrilinearity’ in southwestern Congo.47 On the other hand, they support his idea of the House, a cluster of kin, clients and others around a dynastic core, as the basic social unit of Central Africa. A nineteenth-century Kongo village was such a House. The structure of Kongo society is a network of patrilateral links between matrilineal nodes, all subject to constant political negotiation, with no definite boundary and no center. In the nineteenth century, it generated oligarchical districts (nsi, pl. zi-) in which linked lineages of the free helped to keep each other’s ‘slave’ lineages in precarious subordination.48 T R A D E, M Y T H A N D M A G I C The energy for all this politics came, in the eighteenth and nineteenth centuries, partly from local agricultural production but mostly from the Atlantic trade, in which the Kongo served as mediators between the coast and the point on the Congo, the Pool, at which it became navigable into the heart of Africa. The theory, the model, one might say the anthropology of this activity is recorded in Kongo stories that have been regarded at times as history, sometimes as myth. Can we not ask of myths, as of history, since they are so close, who is thinking and to what purpose ? Though a myth may not carry the signature of its author, it is surely a product of its time and place. If we drop the assumption that the historical kingdom of Kongo with its capital, Mbanza Kongo, is the necessary point of reference ; cease to read Kongo migration stories as a kind of bungled history of events; and situate them in the places from which they are reported, a different sense of their import emerges. 45 The Plateau Tonga in Zambia have, from this point of view, ‘ three systems ’, which ‘ people can play with’ (Elizabeth Colson in a letter to me, 13 Dec. 1995). In Luapula usage : ‘ Some kinsmen are given two kinship terms depending upon whether one discusses matters of descent or kinship ’. K. Poewe, ‘ Matriliny in the throes of change ’, Africa, 48 (1978), 353–67. 46 When Ego’s mother’s father belongs to a given matriclan, mother herself is mpangi to any man whose father also belongs to that clan, no matter how vague the relationship between these fathers; Ego may therefore properly call that man ngudi a nkazi because he is Ego’s ‘ mother’s brother ’. This usage is not metaphorical. W. MacGaffey, Custom and 47 Government in the Lower Congo (Los Angeles, 1970), ch. 5. Vansina, Paths, 152–5. 48 MacGaffey, Political Culture, 71–2, 119 ; Vansina, Societies, 73–82, 90. 202 WYATT MACGAFFEY Kongo traditions of the past (kinkulu) tell two kinds of story, one on a grand scale and the other more modest. The modest ones tell of local migrations between named places, but the routes of the supposed migrations are the principal directions of nineteenth-century trade and point to the sources of chiefly titles. Titles were often also the names of places – powerful charms (minkisi, sing. nkisi), chiefs, clans and settlements all being aspects of the same complex. In the event of the death and replacement of a chief, ritual retraced the route to an earlier settlement, now a cemetery, where investiture took place or insignia were obtained ; or, we could say, the story provided a roadmap for the ritual. Many stories, old and new, of a quest for spiritual power of a hero, whether magician or prophet, take the same form, that of a journey to the land of the dead.49 North of the Congo River the cemeteries went by a number of recurrent names, including Mwembe Nsundi, corresponding to the diversity of trade routes through the mountains. South of the river, where the trade between Mpumbu (the Pool) and the coast was to some extent controlled until the 1870s by the Kongo king, clans added the suffix Ne Kongo to their names ; their chiefs were often taken to Mbanza Kongo for burial.50 Those that participated in the Nsundi network, most of it north of the river, used Nsundi as suffix; modern ethnographers have assumed that Nsundi was a ‘tribe’. Vungu, in modern Mayombe, was a point of convergence between routes oriented towards Kongo, Nsundi and Loango, respectively.51 Stories on the grand scale describe transitions, often across a river, leading to the settlement of a new country. These stories are not historical but sociological, sketching an ideally ordered society. In eastern Kongo they list the food crops carried on the journey, assign a skilled craft to each clan and list the insignia of the chiefs. One tells that the clans were led by a dog who said nothing, ‘even when spoken to ’; wherever the dog stopped, they camped for the night, and one of the clans settled there. Kabila, ‘to divide, distribute’, is a verb that recurs in these stories to mark the creation of social order. The river that is crossed may be called Nzadi, ‘large river’, and may be identified with an actual stream, but it is a cosmological boundary. The marvels accomplished by the chief to effect the crossing, often full of erotic imagery, announce that this is no ordinary river and promise multiplication and prosperity through right marriage, right eating and right government.52 In that sense, the land across the river provides a space in which to inscribe social theory.53 All this closely resembles, though not on an epic scale, the stories among Luba-related peoples in eastern Congo of heroes who come from across the 49 50 MacGaffey, Religion, 107–16; idem, Political Culture, 72–5. This practice was unknown before the late eighteenth century. Hilton, ‘ Family ’, 204. 51 J. Janzen describes the Nsundi network, giving more credence than I would to ‘ conquest ’ stories. Central control of the northern network from the original Mbanza Nsundi must have disintegrated by the mid-eighteenth century at latest. J. M. Janzen, Lemba, 1650–1930 (New York, 1982), 61–72. 52 The inverse forms of precisely these categories – promiscuity, cannibalism and anarchy – were used in European writing about Africa to characterize the absence of civilization there. D. Hammond and A. Jablow, The Myth of Africa (New York, 1977). 53 MacGaffey, Political Culture, 207. CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 203 river to introduce civilization as right marriage, right eating and right government.54 In both east and west, the elsewhere from which the king comes is a land of spirits (Bupemba, Mpemba, Upemba), although it may be identified with a geographical location. It is a place visible to diviners in the reflecting surface of the water; in the form of a cemetery, a cave, a grove or a pool, it is a place of testing and investiture for chiefs and other persons whose special powers are signified by white kaolin clay, mpemba. The initiation rituals of chiefs retrace and recapitulate the migration stories of the myths. In much more detail than it is possible to recount here, Kongo chiefship rituals read like a reduced or provincial version of those found among Luba.55 Given these similarities across the Congo basin, we may explore the nature of the linkages that integrate commonwealths and their neighbors. The positivist bent of Africanist historians inclined them to write what McCaskie calls ‘barebones political history’ and to pay no attention to culture, belief and religion – in short, to turn their backs on the typical concerns of anthropology.56 ‘States’ and ‘empires ’ emerged, endowed with more or less rational-legal means of imperative control and bearing a strong resemblance to early modern European kingdoms. Recent writers, however, have described these entities as not only less structured but as constituted by imaginary powers that for lack of a better word I will call ‘magical ’. According to M. Roberts and A. Roberts, for example, what came to be known in colonial times as the ‘Luba empire’ in eastern Congo should rather be thought of as ‘a constellation of chieftaincies, officeholders, and sodalities that validated claims to power in relation to … a largely mythical center ’.57 The highly secret staffs of Luba chiefs are mnemonics for migration stories that might as well be Kongo, telling of the journey from the royal center to the owner’s village, by way of unelaborated sections of the shaft, representing uninhabited savanna, and lozenge-shaped sections, representing settlements (dibulu ; Kikongo, mbanza) along the way. Such insignia are prestige items, but neither ‘prestige ’ nor ‘insignia ’ is adequate to capture the potency of these magical composites: ‘Memory, medicines, prayers, and prohibitions are implanted in a staff, rendering it a powerful device for curing and protection’.58 The owners of staffs and other potent devices, which were not just signs of the presence of power but active components of it, were themselves magical objects. The ‘tribute ’ the owners gave to superior chiefs from time to time had economic value but was primarily significant as bringing into being the relationship that empowered both parties. In Kongo, the migration stories may correspond to actual migrations of small groups, but they are really about the linkages that made trade possible. Over short distances, individual security was provided by kinship ties, 54 L. de Heusch, Le roi ivre, ou l’origine de l’état (Paris, 1972). P. Petit, ‘ ‘‘ Les charmes du roi sont les esprits des morts’’ : les fondements religieux de la royauté sacrée chez les Luba du Zaire ’, Africa, 66 (1996), 349–66; MacGaffey, Political Culture. 56 T. C. McCaskie, ‘ Empire state: Asante and the historians ’, Journal of African History, 33 (1992), 469. 57 M. N. Roberts and A. F. Roberts (eds.), Memory: Luba Art and the Making of 58 History (New York, 1996), 28. Ibid. 164. 55 204 WYATT MACGAFFEY themselves created by marriage. Chiefs and big traders also formed marriage alliances, a practice followed by the Luso-Africans in Luanda. Large continental caravans were armed for their protection, and monarchs, where they existed, could provide security by administrative means backed up by force. In parts of northern Kongo where there were no monarchs, market cycles controlled by committees of chiefs functioned as governments, as did the Lemba association. All these authorities relied on magical devices (minkisi). An English trader reported in the 1880s, ‘These fetishes play a most important part in regulating conduct of individuals or families – nay, intertribal feuds are settled by the same means, decisions enforced, disturbances quelled’.59 So successful were minkisi in regulating commercial contracts on the Atlantic coast that the French and Portuguese governments found it necessary to confiscate them; the Belgian trader Delcommune had one carried around the markets to denounce some of his employees who had decamped with stolen property.60 A story told in Mbanza Manteke, which is south of the river, gives an impression of how linkages were created : At a time when there was no invested chief in Mbanza Manteke, and they were not under the Ntotila at Mbanza Kongo, Na Bikadyo decided to go to Mbanza Nkazi to buy the chiefship from the Mbenza chief there, who already belonged to the kingdom of Kongo. She brought gifts of goats, chickens, money, leopard’s teeth, a leopard skin, bracelets, anklets, and a buffalo-tail whisk. At Mbanza Nkazi she was to hide with all these things under a blanket until everything was ready. As she was about to emerge she began to menstruate, and was forbidden to show herself. She said, ‘ I have already paid for the title, so let it be given to my sister’s son Na Mpyoso Nsakala Nangudi ’.61 In other words, to belong to the ‘kingdom ’, one acquired a title in exchange for tribute, and equipped oneself with appropriate power objects. The center of this kingdom was a real but at the same time an imaginary place, a ‘mythical center ’. Investiture with a title, as this very minor example shows, resembled the constitution of an nkisi, with seclusion in a special enclosure (or merely under a blanket). The ritual composition of an important nkisi retraced its origin through a succession of banganga (owneroperators of the nkisi) to a founder who emerged from the water. The composition also served to consecrate the nganga himself as part of the apparatus necessary to mobilize the nkisi. Just as the nkisi acquired aspects of personhood, so its nganga became in some respects an object, both of them figuring in a chain of agency extending from the dead to the here and now.62 Invested 59 R. C. Phillips, ‘ The social system of the Lower Congo ’, Journal of the Manchester Geographical Society, 3 (1887),154–69. 60 This nkisi was expensive to rent from the chief who owned it; it now has a new career as a distinguished work of African art (No. 7943) in the Royal Museum of Central Africa, Tervuren. A. Delcommune, Vingt années de vie africaines, 1874–1893 (2 vols.) (Brussels, 1922). 61 Tradition of the clan Nanga Ne Kongo, as told to Ruth Engwall in the 1930s. I heard a briefer version in 1965. The story of the would-be chief who menstruates at the wrong moment is a folkloric cliché in Central Africa that explains why women are not chiefs. 62 MacGaffey, Political Culture, 80. CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 205 chiefs across Central Africa have the same dual nature, both person and object, mediating between the permanent and the transient. POWERFUL OBJECTS In the nineteenth century, scholars called all such institutions, devices and processes ‘magical ’ and ‘superstitious ’; in the twentieth century, the vocabulary of religion was applied to them. De Heusch is only one of those who have tried to capture the magical aspect of kingship with the term ‘divine ’. Part of the difficulty with ‘divine kingship ’ is that the traits that allegedly constitute it are independently variable. Substituting ‘ sacral’ for ‘divine ’ is no great improvement, because the vocabulary of (modern) religion, including ‘holy ’, ‘spiritual ’, ‘ worship’ and ‘ supernatural ’, is inappropriate to Central African thought, as missionaries discovered when they tried to find equivalent terms in Bantu languages. De Heusch’s concept of the king as corps-fétiche conforms better to Central African practice than Frazer’s ‘divine king ’.63 Central African religion is technically oriented, expected to produce practical results. Twentieth-century anthropology, reluctant to endorse the association of ‘magic ’ with irrationality, declared that such rituals were in fact ‘expressive ’ rather than instrumental, thus ignoring their manifest intention. Kopytoff remarks on the ‘puzzle ’ presented by the ‘despotic’ character of even small-scale chieftaincies, in which, despite the rhetoric of absolute, ‘sacred’ powers, the chief’s ‘real’ powers may be not much more than those of a successful arbiter ; in extreme cases, he is a figurehead, a manipulated object. Many scholars have in effect reified their own difficulties with what Kopytoff calls ‘the strangeness of the idiom ’ by making ‘divine kingship ’ a discrete, diffusable object. Kopytoff’s own solution to the puzzle is to introduce a space in time between ‘a cultural inventory of symbols and practices that were brought from a metropole ’ and a reconstruction of the model in the particular circumstances of a frontier situation that tempered and limited its ‘despotic ’ character on, as it were, this side of the river.64 In fact, the discrepancy between the imaginary absolutism of magical power and the constrained reality of secular authority is a problem of our own thinking – or at least of our thought in the self-consciously rational-scientific mode of scholarship. The only way out of the resulting embarrassments is to admit that our own societies also understand power at least partly in ‘magical ’ terms. Thousands of pieces of African magical equipment are now recognized as ‘art’ ; their original uses are best understood in the framework of recent advances in art theory related to the power of images. David Freedberg shows how the need to deny that images were seen as agents, and therefore could be and were agents, exercised Christian theologians from the beginning. Inheriting the problem, art theory from the eighteenth century onwards constructed a set of ideological devices to deny the power of artworks and 63 De Heusch, Le roi de Kongo, 24 ; M. Augé, Le Dieu objet (Paris, 1988); J. Bazin, ‘ Retour aux choses-dieux ’, Le temps de la réflexion, 7 (1986), 253–85. 64 Kopytoff (ed.), Frontier, 34, 52, 64. 206 WYATT MACGAFFEY 65 neutralize their effect. Denying their agential quality became a marker of rationality and even a criterion for distinguishing civilization from barbarism; the classical norm in art ‘ cast the primitive as the dark image of itself ’.66 In a highly technical theoretical breakthrough, Alfred Gell shows how (‘ art’) objects are implicated in chains of agency.67 These related approaches explicitly do away with ‘ magic’ and the invidious distinction it implies between the ‘primitive’ and the ‘advanced’. The agency of artworks fits them for political functions ; acknowledging that objects become powerful when power is attributed to them helps us to bridge the gap between the vocabularies of religion and political science. Political power in traditional Kongo operated through a diverse but unified system that included chiefs and minkisi, which were not only mutually supportive but to some extent interchangeable. Objects derive their potency from their participation in social relations ; mediating between persons, they become person-like. In a modern context, those relations are often easily set aside as ‘ symbolic’; in a foreign environment, as ‘ritual ’.68 CONCLUSION Is it possible to write history after acknowledging the messiness of the frontier? Lévi-Strauss tells us repeatedly that mythical thought works with diminished totalities, discrete entities that can be arranged in an orderly and thus ‘scientific ’ account of the world. We have noted several such discretionary moves: reducing the flux of history to a static set of tribes ; reducing African regulatory systems to two types, and reducing both to administrative rules by excluding political and ritual factors ; reducing social structures to a rule of unilineality. All these reductions are achieved by focusing on what look like rational-legal elements to the exclusion of others, at the same time implicitly reducing the human actor to a cipher. In West Africa, where the the volume of ethnographic and historical studies is larger and richer than in Central Africa, more attention has been paid to the political and economic roles of rituals, priestly associations, shrines and oracles. The Hamitic myth was enabled by the imposition on Africa’s multiplicity of a simple grid of discrete categories, the black and white races, distinguished by physical appearance and by knowledge of the idea of ruling, or lack thereof. This grid was superseded by a more complex one that divided the continent into eternal tribes, but the earlier dichotomy survived in benign form as a distinction between states and stateless societies. Our search for usable myths has introduced other reductions. Fortes and EvansPritchard thought that it would be scientific to describe political systems as mechanisms, in abstraction from their cultural idiom. This view is 65 D. Freedberg, The Power of Images: Studies in the History and Theory of Response (Chicago, 1989). 66 F. S. Connelly, The Sleep of Reason : Primitivism in Modern European Art and Aesthetics, 1725–1907 (University Park PA, 1994), 9. 67 A. Gell, Art and Agency: An Anthropological Theory (Oxford, 1989) ; W. MacGaffey, ‘ Astonishment and stickiness in Kongo art ’, Res, 39 (2001), 137–50. 68 W. Arens and I. Karp (eds.), Creativity of Power (Washington DC, 1989) ; D. Fraser and H. M. Cole, ‘ Art and leadership : an overview ’, in Fraser and Cole (eds.), African Art and Leadership (Madison, 1972). CHANGING REPRESENTATIONS IN CENTRAL AFRICAN HISTORY 207 characteristic of our modern, reduced sense of the political; we restrict it to the use of material resources and secular or ‘real’ powers by leaders, officials, pressure groups and armed forces. Such agents may, we admit, dress up or enhance their powers and claims with ritual, myth and references to the supernatural, but these effects are add-ons borrowed from religion rather than essential to politics. John Janzen complains that there is little conversation between those who study cult cycles and those concerned with state formation, despite the fact that the underlying social processes are similar.69 In northern Kongo, on the routes between the coast and the Pool, the Lemba ritual association, about which Janzen is writing, was probably more important than chiefs in regulating both local disputes and long-distance trade. It is difficult to admit that an African ‘religion ’ is itself a political system. Our disciplines have developed separate vocabularies and conceptual traditions for discussing two kinds of power, the real and the imaginary, that are not always distinguishable in our own environment. In dealing with Kongo ideas about power I have found this dichotomy intolerable. Power in Central Africa, as Fabian remarks, is understood as a personal property, ‘tied to concrete embodiments, persons and symbols, rather than to abstract structures such as offices, organizations or territories ’.70 We have abandoned races, tribes, empires and segmentary lineage systems, imagined objects that made good myths for a while. I have critiqued matrilineal descent, divine kingship, the naturalistic fallacy concerning the origins of descent groups and the myth of real kinship. On the other hand, much has been done since 1960 to fill in the void that was Africa, defined once upon a time by the absence of history, government, art and philosophy. Even the social sciences are making an appearance, as affinities reveal themselves between myth on the one hand and history and sociology on the other. The last great mythical entity, ‘Africa ’, is being demolished, or at least questioned, by studies focusing on the Indian Ocean and the ‘Black ’ Atlantic.71 Meanwhile, the fantasies of Cuvelier, disseminated in Kikongo in the mission bulletin Ku Kiele, are now regarded by Kongo intellectuals as traditional knowledge handed down from ancient times ; the used clothes of social science and indirect rule provide uniforms for revivals of the Kongo Kingdom and the Luba Empire. 69 Janzen, Lemba, 21. J. Fabian, Power and Performance: Ethnographic Explorations through Proverbial Wisdom and Theater in Shaba, Zaire (Madison, 1990), 25. 71 M. W. Lewis and K. E. Wigen, The Myth of Continents : A Critique of Metageography (Berkeley, 1997). 70