Dual Classification Reconsidered: Nyamwezi Sacred Kingship and Other Examples (Atelier d'Anthropologie Sociale)

DUAL CLASSIFICATION RECONSIDERED

This book is published as part of the joint publishing agreement established in 1977 between the Fondation de la Maison des Sciences de l'Homme and the Press Syndicate of the University of Cambridge. Titles published under the arrangement may appear in any European language or, in the case of volumes of collected essays, in several languages. New books will appear either as individual titles or in one of the series which the Maison des Sciences de l'Homme and the Cambridge University Press have jointly agreed to publish. All books published jointly by the Maisons des Sciences de l'Homme and the Cambridge University Press will be distributed by the Press throughout the world. Cet ouvrage est publie dans le cadre de l'accord de co-edition passe en 1977 entre la Fondation de la Maison des Sciences de l'Homme et le Press Syndicate de l'Universite de Cambridge. Toutes les langues europeennes sont admises pour les titres couverts par cet accord, et les ouvrages collectifs peuvent paraitre en plusieurs langues. Les ouvrages paraissent soit isolement, soit dans l'une des series que la Maison des Sciences de l'Homme et Cambridge University Press ont convenu de publier ensemble. La distribution dans le monde entier des titres ainsi publies conjointement par les deux etablissements est assuree par Cambridge University Press. The original French version of this book was published in the series Atelier d'anthropologie sociale Editor: Louis Dumont, ficole des Hautes fitudes en Sciences Sociales, Paris This series will include primarily books which have been written by authors involved in the 'Cooperative Research Programme No. 436' of the Centre National de la Recherche Scientifique, although it will also be open to contributions from other authors with a similar perspective. Books in the series will share a common orientation combining the development of theory with ethnographic precision. They will not be limited either geographically or thematically. Most of the books in the series will be published in French, although some may be in English. Also in the series Cecile Barraud: Tanebar-Evav: Une societe de maisons tournees vers le large Raymond Jamous: Honneur et baraka: les structures sociales traditionnelles dans le Rif Andre Iteanu: La ronde des echanges: De la circulation aux valeurs chez les Orokaiva Wauthier de Mahieu: Qui a obstrue la cascade: Analyse semantique du rituel de la circoncision chez les Komo au Zaire

DUAL CLASSIFICATION RECONSIDERED Nyamwezi sacred kingship and other examples

SERGE TCHERKfiZOFF Chefde Travaux, iZcole des Hautes Etudes en Sciences Societies, Paris

TRANSLATED BY MARTIN THOM

The right of the University of Cambridge to print and sell all manner of books wos %TQfit€Q hy

Henry VIII in 1534. The University has printed and published continuously since 1584.

CAMBRIDGE UNIVERSITY PRESS CAMBRIDGE NEW YORK NEW ROCHELLE MELBOURNE SYDNEY & EDITIONS DE LA MAISON DES SCIENCES DE L'HOMME PARIS

CAMBRIDGE UNIVERSITY PRESS Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao Paulo, Delhi Cambridge University Press The Edinburgh Building, Cambridge CB2 8RU, UK With Editions de la Maison des Sciences de l'Homme 54 Boulevard Raspail, 75270 Paris Cedex 06, France Published in the United States of America by Cambridge University Press, New York www. Cambridge. org Information on this title: www.cambridge.org/9780521105354 First published in French as Le roi Nyamwezi, la droite et la gauche: Revision comparative des classifications dualistes by the Maison des Sciences de l'Homme and Cambridge University Press, 1983, and © Maison des Sciences de l'Homme and Cambridge University Press, 1983 First published in English by the Maison des Sciences de l'Homme and Cambridge University Press, 1987, as Dual classification reconsidered: Nyamwezi sacred kingship and other examples. English translation © Maison des Sciences de l'Homme and Cambridge University Press, 1987 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. This digitally printed version 2009 A catalogue recordfor this publication is available from the British Library Library of Congress Cataloguing in Publication data Tcherkezoff, Serge. Dual classification reconsidered. Translation of: Le roi nyamwezi, la droite et la gauche. Bibliography. Includes index. 1. Nyamwezi (African people). 2. Left and right (Symbolism). 3. Folk classification. 4. Duality (Logic). I. Title. DT443.3.N93T3513 1987 306'.012 87-6618 ISBN 978-0-521-30895-3 hardback ISBN 978-0-521-10535-4 paperback

Contents

Preface Translator's note Introduction

vii ix 3

1 Right/Left oppositions and the 'pre-eminence of the right hand' Symmetry and Hierarchy Robert Hertz: 'The pre-eminence of the right hand' Rodney Needham: 'The left hand of the Mugwe'

PART ONE

3 8 12 15

Nyamwezi classifications

2 Nyamwezi classifications: right or left?

29

Right/left binary opposition Right or left?

34 38

3 Right and left: the distinction between levels Right/left, man/woman, even/odd Black/white The hierarchy of levels

4 Hierarchy andpower: kingship captures the kings King/priest Hierarchy/power

P A R T T W O Comparison

5 The same questions asked of other examples Dogon twins Right and left in China Right and left in Osage ceremonies 'Do dual organisations exist?'

43 43 51 61

69 69 76

and conclusion 89 89 92 98 106

vi

Contents

6 Towards a comparative study ofdualist symbolism Hierarchical opposition The search for values The change in level

113 113 118 12 2

Notes

132

Bibliography

151

Index

155

Preface

This comparative study belongs to a collective research project directed by Louis Dumont and Daniel de Coppet and based at the R C P 436 of the Centre national de la recherche scientifique in Paris, which

considers the relation, in a given society, between the patterns assumed by the hierarchy of values and those which feature in the circuits of ceremonial exchange. It is based upon a shorter, unpublished study (1977: Diplome d'etudes approfondies, under the supervision of Marc Auge, at the ficole des Hautes fitudes en Sciences Sociales) of Nyamwezi classification, which had made use of the method developed by Masao Yamaguchi in his study of Jukun symbolism (see chapter 4 note 7 below). These arguments were discussed at the RCP's seminar (Atelier d'anthropologie sociale), and also at seminars led by Louis Dumont and by Marc Auge. I would like to thank the participants at these seminars for their suggestions, which have enabled me to broaden my discussion. I have drawn heavily upon Louis Dumont's works in my own analysis, and his comments, his encouragement, and the interest he has taken in my research have played a crucial part in stimulating me and in helping me to complete this book. My intellectual debt to Claude LeviStrauss is self-evident. I would never even have conceived of this book had it not been for the ever-renewed sense of discovery that my reading of The savage mind, and of the other pages devoted by him to the logic of classification, has aroused in me. It is Claude LeviStrauss's work which provides the context for the investigations into aspects of the structural method that I have sought to develop here. I would like here to thank Cecil Barraud, Daniel de Coppet, Marie-Jo Jamous and Raymond Jamous, who have very kindly read through various drafts of this book.

Vll

Translator's note

The reader should bear in mind that versions given here of passages from fimile Durkheim, Marcel Mauss and Claude Levi-Strauss do at times differ slightly from the published translation. M.T.

IX

Introduction

Right/left oppositions and the 'pre-eminence of the right hand'

The opposition between the categories of 'right' and 'left' is a familiar and quasi-universal feature of the 'symbolic' classifications of human societies. A given object, depending upon its nature, is placed upon the right or left of a house. A human being, depending upon whether he or she is a man or a woman, is buried lying on his or her right or left side. A different status is accorded to the places on the right and left of the master of ceremonies or host at feasts and banquets. Still more generally, people take care not to confuse the gestures which should be carried out with the right hand with those which involve the left hand. There is, for example, an opposition between amorous caresses and bodily hygiene or, in the case of our own customs, one can distinguish between a respectful greeting, where one invariably uses the right hand, and a friendly wave, where either hand may be used. Oppositions of this kind were already being studied in the late nineteenth century, and scholars were mainly concerned to account for the special value accorded the right hand. Research was not, however, concerned with the social aspects of this one-sided attribution of value. It aimed, rather, to add further refinements to natural explanation. Thus, in accord with Professor Broca's findings, an asymmetrical development of the nervous centres was held to render the greater skill of the right side inevitable: 'We are right-handed because we are left-brained' (Hertz i960, p.90). There has long been a sociological awareness of these problems, for Robert Hertz's study of 1909 (1970), in which he linked the 'Pre-eminence of the right hand' with the study of 'religious polarity', followed on the heels of the medical debates of the 1890s. What is the point of reconsidering these questions now? Once Robert Hertz had highlighted the sociological aspect of the problem by noting that societies did not correct the slight natural inequality of the two hands but rather persecuted the left hand quite openly, so

4

Introduction

despising it that 'right-handedness' became an 'ideal to which everybody must conform' (p. 93), and once Claude Levi-Strauss (1966) had shown how the 'savage mind' used binary oppositions in classificatory schemas, a firm framework for the solution of this question would seem to have been found, one which would simply require a range of further illustrative examples. But to accept this resolution of the problem would be to disregard the vast and shifting domain which Hertz, by placing the right/left opposition on the same level as the sacred/profane 'religious polarity', had discovered. Recent writers have thus thought to use it as the basis for a theory of the 'political' arid the 'religious'. As I shall show below, it was Rodney Needham who argued for this extension of the question, in i960. There was, however, something reductive about this contribution. Hertz's article did undoubtedly suggest a possible extension of the debate, but in a different direction. It called upon researchers to give a more precise account of the manner in which classifications are defined by a particular social norm, i.e. in Hertz's terms, by the variable form of the 'sacred/profane' relation. Indeed, Hertz had argued (pp. 93-4) that one should study the complex relation between the general £cole Sociologique model (the sacred/profane opposition) and the particular symbolic realisations which assume a dualist form. On the one hand, there would be a duality of principles which were interconnected and which assumed various forms, with either one being liable to encompass a particular aspect of the other, depending upon the site of the ritual and the identity of the actor. On the other hand, there would be symbolic figures which, Hertz asserted, tend to form two complementary series. But this latter simplification, which is attributable to the limited nature of the data available to the author, has in turn been applied to the sacred/profane dualism. Researchers have noted the existence, in the case of Africa in particular, of binary right/left structures, and have hastily constituted them as two juxtaposed and compartmentalised paradigms, inferring that a similar symmetry is applicable to the sacred/profane opposition. These two principles are then reduced to a complementarity of two 'powers', which the two hands reflect. This would seem to be the path that several English researchers have followed in their attempts to develop Hertz's pioneering study of right/left categories. The most important work in this area was done by Needham. His analysis of the Meru of Tanzania reopened the debate in i960 and in 1967 he published a detailed analysis of

Right/left oppositions

5

Nyoro symbolism and announced his intention of bringing out a collection of articles on this topic (cf. Needham 1960, 1967). The collection appeared in 1973, with a long introduction by Needham himself. Its title, Right and left, shows quite clearly that it is concerned with the same material as Hertz was. But the conclusion of the i960 article, which was to serve as a point of reference for subsequent researches, was devoted to a 'wider5 theoretical significance' . . . related to the 'studies of Georges Dumezil'. In the last analysis, it is the opposition between 'political power' and 'religious authority' which is studied in this article, and which serves as a working hypothesis in the following article (1967), as Needham notes in his introduction to it. Several of the contributors to Right and left do not extend the argument in this manner, and do not attempt to define two 'powers'. The majority of them, however, acknowledge the validity of the binary method in the study of classifications, and this is a problem in itself. This debate is therefore a fundamental one. It touches upon the articulation between what we term 'power' and the 'sacred'; it undermines ethnocentric approaches to the balance of powers and counter-powers (king/priest; chief/orator in Polynesia); consequently, it emphasises the distinction between the principle of symmetry and the principle of hierarchy; it raises the question of the expression of values in the 'symbolism' of a society. In short, it suggests a possible theory of the symbolic. Evans-Pritchard, in his preface to Right and left, implies that, although the road is long, there are already a good number of markers along the way. If we are to assess this claim, we will need to present a detailed analysis of the distance covered by Needham as he moves from Hertz's 'religious polarity' to the dichotomy in Meru symbolism. This trajectory provides the occasion for the emergence of what will be called the binary method, where dualism becomes a simple logical complementarity which leaves no room for contradiction. This is a crucial point. All accounts of traditional classification point to the existence of contradictions, but modern logic finds them unacceptable. In 1966, Victor Turner, a specialist in symbolic anthropology with a wide experience of the subtle configurations of traditional thought, chose to interpret the variety of values assumed by the colour black in the Ndembu world in terms of a difference of contexts, and not in terms of a difference of levels of value within the unity of one and the same system of representations (Turner 1966).

6

Introduction

Thus, from Hertz's essay at the turn of the century up to the present day, those analysing dualist classifications have sought to discover, behind empirical variety, a coherence based upon unity and upon the non-contradiction of the signifier-signified relation. But one does not have to cast 'savage thought' in the very particular mould of our own logical models in order to acknowledge its coherence. If we set aside our a priori assumptions, we find that 'savage thought' in fact enables us to grasp the peculiarity of modern scientific thought. I shall thus attempt to show that reversal is not a 'contradiction', but rather a commonplace and basic figure of the hierarchical thought which characterises holistic societies (see below). Before examining the development of Needham's argument, I shall briefly summarise Hertz's position. Hertz's study contains several different lines of argument, and is not simply concerned with the complementarity of the universal series of right and left. In order to criticise Needham's approach I will have to refer to the Meru ethnography. Of course, if I am to substantiate my criticism of the binary method, and advance an analysis in terms of levels, I shall obviously go on to invoke other examples also. Yet I do not regard an analysis in terms of levels as just another analytic strategy, which would be necessary for certain societies but which would not invalidate the binary method in other cases. Indeed, when one subjects the dualist oppositions presented in Needham's study to close scrutiny, it would seem to be just as difficult to apply a principle of symmetry to Meru symbolism as it would be to that of other non-modern societies. I shall extend my investigation of Hertz's and Needham's arguments by considering the symbolic classification of the NyamweziSukuma of Tanzania, which, though little known, is well documented and resembles the Meru one. I shall not give an exhaustive account of their system of thought here but shall confine myself to a brief summary. If we consider a few of the oppositions often invoked in studies of dualism (right/left, male/female, even/odd, black/ white and, in addition, king/priest), we find that we cannot construct a binary table without seriously distorting the system. Indeed, if one refrains from imposing a binary choice upon the symbols (to the left or to the right), one becomes aware of hierarchical constructions. The poles of each opposition are not in the same relation to the whole to which they refer. Different values organise hierarchised symbolic levels. Passing from one context to another

Right/left oppositions

7

will therefore sometimes involve a change of levels and reversal will be meaningful too. For the Nyamwezi, a major hierarchy of references organises the whole set, namely, the opposition between the local principle of reference to close, named ancestors and the global principle of reference to royal ancestors, between the single dimension, which signifies incompletion, and the double dimension, which presents itself as a totality uniting asymmetrical opposites. The figure of the twin is one expression of this duality. Dogon symbolic classification lends credence to this perspective, for it also opposes the inadequacy of singularity to the sense of completion that duality offers. For the Dogon, however, twins are not just one instance of this duality but the actual principle of duality. Germaine Dieterlen implies that 'complementarity' (a pair of twins of opposite gender) is not a logical operation but a value which underlies the whole set of representations (Dieterlen 1968). This value therefore organises a range of different and hierarchised levels. The binary method does not therefore seem to be appropriate to the analysis of systems of classification organised in terms of right and left. Yet the Nyamwezi example also serves to show that this debate has still wider ramifications. In opting for a hierarchical analysis, as against the binary method, one is not simply employing a more rational instrument of analysis. A hierarchical analysis allows one to consider values. Are 'symbolic5 classifications the expression of a logical coherence whose universal rationality may immediately be apprehended in terms of European categories, or do they require another approach, involving an emphasis upon the particular values of a given society? In order to answer this question, I shall have to extend the terms of the debate and consider Chinese and Osage examples of classification. I had no choice but to refer to these examples, given the elegant studies on the question of right and left by Marcel Granet and Francis La Flesche, which Needham took care to place at the beginning of the collection of articles published in Right and left. Together with Hertz's study, these two articles represent, historically, the first statement of the problem. La Flesche, in 1916, and Granet, in 1933, had, through their choice of terms and their use of examples, provided a basic outline of the discussion. Furthermore, the Osage and, in more general terms, the Sioux, have a key part to play in the development of Levi-Strauss's theory of classification. The reader may recall his denunciation, in 1956, of the 'dualist

8

Introduction

illusion' (Levi-Strauss 1968). Yet, in reading Levi-Strauss's account, one finds oneself wondering whether the error of those who favoured the notion of binary division did actually consist in their not having grasped that dualism, triadism or classification in terms of n dimensions together constitute a transformation group, or whether there is not, over and above their failure to apprehend the whole structure, a still more blatant refusal to relinquish Western European categories of equality and symmetry and an extreme reluctance to come to terms with the hierarchical mode of thought of non-modern societies. The latter represents a difficult but vital task for comparative anthropology, as Louis Dumont (1978) has stressed. This reluctance to acknowledge the role of hierarchy has many repercussions. It causes anthropologists to place undue emphasis upon universal mental structures and to cast out a part of ceremonial life into the domain of'belief. By comparison, those who adopt a hierarchical perspective and thereby constantly refer to the whole, are able to account for the particular values of a given society, for its choices and for its 'ideology'. A proper recognition of hierarchical opposition is vital not only for any interpretation of'the pre-eminence of the right hand' (or of the left hand), but also for the analysis of ritual. Once the variety of symbolic configurations comes to be interpreted in terms of differences of level within a single system of representations, there is no longer any need to invoke an 'anti-structure' in order to account for the existence of'abnormal' figures. Reversal, for example, does not refer to such an anti-structure, and 'rites of reversal' are not always 'rituals of rebellion' or stagings of a 'cathartic' resolution of tensions. More generally, structural analysis would seem to be capable of reconciling the search for a global meaning with a consideration of the functions performed by the various parts. But 'meaning' here involves a hierarchy of values, and the 'functions' involve articulations between the different levels of values. SYMMETRY AND HIERARCHY

It may help to make my exposition clearer if I now attempt to give a more precise account of the nature of relations between right and left. I shall distinguish between the epithets 'dualist' and 'binary', reserving the former for the simple apprehension of the existence of two poles and using the second where a logical condition is entailed:

Right/left oppositions

9

the relation is symmetrical, or else asymmetrical but complementary, but always remaining on one and the same plane. Some studies do not even raise the question of symmetry or non-symmetry, left and right simply serving to name distinctions. In these analyses, a society's symbolism assumes the form of a dichotomous table in which two columns enable one to connect the contrasted items. Whether the author intends this mode of exposition to represent the actual nature of the situation or to have a merely methodological value, it clearly reflects an equalitarian bias. If, on the other hand, one is concerned with the asymmetry of the relation, one stresses the difference in value between each of the poles. Different emphases may then appear. Some writers insist upon the complementarity of opposites, whilst omitting to study the sole thing which enables one to appreciate (and to represent) a difference in value, namely, the question of reference. The union of complementary pairs may sometimes even look like simple arithmetical addition. If, however, one does not exclude the possibility of their being different planes (levels), asymmetry will seem hierarchical. An opposition a/b, where a is 'superior' in a given context, expresses two facts: a is posited as value, as idea-value ('we say or do that' = 'we must say or do that'), and b is then subordinated to it as its contrary. In Dumont's terms, b is 'encompassed'. There is not simply a linear succession of strata from b to a, a gradation one would follow from the lower to the higher floor. Where something is encompassed by something else, i.e. where there is hierarchy, a new relation is established. The discrepancy between the two terms gives rise to a tension, which takes the form of a series of dichotomies in which hierarchy may become less strongly defined, in which the asymmetry may be reversed, or in which combinations with other oppositions may appear. We then have to think in terms of different levels. Reversal (a> b—> b> a), far from being a contradiction, as it would be in our European logic, which is of a mathematical kind, then becomes operative, and indicates a change in level. A change in ritual context and a change in level in the order of values occur when reversal affects one and the same frame of reference. Thus, when a man passes from the pure to the impure pole, he may switch hands for a particular activity. There is also a change in level when the frame of reference itself changes. A Nyamwezi woman may thus, as she moves from a local reference to her own father's lineage to the global

io

Introduction

reference that marriage implies in her society, shift from a plane where the left position is valorised to a plane where the right position is valorised. Lastly, I will discuss totalisation. To pass from a domain in which one is left or right, man or woman, to one in which a conjunction of opposites appears, is to pass from a plane involving single elements to that of the whole set (where this latter is not the sum of the elements but is something that preexists any conceptualisation of the parts and gives these latter their meaning). We shall see how, among the Nyamwezi, 2 + 3 makes 5 in a very particular fashion, just as the conjunction man + woman implies the level at which kingship symbolises the whole of the society. The totalising level, the level of the whole set, differs from that (or those) at which the elements are located. One has to conceive of'five' as being a 'whole' which preexists its dissolution into 2 + 3. It is not the result of an addition. Among the Osage, 6 4- 7 is to 13 what the moieties are to the whole of the tribe. This totalisation, to employ the term used by Levi-Strauss himself in relation to the Osage '13', must not however obliterate the hierarchies 2/3 or 7/6, for these elements are in a differential relation to the whole which is 5 or 13. It is thus, I hope, clear that the 'hierarchy' with which we are concerned here is in no sense a gradation of substances or a stratification of social groups. It is in fact a question of values. My own study of the logic of dualist oppositions will therefore follow the definition that Louis Dumont, on the basis of a comparative analysis, has given of the term 'hierarchy': It is immediately obvious that there are two mutually opposed configurations of this kind: one is characteristic of traditional societies and the other of modern society. In the first . . . the stress is placed on the society as a whole, as collective Man; the ideal derives from the organisation of society with respect to its ends (and not with respect to individual happiness);... In modern society, on the contrary, the Human Being is regarded as the indivisible 'elementary' man . . . Each particular man in a sense incarnates the whole of mankind. He is the measure of all things . . . What is still called 'society' is the means, the life of each man is the end . . . (Dumont 1970, p. 9) . . . most societies value, in the first place, order: the conformity of every element to its role in the society - in a word, the society as a whole; this is what I call 'holism'. On the other hand, other societies - at any rate ours value, in the first place, the individual human being . . . This is what I call 'individualism' . . . On the logical plane, holism implies hierarchy and individualism implies equality . . . (Dumont 1977a, p. 4).

Right/left oppositions

11

Indeed, in a system in which an element is only defined by the place that it occupies in relation to the whole, two elements which are different cannot have one and the same place. Valorisation of the 'whole' implies a hierarchy of elements. If, on the other hand, one encounters a pair of distinct elements occupying one and the same position, they can no longer be defined by their 'place' in the whole set but only by their intrinsic differences, by their 'substance'. Equality, the antinomy of hierarchy, thus brings about individuality. What is the nature of these 'elements'? They are ideas and values which constitute the ideological system as a whole. The etymology of'hierarchy' implies that it is an 'order of the sacred', an order of ultimate values. Likewise, 'individualism' is based upon the individual as value. In order to clarify this point, we need to be aware of a fundamental fact which Dumont, on the basis of an observation by Talcott Parsons, has emphasised: the sociologist Talcott Parsons has the great merit of having brought fully to light the universal rationale of hierarchy (I have italicised certain words): 'We conceive action to be oriented to the attainment of certain goals, and hence to involve selective processes relative to goals. Seen in their relations to goals, then, all the components of systems of action and of the situations in which action takes place, are subject to the process of evaluation ... Evaluation in turn has, when it operates in the setting of social systems of action, two fundamental implications. First the units of systems, whether they are elementary acts or roles, collectivities or personalities, must in the nature of the case be subject to evaluation ... But given the process of evaluation, the probability is that it will serve to differentiate entities in a rank order ... The second implication is the well-known one that it is a condition of the stability of social systems that there should be an integration of the value-standards of the component units to constitute a "common value-system"' ... In other words, man does not only think, he acts. He has not only ideas, but values. To adopt a value is to introduce hierarchy, and a certain consensus of values, a certain hierarchy of ideas, things and people [on the plane of status] is indispensable to social life. This is quite independent of natural inequalities or of the distribution of power (Dumont, 1970, pp. 19-20). The hierarchy of values thus possesses a 'universal rationality'. If modern society is unaware of this fact, it is on a specific plane, which does not render the preceding assertion contradictory but instead enables one to set up a comparative perspective: 'A society as conceived by individualism has never existed anywhere for the

12

Introduction

reason we have given, namely, that the individual lives on social ideas' (Dumont 1970, p. 10). To use the term 'individualism5 is not therefore to assert that society is composed of individuals in an empirical sense. I have in mind Individualism with a capital T , individualism as value and as 'ideal representation'. It is the principle in modern society which corresponds to that of hierarchy, of the valorisation of the whole, in traditional societies. One still has to explain why, historically, Western ideology has broken free of a principle which would seem to be a feature of a wide span of different societies, and why it has 'scotomised' a 'dimension of man' which social anthropologists ought, according to Louis Dumont (1978), to illuminate. This is what Dumont has set himself to do. Let us now return to the question of hierarchy: It can be seen that the original sense of the term concerned religious ranking. We shall keep to this sense here, making it somewhat more precise. We shall admit that, any idea of command being left aside, the religious way of seeing things requires a classification of beings according to their degree of dignity. Yet the presence of religion is not indispensable, for the same applies whenever the differential elements of a whole are judged in relation to that whole, even if the judgement is philosophical as in Plato's Republic. So we shall define hierarchy as the principle by which the elements of a whole are ranked in relation to the whole (Dumont 1970, pp. 65-6).

These preliminary observations completed, we are now free to go back to the beginning of the century, when Hertz first raised the question of the 'pre-eminence of the right hand'. ROBERT HERTZ! 'THE PRE-EMINENCE OF THE RIGHT HAND'

Robert Hertz's study raises various questions. I have already noted that he treats the problem of right and left from a new perspective, and that he abandons anatomy for sociology. Hertz did not mean to deny the effects of the physical factor altogether, but the crucial question for him was: 'why [should] a humanly instituted privilege . . . be added to this natural superiority?' (i960 p. 92). Why did societies not correct the natural inequality by extolling ambidextrousness? Why was this encouragement restricted to music, for example, to playing the piano? Why, this special case apart, are children who wish to use their left hands rebuked and why are left-handed adults ridiculed? Why, in short, is a physiological inequality given a basis in culture?

Right/left oppositions

13

As is well known, Hertz resolves this problem by invoking the sacred/profane opposition, and by studying this 'religious polarity'. But in so doing he highlights different, and sometimes contradictory aspects of the right/left relation. Subsequent studies have, at any rate, laid unequal emphasis on these different aspects of the problem, and it is therefore fitting that I should here specify what they are. The dual opposition is first presented in a 'hierarchised' form, with the right hand dominating. 'The right hand is the symbol and model of all aristocracy' (p. 89); 'organic asymmetry in man is at once a fact and an ideal' (p. 93); 'a rigid hierarchical structure' is established (p. 96); One pole is 'strong and active', the other 'weak and passive' (p.98). Thus Needham, when he discovers the valorisation of the left hand of the Mugwe priest among the Meru, expresses some astonishment at this reversal in the universal hierarchy established by Hertz. However, his study ends by invoking equilibrium and complementarity of powers between the Mugwe and the council of'elders'. Thus, from the beginning of the century, the existence of hierarchy has been implied by oppositions of this sort. But scholars were quick to contest one of the ways in which its significance has been demonstrated. By the very title of his study, and throughout the text, Hertz asserts that asymmetry has a universal meaning. It is always the right, the 'male side', which dominates, and no room is left in this account for reversal and for its structural position. It was Granet who was to go on to point out 'the alternation of preeminence' and to demonstrate its significance. Hertz was, admittedly, quite familiar with reversal. He considered that it had existed at one point in human history. Each person in a 'primitive tribe' regarded the people of his own moiety as sacred, while those of the other moiety were regarded as profane. Value was thereby dependent upon the speaker, and its sign was reversed when one shifted from one moiety to the other. But, in the next stage (according to the evolutionist perspective of the epoch), 'the evolution of society replaces this reversible dualism with a rigid hierarchical structure' (p. 96). From then on, only one side was superior, for everyone. Finally, in wishing to demonstrate the universal meaning of asymmetry, Hertz produces a plethora of examples and thus inevitably mixes up different contexts. He establishes a series of links,firstfrom the right, then from the left, and seems subsequently

14

Introduction

to place more emphasis upon the symmetry and complementarity of the opposed elements than upon their hierarchy (pp. iO4ff.). It is this aspect of Hertz's analysis, whereby the bounds of each ritual context are crossed so as to establish opposed series, which receives most attention from later English researchers: 'The whole universe is divided into two contrasted spheres . . . so on one side there is the pole of strength, good and life; while on the other there is the pole of weakness, evil and death; day and night, east and . . . west . . . high and low, sky and earth' (p. 96). Hertz derives a general opposition from the Maori examples: the right side, 'the strong side', is the 'privilege of the strong sex' (p. 101). The left is 'profane . . . the weak side . . . female' (p.101). For all the acclamation that these formulae have subsequently won, they sometimes obscure the complex ethnographic data to which Hertz referred in the course of his argument, and which have therefore been disregarded. Thus, with respect to the above example, the Maori do not associate 'right' and 'left' with 'man' and 'woman' but rather with the two principles, the male and female, and man is 'compounded of two natures, masculine and feminine' (p. 102). These two principles are associated with the two sides, right and left, of the body. In this establishment of a compromise between the two principles, the reader must perceive the existence of a hierarchised valorisation preventing a binary development. This is what I shall attempt to show in the rest of this book. In conclusion, I would ask the reader to bear in mind that the somewhat schematic presentation of the two universal series was preceded by intimations of the far more flexible nature of the sacred/profane opposition itself. This latter must be kept to the fore if one is to understand the overall system. Thus, if one is considering things from a profane perspective, the sacred appears to be double in nature, with a pure and an impure aspect. On the other hand, if one is considering things from a sacred perspective, the impure appears to be a feature of the profane, and this latter turns out to have a double aspect also (p. 95). The ambivalent status of the impure, together with the variation in the sacred/profane relation, serves to show, no matter how dubious Hertz's distinctions are, that he does not rule out the possibility of accounting for dualist opposition in general in terms of a conception presupposing the existence of several levels. From Hertz's conception of the relation between religious polarity and the right/left opposition, it is possible to imagine proceeding

Right/left oppositions

15

in two directions. One could follow the path suggested by Hertz and attempt to throw some light upon dualist symbolism by invoking general, 'religious' values. This would be to bear in mind that, for Durkheim, such values define a more complex structure than that of binary opposition, in which the distinction entailed is one of 'prohibition5 rather than of complementarity, and necessarily presupposes its contrary, union: '[profane and sacred] which are both prohibited and which, however, never cease to mingle one with the other' (Durkheim 1975 p. 33). One would therefore have one's doubts about the universality of the two complementary series, even though Hertz's argument does in part favour such a notion. One could, on the other hand, give this latter line of argument priority and proceed in the opposite direction. One would then, on the basis of a right/left symmetry, integrate a society's symbolism with a general and compartmentalised dichotomy. One would sometimes end up with a general theory of powers also. Both of these aspects of the second interpretation of Hertz's arguments are apparent in Needham's analysis of Meru symbolism.

RODNEY NEEDHAM: THE LEFT HAND OF THE MUGWE

I propose now to go into greater detail and summarise Needham's own account of the dualist features which Bernardi's ethnographic study (1959) enables him to identify. Eight oppositions suggest a general division: 1 The Meru are divided into sub-tribes, each of which recognises a particular person as their spiritual leader: the Mugwe, who is at once a 'priest', a 'diviner', a 'leader', a 'chief and a 'representative of the founding heroes' (Bernardi 1959). The Mugwe's power lies in his left hand, and he only has to raise it to bring his enemies to a halt. These latter continually try to destroy this hand, and the Mugwe always keeps it hidden beneath a cloak. If someone saw it, he would die (Needham i960 pp. 20-1). 2 The Meru say that they came from the north, which they call 'right hand' (p.22). 3 In the Imenti sub-tribe, clans are described as 'black', 'red' or 'white', depending upon whether their original ancestors, at the time of their mythical immigration, arrived at night, at dawn or during the day. In the distant past, Needham adds (on Bernardi's authority) these colour categories undoubtedly referred to

16

Introduction 'political, social and territorial distinctions'. The Elders note further that there was no real distinction between red and white, and Needham therefore retains only one opposition: black clans/white clans—»arrived at night/arrived during the day (p. 22).

4

5

6

7

8

The Imenti were also divided into two groups, Igoki and Nkuene, which were 'on the north' and 'on the south' respectively. In addition, the white clans seem to have gathered on the north side. The Igoki were 'proud? and always sought to have the first claim on pastures. There was a similar division in the case of Tharaka, another sub-tribe, and the Mugwe was 'on the south'. Amongst the Tigania, the Mugwe was 'on the north'. Amongst the Igembe, yet another sub-tribe, the 'north/south' division only affected the opposition between the first wife's huts and those of the other wives, within a polygynous household (P-23). Women are depreciated. Nowadays, with the Meru under foreign domination, they say: 'we are all women; the real man is the (colonial) Government'. The women are not admitted to the rituals led by the Mugwe either (p. 23). The Meru say that 'The sun rises at the place of Mukuna-Ruku and sets at the place of the Mugwe'. Mukuna-Ruku is a mythical figure which is 'wholly light', and which gives life to the sun. He is also a legendary Arab trader who lived in Mombasa i.e., as far as the Meru are concerned, to the east. Needham infers from this that the Mugwe is associated with the west, and therefore with darkness. He notes that Bernardi's interpretation differs from his own (I shall return to this point below). Needham himself mentions the frequent association of black with the Mugwe (the sacrificial bull and the Mugwe's sacred objects are black) (pp. 23-4). Needham quotes several passages from Bernardi which demonstrate the 'religious' nature of the Mugwe, and also of the tribal government constituted by the Elders. There is therefore an opposition between a religious and a 'political' power, where the latter is conceived in terms of the 'jural and administrative' control enjoyed by the Elders (p. 24). Finally, the temporal opposition between 'first' and 'second' is important. In the origin myth, the Meru had to cross an expanse of water in order to reach what was to be their country. The Mugwe, who led them, divided the waters and sent first a boy

Right/left oppositions

17

and a girl, then a young man and a young woman. Finally, he himself crossed at the head of all his people. This implies that the first to cross are inferior in status, that they are juniors. A myth of the origin of the races is invoked in support of this: all the peoples came from the same place, but black men were born before those with white skins; and the first-born are 'political inferiors' (in the present situation), the Elders maintain. There is, finally, an opposition, again at the level of myth, between honey-collecting and agricultural work: the age-class 'of honey' comes first, with the Mugwe, then that 'of the agriculturalists'. It is well known that the second of these two 'subsistence activities' is more important than the first (as far as basic sustenance is concerned). The complementarity between the sacred and power features here too; honey is one of the Mugwe's insignia, while the Elders hold the rights in land (pp. 24-5). On the basis of this information it is possible to draw up two columns: left south black clans night junior inferior west setting darkness Mugwe f religious 1 [ authority J predecessors [ honey[ collecting

right north white clans day elder superior east rising light the Elders j political 1

I power J successors agriculture

This method gives rise to two sorts of difficulty. First, on the plane of each opposition, a concern to highlight complementarity may sometimes obliterate the references; and, when attention is paid to the problem of reference, it is sometimes located in the society, but without the different levels being distinguished; or again, a universal common sense, which is actually a form of ethnocentrism (cf. oppositions 6 and 8) may be invoked. Second, this heterogeneity makes it hard to see how to comprehend the overall connection

18

Introduction

between the various oppositions. I want now, on the basis of additional observations drawn from Bernardi's book, to comment further on these two kinds of difficulty. The Mugwe's left hand (i) harbours his power because, so it is said, it contains kiragu, & variable symbol, which is frequently not materialised, and which is held to be the common essence of all the Mugwe's sacred objects. The Mugwe is therefore placed 'on the left'. But while this is the emphasis in the Imenti sub-tribe, it is different in the Tharaka sub-tribe, and indeed in numerous other parts of the country (Bernardi pp. 72-3). The Mugwe there is 'a man with a tail', like an animal, and this tail contains the kiragu. The opposition between 'man with an animal tail' and ' "ordinary" man' would therefore seem to correspond, as far as the Mugwe's power is concerned, with the opposition between the left and right hand. But the former opposition can hardly be described in terms of complementarity, for there is the largest possible gap between man and animal. There is, moreover, a complete change of level between a domain which features men and animals as markers and one which has a being which is outside such classes and is neither a man nor an animal. This point will become clearer in what follows, but it already seems reasonable to wonder whether the 'left' aspect of the Mugwe among the Imenti might itself not be a reversal involving a change in level.

The north/south and right/left (2) oppositions would actually seem to be superimposable in certain contexts. But, in situations involving the Mugwe, the relation becomes more complicated. Among the Imenti, the Mugwe is not associated with this territorial and symbolic division, which has anyway disappeared. For the Tharaka, the Mugwe is 'on the south'. Opposite him, the people of the north, who are said to be 'on the right', are descended from the sister of the First Mugwe. For the Tigania, it is the other way round, and the Mugwe is 'on the north'. But in that sub-tribe there is in the south a double of the Mugwe, the Mukiama, who, it is sometimes said, has a 'comparable' but 'lesser' power. For the Igembe, there is no relation between the Mugwe and the north/south opposition, but a double exists, the Kiruria, who is 'comparable but lesser'. On the other hand, one must contrast the Igembe and the Tigania Mugwe, who is such for the duration of an age-class, with the Imenti and Tharaka one, who is Mugwe for life. The information provided by Bernardi is admittedly very succinct (cf. chaper 2), but it is clear that a binary table cannot do justice to its actual complexity.

Right/left oppositions

19

Bernardi and Needham both maintain, on the basis of the three terms black, white and red (3), that a ternary system must once have played an important role. But Needham, in his study of the oppositions, only preserves the white/black contrast that a binary perspective would suggest, noting the superiority of the white. White is aligned with the light, is naturally superior to the darkness and is on the north side, where the people (in the case of the Imenti) are very 'proud'. This perspective, which Needham adopts from the outset, also leads him to make a slight mistake, for he takes Bernardi to say that the white side has subsumed the red, in opposition to the black, which remains on its own, when Bernardi in fact wrote the opposite (there was a confusion between miiru, black, and mieru, white; it was the black and the red which, according to Bernardi, were not really differentiated (cf. Bernardi 1959, p. 9)). If one were to draw any conclusions from this asymmetry between the double pole of black and red and the single pole of white, it would be the black that would be superior. Moreover, one cannot maintain that there is an absolute connection between white and the idea of a natural preference for the light. Admittedly, the sun is 'source of all life' throughout this part of Africa, and the night is the exclusive domain of evil-doing spirits and of witches. But, wherever one encounters this opposition, one merely has to shift ritual context in order to discover cases of black being a sign of rain, i.e. a blessing from the ancestors, with white then being the nakedness of drought and the nothingness of famine, or else the murderous thunderbolt. If we come back to the Meru, and to the Imenti sub-tribe in particular, we are told that white is the colour of the people from the north, who are 'very proud' and who always claim precedence in grazing. But,.Bernardi tells us, it was this same pride that brought about their downfall. They were defeated, in a war, by the red clans and confined to the north (p. 9 n. 1). The 'reds' being associated with the 'blacks', this third argument for the superiority of white would also seem to be questionable.1 The problems regarding the north/south division (4) were touched upon during my discussion of oppositions (2) and (3). I shall refrain from commenting upon the question of the disparaged position of the women (5), for Bernardi's account of this is too brief. I shall simply note that it seems dubious, a priori, to equate the ritual prohibitions which exclude the women from the great ceremonies with their lesser courage in war, the latter being spoken of at the time of the colonisation of the Meru, when they said that they 'had become women'.

20

Introduction

The sun rises in the east, on the superior side, and sets at the place of the Mugwe (6). But, Bernardi informs us, the Mugwe and his house are regarded as immortal, and the sun sets there in order to renew the Mugwe's power each day. Mugwe and Mukana-Ruku are not opposed but are 'parallel5. They are light and immortality, each linked to the course of the sun. Bernardi, for want of any other information, therefore argues, on the basis of what he knows of the Mugwe's authority, for a simple association between the Mugwe and the west (pp. 73-4). Needham does not accept this interpretation but seeks to use the west in order to pass to the idea of darkness which, as is well known, is of less value than the light. The inferiority of the left column, and of the Mugwe, is thereby confirmed, but by means of two hypotheses for which there is no support in the ethnography, namely, that Mugwe = west and that darkness = inferiority. The 'political'/'religious' opposition (7) is, by the author's admission, derived from the prior consideration of all the other dualities. The analysis which results in the attribution of an inferior status to those who, at the time of the mythical immigration, crossed before the Mugwe (8), is only founded by positing the Mugwe as a point of reference. But this is to contradict the whole of the study, which describes the Mugwe as being only one of the two poles, and the less valorised one at that. It would also be doing less than justice to the persons who, on the Mugwe's own orders, were thus the first to cross: four persons arranged as two pairs and two age-classes. Rather than representing a first contingent of the people, there is every likelihood of this particular symbolic figure marking the founding, in a new country, of a new law, that of the Mugwe and of the new society. I shall present several examples of the relation which unites the Mugwe to totalising duality, and the reader will recall that it is common enough in this part of Africa for the representation of two children and/or young people of the opposite sex to indicate, by their presence, that a new principle is established. The harmony created by the conjunction of asymmetrical opposites replaces a past in which there was only death or nothingness or non-society: rites of birth or of marriage, enthronement and funeral ceremonies of chief or of king, and 'reconciliatory' sacrifices after an exchange of murders (as regards the Nyamwezi-Sukuma, cf. Cory 1951; Blohm 1933 text 163; Millroth 1965 p. 68). On the other hand, honey comes before agriculture. One would think, Needham says (i960, p. 25), that the former is not the basic

Right/left oppositions

21

sustenance and it is in fact described merely as one of the staple elements of the diet. But it is with myth that we are concerned here, since it is there that the honey age-class came first. And at this level, it is honey which is presented as the value: it founds the present hierarchy of the age-class (Bernardi 1959, p. 66) and it is the kiragu of the Tharaka (the symbol of power cf. opposition 1) (p. 43); the founding hero is the one who causes honey to be discovered (pp. 65-6). The importance of the hierarchical aspect of each element cannot help but invalidate any attempt to establish a homology between the elements grouped at either pole. The Mugwe is not always associated with the south, the 'white-north' pole is not necessarily superior, and 'darkness' does not have any part as such in the play of oppositions. It is simply spatial proximity in the binary table which has given rise to the notion of a link between religious authority, the (supposed) inferiority of the predecessors and the (alimentary) role of honey, and the table is itself based upon the dubious condition of allowing all symbols only a binary option as far as classification is concerned, either to the left or to the right. There is nothing in Meru society to justify this working hypothesis, and nothing suggests that symmetry or complementarity are valorised figures which serve to orient the symbolic material in this way. Indeed, it is important to stress that Needham constructs his argument as if the symmetrical mode (the equalitarian mode) was prevalent and as if the binary image of a complementarity was in itself a value that a preliminary analysis of the ritual would elicit. Only a presupposition of this sort can account for an approach in which homology is sought at any cost, as it were, and for a conclusion to the analysis of the duality of 'powers' that is quite unwarranted by the facts. The only evidence for this thesis is provided by Bernardi's remark about the 'tribal government' held by the Elders, but one would still be doing scant justice to his far more numerous observations, which I shall quote below, concerning the absolutely superior position of the Mugwe. Needham subscribes to the non-comparative aspect of Hertz's project, which involves the latter in attempting to define the superiority of a 'right' series, irrespective of ritual context, levels of value and the societies themselves. This enables one to define the valorisation of'left' in relation to the Mugwe as a 'problem'. But for Hertz the 'right' derives its superiority from that of the 'sacred'. At this level, the contradiction is absolute and one has to part company

22

Introduction

with Hertz at some point or another. Needham therefore chooses to make the Mugwe's 'religious power' inferior; the right then remains pre-eminent and it serves by contrast to define the 'political power' of the Elders who 'govern' the villages' affairs. The reader will appreciate that my concern here is with the different methods used to approach symbolism, and that I do not wish to impugn the motives of those involved in this debate. But once Needham has assumed that a relation of the right/left type is invariable, he cannot help but interpret a symbolic configuration of the sort presented by the Mugwe in terms of a 'choice'. Given the existence of a contradiction in the data, and a refusal to question the conceptual tools employed, it becomes necessary to invoke an a priori. Even though Hertz did not often specify their context, he gave so many examples of the 'pre-eminence of the right hand' that to have contested this principle would have occasioned too many difficulties. The only option remaining for Needham was to diminish the Mugwe's power and, since the Mugwe was religious, to turn his attention to the 'political'. Given this 'choice', Needham's approach to the other oppositions naturally follows. The questionable interpretation of the data may be explained in terms of what he was already predisposed to demonstrate. The 'duality of powers' does not follow from the analysis of the various oppositions but is a conception which derives from the a prioris of the method used. There is no direct evidence for an association between 'right' and the power of the Elders. In order to demonstrate that there is one, the author has therefore to make a detour and consider other oppositions. This involves employing the unjustified axiom that the whole of the symbolism is a dichotomous unity. One can thus grasp the necessary link that there must be between the employment of the binary method and the search for the duality of 'powers', between equalitarianism and the emphasis upon an autonomous 'political' domain. As Dumont has stressed (1978), the course of science sometimes involves a repetition of the historical process by which the constitution of Individualism had occasioned the emergence of the category of the political. What then are we to make of the Mugwe? If regarding him simply as one of the two poles of a general dichotomy seems to give rise to more problems than it resolves, it may be because the Meru, like so many other groups in this region, have a 'chief or 'king', an ntemi, mwami or mukama, who is not a priest or a man of power but

Right/left oppositions

23

the central term in the social and cosmological system, and who is therefore situated at a superior level of the symbolic configuration. There are several observations in Bernardi's book which serve to substantiate this point. It is worth noting, to begin with, the numerous quotations which present the Mugwe as the sole master of the Meru: 'He is our King, he is like the kabaka of Uganda' (p. VII); he is like 'the King of the English5 (p. VIII); 'he is our God, the custodian of the Country' (p. IX); 'The law of the country is laid down there' (p. 39), the Elders say. They add that 'He is the chief of the Meru and that there was no other' (pp. 40 and VII). He is the repository of Ugwe, an abstract term designating the power of the Mugwe and an entity which represents the totality, the society. The Elders have no equivalent term to designate what would be their 'power' as opposed to the mystical Ugwe. Finally, the Elders openly assert that it is through this Ugwe, of which each Mugwe of each Meru sub-tribe holds a sample, that the unity of all the Meru is realised and preserved (pp. 7-8). Needham is not unaware of these facts, and indeed refers to several of them. The central role of the Mugwe seems to him to be aptly summed up by the metaphor used by the Elders: 'The Mugwe is like a queen bee'. He is, in short, as this image and the above descriptions suggest, the central figure and the bearer of ultimate values. However, to emphasise the distinction between the ritual figure and the holders of'effective control', Needham notes that the Mugwe is a figure 'protected' by the Elders. But there is more to it than this. I would argue that the Mugwe is central (and the Ugwe an encompassing value), for he is marked by the 'two', which is understood to be the totality par excellence, the one which divides into two asymmetrical parts and which, when reconstituted, is something more than the simple juxtaposition of its parts, is in fact totalising duality. One of the narratives told by the Tharaka, one of the Meru sub-tribes, is particularly significant here. The first Mugwe is robbed of his power by his sons. They shut up their father in a large calabash and carry off the magical preparations. They are two in number, but they differ (one was initiated by his father, the other was not), and they share out their booty in an asymmetrical manner: ' . . . one went with a certain amount, and the other went with another amount. One went to Mbere, and one went to Ukamba. The medicines of the man of Mbere and more powerful ...' (pp. 68-g) 2 . The Mugwe is in the end delivered by his people; he

24

Introduction

curses his sons and recreates everything that has been stolen from him. There is reason to suppose that this renewed creation is not a simple return to a prior state. For the theft has at the same time signified an asymmetrical division of the Ugwe; once this latter is recreated, it has an explicitly totalising value. The disappearance of the son is the corollary of this change in level. Indeed, the narrative begins at the moment that the Mugwe appears to confide the secrets of his power to his initiated son. At the end of the narrative, the Mugwe no longer has a son, he is reborn symbolically by the will of his people (who draw him out of the calabash) and he recreates his power: an act of birth, undoubtedly, of an Ugwe which is from then on the value of the whole people. I want to conclude with a story which is also told by the Tharaka sub-tribe, the story of Iboka. He is a culture hero who, along with several others, is a mythical and localised double of the Mugwe (pp. 62-6), and is sometimes explicitly confused with him (p. 65). He was born by falling from the sky one night, along with the rain. Once grown up, he was named chief. Two horns sprang from his head. They guaranteed the establishment and the permanence of the society: Now (the Iboka is told) you have to look after the affairs of the Tharaka people, because those from the mountains (rogoro) destroy them! It happened, then that robbers, maitha, came to destroy, but when Iboka heard of that, it was then that he performed a great miracle. This is the miracle: he took his two horns and he put them into two warriors who were brave (njamba = heroes), in this way: he made them sleep, he slit them open into two parts; then he took one part from one warrior and he put it together with another part of the other man; he took one horn and he put it inside and the two parts became one. The man who was thus produced became a living man (The same he did with the other parts). It was then that the Tharaka were very happy, there was nofightingamong them, and infightingthe others those who were killed arose and pursued their killers destroying many of them (p. 64). The human group thus eliminates war from within itself and becomes society. It is no longer defeated in fights with its neighbours and thus guarantees its own permanence. This founding 'miracle5 is achieved by the chief who is able to divide what is a whole (his authority, contained in his two horns, and the human body, that of the warriors) and to create a superior unity by joining the asymmetrical opposites together again. The Nyamwezi are also familiar with

Right/left oppositions

25

the theme of the divided and reconstituted body, and the reader will find that my brief account (below) of the Nyamwezi 'blood price' confirms the argument advanced here. The two horns, the two parts of the body, lead us to think in terms of a whole. A first consequence is the emergence of hierarchical orderings. A second is to direct our attention to 'symbolic operations' which produce a change of level in the order of values: an asymmetrical division and a re totalisation (reconstitution of the body of the warriors). These operations are sometimes 'materially represented' in the symbolism of the Chief. This would seem to be how it is in the case of the Mugwe. I shall now try to give a more exact account of my hypotheses regarding Meru symbolism. Their sole purpose is to compare and contrast two principles employed in the study of symbolism, but without thereby implying anything in the way of a personal criticism. The examples which I have chosen mainly concern the Tharaka sub-tribe, whilst Needham took the Imenti sub-tribe as his point of departure. Moreover, I have not advanced a complete analysis of the Meru symbolic system, which would be the only possible means of deciding once and for all as to the true nature of the Mugwe's role. But this is not a crucial objection, for Needham's own perspective in his article is a theoretical rather than an ethnographic one. The present study is also meant, above all, to contrast two principles of investigation as applied to the same (incomplete) ethnographic material, as presented in Bernardi's book, and thereby to throw some light upon the contradictions to which the application of the binary principle gives rise. It is meant, in short, to bring out the radically distinct nature of the logic of symmetry, which is natural for us to invoke, by contrast with that of the symbolism of non-modern societies. I have not, however, devoted any space here to a summary of the final section of Needham's article nor to the long preface to the 1973 collection.3 The reader only has to refer to these texts to see that Needham does not attempt to conceal the questions to which his approach gives rise, and to register the qualifications he makes regarding his arguments: some facts, over and above those to which I have referred, serve to place in doubt the supposed inferiority of the predecessors in the mythical immigration (Needham i960 p. 27), along with the Mugwe-black association as a devalorised term (p. 28); the sacred nature of the Mugwe's left hand may be an exceptional case limited to the Imenti (p. 29) etc. The binary table is itself merely an expository device: 'a two-column scheme . . . is a

26

Introduction

mnemonic and suggestive device which simply brings together in a convenient and apt fashion the series of oppositions' (Needham 1973 p. XXIV) A column does not imply 'a category' (p. XXV) and if the Mugwe is on the same side as the women, he is not therefore feminine (Needham i960, p. 26). However, to prevent the whole of the binary analysis from becoming a mere intellectual game, two points should, as the 1973 introduction reminds us, be borne in mind. Binary oppositions are in themselves 'valid' and they can be 'systematically interrelated' by invoking 'dialectical analogy': the terms are not similar (the Mugwe is not feminine) but the liaisons which constitute each pair are analogous (Needham 1973 pp. XXVII and XXIX). I shall argue here that the cost of establishing the 'validity' of each opposition is the obliteration of reference, and that the 'analogy' between these different liaisons prevents one from constructing a structure in terms of levels. Such an approach allows one to take account of linear gradations between terms, but not of hierarchies between relations (a>b>c ..., a>b 'analogous' to c>d, as opposed to [a>b] subsuming the level [b>a]). It is clear that only a structure conceived in terms of levels can provide some explanation of the fact that, for a non-modern society, to class things signifies to order the world according to a hierarchy of values, and not to arrange terms according to an order dictated by the nature of each term. Analysis of the Nyamwezi case will enable me to develop these points further.

PART

ONE

Nyamwezi classifications

2 Nyamwezi classifications: right or left?

In this chapter I shall try to give some idea of the range of forms assumed by Nyamwezi taxonomies. Nyamwezi society is particularly interesting, in that one cannot help but see it in global terms, adopt a 'holistic' perspective, and therefore conceive of it as hierarchical. There are two linked features of this society which deserve special mention. First, the Nyamwezi, like so many living in 'stateless' societies, live entirely enclosed within the whole that relations with their ancestors (the mizimu) define. Tradition, law and values are vigilantly upheld by all the mizimu, and ritual activity essentially consists in sustaining this relation by means of a wide range of sacrifices and ceremonial exchanges (kuhoja mizimu). Second, Nyamwezi society differs from that of some of the other groups described above, in that it enjoys a degree of institutional stability. There is a 'divine' kingship, related to those of interlacustrine Africa, which guarantees the permanence of the society. These two defining features of Nyamwezi society are interconnected, in that the king is the representative on earth of the law of the ancestors. Kingship clothes itself in the attributes of the collective ancestrality, even to the extent of being confused with it. This is the reason why, although I have chosen the most outstanding rituals to elicit the taxonomies which are operative there, I do not propose to drift into formulating my analysis in terms of statutory oppositions ('the king's people' or nobles/commoners) or into considering conflicts in terms of 'power'. I shall instead return again and again, by way of kingship, to a consideration of the most general values of Nyamwezi ideology, namely, those which, in ordering relations between the living and the dead, and with the king as master of ceremonies, allow the society to feel some confidence that it will endure. Let me begin by discussing an opposition which emerges directly from the features of Nyamwezi society described above. It orientates 29

30

Nyamwezi classifications

the whole set of taxonomies. I shall not however treat it as a condition of the analysis, but shall ask the reader to bear it in mind as something which will emerge at the end of my argument. Local reference to close ancestors, the familial ancestors known as badugu, is subsumed within a higher order of reference, to the collective ancestors and royal ancestors whose representative the king, irrespective of his precise familial and dynastic origin, is. Each inhabitant of the kingdom is the child of his badugu ancestors, but also the 'child' of the king who is the 'father' of all. The royal ancestors, strictly speaking, are the dead kings on whose tombs the court sacrifices. To these one must add a number of great and legendary diviners and the culture heroes like Luhinda, to whom reference is made in invocations.1 This category of royal or 'great' {bahanya) ancestors is in some respect the named and visible part of a larger whole, that of the collective ancestors, the mizimu in general, 'the spirits of the dead'. Each person's badugu ancestors are thus integrated into a higher level, where the relation is formulated in terms of livestock and in terms of men. There is both exchange and levy of livestock. Exchange involves both the badugu livestock-ancestors of each group and their circulation, with the latter integrating all the local relations. Levy involves the court in taking livestock from specific exchanges. Where men are concerned, there are not only those who are born twins or in a reversed position (mabasa; kashindye), who are treated as kings and are considered even when alive to be ancestors who may bring prosperity to all, but also murderers (and those guilty of other transgressions sanctioned by the king) who flee to the court after the deed, thus rendering themselves inviolable and becoming, through a second sort of transgression which they then commit,2 surrogate children of the king. The ceremonial circulation of livestock articulates this hierarchical bipolarity. Marriage prices and blood prices, consecrations and sacrifices send livestock on a circuit involving a reference to the other (royal ancestors / familial ancestors). Being above all a 'child of livestock', a Nyamwezi is drawn into this flux in which hierarchy and ceremonial circulation constitute the basis of the society's permanence and of its renewal. I shall not give a detailed account here of the circular models which determine the ritual circuit, but shall simply consider the hierarchy of levels involved in it. I shall begin by giving an account of the duality of reference described above, which necessarily brings about a first distinction between the various contexts. My argument

Right or left?

31

will proceed in two stages. I will begin by invoking a limited part of the ethnographic data, and will adopt a relatively unrestrained approach to the comparison of different ritual contexts. This first stage will naturally give rise to a general left/right dichotomy. I shall then, in the second stage, present the remaining data, and thereby transform the apparently binary nature of each opposition, so as to reveal the existence of a system organised in terms of levels. The inclusion of extensive footnotes is designed to enable the reader to approach the text in two different ways, either to consider the general argument on its own or to consider the ethnographic material in its entirety. The Nyamwezi and the Sukuma belong to the same tribal group, and occupy a large part of the north-west of Tanzania. To the north they may be found as far as Lake Victoria, and their neighbours to the north-west are the Zinza and the Sumbwa. 3 Beyond these two groups, which only occupy a small area, lie the Haya, Burundi and Rwanda kingdoms. The Ha are to the west, then there is Lake Nyassa; to the south there are the Konongo and the Kimbu, to the east there are the Gogo, the Taturu and the Masai. It is a country of high plateaux (1300 m.), of poor soil, with a precarious agriculture. The sorghum and haricot bean crops depend upon there being regular rainfall. Rainfall is concentrated between November and May and thereby defines the period of cultivation. The division of the year into 'rainy season' and 'dry season' is fundamental to this society and serves as a natural framing of a major principle. The dry season is represented as the 'white' period of the blazing sun, and it is the time of marriage, of murder, of war, and of a principle which is itself a bearer of life, but through blood. It is opposed to the 'black' period of the rain, when peace is reestablished with the ancestors and life is renewed by the fructifying principle represented by the rain. The king symbolises the meeting of the two principles. A large number of cattle and goats are also reared, and are used in the ceremonial circuit of the nsavo (the 'price' of murder, of marriage etc.) and of sacrifices. The great royal sacrifices serve in fact to guarantee the return of the seasons and the transition between the dry and damp periods of the year. The Nyamwezi-Sukuma population is very large (two millions at present), but is divided into a great many chiefdoms (butemi or matjalo), which German ethnographers also called kingdoms. Prior to the colonial period, which is the period covered in the present

32

Nyamwezi classifications

study, there must have been around a hundred of them. They differed greatly both in area and in population. If one excepts a number of short-lived attempts in the south to establish a more general rule over the chiefdoms, such as the Empire of Mirambo (cf. Bennet 1971), attempts which may be understood in terms of the imbalances created by the Arab presence, by the ivory trade and the dealing in firearms, they were independent at the close of the nineteenth century. There was no federation and no chief had any authority over the others. Each chiefdom has its own legends of origin; their founding heroes each have different names and followed different paths prior to their installation as chiefs,4 which can however be understood in terms of the same model. Each chiefdom has a hierarchy and a differentiation of status among its inhabitants. They move about quite freely within the bounds of their own chiefdom and are thus independent of the neighbouring one, but this pattern is the same throughout the country. Each chiefdom is a society beyond which there lies the 'forbidden' mwiko exterior, the place of wandering spirits, the zone of hazardous journeys and of war. It therefore constitutes the basic unit of this study. Within this unit, the traditional ethnography distinguishes between the 'nobles' and the 'commoners'. I shall maintain this distinction but shall redefine it. It is not an economic distinction, still less does it have any feudal connotations. Moreover, aside from hereditary transmission, there are several other ways of acceding to a higher status. There is the chief (ntemi), whom I will call 'king', and all those who declare their 'royal kinship' with the reigning king or by tracing their descent back in an unbroken line to a famous dead king of the chiefdom. There are also those who have reclaimed land, attracted other families to their entourage, and who therefore define themselves as 'kinsmen through land' of the king. Finally, there are those who have forced the king to adopt them by breaking one of the major prohibitions, in a ritual form such that these acts represent a symbolic murder of the king.5 Those of commoner status only have a short genealogy, going back three or four generations, and including neither a king nor a reclaimer of lands nor an illustrious transgressor. Each person's relation to his ancestors and to the supernatural is a primordial one. Fortune and misfortune only come from the ancestors (although witches do, admittedly, sometimes contribute to misfortune). The relationship with the ancestors does not follow the rule for the exchange of gifts on a reciprocal basis; it is in fact a

Right or left?

33

perpetual debt. Happiness is a sign that an ancestor has not intervened, and unhappiness a sign that one must 'appease5 this latter (kuhoja). Daily existence and the chiefdom's great annual ceremonies alike involve a succession of libations, consecrations and sacrifices, all of which are attempts, invariably rapidly undermined, to 'restore the peace' {kuhoja). Each person is therefore preoccupied, not with trying to bring these cycles of appeasement to a close but with obtaining a guarantee that he will accede in his turn to the rank of ancestor. It would be regarded as impossible or as mad to desist from 'appeasing' an ancestor each time that misfortune strikes. It would be tantamount to saying that one was no longer a man and that one had no ascendants, or it would imply that these latter were all witches and none of them had therefore become ancestors. The only acceptable aim, and therefore a necessary one, is to become an ancestor oneself. To achieve this, one condition is vital: one must have grandchildren who will maintain a regular cult for their grandparents in a fixed place.6 If one fails to achieve this, one is condemned to be merely a wandering spirit who is debarred forever from eternity. One must therefore live, and continue to 'appease' the ancestors, if one is to see one's own children born and, above all, one's childrens' children. The expectation that each person thus has, from a local viewpoint, is duplicated at the global one by a belief in a kind of harmony of the world, which gives the society the strength to believe itself to be durable and to reproduce itself. The king serves as the guarantor of this Unity in which differences are hierarchised, in which the sequence of life and death dissolves and allows one to perceive permanence, the necessary illusion for the endless quest for 'peace'. On this plane, mhoja (the substantive of kuhoja) will be accessible, if not guaranteed, for it is a gift of the king's. The binary schemas studied below are heavily marked by this duality of reference, a duality which is hierarchised. The first principle, which requires the unrelenting practice of a cult, without any enduring result, which belongs to the law of the familial ancestors and is a local reference, is encompassed by the royal principle, a global reference, which is the permament symbolic assertion of a place beyond disjunctions. To understand these conceptions of the supernatural, and this symbolic guarantee provided by the royal person, one must refer them to another peculiarity of Nyamwezi society, namely, that the king is explicitly the 'father' of all the inhabitants. Each person must

34

Nyamwezi classifications

therefore also make a contribution to the cult required by the royal ancestors, i.e. the dead kings and particularly the most illustrious among them.7 This involves helping to cultivate the royal fields which will provide the sacred beer for the libations,8 and being prepared to accept the royal levies of a few heads of cattle destined later for the great sacrifices at the first fruits and harvest festivals.9 For, whilst it is vital to be at peace with one's ancestors if one is to resist a witch's attacks, to avoid sterility and to prevent one's field being devastated by animals, it is still more crucial that the rain come when it should, that the plants ripen, that a war be successful, that epidemics be kept at bay, and all these are the concern of the great royal ancestors. These beliefs serve as a whole to uphold the difference between royal and commoner status, opposing the royal ancestors ('upon whom everything depends') and the familial ancestors. This is to distinguish commoners, who see the two cults as being separate, and nobles, who, thanks to their 'royal kinship', are able to honour ancestors who are both their own ancestors and the ancestors of all. The system is thus founded upon a double reference: the local principle of badugu ancestors and of a linear and discontinuous temporal succession (ancestor-descendant, grandfather-grandson, an unending quest for mhoja within the repetition of the sacrifices) as opposed to the royal principle of Unity. The double cult, involving familial ancestors and royal ancestors, would seem to be an initial ritualisation of the two principles. It is also the point of articulation between the domain of values and the plane of noble and commoner status, making the king the central symbol of society and the officiant of the cult of those of noble status. 10 R I G H T / L E F T BINARY OPPOSITION

If one surveys the facts regarding right/left, man/woman, even/odd, black/white and king/priest oppositions in descriptions of Nyamwezi ritual, 11 it seems hard to derive a logical order from it, at any rate if one only considers their symmetrical relations two by two. Some of these facts would seem nevertheless to be amenable to this kind of treatment, and lend themselves to a characteristic two-column schema. I shall begin by adopting this approach, before considering the other data. This should be seen, however, as nothing more than a methodological ploy, which I have adopted

Right or left?

35

solely in order to imitate the model proposed by Rodney Needham, and in order to demonstrate its shortcomings. The first stage of the presentation will therefore involve arranging the dualist oppositions in terms of a binary model. One of the main 'relays' is, as is so often the case, the man/woman opposition. We find in this society a feature which is remarkable enough, at any rate as far as Hertz's theory is concerned, and which should have attracted attention to Nyamwezi society before now (setting aside the more general, but less obvious interest of the Nyamwezi case for the study of hierarchy), namely, the association of the man with the left. Ray Abrahams (1967), a specialist in this society for the colonial period, had noted that this was true of the NyamweziSukuma group as a whole, and that this feature also differentiated it from the rest of Tanganyika. Left/right <-» man/woman

The left/right division corresponds to that between the sexes. When one lays a dead person in the earth, care is taken to place one of his arms between his legs and to keep the other one vertical, as one throws earth over him. A man is buried with his left arm raised, while in the case of a woman it is her right arm that is kept vertical (Gass et al. 1973, p. 450; Blohm 1933, text 134). This rite is supposed to maintain the link between the ancestors and their descendants, a link symbolised by the permanence of the family herd. Indeed, people say that 'if the dead person is buried with arms lowered, the cows will die' (Gass et al., 1973, p. 450). With his arm raised and his index finger pointing, the dead man is supposed to be able to 'warn' the living to 'follow his words' and to honour him as he deserves, in order that the cows should 'live a long time' (Blohm 1933, text 134). Patrilateral/matrilateral <-> left/right

This relation to the ancestors adds a further dimension to the man-left/woman-right correlations. Each person can distinguish between those of his ancestors who are 'on the side of the bow' or 'of the head' (badugu ku buta; ku mutwe) and those who are 'on the side of the back' {ku mgongo). These are the patrilateral and matrilateral ancestors, respectively. Living relatives are distinguished, from ego's point of view, by the same terminology, and are divided into

36

Nyamwezi classifications

two badugu ('kinsmen') kingroups, ku buta {ku mutwe) and badugu ku mgongo. If someone has anxieties, either about his health or about something else, and is warned by a diviner that they are caused by the wrath of one of his father's ancestors, he should consecrate a bead bracelet and wear it on his left wrist. If he has to appease a maternal ancestor, he will wear this bracelet on his right wrist. If the illness is a grave one, an animal must be consecrated, in order to make it into a receptacle for the ancestor-'soul', and thus to succeed in 'fixing' this latter. The animal, which has been chosen by following a diviner's instructions, is brought to the sick person. If his father's ancestor is involved, the sick person touches the animal with his left hand, and the group's elder, who presides over the ceremony, makes all his gestures with his left hand (giving the animal something to drink, anointing it with Iwangaa mixture of flour and water). A month later, the sacrifiand (the sick person who had consecrated the animal), being cured, will scoop up a little water in his left hand and throw it over the animal. When the animal dies, !2 a piece ofits hide will be made into a bracelet and will be placed upon the sacrifiand's left wrist. In the case of a graver illness, one conducts a bloody sacrifice; the whole of the first part of the rite repeats the schema of the consecration. When these rites are addressed to an ancestor of someone's mother, it is the right hand which will be used for each of these gestures (cf. Bosch 1930, pp. 94-101). Indeed, the left hand is 'the hand of the bow', just as the paternal side is 'the side of the bow' or 'of the head', whilst the right hand is 'on the side of the back' (cf. Abrahams 1967b, p. 47; Blohm i933> t e x t s 54e> 54h> 54*; B<>sch 1930, pp. 73, 158, 221; see note 14). Patrilateral/matrilateral <-> man/woman

Rites of consecration and sacrifices bring us to the general masculine/feminine opposition. Male animals are sacrificed in honour of ancestors 'of the bow\ and female animals in honour of 'the side of the back\ The second ceremonies tend to concern women, while the first tend to concern men. But each may 'help the others in case of need' (Bosch 1930, pp. 99-100, 128-9). Even/odd «-> man/woman

The opposition between even and odd numbers is also homologous with the above oppositions. A newborn boy is laid upon four dry

Right or left?

37

fibres, a little girl on three. Mourning for a man lasts four days, or six days in certain regions.13 I have no information about the length of time one mourns for a woman, but odd numbers are closely associated with it. Thus, it is obligatory in this form of mourning to replace three stones in the entrance hall (Gass et al. 1973, p. 368). These stones are always arranged by a woman. They are not changed when a man dies, but the fire must be rekindled instead, and the central post in the hut replaced. A pregnant woman knows that, when she has missed her period twice, her pregnancy should last eight months if a boy is to be born and seven months if it is to be a girl (Blohm 1933, text 54a). It is also worth noting the opposition between the different kinds of rhythm that are beaten out on a drum. A man beats in a binary, a woman in a ternary mode (text 103). East/west«-» man/woman - 'head'/'back'

East and west are associated with the difference between man and woman and, as mediated by the 'head'/'back' opposition, with the two categories of ancestor, the patrilateral and the matrilateral. A man is buried 'with his back to the west', i.e. facing east. A woman faces west (Gass et al. 1973, p. 450). A sacrificial victim is stretched out, at the moment of its execution, with its head facing east and its rear towards the west (Bosch 1930, p. 158). The hindquarters, the rear - and the back - the lower part of the body, and the back of the hand, are the 'back' side, mgongo, while the head, the front and upper part of the body, and the palm of the hand are the 'head' side, mutwe. Likewise, the division of patrilateral and matrilateral kinsmen into 'people of the head' (or 'of the bow') and 'people of the back' is said to recall the two parts, the front and the back, of the sacrifical animal. This latter is often cut in two down the middle so as to mark both a rupture and the creation of a higher order tie between two individuals or two groups, particularly in the case of a relation having to do with murder. When the animal is killed, the group 'of the man' holds the animal by the head, whilst the group 'of the woman holds it by its hindquarters and its back paws (Blohm 1933, texts 54h and commentary, 54c, 163).14 I would add that the east is the side of Likube, of the Demiurge, of the Sun, 'of life'. The Sun is the spirit which dispenses light, heat and life; the world is a dwelling which is heated by the fire of the Sun. The west, on the other hand, is 'the country of night'; in

38

Nyamwezi classifications

addition, things that live at night are hostile (Blohm 1933, pp. 139-41; Millroth 1965, pp. 187-8).15 Black/white <-• right/left

Finally, it is worth mentioning the opposition between the colours black and white. Those bracelets which are worn on the left wrist, in honour of a patrilateral ancestor, are made of white beads. Those worn in honour of one of one's mother's ancestors are made of black beads and are knotted, as we have seen, round the right wrist (Bosch 1930, p. 101).

If we use, depending on circumstances, the relays of the man/ woman, left/right or 'head'/'back' opposition, we obtain two series: left man 16 patrilateral 'head' or 'bow' even east white

right woman matrilateral 'back5, hindquarters odd west black

One might end one's account here, emphasising the logical complementarity of these two columns and noting that the binary opposition is a distinctive and powerful analytic tool. RIGHT OR LEFT?

But if one subjects the data to a closer examination, whilst still searching for a general binary scheme, one will find the first intimations of hierarchy, for most of the contexts demand that one or other of the poles be given precedence. But how, in distinguishing the determining contexts which will enable one to choose the superior column, is one to avoid being arbitrary? In order to establish the pre-eminence of the right hand column, one can point to the valorisation of the 'back' and of the colour black. It is by sleeping on her back, an inhabitual position, that a woman conceives twins. This is what the sacred chants of the mabasa brotherhood, which is composed of the parents of twins, says: 'Let us go and sleep together, let us lie in the proper position, Kulwa and Doto [all twins are called by these names], I conceive these my children, lying on my back.' (Cory 1944). This position which is not

Right or left?

39

'proper' - which is actually reversed - is in reality valorised. For a twin birth, so long as the twins undergo all the rites prescribed (which are both numerous and costly for the parents), will guarantee great prosperity for the whole society and enables the parents to join the prestigious mabasa brotherhood. 17 In addition, it is by stretching himself out on the ground and on his back, before his people, that the king calls for rain when drought threatens the survival of all the inhabitants (Cory 1951, pp. 52 ff.). Black is in fact the symbol of rain, representing a sky obscured by clouds. In addition, when a new king is to be chosen, one examines a chicken's entrails: a white stomach indicates drought and famine, but if it is black in colour, the king will be a good one and will know how to bring rain (Cory 1951, p. 7). Here our right-hand column would prevail over the left-hand one. Yet other facts may be deployed to demonstrate the preeminence of the side 'of the bow', that of the east and of even numbers. Then it would be the left-hand column which would be valorised. Man is symbolised by the bow, woman by the midinho, a large wooden spoon and an essential cooking utensil. If a dispute occurs between husband and wife over a particular decision, the man will assert the superiority of his own point of view by asking: 'Are there two people who carry a bow?' (Blohm 1933, text 54I1). During a sacrifice, one invokes the Demiurge by presenting the victim to all four points of the compass. One begins with the east, saying: 'Thou who art the Creator . . . ' ; the other points of the compass have less symbolic value (Bosch 1938-9, p. 90; Gass et al. 1973, p. 388; Millroth 1965, p. 125; cf. below, chapter 4 note 12). I have already pointed out this superiority of the east over the west. Finally, in one's genealogical relationship with the first ancestor, it is more prestigious to be an even number removed from him (bizukulu, 'grand-son', i.e. (211) X great-grandson) than to be an odd number removed from him (ngwanangwa, son, great-grandson, etc.). There are many other contradictory valorisations which I could mention, and I shall comment below upon the problem of the relativity of the context, a factor which prevents one from attributing a position within a hierarchy to a whole column, whether left or right. But I want now to end this first tentative sketch, which is exclusively concerned with binary oppositions, and by presenting some information about the dignitaries of Nyamwezi society, those who enjoy the superior status of'nobles'. One might then suppose it to

40

Nyamwezi classifications

be possible to add an opposition between powers of a 'political' type and 'religious' powers. There is another figure, besides the king, who would seem to occupy a category of his own. This is the high priest, master of the great ceremonies and 'master of prosperity', 'king of peace with the ancestors', ntemi wa mhoja or ngabe. He supervises the divination ceremony by which the king is chosen, teaches him the sacred history of the dynasty and supervises the sacrifices on the royal tombs. He 'sleeps between two men . . . He is shaved with eight razors . . . He eats eight mouthfuls of sorghum and eight bitter bananas. If someone is left-handed and therefore accustomed to picking things up with his left hand, he cannot become ngabe* (Gass et al. 1973, pp. 343-4; cf. below chapter 4 notes 1 and 12). This master of the 'sacred' is thus clearly on the left and even side. Perhaps he carries out the essential ritual acts with his left hand in order to distinguish them from the daily acts performed with his right hand, in a manner reminiscent of the Meru Mugwe. This is why he cannot be 'naturally' left-handed. To complement this figure, one would expect to find a king whose distinguishing marks have to do with the right and with odd numbers. There is evidence to support this hypothesis. Thus, if we examine the various ceremonial payments, we discover that payments made to the court invariably involve the number 'five' (Bosch 1930, pp. 66, 80; Blohm 1933, text 161; Cory 1953, p. 113; Cory 1954, p. 22). On the other hand, the crucial royal emblem, the ndeji (two bracelets), was worn by the king on his right wrist and on his right ankle (Cory ig5i,p. 18).18 Finally, it is on the little finger of his right hand that the king wears the ring made from the hide of a bull solemnly sacrificed at the great feast of the agricultural cycle (Cory 1951, p. 44). I must bring my binary interpretation of Nyamwezi-Sukuma symbolism to a halt here. There would, admittedly, seem to be a set of paired oppositions linked to the universal right/left spatial opposition and, possibly, to a distinction between two types of power. But the facts still to be considered demand that we acknowledge a hierarchy of symbolic levels in which the dualist oppositions are deployed in a different manner. I have already stressed how difficult it is to identify the determining context which enables one to specify the value. But this is actually a false way of posing the problem for, where a hierarchy is involved, a direct link between various contexts becomes impos-

Right or left?

41

sible. Our investigation demands that we oscillate constantly between the situations in which the fundamental values feature, in which the whole of society is staged, and the subtle variations asymmetries, reversals, totalisations - of this or that dual opposition. Once we have specified the references which are essential to a society, and the reordering of contexts which each of these references organise, the difference in value between the poles of a particular opposition may easily be grasped as the disposition of two elements in relation to a whole. It is the manifestation of a relation of order within a given set. Once we have clarified the hierarchy of these references amongst themselves, a new arrangement (e.g. reversal), affecting the terms of a particular opposition, when one passes from one context to another, shows how heterogeneous the references implied by each context are and indicates a shift towards hierarchy. I will present an example of the procedure I intend to follow. I began by describing two principles informing Nyamwezi society's relation to the supernatural, namely, the principle of royal ancestrality and that of close ancestors. The former expresses harmony and union within permanence. The latter requires the temporal renewal of disjunctions (living person/ancestor; endless cycle of ceremonies of 'appeasement'). It is encompassed by the former. In a family in which the man has given a particular bride price to his wife's group, the children 'are his'. 19 At their birth, both boys and girls are presented to the family and to the neighbours holding their respective symbols (bow, wooden spoon) in their left hand, which is the hand of'the bow' and is on the father's side. But at a marriage, where a relation is established which refers to the royal principle of 'the union of the head and the back', the two fiances stood side by side, the woman on the right of the man. Their 'centre', which at certain moments in the rite is personified by an old woman who is a double of the king (see below, chapter 3 note 3), is society. The reference has changed, the girl being 'on the left' in relation to her father. Having become a woman, she is on the right of the husband. Moreover, from then on she carries the wooden spoon in her right hand, during the rituals (see note 14, above). As a young girl she had been a 'wooden spoon'. Once she has become a mother, she is a 'back', a 'right side'. Each opposition involves a constraint and two possibilities. The constraint entails an asymmetry in the value of the terms, in a given context, a constraint which is characteristic of a society in which, in

42

Nyamwezi classifications

general, the hierarchical mode encompasses the equalitarian one. The possibilities involve locating the terms again, at another level, either in a reversed symmetry or as a single term which is their totalisation. My reexamination of each of the dualist oppositions and my analysis of some new data will be conducted in accordance with these principles.

3 Right and left: the distinction between levels

RIGHT/LEFT, MAN/WOMAN, EVEN/ODD

Right/left

If we reconsider this opposition, we find that it does not provide us with a dichotomous model but employs reversal to oppose levels. I have just argued this point in relation to marriage; it is also apparent in the highlighting of situations of the kind that certain important ancestors require, in which right and left are associated. In case of misfortune, it is sometimes appropriate to honour particular ancestors, namely, twins and persons born in a reversed position (the kashindye, who 'are born showing their feet'). There are distinctions within these two categories but together they are clearly differentiated from all the other ancestors. A particular type of majinga calabash is especially associated with them. One fills these mafinga with Iwanga (flour and water) and anoints the sick person. The person who does the anointing is the mhoja (a commoner who is equivalent to the ntemi wa mhoja mentioned above, the master, at court, of'peace' with the ancestors) - on this occasion one of the sick person's aunts. Ordinary anointing, for instance to protect someone who is setting out on a journey, which does not involve the invocation of ancestors of a particular rank, is done on the forehead, throat and sometimes on both shoulders. It is one's father who performs it. For other acts of consecration, of animals, and for sacrifices, the mhoja is an uncle (Bosch 1930, pp. 95, 99-100, 103, 146), but here a woman, an aunt, is chosen. The paternal/maternal opposition recurs, for one's father's sister or one's mother's sister is chosen, depending upon whether the kashindye ancestor or the child born a twin is on the father's or the mother's side. The superior rank of an ancestor is indicated by moistening the sick person's two cheeks, his two shoulders, the middle of his chest, and the middle of his back, right between his shoulders. Within this 43

44 ordinary ancestor

Nyamwezi classifications Table 3.1 slight illness: father's side: mother's side:

bracelet on the left wrist white beads right wrist black beads

serious illness:

superior ('royal') ancestor

father's side: mother's side:

touching and anointing with the left hand with the right hand

kashindye rank:

an aunt on the father's or on the mother's side the two sides of the sick person anointing with one hand

superior rank of twins:

an aunt on the father's or on the mother's side the two sides of the sick person with two hands

superior level configuration, special provision is made for kashindye (reversed birth) twins. The mhoja anoints the two sides of the sick person with the one hand, which is left or right depending upon whether the ancestor is on the father's or the mother's side. If twins are involved, the mhoja uses two hands (Bosch 1930, pp. 120-1). I should add that twins and kashindye^ even those whose families are of commoner status, become ancestors of royal rank (pp. 45, 65, 119; cf. chapter 2, note 2). They belong to the superior domain marked by the number 'two'. There is a further subdivision in terms of this same 2/1 principle, with situations in which two hands are used to anoint the two sides of the sick person being contrasted with those in which just one hand is used for the two sides. The ordinary form of anointing, clearly lower in status, is the one where one hand is used on one side (the front) of the sick person. Table 3.1 will serve as a summary of what I have said so far, including the consecration of a bracelet or of an animal for an ordinary ancestor (on pp. 35-6, 38): In the first frame of reference (ordinary ancestor), only univocal features - such as right or left, black or white - are employed. It is a question of distinguishing the father from the mother or, more precisely, of ordering the two sides hierarchically. The side to be invoked depends upon the identity of the sick person and upon whether they belong, because of the type of marriage they have had (with or without a 'price'), to a house in which the father or the mother's brother is dominant. In the second frame of reference (superior ancestor), there is univocity also, and for the same

Distinction between levels

45

reasons, but it only involves one factor (brother's sister or mother's sister). A reversal, however, occurs: the mhoja, who is usually the sick person's father or uncle, is in this case a woman. Most important of all, there is a dual feature (when kashindyt are involved): the two sides of the sick person. For twins, two dual features are necessary: the two sides and the two hands. This example serves to show how impossible it is to reduce left and right to just one dichotomy. One cannot dispense with general values and with the hierarchy of references. One must be aware of the importance and the form of the relation to the supernatural, together with the hierarchy of principles concerning royal ancestors/familial ancestors. One is then in a position to understand the homology, which is asserted elsewhere, between kashindye, twins and dead kings. In addition, this hierarchical passage sheds light on the man —> woman reversal in the person of the mhoja. Man/woman

Let me recall some of the features of this opposition. I have already noted that it seems to be connected to the opposition between paternal and maternal kin-groups, being correlated with it either directly or through the mediation of the left/right opposition. It is usually the men who sacrifice to the patrilateral ancestors, and it is they who choose the male animals. The women do the reverse. But a closer examination proves this apparent equilibrium to be an illusion. We learnt just now that the person who consecrates is a man or a woman, depending upon whether a 'royal' or an ordinary reference is involved, and irrespective of patrilateral or matrilateral categories. If an ordinary reference is involved, the father performs the customary consecrations and the father's brother or mother's brother (depending upon the ancestor's side of the family) performs sacrifices. If a 'royal' reference is involved, the father's sister or mother's sister performs these things. Furthermore, in sacrifices organised by the court, the victims are always male; the opposition between man, patrilateral ancestors and male victim, on the one hand, and woman, matrilateral ancestors and female victim, on the other, no longer applies. This is a point to which I shall return below.1 Thus, the use of the left/right opposition as an intermediary does not entitle one to posit a direct and general relation between, on the one hand, the two groups of ancestors (the father's side and the

46

Nyamwezi classifications

mother's side, the 'head' side and the 'back' side) and, on the other, the two categories man/woman. I began by describing, as a typical example, the passage from childhood to marriage, in order to establish that man and woman change reference when, abandoning the status of the child 'of one's father', they together become a totality at the moment of marriage: the conjunction of the 'head' and of the 'back'. There is no equivalence between the children, where the boys each hold a bow in their left hands and the girls hold a wooden spoon, and the young newly married persons, who are placed on the left and right of the officiant and who become 'head' (or 'bow') and 'back' side.2 There is a difference in level and this gap is marked by a reversal staged at the crucial moment of the ceremony. The married persons enact an exchange of sorghum flour balls, which is reminiscent of the Christian exchange of rings. But this exchange is not carried out directly. It is the officiant who, by crossing over his hands, presents each person with his sorghum ball in such a way that the ball 'on the left' becomes the ball 'on the right', and vice versa.3 Even/odd

We have seen how a man is represented by the numbers 'two', 'four' and 'eight', and sometimes by 'six'; a woman is associated with the numbers 'three' and 'seven'. If one disregards the different contexts that are involved, there would seem to be a homology between the even numbers and the masculine and the odd numbers and the feminine. But here too a consideration of other factors will lead us to posit the existence of a construction involving several levels. First of all, it is not a simple distinction between two and one, four and three, eight and seven. Two, four and eight suggest a termination, a conclusion or a completion of one, three and seven. When one has undertaken to offer some beer to the ancestors, one prepares it for seven days (it is the woman who is concerned with this). On the eighth day, one can perform a libation (it is the man who does this) (Bosch 1930, p. 162). When twins are born, their family is in a state of grave danger. Numerous rites, in the presence of the king, will however transform this 'danger' into a public blessing. But the parents must bring two other children into the world before they can be completely purified. The first of these, who is the third if one counts the twins, is called sija, the 'remaining one': 'For the birth of twins has brought about a fear which still continues, which persists'.

Distinction between levels

47

A second birth is therefore necessary: this child will be the mhoja, the 'pacifier', and his birth 'brings peace to the family again' (Bosch I93<>> P- JI 9)« Another, or rather a complementary interpretation is possible. We are involved here, with twins, with the number 'two', at a higher level. At the first level, 'plus one' is sufficient to 'complete', if at any rate one has to pass from one to two, from three to four or from seven to eight. At the higher level, and above all in the context of twinship, where the number 'two' is involved at every moment of the rites, one does not count in ones but in twos, sometimes in fours and eights and, as we shall see, in fives, tens and fifties. We therefore complete here by adding two. And anyway this fact is a logical consequence of the coherence of the system: 'plus one' added to two (the twins) will bring us not only to the lower level of completion by means of'plus one', but to the incomplete and devalorised pole of this level, the number 'three'. There is another festival of'appeasement' of the ancestors which lasts three days; on the fourth day, 'peace' comes (Gass et aL 1973, p. 405).4 I ought also to mention the three hearth stones, whose association with the women I have already noted, and which are invariably associated with a cooking-pot resting upon them (which is associated with the men), like 'a crow on the branches of a tree' (Gass et al. 1973, p. 372). The totality of the house is also defined by these three stones, to which must be added the central column (which is associated with the men) (Gass et al. 1973. PP- 367-8). When, with the onset of the new season, the first furrow is dug in the field and one has thus 'torn the earth', one must go back to one's own house and allow a whole day to pass without doing anything at all. On the second day, one can safely proceed with the sowing (Blohm 1933, text 223). It is worth adding the following observation at this point: a man who has a child becomes independent (of his father-in-law in particular), the master of his house, etc., but it is only when he has had two children that he becomes a 'big' man {muhanya), a man whose opinion will be heeded. It is also worth bearing in mind that married persons did not accept the second sorghum ball offered to them by the kanumba (cf. note 3). Thus, two, four and eight are not only distinguished from one, three and seven in that they are associated with men, whereas the odd numbers are associated with women. Two, four and eight also complete one, three and seven, and in a hierarchical fashion, for the level at which 'peace' is re-established with the ancestors is superior

48

Nyamwezi classifications

to that at which impurity was occasioned by the breach of the prohibition. Three plus one becomes a four which, in a sense, is both more and something other than simply three units plus one unit. Two, four and eight encompass one, three and seven, in the sense that a complete state encompasses its incomplete contrary. This mathematics, in contrast to our own, might be written as 4 > 3 + 1. The level attained by the hierarchical operation of completion is more and something other than arithmetical addition. More precisely, 4 = 3 + 1, but this 'four' must be distinguished from the simple sum of four units, so that 4 > 4 X 1. In discussing twins, I noted the superior level of 'two'. One should bear in mind that, in this society, the king is the symbol of the totality and that, as the Nyamwezi say, 'twins are kings'. Numerous rites indicate, on the one hand, the bivalent character of royal authority and, on the other, the relation between this bivalence and the dual nature of twinship. It is not therefore surprising to find that the number 'two' plays a prominent role here. It only features to a very small degree in the man/woman opposition (cf. p. 35), but it has a more important part to play in the complete/incomplete hierarchical opposition, through which is effected the passage, by means of a ritual, to 'peace' with the ancestors. But two is most strongly associated with the king: — —

— —

the king always gives two head of cattle to the diviner whose services he requires; the king levies two head of cattle from blood prices given wholly in cattle. During the colonial period, when the king had ceased to be a symbol of the totality and had become a mere official, he only levied one head of cattle (Cory 1954, p. 10); the king is the only person to carry two spears (Tanner 1957, p. 207); when a man finds a pangolin, he takes it home. The chief of the residential group (kaya) is 'glad' and kills a goat. The man then takes the animal to the chief of the village, who also kills a goat. Finally, he takes the pangolin to the king, who has two goats killed and has the pangolin impaled. It then becomes a public mascot (Wright, 1954, p. 71).

This handful of examples would seem to lend credence to the theme of completion that we have encountered elsewhere. A man who is not king only gives one head of cattle to a diviner; he only owns one lance; he only kills one goat in order to mark the discovery

Distinction between levels

49

of a pangolin. Study of the colonial epoch clearly shows that, with the installation of indirect rule, royal power becomes symbolically univocal: he is a king of sorts, whose authority no longer involves the bivalent aspect which used to characterise it, and who only levies a single head of cattle from the blood prices. 'One' therefore does not suffice, 'two' are necessary. But there is more to the bivalence of kingship than that. In those instances in which one, three and seven were opposed to two, four and eight, the same actor, within the one ritual framework, reestablished 'peace' with his ancestors (but only briefly). There is a change, admittedly, but only in terms of the local reference to familial ancestors. Here, on the other hand, there is a distance which cannot be bridged, and an absolute discontinuity between the royal two and the one that is the attribute of the others (commoners and nobles together; it is not a matter of status). For this two has another nature, which only belongs to society as a whole, at a global level, and which involves reuniting, at a superior level, the asymmetrical opposites. A last example will serve to demonstrate this. A crucial sacrifice, performed in the presence of the king, will halt the vendetta. It involves cutting the animal victim 'in two'. I have already noted that it is after this sacrificial division that the two categories of ancestors and parents are named. This rite serves to reconcile two groups who exchanged murders, and to transform them into a pair, one of them on the 'head' side and one on the 'back' side: one group holds the animal by the head, the other by the rear and, at the king's command, it is 'cut in two'. The 'reconciliation' is performed and must be accepted. For 'head' and back' are only distinguished in order to be joined together again. There is thus a sense in which, at the end of the sacrifice, 'the two groups only constitute one' (Millroth 1965, p. 68). The fundamental value of the number 'two', which is a totalising duality, is thus clear. The superiority of the two implies logically that, in the previous opposition (complete/incomplete), the twofour-eight series encompasses the one-three-seven one and not vice versa {a priori one might have had the notion that two completes three, six to complete seven, etc.). It also implies that it is this series which, in some contexts, denotes the man/woman hierarchy. But if two is the value which thus organises the hierarchy of ordinary contexts, the infrequency of this number at this level may also be accounted for by the fact that it belongs to the superior domain. It is

50

Nyamwezi classifications

an index of global reference which serves to valorise the two-foureight series in local references whilst at the same time featuring very little there. This royal two is also, as I have noted, the two of twins. But how does asymmetry manifest itself in twinship? If new-born twins threaten the whole country at their birth, on account of the monstruous nature of double birth (which is also an imitation or a mockery of royal bivalence), they are a token of prosperity once they have been through the rites. But they have then become different, with one thenceforth being called Kulwa (the 'big'), the other Doto (the 'little') (Bosch 1930, p. 119; cf. chapter 4, note 2). We still have to discuss the number five. It is not to be found in the one-three-seven series. Indeed, as far as the particular symbolic system we are studying here is concerned, it is of another order altogether. There is a sense in which it is not an odd number. It is the unit used in ceremonial prices, and it mainly defines the masumula payment which one makes to the king for breaches of public prohibitions (mwiko: prohibition which concerns the community and which is sanctioned by royal authority): five objects whose actual nature is of secondary importance. The important thing is to give 'the five' (cf. Bosch 1930, p. 66, 80; Blohm 1933, text 161; Cory 1953, p. 113; Cory 1954, p. 22). The great majority of the other payments, which entail exchanges between commoners, are counted in multiples of five; and, in Nyamwezi society, exchange (along with 'price') is located on the superior level - we have had intimations of this in relation to marriage (Blohm 1933, text 161; Bosch 1930, pp. 507-8; Gass et al. 1973, pp. 373-4; Cory 1953, paragraphs 156, 176, 271, 300b, 313b; Holmes and Kadase Mange 1971, p. 136; Varkevisser 1973, p. 104).5 Generally speaking, it is clear that the unit of the 'prices' (nsavo) is the 'five', i.e. the male bovine; the goat, which is used in ordinary situations and whose value is 'one', is conceived rather as one fifth of the 'five'. The 'five' is the image of the king: the male bovine is the appropriate animal for royal sacrifice and the one which, in consecrations, represents the dead kings. A black bull is kept in perpetuity as a consecrated animal: it is the 'elder king', ntemi nkulu, the image of the first king of the kingdom. The convergence of the two contexts confirms the existence of a link between the 'five' and royalty: when the king stays with the main queen, he remains there for five days; with the other queens he only stays two days (Blohm 1933, text 105). For an ordinary man with several wives, the

Distinction between levels

51

lengths of stay are four and two days respectively (Blohm 1933, text 67). Is this five, like the royal two, a symbol of totality? The detail of 'blood price' proves that it is. In cases where this price is given entirely in cattle, it generally amounts to as much as fifteen heads, with the king levying two of them, as I noted above. In other cases, and particularly in those regions in the centre and in the south where bovine livestock is less plentiful, the king orders murderers to bring the 'fifty': various objects and two head of cattle. On the basis of a detailed examination of the rite (cf. Tcherkezoff, 1986), two observations may be made. First, the number fifty does not represent the exact sum of objects and livestock brought. Second, it is a symbolic totality - that of the body of the murdered person - and there is therefore obviously no need for the addition to be an exact one. The king distributes the objects to the kin of the victim and accompanies each gesture with these words: 'This is for the feet of the murdered person, this is for the legs . . . ; that is for the eyes, the nose . . . that is for the chest ' He thus reassembles the whole body bit by bit, recreating the 'head' part (the upper part) on the one hand, and the 'back' part (the lower part) on the other: the objects given for the eyes, the nose, the chest etc. are given to the father, those for the feet, the legs, etc., to the mother. If the five does not belong to the one-three-seven series, it is because it is not a five as opposed to a six, for example, but a union, i.e. a royal index: 2 + 3, 'head' and 'back', male and female. Nyamwezi number symbolism thus seems complex but coherent, so long as one sets aside the laws of arithmetic. There is no linear and continuous progression from one to fifty, but jumps between levels: 1 - 3 - 7 / ( 2 ) - 4 - 8 / ' r o y a l ' 2 - 5 - ( 1 0 - 1 5 - 2 0 - 2 5 ) - 5 0 . There are different values for the same number, depending upon the level: ordinary 2 or 'royal' (totalising) 2.6 There is no symmetry between even and odd: the woman/man opposition only involves 3 - 7 / (2) 4 - 8 and it is hierarchical, 3 ~ 7 < 2 - 4 - 8 . 7 BLACK/WHITE

I shall end with a discussion of colour opposition, in which the changes of level marked by reversal and totalisation also feature. Along with the 'head'/'back' opposition, it is the one which has the largest place in the system. I noted above that the two colours, used for the consecrated bead bracelets called maholero, enable one to

52

Nyamwezi classifications

distinguish between the two categories of ancestor, the patrilateral and the matrilateral. Both the disjunction between the two colours, and their relative symmetry, must be emphasised. We have to do here with a brief binary series: —

—

the men consecrate and sacrifice to their ancestors 'of the head' (i.e. their patrilineal ancestors) male animals and white beads; they wear, on their left wrist, bracelets made of these beads or of the skin of animal victims; the women invoke their ancestors 'of the back' with female victims and black beads; they wear maholero jewels on their right wrists.

This dichotomy is valid; it refers exclusively to a single context, that of familial sacrifice. This disjunction disappears, however, in the case of royal sacrifice. This symmetry, which is defined as such relative to the characteristics of the superior level (see below), does not therefore imply that men and women have equal status in their relations to the ancestors. The 'head' ('bow') side would seem to dominate the organisation of the sacrifice (the victim's head points east) and in the sharing of conjugal authority: 'Are there two persons carrying a bow?' (where a marriage involving virilocal residence is concerned, which implies that a part of the bride price has been paid). It also dominates the transmission of individual prohibitions (mgilo): children only have to respect those who come to them from their paternal ancestors (Bosch 1930, pp. 54-6) .8 To arrive at a clear understanding of the different values attributed to the colours black and white, we must examine those contexts in which the collectivity is directly involved. We have repeatedly seen that this is true of the royal domain. Two configurations feature in it, namely, the supremacy of the colour black and/or that of the combination of black and white. I noted in the introduction that the Nyamwezi year is divided into two periods, the dry season, symbolised by the colour white, and the wet season, symbolised by black. It is the whiteness of the sun which dries things out: as we have already learnt, white spots on the body or entrails of a chicken examined during a royal enthronement give promise of'a dry earth'. In the initiatory hut (see below), this season is represented by a white circle. The colour black, on the other hand, represents clouds swollen with rain: if the stomach or the crop of the chicken are, in certain parts, very dark 'like rain clouds', a reign

Distinction between levels

53

with an abundance of rain is guaranteed. In the initiatory hut, this season is represented by a black circle. The sun may well be 'light and life', but Nyamwezi definitions of the king place much greater emphasis on controlling the rain. The majority of royal rites involve a staging of this theme, even when the rite in question is not explicitly aimed at invoking the rain (enthronement, twin birth, expulsion of epidemics). This is the reason why, when divination is used to choose a king, black prevails over white. In the hierarchical list of animals associated with the court, one first encounters the lion, which represents the king, then the leopard and the lynx, which represent high-ranking dignitaries (ng'wanang'wa and mwizukulu), then the jackal, which represents the 'people of the court' (ban'ikulu). A hunter who has brought the hide of a lion to the king is rewarded with a black cloth in return. 9 Someone who brings a lynx to a mwizukulu receives white pearls. We can thus discern the outlines, on the collective plane, of a hierarchy in which black > white. I want now to compare familial and royal sacrifices. In the former, we have already seen that the animals are male or female depending upon the category of the ancestor and/or that of sacrificer involved. The colour of their coat is not irrelevant either, in that the diviner always prescribes a particular combination. Although many different ones would seem to be possible, we can discern a logic. For ordinary cases, the diviner prescribes a black caprine; for more serious cases, he prescribes a coat 'of two furs', black with white spots. This latter appearance is obligatory for the sacrifice of a bovine, which is itself only prescribed in serious cases; it is sometimes required for a caprine. We shall see how familial sacrifice cannot in fact involve a completely black bovine. The coat 'of two furs' is sometimes defined more exactly, and its asymmetry is further accentuated: a single white spot on the forehead is added. This latter detail is more frequent, however, in royal sacrifices than in those performed by the local groups, in which the sacrifice of a bovine is a rare and extremely serious affair (Bosch 1930, pp. 94, 128; Bosch x 938-9> P- 8 9; Millroth 1965, p. 160). In royal sacrifices, the victim must always be male. Caprines do not feature. It is rams and bulls that are sacrificed. Royal sacrifices never involve the black animals used in the ordinary ones. It is replaced by the animal 'of two furs' and, most crucially, with one white spot on its forehead. The commonest royal sacrifices would seem to be those that the king offers to his close ancestors, his badugu,

54

Nyamwezi classifications

before the familial altars of the court. Furthermore, for 'the solemn sacrifices which are held at the kings' tombs to demand rain', totally black animals are chosen. In addition, the cult objects and the participants' clothes are black. Sometimes other constraints are added, which distinguish royal sacrifices still further from the black goat of the ordinary familial sacrifice: the victim, whether, be it remembered, ovine or bovine, must have been born at night (Bosch 1930, p. 128; Millroth 1965, pp. 135-7; C o r Y I95 1 )We thus have an initial hierarchy, in which two colours > one colour, 'two furs' > 'one fur', double pole > univocal pole, of the sort that we have already encountered in relation to twins and reverse births (the system of hands for the anointments) and in relation to the numbers five / / two / three. This totalisation is reversed in the extreme case of solemn sacrifices: sacrifices for rain, for enthronement, for royal funerals, for the new fire, and against epidemics in which expulsion of threats of drought remain crucial, require the dominance of a single pole, the black one. The 'two furs' / 'one fur' opposition is a hierarchy. From our analysis of Nyamwezi number symbolism we know that it is not by chance that the number 'two' features here. It is also significant that we encounter the expression 'two furs' for different kinds of animal coats, those in which the colours are relatively evenly distributed (with the head and flanks black and the back and stomach white (Millroth 1965, p. 160)) and those which have a marked asymmetry (a white spot on the forehead). The 'royal' nature of this kind of duality (which may be encountered in familial sacrifices, but always in a superior position) is further emphasised by the image of the dog. For the diviner will sometimes prescribe the sacrifice of a dog (cases of this are rare, and are not commented upon). The animal to be sacrificed must have 'four eyes', i.e. a clear spot above each eye. This is the same dog that accompanies the king during his initiatory retreat at the beginning of his enthronement (Abrahams 1967, P- 123). Other examples further emphasise the importance of the double pole. A white snake is considered to be an ordinary ancestor. 'That's our ndugu (our parent)', they say. But if the snake is black with a white spot on the head, it is a dead king, and great joy is thought to be in store for the whole community (Bosch 1930, pp. 81-5). If we consider the beginning and the end of the reign, we find that the first act is linked to the colour black and the last to white, or more exactly, to white and to black. In fact, once the king has been

Distinction between levels

55

designated such, he is seized and wrapped up in a large black cloth. He is hit with sticks by court dignitaries, who say to him that from then on no one, under pain of death, shall strike him. It is this phrase that invests him with royal authority (Cory 1951, p. 9). When he is old and sick, he is strangled and placed in the ground with his arm in a raised position. When he is completely covered up and only his hand obtrudes, eight grains of white sorghum are placed in it. But black still features, inasmuch as the shroud is also a large black or blue cloth (these two colours seem to be treated as equivalent) or the hide of a black bull (Cory 1951, pp. 9, 11; Blohm 1933, texts 97, 147)Confronted with this royal duality, an ordinary man may also have to pass from the univocal to the dual. In the consecrations and sacrifices described, he will address several ancestors of a category (patrilateral or matrilateral) and will simply mention the other category ('and all those on the other side'). But if he wishes to invoke all his ancestors on both sides directly, he will consecrate a sash of black and white beads'(3 blacks + 3 whites + 3 blacks + 3 whites, etc.) (Bosch 1930, pp. 100-1). However, this duality is not marked by asymmetry; this is only to be expected, I believe, because we are here on the inferior plane, that of familial sacrifice. This type of sash occurs as an iholero consecrated to paternal or to maternal ancestors. In the first case, all the beads are white and the sash is worn on the left arm (and on the right shoulder); in the second case, the beads are black and the sash is worn on the right arm (and on the left shoulder). The black and white sash is said to be worn 'sometimes on the right shoulder, sometimes on the left' (Bosch 1930, pp. 100-1). In addition, it is significant that the beads are grouped in alternating threes. Just as three is incomplete in relation to four, two and still more in relation to five, so too three is the only one of the first five numbers which is presented as composed of similar elements. People know that 'three' is the feminine number of the three stones in the hearth, and they refer to the 'riddle' (kalagu): 'All my children resemble each other?', the correct answer being: 'The three stones of the hearth'. Let me emphasise again that this symmetry should be regarded as being at a lower level in relation to that of the dual figures, such as the black animal with the white spot on its forehead, or the succession of the two seasons of the year, where the black-white conjunction is presented as a black/white hierarchy (see below). On the other hand, in relation to the sacrifices made to a single group of ancestors, with either white or

56

Nyamwezi classifications

black beads, in relation to this unicity, the conjunction recovers its superiority: it is in serious cases that the diviner advises one to invoke all the ancestors 'of the two sides'. I have already stated that examination of the colours of the sacrificed animals uncovered a reversal of the hierarchy that I have just considered. The great royal sacrifices only allow animals whose colour is black. These are also the only animals which the king can dedicate to a dead king: the animal is consecrated, it is untouchable and is addressed as if it were an ancestor. The first of these consecrated animals is 'the elder king', the black bull which represents the first ancestor and before which one performs the great rituals of the agricultural cycle (Cory 1951, pp. 43-4; Millroth 1965, P- 159)How is one to interpret the superiority of the colour black? First, there is the logic of the system. As soon as the two colours / one colour opposition is at work in the domain of ordinary ritual, creating a subdivision there, it is natural that the superior plane, where a subdivision has also been created, should also feature the same opposition, but in the reverse order. One might have anticipated this configuration if one had known that the other possible structure, involving one colour (white) was already occupied, so to speak, by the great diviners' sacrifices.10 Second, we need to consider the particular nature of the solemn rituals directed by the king. They are always concerned, whether explicitly or implicitly, with ensuring that the rains come. Rain is an aspect of reality which is both crucial and precarious. For prosperity is based upon cultivation, and the coming of the rains is an expression of the approval of the royal ancestors, upon whose mood every single person in the country is supposed to depend; but rain is precarious and the presence of an object coloured red - the colour of blood - is enough to prevent it from falling. We know that the rainy season is the season of prohibitions, so to speak, the foremost of which is the prohibition against shedding blood. Being thus a season which is both of great value and of extreme frailty, which involves a disavowal of a part of the real (blood which flows; the powers of death of the 'white' period) and, most crucially, the need to disavow it, so that only the colour black must feature on the animal. The commentary which the Nyamwezi themselves habitually give only provides the analogical and positive explanation of the problem, namely, that rain clouds are black. It does not account for what is implicit in it, i.e. the prohibition regarding the presence of white.

Distinction between levels

57

Once we are concerned with prohibition (a topic which I cannot analyse here), classification acquires an ethical dimension and becomes a cosmological matter. If I may, however, briefly summarise the position, as regards the plane of hierarchical value, black is the encompassing element, rain is the crucial fact or value in the Nyamwezi conception of royalty and in the annual cycle of the kingdom, and the rainy season is the totalising period (see below for the final point). But on another plane (which I shall term the plane of'power'), the crucial factor is the success of the 'white' rites: the king's settlement of disputes which cause blood to be shed and which involve two commoner groups or one of these groups and the court (in the case of witchcraft accusations). It is worth noting here that it is a bracelet of white beads that an aggressor places on the left wrist of the person whom he has wounded; if this latter then dies, this bracelet will serve as a reminder of the murderer's blood-debt (Bosch 1930, p. 436; Cory 1953, p. 10). The ordered sequence of the two seasons, in which the black period encompasses its contrary, the white period, in which respect for the prohibition on blood and a successful relation to the ancestors are followed by the lifting of the prohibitions and by conflicts which place royalty in danger; in which the success of the passage depends upon the performance of very particular sacrifices which, upon this occasion only, display a conjunction (human sacrifice with separation-union of the head and the body on the model of royal funerals; sacrifice of a black bull and of a white goat - the only instance, it seems, of a white animal featuring in royal rites), in short, the hierarchical alternation of the two seasons is the basis of royal authority. A mixing of colours in a ritual associated with the rain would be disastrous. The prohibition which both distinguishes the two periods and causes them to be linked one to the other would thereby lose its meaning, and with it would disappear the temporality which permits the natural envirionment of the Nyamwezi to have a signification. It is important to realise that we are concerned here with rites which belong to a cycle, in which the distinction is sanctioned by prohibitions and is conceived as a sequence. So far I have described familial sacrifices or other rites which could occur upon any occasion or which, in any case, were not placed in a cycle of this type, although a global circulation, which is established between living and dead, passes through and links all the sacrificial rites. If one sets aside the timescale of a single season, in order to recover an image of the whole, one arrives at the global image of the conjunc-

58

Nyamwezi classifications

tion of the two seasons, the black and white, as a totalising figure. But one then has to jettison the example of the coats of the sacrificed animals. However, as I have already suggested, if one considers the rituals of transition between seasons (victim cut in two; presence of black and white) there is no need to set aside all references to sacrifice. Two configurations then emerge, one of which is centred upon totalisation and one upon reversal. Totalisation

The duality > singularity hierarchy differentiates between ordinary victims, in black coats (in the context of familial sacrifice), and black and white victims, who have coats 'of two furs', and between the appearance of the white snake, an ordinary ancestor, and the black and white snake, a royal ancestor. But totalisation depends upon asymmetry and there is also the hierarchy in which symmetry < asymmetry, disjunction < conjunction, and the alternation in singularity < duality. Thus we have: — —

—

ordinary familial sacrifice: the pearl bracelets may be black or white in colour (disjunction and symmetry); familial sacrifice in more serious circumstances: one invokes all ancestors on both sides, with a black and white sash, but one in which symmetry dominates (conjunction but symmetry); royal sacrifice: the animal's coat must be black in colour, with one white spot on the forehead; the same configuration is a feature of the royal snake-ancestor (conjunction and asymmetry).

In addition, the royal sacrifices, even if they are not directly concerned with the order of the seasons (the solemn sacrifices) and refer to the relation between the king and his various close ancestors, his badugu, do however have this annual cycle as their backdrop, since it serves to define what kingship is. Moreover, taken as a whole, this cycle appears as the conjunction of the black and the white: this conjunction is explicitly opposed, as we shall see, to the disjunction between the black or white pearl bracelets used in familial sacrifices. Reversal

Reversal is another figure which emerges in the context of royal sacrifice. Indeed, it enables one to differentiate between the solemn

Distinction between levels

59

sacrifices and the other royal sacrifices. The univocal pole - black becomes superior but everything, both objects and clothes, must then become black. In the case of familial sacrifice, the same configuration enables one to distinguish between a relation with a witch (a living one or an ancestor) and an ordinary relation. In the latter case, however, the univocal context is devalorised. This superiority enjoyed by black, at the superior level, bring us to the general figure of colour reversal. If we now set the dual/univocal opposition to one side, the major reversal undergone by colours, in a context involving one's relation with the ancestors, is as follows: superior level: inferior level:

black >white white >black

If I complete my ethnographic account, this will become clearer. Up until now I have referred to the colours of consecrated or sacrificed material, i.e. jewels and domestic animals. But the ritual performed in family sacrifices depends upon a complementary operation of destruction, or upon an operation that is homologous to destruction: the anointing with Iwanga (white sorghum flour mixed with a little water). The whiteness of Iwanga is emphasised; it is prepared in a cup which is called 'white cup' and upon which, contrary to custom, no decoration appears. It is entirely white. This is because white sorghum is 'a symbol of happiness' and 'of harmony'; it enables one 'to ward off evil' and it is 'a good omen'. It 'makes the ancestors happy, and they then bestow mhoja'; through being anointed with Iwanga, 'one wins recognition from the ancestors' (Bosch 1930, pp. 115, 127, 372, 380, 449; Blohm 1933, pp. 164-5; Millroth 1965, pp. 161, 180-1; Gass et al. 1973, p. 377)- The colour white characterises one's relation to the ancestors and serves to differentiate it from one's relation with witches. Further evidence for this distinction is provided by that form of divination in which a chicken is used, which is employed when commoners are ill (Bosch 1930, pp. 220-1; Blohm 1933, text 146; Millroth 1965, pp. 138-40; Cory i960, pp. 22-4). Points or lines which are white-coloured, or which are lighter than the rest of the skin, indicate that one has neglected to offer sacrifices to such-and-such an ancestor, or that an object consecrated to an ancestor must be cleaned or replaced. Black or 'darker' signs, on the other hand, indicate an attack by a witch.11 Thus, on the plane of family ritual, 12 in divination with chickens, white is distinguished from black and is superior to it, just as the

60

Nyamwezi classifications

Iwanga which enables one to establish a relation with the ancestors is distinguished from the 'heinousness' of the witches.13 The witches are only familiar with murderous action. There is a sense in which this is true of the ancestors too; but by indebting oneself to them, one can wrench the benefits of life from them. The debt process enables one's relation with the ancestors to be a constructive one, and this makes it the opposite of the witchcraft relation.14 The opposition is made complete by the fact that the witches are generally women, often on the maternal side, the 'back' side (and, as far as the maholero bracelets are concerned, the 'black' side). The whiteness oflwanga, for its part, is reminiscent of that of the paternal beads. It is worth noting that there is a correspondence between Iwanga which, in a general sense, 'purifies' (kwezya), and the white beads 'on the head side' which also 'purify' in terms of a relation between two groups. These are the beads which a second wife, upon her arrival in a family, must place around the neck or the wrist of the children of the first wife, in order to protect them from the dangers to which their father's new relation will subject them. 15 They are also the beads, mentioned above, that an assailant will hang upon the hair or around the wrist of the person that he has wounded, both to ward off the threat of death and to mark the possible 'blood debt'. Thus the white/black <-» relation to ancestors/witchcraft opposition implied in the treatment of the chicken may, although one cannot, because of the slipping of the contexts, assert it, be associated with the opposition of the beads and the 'head' and 'back' sides, in order to mark a pre-eminence of the former side. Lwanga also plays a part in royal sacrifices. But, on the one hand, it becomes the inferior pole in a system of anointings and libations dominated by mla (python dung, which is also white), in accordance with an opposition of levels between 'cultivated' (sorghum) and 'wild' (the python). 16 On the other hand, the Nyamwezi, whilst they emphasise the importance of the part played by Iwanga in commoner sacrifices, describe royal (solemn) sacrifices in terms of the crucial presence of black, even ifIwanga is present too. It is significant in this respect that Iwanga is dominant in those sacrifices performed by the king in his role as family head (rather than as head of the community); in which the ceremonies are no longer carried out on the royal tombs but on the family altar close to the royal appartments (Blohm 1933, text 103). Black dominates the superior level. This is apparent in Nyamwezi conceptions of the royal chicken, from the entrails of which one may

Distinction between levels

61

read off an opposition between the murderous drought, which is 'white', and fertilising dampness, which is 'black'. Royal ritual as a whole confirms what the chicken teaches us. If we compare the rites performed during the two seasons, and if we examine those which guarantee the transition, we find further evidence for black's dominance. But, if we wish to bring out the hierarchy of the seasons, all we have to do is to refer to Nyamwezi conceptions of the 'year' (mwaka). We then encounter the hierarchical relation in its simple form: a whole composed of two terms, with one encompassing the other. A year consists of twelve months, six of which are the dry, or kipwa, season, and six of which are the rainy season. The latter is spoken of in three different ways: strictly speaking, it is kidiku; sometimes it is simply termed mbula, the 'rain'; but usually it is called rnwaka, i.e. the 'year'. Thus, if one wishes to say that one has remained two years in a particular place, one says that one has remained there for 'two rains'. More clearly still, the 'rain' signifies the whole year and encompasses its contrary, the other 'half, for people say that the year begins with the rainy season and, likewise, when the rains have fallen, people say mwaka gwadja, 'the year has ended' (Blohm 1933, p. 142). THE HIERARCHY OF LEVELS

One question, however, still remains to be answered. Can one speak of 'reversal' in the hierarchy of colours when comparing two domains which appear to be distinct, namely, that of sacrifice and the relation with the ancestors at the inferior level, and that of the two seasons of the year at the superior level? And even in the case of the chicken, whose dissection still belongs to the domain of divination, can one compare the 'relation with the ancestors' / 'witchcraft' opposition, on the one hand, and the 'bad king' / 'good king who knows how to bring rain' opposition on the other? Likewise, in speaking of 'totalisation', can one oppose the configuration of the two seasons, whose succession constitutes the year, to the disjunction which separates the white-paternal and black-maternal (bead) bracelets. There are several good reasons for doing so. The opposition between the seasons still belongs to the context of the relation to the ancestors. The rainy season is a 'black' season (I am not certain that this expression is ever used, but the season is represented by a black circle, as opposed to the white circle which symbolises the dry season - see below) because the sacrificial rites

62

Nyamwezi classifications

which introduce it are, so to speak, black, in that victim, paraphernalia and garments etc. are all black. During these rites, the ancestors are invoked but there is no opposition between paternal and maternal ones, or between white and black beads. The king invokes 'all his kin5 and, generally, he stands on a tomb and invokes a dead king. Likewise, one cannot expect to come across the sacrifice/ witchcraft relation in royal ritual for, once he is enthroned, the king no longer has to defy witches. His role is to 'thank5 those who, having killed a witch in their village, come to the court to 'pay the five5. But the reader will most readily grasp the continuity between the contexts compared here, and the pertinence of the arguments presented above, by considering an outline picture of Nyamwezi initiation, which presents, in two levels, the elements we have been discussing. The Nyamwezi have no 'tribal initiation5, in that the young people do not pass through a single, compulsory ritual, at a given age, in order to accede to the status of adult. Initiation is not compulsory, the rite is not the same for all, and age is of little significance. Indeed, various associations, often erroneously described in the traditional ethnography as 'secret societies5, have the right to stage initiations. These associations contain a range of different status distinctions, linked to age-classes. Additional variables include the option of taking on an apprenticeship and that of paying, in goats or in pots of beer, for one's right to climb the rungs of the ladder. Apprenticeship is important because these associations are all specialised, and concern either divination or the training of snakes, or the hunting of porcupines, or dancing or even stealing. One is free to choose one's association, the important point to note here being that in principle each person belongs, at one moment or another, to one or several of these associations and that entrance to them is marked by a collective initiation (the postulants of a particular period are reassembled) which, in spite of the diversity of the various associations, clearly indicates a common model (Cory 1946; Bosch 1930, pp. 169-202; Gass et al. 1973, pp. 400-29; Millroth 1965, pp. 140 et seq.; Blohm 1933, pp. 39-40). Thus one can speak of initiation among the Nyamwezi-Sukuma, although their society has been described as lacking this institution. The ceremony, which is spread over several days, is a very rich one. It partly turns upon a spatial opposition between a stay in the bush (and the ordeals undergone there) and a stay in the initiatory hut, a large house where the initiands are shut in and sleep on the

Distinction between levels

63

bare earth. In order to give some idea of the arrangement and the decoration of the inside of the hut, we shall follow in the steps of the initiands. On entering the hut, one's eye is caught by two large pots, which are placed on the bare earth and which constitute the only furniture in the hut. The first pot contains an axe, which is laid upon some stones. A copper bracelet is placed just underneath them. The pot is called ngosha (= mugosja), 'the man'. The axe is equivalent to the bow and represents the masculine sex. The copper bracelet is an object that one inherits and which is passed down, Cory tells us, from father to son. He also states that 'the bracelet indicates the presence of the paternal ancestors'. It is obviously an iholero bracelet of the 'head' category. The reader will recall that the copper bracelets are consecrated to the ancestors, just as the pearl bracelets are. The second pot is also placed on stones. It contains a wooden spoon and, under the pot, a lupingu shell is visible. The pot, as one would expect, is called nkima, 'the woman'. The lupingu shell is one of the list of jewels which may be consecrated to the ancestors. It is cut into a triangular shape and threaded on to a string from the mhoja tree (the tree 'of peace with the ancestors'), which one hangs around one's neck. It can be consecrated for either the 'head' or the 'back' side. In the present case, we learn that this shell is said to represent the maternal ancestors. The wooden spoon, as we already know, symbolises the feminine principle. It is crucial to note that these two pots are laid upon the ground, with those entering the hut apprehending them together in terms of the disjunction which they illustrate: men - masculine - paternal ancestors / women - feminine - maternal ancestors. If one were then to raise one's eyes to the conically shaped roof of the hut (which is round), one would see a large number of white spots daubed on the rough cotton. These are the stars, and the roof is called the 'firmament'. In the starry sky, everything is uniform and is confused; the stars all resemble each other and are countless in number; there are no oppositions and therefore no pairs emerge. Which is undoubtedly why the 'first ancestors', who are located 'on' the celestial vault, or even above it (in contrast to the close ancestors, who are beneath the earth or who wander in the woods and grottos, to be near to their descendants), are represented as being 'undistinguished' beings, in the literal sense. They are 'those with one foot': in fact they have, we are told, only one eye, one ear, one hand and one foot (Bosch 1930, p. 46; Gass etal., 1973, p. 391).

64

Nyamwezi classifications

Between sky and earth, between the One of origin and the distinctions so crucial to human reality, between those who only have one hand and those who either have the 'bow' hand or the 'back' hand (left and right, respectively), the walls of the hut have been erected. Without them the vault of heaven would not be held up and would come crashing down to the ground; the sky would not be able to 'cover the earth like the lid of a pot' (Gass et al. 1973, p. 377), thus ordering a space between two immobile zones, a celestial one involving perfect fusion, and a terrestrial one, involving a forever rooted disjunction. There is room for some movement on these walls and in this space; not in order to 'supercede' the opposition between the warrior's axe and the housekeeper's spoon, between man and woman, as those who espouse a dialectics quite alien to hierarchical logic would have it, but in order to hierarchise in one movement.17 If one faces the door of the hut, looking still from the inside, one sees various drawings on the plastered walls. But they cannot all be taken in at once; one has to move around, looking at them in a particular order. One must begin at the left hand side of the door and follow the wall round, thus describing a circle. One then comes across three snakes, which are red, white and black respectively (we are in the initiatory hut of the snake charmers) and further on, in order: — — — — — —

the moon and the star 'which comes a little before dinner', the evening star; the sun and the star 'which kills (the night)', the morning star; kipwa: a white circle, representing the dry season; kitiku: a black circle, representing the rainy season: a broad red streak, representing the dawn; a rainbow, composed of three red, white and black concentric semi-circles (unfortunately we are not told if these colours are to be read from the outside in, or vice versa).

First there is the night, which is black, then there is the day, with the whiteness of the sun's rays. Finally, there is the dawn, which is red, and which serves as a transition; there is also the red-white-black rainbow, which joins them all together. Between these two groupings, there is the succession of the seasons. There is thus an order, involving a separation of day and night, an intermediate image (dawn), then a global image (the rainbow). We might have intimated this from the turning movement required of the observer in order to perceive the drawings. The two seasons are therefore

Distinction between levels

65

presented in their sequence, from the dry to the rainy season, from that of death to that of life. We should note that, in the repeated sequence night—»day and drought—»rain (1-2 then 3-4), the movement is identical (from lower to higher) and the chromatic order alternates: the night, black like darkness and witchcraft, is followed by the day in which the sun arrives and 'works hard to heat up the house again5 (Blohm 1933, p. 140) with its white heat, just as the 'white' drought, which carries its own threats with it, is mercifully followed by the 'black5 rain, which guarantees life. This makes it clear that 'analogy5, in the sense in which I criticised the use of that concept in relation to Meru symbolism, would lead one to confuse the harmful blackness of the night-time witch with the harmless blackness of clouds swollen by rain, or the comforting whiteness of the morning sun, of the sun of a single day, with the mortal whiteness of the earth burnt and cracked by the sun of a whole half year. White and black may not be defined univocally, and in wishing to attribute a single place to each of them, one gums up the hierarchy of the levels. The initiatory hut displays this hierarchy in space and in time: —

—

below, the two pots: man, woman; above, on the wall, the two black and white circles. Below, the two consecrated jewels: one a 'head5 jewel, the other a 'back5 one, entirely homologous with the white or black bead bracelets (Bosch 1930, p. 101); above, the two seasons; below, the immobility of the disjunction; above, the sequence of the seasons, the succession of two seasons which are different but which together form the totality (the year). We are not presented with one season - like the 'ow^-footed5 ancestors which would turn the year into an eternity, a timelessness; nor with two seasons at the same time - like man and woman, or the 'head5 ancestors and the 'back5 ancestors - which would prevent alternation and would impose the same immobility. There is instead an ordered succession from the inferior to the superior and vice versa.

The opposition between the royal chicken and the one used in familial sacrifices, and that between the colours of the seasons and those of the bracelets, would thus seem to be justified. They belong to one and the same model, which is only fully illustrated in ritual, in one of a number of ceremonies (along with the royal harvest and coronation rites) in which a passage is affirmed for

66

Superior level

Inferior level

Nyamwezi classifications Table 3.2 1

2

Left/right

Woman/man

3 Left/right

Sacrifice', anointing with Iwanga

Sacrifice: the officiant

'royal' ancestor: 2 sides and/or 2 hands

a woman (on the paternal or the maternal side)

Marriage: position of the fiances, with the husband on the left and the woman on the right of the kanumba

ordinary ancestor: right or left (depending upon whether the maternal or paternal side is invoked)

A man (on the paternal or the maternal side)

REVERSAL of sorghum balls Birth: presentation of children; boy and girl on the left of the father

4 Woman/man

5 Completion by + 2 / +

6 5 : 2 and 3 / 2 or 3

1

Superior level

Inferior level

Superior level

inferior level

Marriage: arrival of the Twins and second birth after that of twins fiancee, reference to the myths (cf. chapter 3 note 3)

The king; the law of exchange

(5 is the basic unit employed in calculating 'prices') 2 + 3: a whole, a totality of two asymmetrical elements (masculine and feminine)

In the myth, man creates nothing, woman creates society (the mortal condition and the art of sacrifice; fire and seasoning)

the 'two' (superior; that of the twins) + 2 enables the mhoja

Husband's and wife's tasks

Sacrifices, rain magic, Pregnancy, birth: house construction, sowing funerals, mourning

In ritual and daily work, the man fights, hunts, clears the ground and lights the fire, whilst the woman sows, ploughs and cooks

(cf. p. 47 and chapter 3 note 4)

7 Female/male

8 Left/right

1

2

3 + 1-+4 7 8

2 or 3 (man or womam)

9 Black/white

Sacrifice: the sex of the Sacrifice: ring of skin victim royal animals which rings always placed on are always male the king's right hand

Sacrifice: the victim's coat

animal always black

animal black with one white spot on the forehead

ordinary animals, which are male or female (depending on whether paternal or maternal ancestors are involved)

blackwhite animal

black animal (nonbovine)

on the left or on the right (depending upon whether paternal or maternal ancestors are involved)

Distinction between levels

Table 3.2 (cont.)

Superior level

Inferior level

10

11

Black/white

Black/white

The initiatory hut

the

annual cycle/familial sacrifice Black AND white (the seasons: the two circles) Black OR white (the maholero jewels for the maternal or paternal ancestors)

Divination with a chicken:

royal rite/familial rite Black>white (rain>drought) White>black (ancestors>witches)

each person and in which, in addition, the whole community is involved. I shall now present a table summarising the oppositions involving colours, so as to compare them with those elicited in my examination of the other pairs of poles. We are now in a position to compare the different contexts, not in order to establish an analogy between pairs of terms but rather between pairs of levels (Table 3.2). The reader will readily grasp that this table, in spite of its separation of superior and inferior elements, is incompatible with the dichotomous exposition presented in a 'binary table'. First, it summarises situations which may sometimes call for intermediate configurations (the system of anointing; the system of animal coats). On the other hand, we are here concerned to compare the operations which, for each hierarchical opposition, articulate the inferior with the superior level. The analogy does not so much involve the various figures with their various terms as the transformation affecting the terms. It is not the terms that are opposed, nor even dual relations (a>b compared with c>d), but relations between levels ([a>b] encompassing [b>a] compared with [c>d] encompassing [d>c]). The various figures involve two operations, totalisation and reversal, the first being further divisible into 'totalisation' (or 'hierarchy in a broad sense') and 'strict hierarchy': totalisation: a AND b —» a OR b (columns 1,6, 10); strict hierarchy: always a —» a OR b (columns 4, 7, 8, 9); reversal: a > b —> a < b (columns 2, 3, 11). Having named and differentiated these operations, we should disregard their particular characterisations and try to unify them. Thus, totalisation may be expressed either as unity/alternation or as con-

68

Nyamwezi classifications

junction/disjunction or as duality/singularity. Indeed, to oppose the '2 OR 3' level to the '2 AND 3' one, for instance, is to oppose alternation (2/3), disjunction (2 or 3) and singularity (either 2 or 3), on the one hand, and Unity (with 5 constituting a whole), conjunction (2 AND 3), Duality (with 2 and 3 constituting a whole), on the other. Strict hierarchy, for its part, is the singularity/alternation opposition, only a/a or b. But singularity (a only) is here conceived as a whole. It becomes an encompassing Unity, with a encompassing b. This configuration may also be identified in a Chinese example which makes 'encompassing' explicit (in the Yang/Yin opposition, Yang is both the whole and a part: Yin and Yang both derive from Yang). Thus, strict hierarchy (singularity/alternation) rejoins totalisation (Unity/Alternation). What about the third operation, reversal? Reversal would seem to differ from the Unity/alternation model, and European logical discourse does not allow for its inclusion within the previous category. This is why I have devoted so much space to the example of colours. For, within this series, analysis has clearly shown that the configurations 9 (strict hierarchy), 10 (totalisation) and 11 (reversal) are three different expressions of one and the same model, the unity of this latter being demonstrated by that of the contexts involved. We have not left the domain defined by the relation to the ancestors, it being understood that initiation refers explicitly to it.18 The columns 9, 10 and 11 are the expression of a single relation: the black/white hierarchical opposition. Once the theory is

accepted, we merely have to specify that an opposition alb is 'hierarchical' in order to imply the possibility of encountering the three configurations illustrated by columns 9, 10 and 11. This conclusion is particularly significant where reversal is concerned. In our logic, a contradiction would be involved: if a is 'greater than' A, the relation cannot be reversed without affecting our original definition of the terms. As these latter are defined by themselves, as substances, the change marked by reversal would mean a change of the terms themselves. If the terms are still the same, there is 'contradiction'. One cannot have both a>b and b>a. In hierarchical logic, however, a term is only defined by its place in the whole, i.e. by the value assumed by it at the level at which it is placed. Reversal is therefore no longer a contradiction but the sign of a movement. It indicates that there has been a change of level and that a, which was superior, is now inferior. It is time now to give a more precise account of the Nyamwezi image of kingship.

4 Hierarchy and power: kingship captures the kings

My critique of Needham's analyses has led me to question both the general definition of binary opposition and the narrower conclusion it entails regarding the 'duality of powers'. I shall therefore conclude my analysis of Nyamwezi oppositions by enquiring into the possible existence of an antithetical pair of 'mystical powers' and 'political powers'. This involves distinguishing between two types of question which are often confused: what is the nature of the opposition between the figure of the 'Chief and that of the 'priest'? What is the relation between power and values? Not only must we distinguish between these questions, we must also avoid comparing them in the name of a supposed homology. KING/PRIEST

Royal authority is defined in terms of mastery of the yearly cycle. At the beginning of the year, the king 'calls' the 'black' rain. Sometimes a human victim, treated as a symbolic equivalent of the king, is put to death (Cory 1951, p. 35). If, however, the rain is too long in coming, the king himself must be killed, or expelled, or replaced. One may also attempt to thrash him, to the point of tears. For the latter, it is said, being royal drops of water, 'will attract' those other drops of water that all await for their fields and that the royal ancestors in their wrath are holding back (Abrahams 1967, p. 34). The queen and the priest can then 'bring forth' seeds. Then, during tillage, the king and the priest supervise sacrifices which guarantee that the crops will mature. All these rites are performed on the orders of the royal councillors. When the weather is threatening, they say to the king: 'You have neglected [in the sacrifices] such-and-such of your predecessors: sacrifice to him!' During tillage, the king is literally covered with prohibitions. Then, when the sorghum is ripe, the king - mainly in the guise of the bull 'elder 69

70

Nyamwezi classifications

King' who guarantees the presence of all the previous kings - and the priest bless, through the sacrifice of the royal bulls, the whole community's fields. Each person may then harvest his own plot. The prohibitions protecting the earth and sanctioning the deeds and gestures of the kings are lifted (Cory 1951, p. 43). Then, during the dry season, the 'white' season, the king becomes the person whom traditional ethnographers refer to as the holder of 'judicial authority' (Cory 1954, pp.8ff., 2gff.) and the arbiter of conflicts. Indeed, he is the person before whom and to whom, in certain circumstances, one must 'pay' the 'price' (nsavo) of a murder, a theft or of any sacrilegious act, but according to a fixed code that the king simply recalls upon each occasion, making no allowances for property or intentions and proclaiming no judgement upon the wrongdoer. The 'price' consists either partly or entirely of the cattle which thereby serve to provision the court and the royal sacrifices: when the price is not paid at the court but to a hostile group (murder or theft amongst commoners), a part of it is levied by the court, either in kind or through the metaphor of a sacrifice which the guilty party is ordered to perform on the spot. Kingship, as principle and as value, is - in order- 'black and white'. But the priesthood is restricted to the 'black' side. Essential though the priest's role may be during the black season, for the reckoning of the 'prices', all he does is to introduce the plaintiffs and the guilty into the presence of the king. The shaven/unshaven opposition is also illuminating. A change in season is marked by the shaving of the king. During the whole of the damp period, he is forbidden to have his hair cut; at harvest time, he is shaved with great ceremony and the gesture signifies that one may then reap the sorghum and that, in addition, the prohibitions are from then on lifted. The ngabe, for his part must always wear his hair long. This may be why1 the ngabe is associated with the numerical series 2-4-8, which is superior to that of the incomplete numbers 1-3-7 (even absolutely superior, in that it plays no part in the context of the feminine/masculine opposition—* 3/2-4 and supercedes it), but inferior in position to the 'five'. Likewise, the ngabe would be associated with the left, as opposed to the king, who is associated with the right (the royal emblem, ndeji, the two lances), according to a hierarchy in which right > left, which is characteristic of the superior level and the reverse, as it ought to be, of that of the inferior level (left > right). How does the opposition between the sexes express the royal

Hierarchy and power

71

principle? Whilst the opposition of numbers or of colours allows of many combinations, in which totalisation readily features (black AND white, the 'five'), the opposition of the sexes is more resistant to it, in ritual at any rate (whereas it is otherwise in myth, where androgynous figures may appear). One can begin to account for this state of affairs by referring to the figure of the twin which, in the case of a mixed birth, enables one to represent, within the unity of a birth, the duality of the sexes. But, for the Nyamwezi, twins are important because they are the 'two', and the figure of the mixed pair is not isolated as such.2 This example apart, how may the masculine/feminine opposition be placed, without a third term, in the royal domain, where it is entirely fitting that singularity be warded off? Is the totalising figure impossible and is reversal therefore the only option? The reader should first of all consider and recall to mind the figure of strict hierarchy, where singularity assumes the form of encompassing Unity. There is a correspondence between the image of the royal sacrificial victims, who are 'always male', and the representation of a king who is 'always male' even if he is a woman (which was sometimes the case). The Nyamwezi account for the former obligation in terms of the latter. There is an opposition between the alternation of the sexes known to ordinary humanity and the eternal definition of the king. Likewise, we have noted that it is always on his right hand that the king wears the consecrated bracelets, which are emblems of his status (the ndeji, whose other part is on his right ankle) or marks of a recent sacrifice (the bracelet or ring of hide). At the king's death, this same figure of strict hierarchy features. He is buried with his arm raised, just as any other individual is. The left/right <-» man/woman alternation does not concern him, however, and he is always buried with the same arm raised.3 Yet the king's death is marked by a double sacrifice and the image of the totality also affects the nature of the pair of victims, for they are boy and girl, laid at the head andfcet of the king, forming his two 'pillows' and caring for him in the beyond, the boy going out to cut wood and the girl concerning herself with the fire and with cooking, assuming all responsibility, for ever, for the masculine and feminine tasks required by their royal master. When the king had died, the councillors of the kingdom, before announcing the news, would choose two children and would go in their company to the person whom divination had designated as the new king. They would throw the great black cloak over him and everyone would then return to

Nyamwezi classifications

72

Table 4.1 Totalisation

Reversal

'Five' Hierarchical conjunction of the two seasons, guaranteed by the royal ancestors: black AND white feminine AND masculine Superior level ,-,„,-, ,„,..„,„ ,, .. , THE 1 WINS (are kings x) the TWO anointings)

hands

(system

of

The chicken used for royal divination, solemn royal sacrifices, royal rain rites, the 'royal' twins, the royal ndeji bracelet: black > white the animal is always black and always call for rain on the back conception of twins whilst lying on

the king is always male AND always the back 'child of woman' (cf. chapter 4 note 3) bracelet always on the right for the royal ancestors are always 'of the back' (cf.

chapter 4 note 3) The chicken divination: white > black

Familial sacrifice: 3 OR 2 bracelets are black OR white Inferior level X l l l v l lV/1

lv

V

a n d are w o r n o n the right O R

left

vl

anointings and ritual gestures performed on the right OR left

used

for

ordinary

close ancestors > witches head — east — male > back

for the ancestors 'of the back' OR 'of the head1

the court. The two children were then strangled and placed, according to precisely defined rules, in the royal tomb. The girl was seated, with her legs extended, and the dead king's head resting on her chest, and the boy was placed at the king's feet and facing him. An execution is thus required for the establishment of this spatial organisation which, from this point on, is immutable and in which totalising duality can again be identified, in which the two children are linked by the royal body, in which they are placed on both sides of the dead king, who thus becomes the axis of this configuration, the frame of reference for the opposition of the categories man/woman (Bosch 1930, pp. 496-502; Cory 1951 pp. 5-32; Blohm 1933, texts 97> H7) • It is important to note the opposition between this and the other human sacrifices carried out in the royal domain. When there was a drought, a man was sacrificed. Likewise, it was a boy who was ritually suffocated and trampled by the king and the army before their going off to war. Only a royal funeral required a pair. 4 Thus, the opposition between the sexes also features the totalising figure characteristic of the superior level. Table 4.1 gives a single schema,

Hierarchy and power

73

which groups together all the terms distributed across both levels and recapitulates the information already given in Table 3.2. It is thus apparent how, at the end of the analysis, we again encounter the duality of references mentioned in the introduction. Each binary opposition is shown to be a hierarchical opposition, therefore upon each occasion involving levels. If each of these oppositions only appeared in a single context, irreducible to those of the other oppositions, an analysis would properly be limited to that one context. But this is not true of the Nyamwezi case. The majority of our examples derive from sacrificial ritual and, in a broader sense, from rites in which the Nyamwezi renew their relation to the ancestors.5 I have therefore felt justified in linking the various hierarchical oppositions so as to bring out what in a sense are two super-levels (superior level, inferior level). These latter do not represent a particular hierarchical opposition, as is the case with colour, number or points of the compass. These levels emerge at the end of my grouping together the oppositions. But uncovering them enables us to grasp the nature of the connections between the two frames of reference, the 'ordinary' and the 'royal', the local and the global, that of the 'familial' ancestors and that of the 'royal' ancestors. The statutory opposition between the king (with his court) and his people then appears in its particular reality, in the society's order of values, in the system of idea-values and in ideology. There is no tautology involved for, upon each occasion, the superior level is not defined by the fact that the example derives from a royal rite but by the totalising or reversed nature of the level which emerges. In the case of the former (totalisation), my bias has been to insist upon placing a configuration of 'a AND V type at the superior level when it was opposed to the 'a OR V type. This was something of a wager: the Nyamwezi represent a society of the holistic type, in which a part is only defined by its relation to the whole and is subordinated to it, and I calculated that this wager would be justified, as the analysis proceeded, in that it would enable us to integrate into one and the same coherent pattern, and without being forced to reject 'contradictory' parts, the whole of the ethnographic data. In the latter case (reversal), I have sometimes used the royal context of the rite as a justification for placing one of the configurations at the superior level. This is not to say that I took the superiority of kingship for granted, on the basis of a 'power' whose

74

Nyamwezi classifications

content was not placed in question; I merely noted that a given rite concerned the whole community and was thus to be considered in terms of a wider perspective. This short cut in no way affects the complete proof that one might have offered. The latter, a far too cumbersome thing, which would not have carried conviction until the final page, would have consisted in isolating, for each case, an opposition between configurations without drawing any conclusions as to the superiority of one over the other; then in grasping, as I have done here, that the numerous figures featured in the various rites could be aligned with one or other of these kinds of configuration; and finally in giving various examples, linked to particular figures, whereby the Nyamwezi account for value: twins are among the more important ancestors and, in contrast to other commoners, their memory does not fade; royal sacrifices concern the most important misfortunes, epidemics, drought, which involve the risk of death for all; the commoners do not sacrifice black bovines whilst, in both myth and exchange system, they are the superior animals; nobles prefer animals 'of two furs', etc. Only then would one have pointed to the superiority of a whole group of configurations, so as to grasp that it groups together the collective rites, the royal rites, thereby affirming the superiority of royalty along with the content of this superiority: not a political power but a totalising symbolic role, a guarantee of time (the return of the 'year') and of space (establishing for all the inhabitants of the kingdom a firm relation with all the dead). We thus return to the wager that my approach to totalisation implied. As far as kingship is concerned, what I have argued here is self-evident for anyone who conducts even a cursory examination of Nyamwezi ritual life, namely, the central role of the court (even though nothing is presupposed about the content of this central role). Indeed, in both the present study and in a work in progress on Nyamwezi ritual, I am concerned to invoke a minimum of sociological presuppositions in accounting for the fact of kingship, for these are only too liable to turn, in a given instance, into prejudices regarding 'power'. My working hypotheses did not rest upon the expectation that kingship would, through the forms it assumes in ritual, occupy the superior level. The domain of familial ritual could well have occupied the 'a AND b' position, with kingship occupying that of the 'a OR tf position. One would then have had to enquire as to why it was that, in spite of the ostensibly crushing presence of royal ritual, Nyamwezi ideology involves the subordination ofbutemi

Hierarchy and power

75

to another domain. Here, on the other hand, the approach which I have adopted enables me to affirm that the central role enjoyed by the court is indeed such in ideology. The court's role depends upon the totalising position which royalty holds with regard to the plane of values.6 Kingship in its actual principle (butemi), a kingship which is termed 'sacred' and is here described as 'totalising', is defined by the complete set of'superior' positions that examination of each pair of poles has brought to light.7 I want now to reconsider the image of the royal person that this eternal kingship represents, so as to conclude with the two operations which feature in the table of levels, namely, totalisation and reversal. Kingship is both 'totalising' and 'reversed', and the homology between these two features is once again apparent. Indeed, it is striking to note that the king, insofar as he 'is' a pair of twins and a reversed-born, is explicitly the image of these two figures. We have already noted that, for the Nyamwezi, 'twins are kings' and vice versa. The king, as image of totality, resembles the pair of twins. Is he also an ipiage of reversal? I have already noted that the reversed-born, the kashindye, are, along with twins, situated at the superior level of the system of anointings. In addition, their birth is attended by the same ritual as follows that of twins; the king is 'paid' and the royal drum is borrowed to accompany the birth dances. The kashindye, like twins, belong to the category of'high men' (bahanya) (Blohm 1933, text 214). It is worth emphasising, also, that the kashindye are something more than the ethnographers' commentaries would have one believe. They are not merely 'abnormal' beings who are born in an abnormal position and who are therefore 'magical' beings. Hunchbacks, or those born ill, should not pass through a royal ritual (Blohm 1933, p. 164).8 In addition, a precise prohibition makes the situation quite unambiguous. What happens if 'one lights a log by its large end (instead of lighting the smallest extremity)'? 'One will give birth to a kashindye.' The kashindye are indeed 'reversed'. Now, the king may be addressed, honorifically, as kashindye. In a narrative regarding 'blood price', we find the two parties involved leaving the king, after the conclusion of the rite, saluting him respectfully with the following words: 'Master of the meat! Reversed-born!' (Bosch J 93O> P-^5; Blohm 1933, text 163). Both the reversed and the totalising figure are 'superior' and both define kingship. The reader will now have some conception of the full complexity, and of the coherence also, of a hierarchised dualist system. Is this

76

Nyamwezi classifications

coherence 'in several dimensions' peculiar to the Nyamwezi? Are we a long way from what the Meru system, as described by Needham, might conceal? It would seem not. When all is said and done, we have simply clarified what it means for a society to say that butemi, kingship, is identified with the whole of society, that it actually is that 'whole'. What if this were also true, mutatis mutandis, of the figure with whom this criticism of binary method began, the Meru Mugwe? Millroth asks such a question, starting with the Meru and then going on to consider the Nyamwezi (-Sukuma). I shall end this section with his words: 'Rain-fall is of major importance if men and animals are to survive ... one of the main duties of the chief is to see that rain falls in its season. In fact, this is so important that it must be considered one of the heaviest burdens of the ntemi. The village headmen acted as rainmakers in their own areas, but the chief was responsible for the well-being of the whole chiefdom. Bernardi writes of the Kikuyu in regard to rain-making: "The Mugwe is greater than an ordinary elder, and his power extends beyond the limits of a single clan, even his own. What the Mugwe does, he does for the whole of the country." The same applies to the Sukuma chief (Millroth 1965, p. 178, quoting Bernardi 1959, p. 113). HIERARCHY/POWER

My investigation into Nyamwezi dualist classification ends here. There is one question, however, still to be answered. This concerns Needham's conclusions regarding the duality of'powers'. We have defined Nyamwezi kingship on the plane of values, but have we therefore defined royal power? It would be somewhat startling if we were to reply in the affirmative, for to do so would be tantamount to confusing 'hierarchy' and 'power'. These two terms themselves constitute a relation which will bear examination. The 'hierarchy'/ 'power' pair is the last opposition in Nyamwezi society to be discussed here. But it is important to realise that we are no longer concerned here with an opposition within the society's symbolic system but rather with an opposition, albeit a fundamental one, entailed by our approach as anthropologists to the society.

In other words, in identifying 'royal' value, in marking out the superior position of kingship in terms of its principle, we still have not yet clarified the actual place of 'power'. This approach is a natural consequence of grasping that any enquiry into 'power' in itself cannot help but be ethnocentric, in that the only available

Hierarchy and power

77

point of reference is our own society and our own conceptions of 'power'. 9 Comparative anthropology, on the other hand, aims at a clarification of the relation of power to values, thereby contrasting the individualist universalism of our own perspective, which is partly concerned with an enquiry into 'power', with an awareness of the holisms of particular societies, and therefore of the hierarchy of values (cf. Dumont 1978). In order, then, to grasp what power is, we must now suspend our analysis of the hierarchy of levels and attend instead to royal ritual and to the temporality which it establishes and which, in turn, conditions it. One must also attend to an opposition which is only properly constituted by a movement from outside to inside, but this is not to be understood as yet another binary (outside/inside) classification but rather as an alternating passage from the centre to the periphery and from the periphery to the centre. We must consider contrasts which are peculiar to our own ideology, such as that between the master and servant in a ritual, between the one who has control of it and the one who is simply a wholly passive performer. These questions are not, however, central to the present discussion and I shall only tackle them in broad outline, basing my presentation upon a number of facts derived from the two great royal ritual cycles, that of funeral and enthronement, and that of the seasons and cultivation. Nyamwezi kingship is, in principle, immutable. It is 'black and white', with black first and white second, and it is thus hierarchised in space and in time; it is opposed to the inferior level of 'black or white'. But the holder of the title does not remain unchanged from one season to another. If kingship (butemi) is 'black and white' in the succession of the seasons that it organises, in the mwaka year whose flow and repetition it brings about, the king (ntemi) for his part clearly gives the impression of being the master only of the 'white' slope of the year, and the servant, so to speak, of the 'black' slope. He is thus servant of a court which is both his palace and his prison, and 'father' of all the inhabitants of his kingdom but obeying the orders of the great courtiers who surround him at all times, instructing him as to sacrifices he should perform when he has 'neglected' his ancestors, the dead kings. The king himself is put to death in substitute form when, as we have said, in times of drought, a man from the court is sacrificed and his body treated as that of the king is at his funeral. If the drought lasts, the king is actually sacrificed (or banished).

78

Nyamwezi classifications

At his enthronement, the king is literally captured by the councillors. Henceforth, for the rest of his life, he can no longer leave the court, except in the case of ritual obligations, when he is totally surrounded by councillors. As I have already noted, he becomes the person to whom one gives orders, when necessary, to coax the royal ancestors into granting rain and good crops. There is a sense in which he is from then on the prisoner of'black' value, as his position at the very beginning of his reign indicates, in that he is wrapped in a black cloak, and at the last moment of his reign, when he is enveloped in a black shroud, confirms. Kingship here captures the kings, for they are wrapped in black cloaks and shrouds which are often made of the hide of a royal bull, and this animal and its fellows represent the various dead kings, the complete series of previous kings and therefore the continuous existence of royalty. It is before the eyes of the first of these dead kings, of the primordial ancestor the 'elder king' black bull - that the reigning king conducts rituals, especially the harvest festival. The continuity of kingship, of which the kings are at the same time the victims and the symbols, is expressed at every moment of the enthronement and funeral rituals. The king's death is kept secret until his successor has been named. More clearly still, it is stipulated that the new king should arrive when the royal corpse, on its way to being buried, leaves the court. It is held that at the very moment that one passes through the back gate, the other should enter in at the front gate. The funeral is also marked by a particular treatment of the body. The head is detached from the trunk; the skull serves as a sacred receptacle for the anointings that the new king will perform, and the dead king is buried with his predecessor's head, which had itself been used in a similar way (Cory 1951, p. 6). Cory speaks of the continuity of the 'spiritual power' thus manifested, beyond the individual persons of each of its representatives. These terms do not seem to be exact translations of Nyamwezi expressions but their use is not unjustified. Even if there is no image here comparable to that of the Shilluk Nyikang, who joins the north and south of the country (the permanent presence of the founding king is guaranteed by the sacred bull, the 'elder king'), it is apparent that kingship is only complete when it is defined across two intervals of royal generations, with the joining of a head and body belonging to two successive kings.10 The continuity of kingship, which is emphasised by this joining of a king and his successor in death, reminds us of what was noted in the introduction, namely, that kingship expresses the

Hierarchy and power

79

presence of the ancestors and of their law, the osmosis between the living and the dead, and the force of diachrony. As for the reigning king, he has beside him a cranial calotte which serves as a basin for the lion's fat, a symbol of kingship, with which he knows that he will himself be covered on the day of his funeral. This prisoner king, who remains alone at court with the women when everyone leaves for war, hears, when struck by the councillors and wrapped in the cloak, that it is the last time that he will be struck by human hand. But this is on condition that he knows how to give the diviners the 'strength' to cause the rain to fall. If he does not, he will again be struck by the councillors, or even sacrificed. At any rate, when he is old, he will receive the mortal blow and will be laid in the earth upon the 'pillows' (the two children) as a sacrificial victim (which is laid upon the 'bed' of sacrifice, with the consecrated object being anointed and also placed upon a 'pillow'). Finally, it is the king who, during the period of tillage, is subject to the most onerous prohibitions. No other Nyamwezi undergoes so great a subjection. He alone has to submit to ritual sanctions in time of drought; if the rain is late in coming, all other Nyamwezi will emigrate;11 the king alone is forbidden to do so, and suffers the consequences of this exodus at the hand of his courtiers. The words that the king hears on the first day, and which declare that he is immune from blows are not, however, deceitful, for they refer to another domain of royal authority, the 'white' side. The system of prohibition is modified there. The king is no longer the sole victim of the sanctions, but, on the contrary, the sole person who is exempt from sanctions. At the same time, the scope of the prohibitions has altered. During tillage, not a hair, nail-paring, dejection, or drop of blood must touch the soil, but it is the king's body and the soil of the court that are the predominant concern here. During the dry season, the king enforces respect of the prohibition on the shedding of blood (of other people), of the prohibition on the theft of cattle, the prohibition on the breaking of the royal objects of the ancestor cult, which hang at the entrance to the court and within everyone's reach, the prohibition on giving birth to twins etc. These prohibitions apply the whole year round, but insofar as the acts which they penalise generally occur in the dry season, they are indeed prohibitions belonging to the 'white' side. Most important, if these transgressions occur in the season of tillage, it is the custom to bring work in the fields to an immediate halt, until the guilty persons are 'blackened' (purified through the payment of

80

Nyamwezi classifications

a 'price' and then sometimes smeared with soot on their faces or dressed in black cloth). The 'black' phase, brought to a halt by a 'white' act, may then resume. The white side is thus both a period of time (the dry season) and a domain of royal activity (the judging of acts in breach of the royal prohibitions). The penalty is the payment of a nsavo 'price', itself carrying the death penalty if someone refuses to pay. The king then unleashes his 'dogs', the ban Hkulu, who seek out the guilty. 'What are these people that deny me?' asks the king of those who are not prepared to pay. They are sought out and they then remain at the court until they have settled their 'debt'; 'Who is this who refuses?', he again asks, adding: 'Has he a king among his relatives? If so, let him come and we will fight!' (Blohm 1933, texts 55b, 56). Everyone, including even the greatest royal dignitaries, has to submit to the system of'prices', with a scale of payment according to status. The king penalises people and he is the only one who can sentence someone to death, who can render a murderer untouchable, and the only one to receive a part of the 'prices' paid among the various groups. Thus, kingship, which is single and immutable by virtue of the absolute continuity of its successive representatives, guarantees the cosmic order - the return of rain - and the political order - mastery over death - but by contrasting two different images of the royal person, who is by turns a servant in 'black' and a master in 'white'. The unfolding of the annual cycle makes this opposition more specific. In the first period, the king is symbolically sacrificed; he is reborn at the harvest festival (the sowing recapitulates the royal funeral; harvest recapitulates the enthronement ceremony). The king rules at this time by his absence and by his passivity. In the second period, the king rules by armed force, by his control of the 'prices' (he is the only person who can override the prohibition on the shedding of blood, a prohibition he strives to uphold). Thus, if kingship is 'black and white', with there being a black > white hierarchy on the plane of values, royal power only has a place on the 'white' side. The king is the 'victim' of the sacrifices that the priest supervises but he is the 'devourer' of the guilty persons that the latter points out to him. From the 'black' point of view, the court is the sacrificial altar of the whole society, at which the king, who is the representative of eternal kingship, joins with the priest in sacrificing to the royal ancestors for the prosperity of all; but he is the first victim and the first sanctioned inhabitant of kingdom. From

Hierarchy and power

81

the 'white5 point of view, the court is the place where those who have broken a prohibition come to be 'eaten' by the king (Bosch 1930, p.91). These distinctions may enable us to understand a number of features respecting kingship which ethnographers regularly point out in their historical introductions, though invariably expressing surprise as to the contradictions which these points occasion. The founding ancestors of each dynasty are foreign conquerors, as I noted above (cf. chapter 1 note 4). But Nyamwezi social organisation carries no trace of this conquest and of the massive status distinctions which might have resulted from it. One must therefore suppose that these conquerors were 'peaceful' and, above all, that they arrived in the designated place when the various autochthonous groups of the region had become numerous and had come to feel the 'need' for an institutional order. Cory, for example, develops this argument in the prefaces to his various works, adding that the conquerors were completely 'absorbed', since they would seem to have lost their language. But is it not curious that the origin myths and legends make no reference to a pre-monarchical state of society? Kingship appears to be as ancient a thing as humanity itself, and the origin legends of the various dynasties refer either to an arrival in a desert land or, more commonly, to the usurpation of an autochthonous king's power by means of magical exploits and thefts of the royal emblems (the ndeji bracelets). Instead of imagining completely 'absorbed' 'Hamitic' invasions, it would seem more reasonable to suppose that the Nyamwezi do not place kingship, without which society is inconceivable for them, on the same plane as the kings, who are 'devourers' come from the outside. The first conception is expressed by a type of mythical discourse in which a hero, who is a child and a traveller or a hunter, succeeds in killing a monster who is depopulating the earth or in putting to death for the first time a black bull. The animal is sacrificed. Before dying, the monster explains how he must receive the last blow and how he must then be cut up. From his body, which is treated as the sacrificial victim will later be treated in ritual, there then emerges humanity, which is born or reborn in this manner (Ikombe 1943; Millroth 1965, p. 193; Spellig 1927, pp.203 et seq.). The hero becomes the first king (he may already be a king, who has come from elsewhere, and who has his people 'reborn' here). The second conception is expressed by the legendary discourse which follows on from the previous one or may sometimes even enclose it

82

Nyamwezi classifications

and precede it. Foreign origin is much emphasised and the politics of alliance and the succession of the first kings are all described. But the founding act itself, when described in detail, is the same as in the first schema. It is a ritual and magical act, sometimes in the guise of a ruse and of a theft. Here we touch upon the highly fruitful opposition between 'inside' and 'outside', between eternal kingship and the conquering kings. In response to the centripetal movement ruling the legendary narratives, royal ritual harps a great deal on centrifugal movement: twins are buried at the frontier; anointment at their birth requires a sorghum flour stolen from the neighbouring kingdom; a particular class of worthies, having helped at the king's enthronement, must then go and live at the outermost edges of the territory, with there being an absolute prohibition upon seeing the king again, for he might well die as a result etc. To give an exhaustive account of this point, a separate study would be necessary.12 I shall now try to state more precisely just what 'the relation of power to values' is. The hierarchical black/white relation establishes the following values (symbol-values): — —

—

Black + white: the year is ringed; Black: rain; the rainy season which 'is the year'; the fertilisation which brings things to life again; the art of sacrifice: the dominant activity of the 'black' season, a rite in which black is dominant at the superior level; White: drought; the dry season; death which is not followed by a production of life (by contrast with the sacrificial death kuhoja); the acts of death which do not in themselves produce life: murder, the procreation of twins (who, it is said, bring drought).

Power is neither 'black' nor 'white'. All that one can say is that it is wielded during the 'white' period. What of law? Further distinctions are required here. The law which is operative in the 'black' period, the law of the rain, is the Law of the world and the Law of the royal ancestors, upon whom the rain depends. It is a cosmic Law which the kings are able, because of their access to a transcendent outside which opens out into the universal, to play a part in enforcing. On the other hand, the law of the 'white' period, the law of spilt blood, is one that each king decrees within his own kingdom. Indeed, even if it is common to all the Nyamwezi kingdoms, the myth has it that 'one day the king woke up and said: "Henceforth all those who . . . ' " , and then

Hierarchy and power

83

promulgated the prohibitions regarding murder and twins. The law of the 'black' period encompasses the law of the 'white5 period. The law of the 'white5 period has a monopoly of the power regulating the use of legitimate force (the sending of guards to force people to 'pay 5 ). The encompassing of the law of the kingdom by cosmic Law, which means that the place of royal power is circumscribed by the religious order (the law of royal and collective ancestors, the art of sacrifice), is confirmed by, and assumes a visible form in the ceremonial circuit. To enforce respect of the law is called kutema. This verb corresponds to the terms butemi (kingship) and ntemi (king). It is not used with reference to the holding of a kuhoja (sacrifice). There is a sense in which kuhoja is homologous with kutema, but on the plane of the Law. One rings the year by means of kuhoja, which serves to link the ancestors with the endeavours of the living. If one then raises the question of power, and enquires as to what it is in the exercise of kingship that is equivalent to the use of'legitimate force5 and concerns the individuality of the man of power,13 one will find the answer in kutema rather than in butemi as a whole, so long as one stipulates from the start that kutema is in the service of kuhoja. Indeed, it becomes apparent, through our analysis of the settling of acts of transgression, that kutema activity is integrated into a cycle such that it is only one part of a whole whose essential purpose is still kuhoja. The payments demanded by the court for these transgressions turn quite explicitly into offerings sacrificed upon the altar of the whole society ('paying5 will then be expressed through kuhoja equivalents; the 'prices5 in cattle come to fuel the royal sacrifices). Thus the dual oppositions, such as black/white, do not in any sense allow one to conclude that there is a 'duality of powers5, even if one hierarchises these latter. The totalising configuration which is applied to society does not involve an addition of 'powers5, but a structure of ideological levels on the inside of which power may, on occasions, have a place. By way of conclusion I shall recapitulate what I said at the start of the present section. If the 'inside5 and 'outside5 conceptions of kingship seem to be mixed up in myths and narratives, if, in a general sense, the contrast between the Law of the 'black5 period and the Law of the 'white5 period did not feature at every stage of my earlier investigations concerning dualist oppositions, it is because we are no longer concerned here with a distinction of values which would be explicitly named and which would form a dualist system of levels. On the one hand, I drew attention to the temporality of the annual

84

Nyamwezi classifications

ritual cycle. This was not so as to oppose it as such to an immobile disjunction, which would indeed involve an opposition of levels, but so as to enter into this temporality and to follow the unfolding of the cycle by passing from the season in which the king as individual is the servant of the 'black5, to that in which he is master of the 'white5. Value, for its part, still dominates the whole of this sequence. On the other hand, I distinguished between systems of penalties by opposing them one to the other. Thus I contrasted the 'black5 king, who is the main object of the penalties of the Law, with the 'white5 king, who is the master of the law, and I separated the 'black5 penalties, which apply to relations with the (royal) ancestors, from the 'white5 law, which defines rights regarding death (sacrifice apart). These distinctions do not arise as such from the society - if we except the contrast between the colours of the seasons, with which we began and they appear when we apply our criteria to the whole range of ritual facts. With this methodological account, which obliges us to distinguish between 'hierarchy5 and 'power5, I will bring my enquiry into 'the duality of powers5 to a close. There is nothing surprising about discovering that power enters into the temporality of the royal ritual cycle, for power is mainly constructed on the basis of the temporal dimension. But we should be chary of inferring, on account of its apparent institutionalisation, that the position of power is identical to that of value. This is something that we already knew, to some extent, from Melanesian societies (cf. Iteanu 1983), where structure, which is widely fluctuating, makes it necessary to give considerable emphasis to time. This has been less commonly noted of the African sacred kingships, where value in the singular (in the present case, the two seasons; black + white; 'totalising5 kingship) has been identified with power (in the present case, the application of the law of the 'white5 period; the 'devourer5 kings). In the present case, the relation between the order of values and the inscription of power in the whole is a contiguous one, for the hierarchy/power opposition assumes the same form, on the plane of colour, as the global hierarchical opposition does, i.e. black/white. But the order of values is not identical to that of power, for the 'white5 aspect of royal authority (so named by me because these rites constitute the basic material of the 'white5 season), in being opposed to the 'black5 aspect, does not reproduce the opposition of levels of the sort that was apparent in the symbolism of divination using chickens: [(white the ancestors) > (black «-» witchcraft)]

Hierarchy and power

85

encompassed by [(white <-» dry earth) < (black <-» rain clouds)]. Indeed, there are not two levels, one of which would serve as the site for power as such to become superior (to the kuhoja law, for example), thereby reversing the global order. At the most, this contiguity enables us to appreciate that kingship, which is value on the one hand and 'power' on the other, is 'sacred' to the extent that it is the articulation (but not the superimposition) of power and values, wielding power in one of its phases, that of the 'white' season (whose place is anticipated in the hierarchy of colours), and inscribing this power in a hollow within the order of values.

PART

TWO

Comparison and conclusion

5 The same questions asked of other examples

DOGON TWINS

The Nyamwezi example has shown how important a role twins play in relation to the value that duality may represent. Duality is not, however, unity times two but a conjunction of asymmetrical opposites. It may therefore prove useful to round off this discussion by giving a rapid account of a study on 'complementarity', where the latter also appears as a value which organises levels. In a conference held in 1967 on the theme of 'Norm and laterality', Germaine Dieterlen used the Dogon origin myth to remind her listeners that 'complementarity' is at the basis of the whole 'Sudanese cosmological system5 (Dieterlen 1968). This is not, however, just a complementarity on the logical plane alone, with right + left, sacred + profane or man + woman, but a symbolic configuration which engenders a system and which must be defined in terms of value. On the one hand, there is 'the ideal image of the universe': two twins of opposed sex, a 'non-antagonistic complementary duality'. On the other hand, there is the element which is encompassed, singularity as failed duality: the birth of the 'prematurely born, conceived as a twin but having lost his female twin forever'. This 'antagonism' or, as I would put it, this hierarchical opposition, affects the meaning of both mythical narrative and ritual organisation. The narrative is a familiar one. First, there is a double genesis, with the first one failing and the second succeeding. God creates 'the smallest thing', in the form of a grain, but decides not to build up the world from it. The grain is single. It is then with two grains, twins of opposed sex, that the world is created. Then the 'original egg' splits, with two placentas giving birth to two twin fish which multiply, still in the manner of the pair of twins. Significantly enough, complementarity seems from the very beginning to be in opposition to 89

90

Comparison and conclusion

singularity, for the single grain is not creative. On the other hand, the twins are of both genders and asymmetry is therefore already a feature of the complementarity of twins, which subsequently becomes the basic pattern of creation. We then arrive at the asymmetry which founds society. The fishes multiply, but one of them is born before its time, 'before God had formed its necessary complement, namely, its twin'. As we know, it will, on account of its singularity, become the fox, the symbol of lack - it will always be in search of its twin - the symbol of revolt and of Disorder. In opposition to this, God created Nommo, symbol of Order and life; he is at the superior level, because a change of level will have occurred. Nommo is a fish that has been sacrificed and then resuscitated in human shape, and also a true twin. The opposition which founds the social order is now in place: singularity
Other examples

91

married, previously marked by a lack (that of not being 'born a twin5), to become real twins 'at the moment of sexual union and through exchange of the contents of the collarbone' (p. 149). A further nuance is added to the twin birth / single birth opposition. A mother with two twin embryos in her womb who breaks a prohibition during her pregnancy will cause the two foetuses to fuse. She will give birth to a left-handed person, if he is the avatar of two male twins, and to an albino, if twins of both genders had been anticipated. Left-handed people and albinos have a very special status, quite different from that of real twins and of ordinary non-twins. Discussion of this special status brings us to another problem. But I want first to emphasise the relevance of this brief discussion of the Dogon example to the present argument. 1 It is worth noting, first of all, that for Dieterlen the above definitions do not refer to a logical complementarity of the sort at work in arithmetical addition, but to levels and values. I would go on to point out that myth does not serve here as a mirror that would provide some explanation of the rite for, in either case, the symbolism seems to be immediately differential. In both domains, the principle of twinship imposes an order: second genesis/first genesis, twin born at the proper time/ premature twin, sacrificed and resuscitated fish/premature fish; twin birth/single birth, man/woman and, depending upon the content of the collar-bone, right-handed/left-handed, pigmented/ albino. In the case of the last two oppositions, an extraordinary status is involved. Thus, a left-handed person, if he fights in the front line, will guarantee success in war; and an albino will, by being the preferred victim, assure the success of a sacrifice. We would seem to be faced here with something akin to the logic of Nyamwezi ritual (see below). A hierarchical opposition (twin/non-twin) creates a gap in levels in which the ritual is inscribed. At the centre of the rites there are ambiguous objects whose nature in fact enables some movement to occur. In the case of left-handed people and albinos, there is a threefold ambiguity. They participate at two levels: they both are and are not twins, since they are avatars of twins. They are the result of a transgression (their mother has breached a prohibition) which has given rise to an unthinkable fusion: they represent a unity both stronger even than the pair of twins, since they are a fusion of it; but as a consequence they are also unique. Finally, it is their execution (intimated in war, asserted in sacrifice)

92

Comparison and conclusion

which displays their might, in that through it they bring about victory and success. There is, as I have noted, a subdivision between left-handed people and albinos. The former are avatars of male twins, the latter are the avatars of perfect complementarity, that of twins of both genders. The reader will recall, in the Nyamwezi case, the subdivision in the superior domain, between reversed born and twin-births. The rites accompanying the birth of these beings are also a staging of ambiguity in which the 'two' and the 'one' are mingled. It is possible to glimpse the sort of analysis allowed by a hierarchical perspective, when dualist oppositions are resituated within the order of values and when, as in the present case, the 2 / 1 opposition is understood as a conjunction as opposed to a lack. Twins presented as value refer on the other hand to the encompassed level of singularity. Twins, considered as the maximal expression of ambiguity in the relation between same and other, create the movement of the rite. Hierarchy, understood as a symbolic construction and as a ritual principle, is therefore at the heart of the system of beliefs and practices.2 RIGHT AND LEFT IN CHINA

In this, the final stage of my analysis, I will discuss Granet's text on 'Right and left in China' (1973) and La Flesche's (1973) treatment of the same topic in relation to Osage society. I have several reasons for being concerned with these texts. It is important, first of all, not to restrict the discussion to Africa. Hierarchy is not peculiar to any one geographical area or civilization. It is a characteristic feature, to a greater or lesser extent, of all societies that Dumont has described as 'holistic', i.e. of a large number of non-modern societies. Furthermore, Granet's and La Flesche's texts, written in 1933 and 1916 respectively, along with Hertz's analysis, constitute the classic early studies of left and right. They are as rich as Hertz's study and, in addition, they provide clearer formulations of the problem of reversal and of reference. This is not at all the remarkable premonition that one might at first judge it to be, in fact it is quite the reverse. These texts, from the early twentieth century, gave some indication of the simple nature of the right/left problem: only subsequently was this clarity obscured and this simplicity forgotten. These virtues are attributable to Granet's and La Flesche's respect, in their presentation of their examples, for

Other examples

93

the order of values, and I have already emphasised how necessary such an approach is. I hope to show how important La Flesche's treatment of the question of values is by using Levi-Strauss's commentary as a basis for a further discussion of the Osage case. Levi-Strauss employs the Winnebago case (which closely resembles the Osage case) to enquire into the nature of the 'dualist illusion' and also uses Osage ethnographic material to analyse the concept of 'totalisation' in classificatory schemas. In Osage, as in Nyamwezi society, mathematical addition - totalisation - involves levels; there is also the hierarchy i3>6 + 7 (society / the two moieties), which resembles the one which characterises Nyamwezi royal value: 2 > i +1 and 5>2 + 3Levi-Strauss's contribution to the study of the taxonomies through which the 'savage mind' operates also enables us to ask a question which lies at the heart of the debate on right/left classification. Does the 'dualist illusion' of the proponents of the binary method consist in treating non-modern thought as a faltering intimation of scientific classifications, and in reducing the manifold and pluri-dimensional configurations which are actually at work to schemas with two poles? And may it not be just as much of an illusion to identify in a layered and multi-level construction a universal code and a network of logical interconnections, i.e. a model which is admittedly more effective than a binary table, but which does not therefore entail any more of a conception of hierarchy? We shall see that Levi-Strauss puts his finger on the problem when he specifies that, in a given society, each person is not free as regards his access to the stock of collective representations. However, because he advances an autonomous analysis of the logical relation linking the different levels of representations to each other, Levi-Strauss does not succeed in barring the way to the rationalist illusion, which is ever liable to surface in contemporary analyses of symbolism. According to this sort of rationality, the notions of opposition and of complementarity can only have a single meaning, whether one is in the context provided by the mathematical thought of modern European society, where notions of equality of status predominate, or in that of traditional societies, where hierarchy has a part to play. I shall begin with the Chinese example. Writers often refer to Granet's text when treating the problem of reversal. While Hertz sought to establish the universality of the pre-eminence of the right

94

Comparison and conclusion

hand, Granet clearly demonstrated that there was an 'alternation of pre-eminence' within a given society. But this is not the salient feature of his exposition, which would then be nothing more than a logical demonstration that binary schemas are always liable to return. Granet has not simply replaced a mechanistic and static model with a structural one which, although it takes reversal into account, still has the coherence of a logical system. If reversal occurs, it is because the social whole is composed of 'correlative positions and qualities' which require numerous combinations. Thus, in passing from one position to another, directions may be reversed; several oppositions may be involved simultaneously (left/ right + high/low, for example; see below). But above all, these 'positions' are given in the 'representations' (the values) which organise the system and which are many in number: the pure or impure, profane or sacred positions are various and cannot all be grouped within the dichotomy of 'religious polarity' suggested by Hertz. Mauss was not mistaken in this respect, as is proven by the contribution he made at the end of Granet's delivery of his paper: 'Granet's advance is to have introduced mythology and representation into all this . . . this consideration in fact serves to account for the nature of rites' (Mauss, 1968, pp. 144, 146). Granet therefore attempts to give a more precise account of the different contexts and to highlight the whole from which each opposition derives, along with the kinds of reference involved. There is thus nothing surprising in our finding clear instances of hierarchical oppositions, even though Granet does not use the term himself. I shall examine a number of these here, reordering them in such a way that they serve to further my enquiry into the nature of hierarchy. Having considered the question of reversal, I shall go on to study the combinations of oppositions which give rise to such reversals, and I shall then analyse hierarchies of levels. The main point at issue in Granet's essay, if his introduction and conclusion are to be relied upon, undoubtedly concerns reversal as alternation. But it is not the only aspect to be considered, and I shall therefore limit myself to two instances. On the one hand, there is the Yin - right - female association, which is opposed to the Yang - left - male one. In salutations, boys cover the right hand with the left one, whilst bowing at the same time: 'They conceal the right and present the left'. Girls do the opposite. Yet, in periods of mourning, men reverse the order of the hands and cover the left hand with the right. Value is located in the Yang side, which engenders (encom-

Other examples

95

passes) the Yin; as a corollary to this, left dominates the general cosmology. But a witch is Yin and engenders Yang. Witchcraft dances have a step in which the right foot predominates. I will now go on to consider the more complex aspects of the problem. The example of oath-taking provides us with an insight which is already of a structuralist type. There are two forms of oath. In the first, the two partners clasp right hands; the corresponding ideogram consists of two right hands. This is the case with oathtaking in marriage or in war, which is sometimes followed by a blood-pact, the blood being taken from the right arm. In the second case, a 'solemn' oath is sworn 'under the eyes of the gods', and a sacrifice is performed; one takes a little blood from close to the victim's left ear (the ear has an important part to play in relation to the 'vital breath'). In the first case, the right prevails; in the second, the left does. But, Granet informs us, this is because another opposition, between the upper and lower part of the body, is involved (the hands belong to the lower part of the body). This latter opposition is itself linked to the other (Sky/Earth; east/west) oppositions. It is not so much that these oppositions all lead to one and the same dichotomy, but rather that their combination accounts for reversal. The Chief faces south and thus determines the relations between left and east and between right and west: the east is on his left. On the other hand, Sky and Earth, which were originally placed one above the other, as they should be, have shifted in such a way that an asymmetry in relation to their axis has been produced. The structure of the world is the same as that of the house of the Chief. It consists of a roof, which is round like the sky, a floor, which is rectangular like the earth, and one or more columns connecting the sky to the earth. A solitary column represents the Chief himself, whilst four columns stand for the ministers or the mountains situated at the four corners of the world. But, in a famous myth, a minister uprooted a column. Sky and Earth were thrown off balance, tipped over and shifted a little way apart from each other. An east/west hierarchy then appeared in the Sky, and a reverse west/ east hierarchy on the Earth. Finally, I ought to mention the relation to the body: the head, which is round, represents the Sky; the feet, which together describe a rectangle on the floor, represent the Earth. We can now return to the question of oath-taking: for the upper part of the body, the left is favoured (upper — Sky — east — left), whilst arms and hands are defined as belonging to the lower part of the body and the right side is therefore favoured.

96

Comparison and conclusion

An additional point ought perhaps to be made here. This detour by way of 'representations' demonstrates the coherence of the system and accounts for the form that reversal takes. Yet, in itself, reversal makes it plain that a change of level has occurred. We shall see below that the relation between upper and lower, Sky and Earth, breath and blood is a relation between encompassing and encompassed element. The difference between the upper and lower part of the body, which corresponds to a difference in ritual context (oath of peace—» 'solemn' oath), therefore represents a change in level (since Sky encompasses Earth), and the accompanying reversal is therefore to be expected. I shall now give a quick gloss on the Sky/Earth opposition. It would be easy enough to draw up a binary list in which Sky = upper = south = Yang = front, etc. as opposed to Earth = lower = north = darkness = Yin = back. Yet Granet strives to reconstitute the global structure of these representations. We have seen how the asymmetry between Sky and Earth - and, for each of these poles, between east and west - referred to the structure of the Chief's house and to the introduction of disorder occasioned by the uprooting of the column. Likewise, Granet states that the image of the world is a faithful reproduction of the ritual staging of the Chiefs reception of his vassals. The Chief keeps his back to the north and looks up at the Sky; the vassals, when prostrating themselves face north and press their foreheads against the Earth. Facing south, the Chief receives the totality of the sun's rays (the Yang light) on his face which, in opposition to his back, also becomes Yang. Reference should be made here to the crucial myths regarding the Sky and the Earth: the Sky (which is Yang and male) 'covers, clasps' (encompasses) and presses the Earth against its chest; the latter, on the other hand, presents its back and 'carries things on its back'. The Earth, being the mother, supplies the blood; the Sky, being the father, supplies the breath. This general opposition, presented as a hierarchy, then allows a whole range of different reversals to occur (in a famous book of medicine, the back becomes Yang and, as a consequence, the heart leaves the chest). What form does the totality take in these various oppositions in which only two terms are present? Breath and blood obviously refer to the body, and the former is presented as more vital than the latter. But here the representations involve no emphasis upon any particular level within the body as a whole. Indeed, breath and blood are like the Sky and the Earth, and these latter constitute a dyadic

Other examples

97

totality which is hierarchised, so that the one encompasses the other. We should, however, bear in mind that society is defined by the link between these two poles, at least as far as authority is concerned, for the Chief is the column which connects Sky and Earth. But the relation is an ambiguous one, for a column was shattered (uprooted) and the order (of asymmetry) was thereby established. Finally, this Sky/Earth encompassing implies that the opposition between Yang and Yin is not a binary one. It is not surprising to discover that Yang is in fact a composite of Yang and Yin. Likewise, the odd numerical order is a composite of the even and the odd one. The reader will also recall that the African examples analysed above featured, upon more than one occasion, another figure, in which two terms refer to a bodily referent which appears as such within the representation, being situated at a superior level. This problem of the 'triadic' figure will be frequently encountered in Osage classification, but we have already found that right, left and 'centre' are not situated at the same level and that the 'centre' may be the body as a whole. The representation of the Chinese Chief is not dissimilar to this classic schema, in that he is the axis, or rather the whole - the superior level - which hierarchises inferior terms. Men are placed in the west and face the rising sun; they are therefore associated with the east. Women are placed with their backs to the east and are associated with the west. The Chief, who always faces south, therefore has the men on his left (to the east) and the women on his right (to the west). It is also said that the Chief is Yin and Yang. It is common in the symbolism of royalty for the Chief to serve as a 'centre' for the division and hierarchisation, in relation to himself, of the categories. But my purpose in discussing this instance of it here was to contrast it with other ternary schemas which are unusual and which clearly display the left/right hierarchy. The Chief is an archer who rides in a chariot and who is always careful to face south. A red flag representing the sun and the south is used in military campaigns, to enable the chief to move about as much as is necessary. There are two men, apart from the Chief, on the chariot: the driver, who is in the middle, and a spearman who is a vassal. The spearman is right-handed and is placed on the right side of the chariot. The Chief is on the left, in the place of honour. Granet does not, however, comment upon this interesting combination of the south/north and left/right hierarchies. The chariot, which moves towards the south, and which is the royal chariot, is thus the point of

98

Comparison and conclusion

reference. Just as the Chief himself does in other cases, the chariot hierarchises left and right, with the Chief being placed on the left. The 'centre' is not in fact the central person (the driver), and the left side encompasses the remainder, i.e. both the geometrical centre (the driver) and the right side (the spearman). Likewise, the Chief is represented by three dignitaries, the 'left', the 'centre' and the 'right' one, and it is the 'left' dignitary who is honoured the most. This also allows us to undertake a study of the relation between classifications and royal symbolism. We have already seen how the Chief is sometimes ranged on the side of the encompassing element, Yang, that he has all the ambiguity of a mediating term (the column between Sky and Earth), and that he can be both a centre (between east and west, between men and women) and a whole (Yin and Yang). We have in fact already learnt that it is in the nature of the encompassing pole, in the case of the hierarchical opposition a/b, to be both a and b. The Chief is therefore both Yang and Yin-and-Yang. In the man/woman or right/left oppositions, the whole is the reference point, placed both between the two poles and at another level, which includes both of them. As for the uprooted column, it introduces on to the plane of myth the temporality of action, which destroys symmetry and creates asymmetry, and which thus articulates the different levels. 'Alternation' thus gives rise to encompassing/encompassed hierarchical oppositions and to the symbolism of authority, so long as the various terms are not isolated from the mythical, ritual or literary contexts in which they appear. It is Granet's great virtue to have shown that ' . . . so long as it is a complexus of positions, of powers and of purities that is operative - rather than one in which there is power without purity or power without opposition - the simple right-left division is then employed as an element of the whole' (Mauss 1968, p. 147). RIGHT AND LEFT IN OSAGE CEREMONIES

The African and Chinese examples have served to show both the diversity of the classificatory levels and also the profound unity of the underlying structural principles. I shall conclude my analysis of the principle which orders these levels by discussing the Osage example and thereby contrasting logical relations (as they are conceived in a universalist perspective) and hierarchical relations. I have already noted the importance of the Sioux example in

Other examples

99

Levi-Strauss's work, along with the brief discussion of the Osage system in chapter 5 of The savage mind (Levi-Strauss 1966): —

—

—

The general system of classification is tripartite: 'The Osage used to divide beings and things into three categories, one of which was associated with the sky (sun, star, crane . . . ) , one with water (mussel, tortoise . . . ) , and one with dry land (black bear and white bear . . . ) ' (1966, p. 59) Some contexts are organised in terms of dual oppositions: 'The rising sun emits thirteen rays, which are divided into a group of six and a group of seven, corresponding respectively to right and left, land and sky, summer and winter' (1966, p. 59). There are in fact a wide range of different schemas, binary, ternary, quaternary etc. which derive from 'the major opposition' of Osage thought. This concerns the relation between the two Sky and Earth moieties, the latter being subdivided into Dry Land and Water (1966, p. 143). I have already noted the even/odd and right/left oppositions. The points of the compass integrate the subdivisions of the Earth moiety: north is to south as Sky is to Earth, and east is to west as Dry Land is to Water (1966, p. 142).

Levi-Strauss draws several conclusions from the Osage material. I shall not lay undue emphasis upon his main argument, which recurs throughout the book, concerning 'the mutual convertibility of "concrete" and "abstract classifiers'" (1966, p. 142) and the critique of 'totemism'. The importance of these passages is common knowledge. There is, however, another theme, which is of direct relevance to the present discussion. Extending his enquiry into the existence of dual organisations (Levi-Strauss, 1968, French ed. 1958), LeviStrauss shows that, in the Osage case also, dualism is often only apparent, and that beneath it a ternary principle may be at work. Like the Winnebago or the Omaha, they have a circular camp which is divided by an ideal diameter into two moieties. But one of the Winnebago moieties has twice as many clans as the other, and one of the Omaha moieties has two chiefs and the other only one; in the case of the Osage, as I have noted, a moiety is subdivided into two sub-moieties (Levi-Strauss 1966, p. 140). Thus, depending upon circumstances, one and the same group may bring its binary or its ternary aspect to the fore (Levi-Strauss, p. 141), and one can study how in its thinking it overcomes the logical problems

i oo

Comparison and conclusion

involved in passing between schemas which have a different number of poles. In terms of the general perspective of Levi-Strauss's book, his trajectory is clear. He has wrenched 'primitive' thought from the prelogical matrix in which it had supposedly been enclosed before. But, in pursuing the dominant line of argument of the present work, we are bound to enquire into the specific nature of the levels at which each pole of the relations in question is situated and, more generally, to analyse the difference between the domains in which the various modes of classification, whether 'binary', 'ternary' or 'quaternary', are operative. The African example has given us cause to argue that this latter difference could well be a radical, indeed a hierarchical one, to the very extent that the opposition between the various modes actually assumed the form of a difference between the level of the totality ( 5 / 7 2 / 3 ) and that of the activation of certain of its elements (even / odd, right / left). The information given by Levi-Strauss concerning this problem, at the beginning of chapter 5 (1966), is too general for my purpose here. The upholders of the totemic theory had opted quite arbitrarily for one particular level, that constituted by natural species, and Levi-Strauss reacted against this, as I noted above, by insisting that the various levels were interchangeable. But, in doing this, he did not close the door sufficiently firmly against a systematic use of logical short-circuits between levels, the misleading consequences of which we have seen in our discussion of Meru symbolism. A 'level' thus does not take on the hierarchical meaning which is proper to it and which is revealed when one treats classifying as an ethical operation which is governed by the order of values. One can shed some light on this problem by tackling the question of the passage between different levels directly, and by analysing the concept of'totalisation'. Levi-Strauss has demonstrated the importance of this concept in the very same passage of The savage mind: it is by successive 'detotalisations and retotalisations' (p. 147) that the system progresses, changes plane and may, on the basis of a binary opposition, attain to n dimensions. This is how it is with the Osage conception of the number thirteen: —

—

the thirteen totalises the two moieties (one being associated with six, the other with seven), right and left, south and north (Levi-Strauss 1966, p. 145). It has a mystical value; the rising sun has, as I have already noted, thirteen rays,

Other examples

—

101

divided into two groups, one of six and one of seven, with their corresponding associations with moieties, points of the compass and seasons (p. 145); noteworthy actions performed by a warrior should be thirteen in number and, in the case of the dry land sub-moiety, are represented by the thirteen footprints of the black bear, and in the case of the water sub-moiety, by thirteen willow trees.

Commenting upon the second point, Levi-Strauss does admittedly point out that the rising sun (thirteen rays) effects a union between Sky and Earth (6 + 7). But the sun is also a star, and as such assigned to the Sky moiety, and the thirteen is thereby associated directly with this moiety. The third point tells us by way of war, assigned to the Earth moiety, of the association of thirteen with this moiety. In addition, this thirteen follows the subdivision of the moiety and characterises the warriors of both sub-moieties (dry land, water). Thus, thirteen, which is an expression of the totality (first point), in a dual form (left + right etc.) is also 'scaled down under the impact of the ternary schema5 (1966, p. 146) - Sky moiety and Dry Land and Water sub-moieties - so as to be assigned in its own right to the sky, to the earth and to the water: 'There is a "thirteen" of sky, a "thirteen" of land, a "thirteen" of water' (1966, p. 146). Levi-Strauss goes on to refer to other codings, such as coding by species, where two groups, again employing the opposition between six and seven, are duplicated and thus raise 'the number of units' to twenty-six. These different codings are seen as so many 'preconceived grids' (1966, p. 149) applied to the full range of empirical situations. But it seems to me that this general conception of 'totalising instruments', such as the thirteen, obscures a dimension that is essential to them, namely, the very one that gives them the power to function as totalities and which makes them into signs of reference. Such an approach in fact favours the comparison of different codings - numbers, species etc. - in terms of the nature or the quantity of the units involved and in terms of the dimensions of the basic schemas (binary, ternary etc.). If, in addition, one disregards the contexts in which Levi-Strauss's statements are made (i.e. the criticism of the according of a special privilege to the level of species), one's symbolic analysis of a particular society can be seriously vitiated. One may be defining one's analysis as a structural one and merely considering, in a symbolic whole, the

102

Comparison and conclusion

nature of its elements and its cardinal (number of elements), but if so it is important to remember that such a whole cannot be defined without including a wide relation of order (^). We have seen how some of the works criticised above succumb to errors of this sort. I therefore wish, in relation to the number thirteen in Osage symbolism, to posit the existence of an encompassing superior level, and to argue that, if the thirteen is this totality (first point), its other manifestations, even if they only concern a part of the society, are also an expression of the supreme reference. There would not be a 'thirteen5 for the whole and a thirteen for sky and for earth, inasmuch as the relation between a whole and one of its parts is hierarchical. At this point we must turn to La Flesche's actual text (1973), where, in his investigation of dualism, we may learn more about the thirteen rays of the rising sun and the thirteen noteworthy actions of the accomplished warrior. At the great Osage rituals, the two moieties are always ceremonially ordered as follows: North BA7654321 TSIZHU ('SKY')

West ('WATER')

765432 I

—>

East

HONGA ('EARTH') WAZHAZHE HONGA ('DRY LAND')

765432 I South

The members of the two moieties sit opposite each other, in two parallel lines extending east and west. The intermediate space, 'between sky and earth', is 'the path travelled by the sun' (p. 33). The officer stands in this space, facing east and, La Flesche adds, symbolises the unity of the tribe, in that he has the Sky moiety on his left and the Earth moiety on his right. The sun is also, in a general sense, the creative principle. A man contemplating it as it rises is venerating the source of all life. La Flesche also states that life derives from the dynamic union of the forces of the sky and of the earth. The sun is indeed a link of this kind, symbolised by the central path in the spatial organisation of the ceremony. There is also no direction more sacred than that of the rising sun, and the officer must always look that way (the west, on

Other examples

103

the other hand, is the place of one's enemies). The sun represents a superior level; the east is therefore marked by the number thirteen, which is the number of the sun's rays. Other rites are marked by the symbolic properties of the sun, and we find that it is something more than just a star, assigned as such to the Sky moiety. When it rises, it is neither 'of the sky' nor 'of the earth' but the thing that encompasses sky and earth, and gives them meaning, by defining and connecting them. In the initiation ceremonies, an eagle's downy feather is worn by the master initiator: '[it] symbolises one of the sun-pillars, sometimes seen on both sides of the sun as it begins to rise above the horizon. This feather is spoken of as the sun-pillar at the right or the one at the left side of the sun, according to the side of the tribal division to which the initiator and the candidate belong' (p. 34) (the italics are mine). What of the thirteen that designates the number of exploits of the accomplished warrior? War is assigned to the Earth moiety. But does not the accomplished warrior accede to a superior order transcending the division from which war derives? Has not the one with thirteen exploits to his name come to that prestigious 'solar' place where six and seven, sky and earth are united? A part of the recitations during the warrior's initiation rites develops the myth of the bear and the beaver. They tell us how these two animals (associated with the dry land and water sub-moieties respectively) have informed men as to the number of exploits that they have to perform. I shall give a very brief summary of these recitations. At the end of the summer, the bear looks for a cave in which to hibernate. He finds one and enters, hugging the right hand side. He remains there for seven lunar months. Upon waking, he drags himself towards the exit, taking care to remain on the right hand side feeling his strength coming back with the air of spring, he goes and makes seven steps. Contemplating the footprints that he had just made, he said: 'When my children, the Wazhazhe and the Tsizhu [i.e. the people of the two moieties (p. 149)] go forth against their enemies who dwell toward the setting of the sun, they shall strive to win honours, equal in number to these footprints.'3 Then he goes back into the cave, this time hugs the left hand side and moves six steps away. He then utters the same words as before. Next he heads for the beaver's dwelling. The second narrative then begins: the beaver, overcoming numerous difficulties, cuts down thirteen young willows and takes them home. In cutting them down, he takes care to make them fall in

104

Comparison and conclusion

the direction of the Osage's enemies (the west). When he has carried them all back, he places seven of them on the right of the door of his house and six on his left. During his labour, he says: 'This act of mine is not without purpose. My children, the Tsizhu and the Honga, shall use this sapling to count their war-honours' (p. 39). Thus, the bear and the beaver, who are manifestly dry land and water animals, have here assumed a superior, totalising value: the members of their whole tribe become their 'children'. The combination of references to both 'left' and 'right' (the two sides of the bears' cave and the beavers' dwelling), and union of the six and the seven, clearly point to a level which is no longer just the level 'of the sky' or 'of the earth'. One and the same thirteen qualifies the warlike exploits, the rays of the rising sun and the symbolic union of the two moieties. The level of the totality is involved in each of these instances. This is precisely what the continuation of the narrative quoted by La Flesche states: the bear and the beaver, animals of the Earth, have indicated the requisite number of exploits. But the Sky is not to be outdone, for it too has specified this number, in 'taking' thirteen rays from the sun, seven from the right and six from the left, 'for the use of the warriors in counting their warlike deeds' (p. 39). The warrior, who is thus triply imbued with this global value, may then place himself at the centre and rejoin, in the ceremony, 'the path of the sun'. He is placed in the middle, facing east, and thirteen willows are laid at his feet, seven to his right and six to his left. My own perspective will therefore be as follows. Rather than envisage a scaling down of the totality along the whole length of the chain constituted by its elements, which would leave the whole and the constituent elements at the same level, I shall argue that both of the two moieties, in their relation to the rising sun or to the state of the accomplished warrior, affirm that they each belong to the totality ( 6 + 7 = 1 3 ) , whilst at the same time they designate the superior level of this latter on the plane of cosmology and of status. The ceremonial sun is more than the sun that is a star in the Sky; the accomplished warrior, at least in the ritual in which his exploits are counted, is more than a representative of the Earth. The symbolic grammar of Osage thought does involve reference. The rules governing passage between levels, between binary and ternary schemas, imply a horizon of reference. Totalising 'tools', such as the 'thirteen' discussed here, provide us with a particularly good vantage point from which to scan this horizon.

Other examples

105

La Flesche's text is thus of particular value, for it sheds light upon the important notion of'passage through totalisation'. But I would like to emphasise, by way of conclusion, that this study is also of fundamental importance for right and left symbolism. Before Granet's study, and in contrast to the later studies of African systems of classification analysed above, this text demonstrates that the opposition between the two hands only has a meaning if one relates it to a body and if, when one introduces, alongside the right/left opposition, that involving points of the compass, one specifies the direction in which the body is facing. If one can speak of left and right, and if one can state that the Sky moiety is 'left' (and 'north') and that the Earth moiety is 'right' (and 'south'), it is because, on the ritual plane, in the organisation of ceremonies, a central path 'of the sun' (east-west) is defined, upon which the officer is placed facing east. Sky and Earth are then on his left and his right respectively. It is also, and above all, because, on the cosmological plane, the tribe is 'a man' whose body 'unites' the two sides of sky and earth: 'the Osage of the olden time, when forming their tribal organisation, divided the families of the people into two parts to represent the interdependent sky and earth, finally they moulded the two parts into an inseparable body and made it to symbolise a man standing in a position facing the sun' (p. 40). There is nothing startling about this. Just as for a long time now secondary school pupils have had to include, in their representations of electro-magnetic fields, He bonhomme d'Ampere*, who orients the schema, so too a bilateral symbolic organisation is inconceivable without a body serving as a reference point, without an image of the whole. We are now in a position to understand reversals. When one examines, amongst the Osage, the encompassed domain of warrior activity (on a general plane and not on that concerned with the status of the accomplished warrior), one perceives a reversal. In the ceremony preceding a departure for war, the 'symbolic man' and the officiants, in this case the war chiefs, turn to the west. The members of the two moieties are still arranged on either side of the central path but the dualism is reversed, with the members of the Sky moiety being placed in the south and those of the Earth moiety being placed in the north. Indeed, the Osage have not simply conceived of the 'symbolic man' as a static principle of conjunction: 'They did not leave him there like an immovable statue, but attributed to him a mind capable of thinking and a body having the power of motion,

106

Comparison and conclusion

as shown in the movements of the people when performing the great war ceremonies' (p. 40). One could hardly hope to find a better example to illustrate the argument that classification cannot be understood apart from reference, and that it is the operative nature of the taxonomy which causes movement to occur, i.e. on the plane of the model, allows changes in level to exist. 'DO DUAL ORGANISATIONS EXIST?'

I have used the analyses contained in The savage mind to comment upon La Flesche's essay, and I have tried to show how the data contained in both make the Osage example, so long as we bear in mind the difficulties of comparison, a kind of guarantor for the Nyamwezi one. I ought also to point out that, six years before The savage mind, Levi-Strauss had emphasised, with regard to 'dual organisation', how urgent it was to develop a form of research which would situate itself at a more general level than that of dualism alone, and which would then demonstrate the illusion, the 'logical subterfuge', that this kind of organisation represents (Levi-Strauss 1968a, p. 154). Several of the examples presented by Levi-Strauss in this essay serve to show that a number of the antitheses are 'falsely symmetrical'. Thus, the Sioux moieties, apart from the Sky/Earth + Water asymmetry already described (and the ternary schemes derived from it), are sometimes opposed in terms of the univocal/ambivalent nature of their mythical and ritual functions (p. 154) or in terms of such oppositions as being/becoming, synchrony/diachrony etc. (p. 153). As for the numerous 'homogeneous' antitheses, such as right/left, summer/winter, high/low, they are often accompanied by hierarchies: elder/junior, noble/commoner, etc. (p. 139). Indeed, dualism often rather awkwardly conceals a 'concentric' system, in which the two elements represented by concentric circles are 'arranged with respect to the same point of reference - the centre' (p. 140). Generally speaking, Levi-Strauss remarks, dualism presents itself as an inextricable mixture of dyadic formulae, whether symmetrical or hierarchical ('concentric'), and of triadic formulae. It seemed appropriate to remind the reader at this point of this warning regarding the dualist illusion, which was published four years before Needham's study of the Meru. However, the relation that Levi-Strauss seeks to establish between these formulae, and which the analyses in The savage mind

Other examples

107

Figure 5.1 huts

Dancing ^ ground % H Chiefs yam house DChiefshut*

o D

•

•

D O

Figure 5.2

will express a few years later, does not allow one to develop the hierarchical perspective suggested by one of those formulae, that involving concentric structure. Amongst the representations that he studies, he distinguishes between 'diametric' dualism, 'concentric' dualism and triad (p. 150). The first kind of structure, called 'diametric' after the circular Winnebago camp divided into two parts, is by its very nature 'imbued with reciprocity' (Fig. 5.1). It is conceived as the result of a symmetrical and balanced dichotomy. It may continue to be symmetrical and balanced throughout, but this is exceptional. More commonly, the connotations of the two terms are unequal: elder/junior, etc., and there is thus an apparent contradiction with the principle of reciprocity (p. 151). Concentric

108

Comparison and conclusion

structure, for its part, allows asymmetry direct expression. This is the case with the Trobriand village (pp. 136-7). At the centre there is the sacred plaza, with the dance ground, the chiefs hut and the chiefs Yam house. Around it, in a first circle, are all the yam storehouses, 'sacred in character and the object of many taboos' (Fig. 5.2). There is an alley, then a second, larger circle containing the huts of the married couples. Malinowski informs us that this is the profane part of the village. The centre/periphery - sacred/ profane opposition also extends to raw-stored/cooked-consumed, bachelors/married, chiefs place/place of the others, male principles/female principles, etc. Finally, ternary schemas are often tied to the first two structures. Thus, the sacred centre of the Bororo village (p. 142) is on the one hand the central pole of an opposition of the same type as is applied above to the Trobriand village. On the other hand, this centre is itself divided into three parts, two of which integrate the 'diametric' opposition of the moieties - they are each of them reserved for one of the moieties - and the third of which symbolises the unity of the whole village. Levi-Strauss's main concern is therefore to demonstrate that the concentric structure possesses both a binary aspect (centreperiphery; small circle-large circle) and a ternary aspect (the two circles •+• the centre or the two circles + the outside), which enables it to play the role of logical intermediary between diametric dualism and triadism. But the 'centre', which was responsible for the asymmetry, then becomes just one term amongst others: one of three. It is no longer a reference. The notion of 'logical transformation' dominates the argument, and it is therefore far more concerned with the number of poles involved than with the levels at which they are situated. As well as serving as a reminder of the absolute distinction between the rare cases of symmetry that do occur and false dualisms (p. 154), the notion of triad is used in relation to figures where the poles can as well be the whole, the set, as one or other of its parts. In the Winnebago case, the triad is Sky/Earth + Water (as amongst the Osage) (p. 155). It is a hierarchical opposition for, as LeviStrauss reminds us, a univocal function (the preservation of the peace) and a bivalent function (war + policing) are opposed in it, but it is above all an opposition between elements of the whole, here absent, that the society is. In the Osage example, this whole was the 'thirteen' and 'the symbolic Man'. In the Trobriand example, on the other hand, the centre (which serves as the image of the whole), a

Other examples

109

sacred and public place, is opposed to the two circle of storehouses and individual huts, whilst at the same time, by virtue of its sacred character and by virtue of the presence of the central storehouse, it hierarchises the two circumferences - with the storehouses being nearer and the huts further away. And in the whole that the centre of the Bororo village constitutes, the part that symbolises unity (the whole), the bororo, is represented on the schema as enveloping, and as encompassing symbolically, the two other parts associated with the two moieties (p. 141).4 There are two orders of problem here. First, one has to demonstrate, as Levi-Strauss's essay in fact does, that there is nothing illogical about the thought of a society which handles diametric, concentric and ternary schemas. They are all structurally interconnected, and the members of that society, depending upon their situation or the group to which they belong, can bring one or the other representation to the fore. There are rules, however. LeviStrauss notes that a particular type of representation is sometimes used by a particular group in the society (pp. 133-5 a n d P- 146) and that there is a 'not unimportant' freedom of choice (p. 146). Yet if there is in fact a not unimportant freedom of choice as regards representations, the above argument implies that the symbolic structures employed are not merely differentiated in terms of temporary context or in terms of the status of the person employing them but that, in addition, their own configuration, and their relations with each other, also involve hierarchisation. Each may be transformed into the other on a purely logical plane, but is irreducible to each other as far as the hierarchy is concerned. To bring a diametric opposition or a concentric schema to the fore is to emphasise the egalitarian perspective (diametric type) or, on the other hand, to imply a hierarchy (concentric type). It is, at any rate, a question of emphasising the implicit or explicit nature of the reference that is involved: it is implicit in the case of the diametric type, where reference is made to the total circumference, and explicit in the case of the concentric type, where the reference is given material form through a 'centre'. Depending upon the 'shape' of the global system of the society's values, one does or does not carry out a reversal, one chooses a level. We must therefore classify the different formulae. (1)

Pure diametric dualism. This, we are told, is very rare. It can only be characteristic, where it is in a dominant position, of wholly

11 o

(2)

(3)

Comparison and conclusion egalitarian ideologies, such as the modern type, and may perhaps be restricted to particular domains. One example would be the opposition in mathematics between positive and negative whole numbers. The centre, zero, becomes a mere point of division, the middle point of a straight line. This formula must be distinguished from all the others. Diametric dualism of the right/left type. This is very common. It is generally a hierarchical opposition, with an implicit reference, as the whole of the present book is designed to show (or else with an explicit emphasis, which features largely in the thought of the society, on a simple symmetry linking right and left; which would bring us back to the pure diametric type described above). Concentric dualism. This opposes the centre to the periphery, with one or several circles, and makes the reference explicit. Indeed, it represents it. Represented in the enveloping circumference, it corresponds to the symbolic idea of encompassing; materialised in the centre, it creates a 'field of attraction'. We must in fact distinguish between the geometric model of hierarchical opposition and the indigenous spatial representations.

The former may take the form of concentric circles arranged around a common centre, although I ought to make it clear that there is not a linear order of distance from this centre, which would replace the hierarchy with a gradation. Or, to put it more exactly, simple hierarchical opposition (encompassing/encompassed) is represented by two concentric circles. The exact nature of the representation of the centre does not matter. It should be thought of as the limit, which is by definition out of reach, where the two circles join after an infinite play of mirrors, defining, through a series of successive hierarchical dichotomies, a host of inferior levels, with the hierarchy gradually being less and less clearly defined. The spatial representations of the group, village or society, as encountered by the observer, will often be quite clearly represented by the centre/periphery opposition: a hierarchical opposition whose form depends upon the context. The totality is constituted by the centre and by the periphery; as, on each occasion, it is not the logical sum of the two poles, but the globalisation of the centre/periphery gaps, in which the combinations of ritual movements of integration and expulsion will at crucial moments reaffirm the totality. Royal

Other examples

111

enthronement in interlacustrine Africa consists both in bringing the king to the centre of the country, to the court, and in shutting him up there, and in sending some doubles of the king to the frontier, who are forbidden to return to the royal court and who must stay on the frontier forever. These centre/periphery oppositions do not pose a different problem from that raised by right/left ones. We thus come back to what seems to me to be the major distinction regarding dualist 'formulae'. On the one hand, there is the 'pure diametric' (i), which may be further qualified as 'distinctive opposition', as 'symmetrical opposition' or as 'binary opposition'. On the other hand, there is hierarchical opposition, which may take the simpler and commoner form of an opposition between whole and elements or the complex form of an opposition between element and element (with the two elements not being in the same relation to the whole), and which may appear as concentric (3) (encompassing/encompassed) or as diametric (2) (right/left; asymmetrical diametric opposed to 'pure diametric' (1)). Finally, there is the triad, which is no different; opposing two levels, it unites (2), in the Winnebago or Osage case of an opposition between moieties and sub-moieties, Sky/Earth + Water; or (3), in the example of the Osage numbers 13/6 + 7, and in the example of Nyamwezi royal value, where the 'two' is more than, and something other than the sum of two units (2>i + 1). The appearance of one of the above formulae, in a given context, and in a particular society, serves as a basic sociological pointer, so long as one has clarified the hierarchical position of the type of representation involved with respect to the global orientations of that society. Which is not to say that symbolism is a 'reflection' (or an 'inversion') of social relations, and that a given status only allows of a given representation. Quite the reverse. But, whatever the status of the actor, one has to consider the representations he sets in motion to judge whether a maintenance or a change of symbolic level is involved and, although there is no predetermined relation between representations and values, one has to decide whether the former are in accord with the latter or entail a rejection of them. Only when we have considered the result of such an enquiry in terms of status are we in a position to say something further regarding the laws of the ideological system of the society. Let me reiterate this point. Depending upon the pattern of the society's global system of values, the choice of a particular type of representation represents the choice of a level at which one is placed,

112

Comparison and conclusion

whether one intends to place oneself there or whether one is obliged to do so. It is, as structuralist enquiries have conclusively demonstrated, a logical act, within a coherent set of transformations, and a functional act, or what I would term 'a valorised act', be it positive or negative. It is not 'functional' in the sense that there are needs that must be satisfied or in the sense that there is a 'complementarity' which guarantees equilibrium, but because it is a fact-value and an idea-value. In a hierarchical context, the occurrence in a superior position of representations deriving from the 'pure diametric' type may be taken to indicate a reversal. In a context based upon egalitarian logic, on the other hand, it is the occurrence of the 'concentric' type which indicates a change in level. We are concerned here with a meeting of two planes, a plane of meaning whose logic becomes hierarchical, and a plane of function, whose law of complementarity becomes a law of value. It may be that this meeting is the occasion for the hollowing out of a path which enables us to apprehend symbolic efficacy directly.

6 Towards a comparative study of dualist symbolism

The various paths we have taken across the range of classificatory schema bring us to a small number of cross-roads in symbolic analysis. Though few in number, these cross-roads are hard to identify, and resemble the 'crossings of the roads' in the bush where the Nyamwezi sometimes feel the need to call upon their ancestors with sacrifices and, when a risky journey has removed them from familiar landscapes and daily certainties, to ask questions of the beyond. HIERARCHICAL OPPOSITION

We have contrasted binary opposition and hierarachical opposition, with

symmetry and complementarity characterising the former, and asymmetry and the existence of different levels defining the latter. In the case of the former, the opposition aid ('a/not-a') corresponds to a logic of addition: a+d=d+a= the set (or = zero), no matter what difference in value is allowed between the two poles. The totality is then simply a 'complementarity', such that one effectively intuits a 'homogeneity' between a and a, right and left, summer and winter, black and white. In the case of the latter, commutativity (a + d = d + a) is no longer meaningful, 'a' is posited as value, and thereby defines an actual domain for its contrary, 'not-a'. This domain is delimited by that of a and encompassed by a. We are no longer on one plane only, but have to take two different levels into account. The totality is composed of two terms, but it pre-exists them and gives them their meaning. Let me now give a more precise account of the notion of'totality'. I shall then go on to to the same for 'level'. Let the hierarchical opposition be alb, it being understood that b is not defined as a complementary but as an encompassed term. The totality is composed of two terms, so that S = {a, b}. This is not the same as "3

114

Comparison and conclusion

writing S = a + b\ the set is not a collection or a container, but the whole in relation to which a and b are ordered. This symbolisation serves, in general terms, to identify the hierarchical principle. But since it is in the context of ritual that classifications are apprehended, we must consider the totality as it actually manifests itself in representations. We have learnt from the examples presented above that various configurations occur. Thus, amongst the Osage, right and left, in relation to the two moieties, refer to the body of'the symbolic Man', which represents the unity of the tribe. On the empirical plane, the totality is, admittedly, composed of the two moieties but, on the symbolic plane, it is represented by the body of the Osage 'Man 5 . It is therefore the two poles of left and right which are most apparent, and it is their reversals which strike us first, but it is by referring to the body of the Osage 'Man' that we will understand what these reversals mean. The numerical classification consists of three terms: 13/7/6. The totality (the 'thirteen') here participates as such in the hierarchy and informs us that the fundamental opposition is the one which hierarchises the set with the element. The thirteen implies a cosmological and social value (the sun; the accomplished warror) and derives its symbolic efficacy from the fact that it is a totalising duality. For it is the union of two asymmetrical opposites, the two 'moieties', one of which is subdivided and one of which is not, each one respectively symbolised by the six and the seven. There is a sense in which the gap between the value of the 'six' and that of the 'seven' differs from the gap between these two values, on the one hand, and the value of the 'thirteen' on the other. The Chinese example likewise served to show that left and right are the two sides of the Chief, which thus divide the men from the women and hierarchise the left in relation to the right. In addition, we noted that Yang and Yin do not refer to a body. Both of them constitute the totality directly, which, however, is actually defined as hierarchical: Yang engenders Yin, Yang = {Yang, Yin}. Thus, to come back to the Chief, he is to his vassals as Sky is to Earth, as south is to north, as light is to darkness (as regards the arrangement within the palace). I shall now try to give a preliminary summary of the argument so far. Hierarchical opposition articulates two terms, encompassing/ encompassed, with the totality being composed of these two terms. I have in mind here the Nyamwezi colours (black and white), the Nyamwezi 'royal two' in its opposition to singularity, Sky and

Towards a comparative study

115

Earth, Yang and Yin, left and right (as in the Chinese example, where one has [left] > [centre and right], as in the spatial arrangement of the three dignitaries on the chariot), as well as the general oppositions between global and local and between set and element (along with the pure/impure opposition, which is characteristic of a hierarchical order). The 57/2/3 (Nyamwezi) and 13//7/6 (Osage) systems, and the 'symbolic ManV/right/left (Osage) and Chief//right/left (Chinese) systems actually belong here too. The level of the totality (five; thirteen; Man; Chief) therefore features both explicitly and separately in the representations but does not therefore result in another type of symbolic construction. A simple account of the situation would run as follows: those oppositions which directly express a hierarchy of values take on the basic encompassing/encompassed form; those oppositions which make some appeal to the universal representations of the body, more generally to those of space, and sometimes to those of colours or of numbers, etc., install a system which has its own formal constraints, whereby a right/left pair may oblige one to consider the body-image in which right and left are ordered. We have here the notion according to which every duality refers to a unity, just as every unity breaks down into two poles, at any rate if one is situated within a system. But we know how an observation of this sort must be further qualified. We succumb to the binary illusion when we concentrate exclusively upon these passages between duality and unity, or when we are only aware of the 'third term', thereby forgetting that the relation between the two initial terms unfolds on two distinct levels and that the 'third term', if there is one, must itself be defined in terms of the level in which it is involved. Besides, the 'totalising' third term is not obtained at the end of a dialectical process. On the contrary, it pre-exists its elements (cf. chapter 3 note 17). There is a more general sense in which the binary illusion may be said to derive from a constraint which our own ideology formulates. If one pays due attention to the oppositions advanced by Dumont, when he is defining the tasks of a comparative social anthropology, one readily grasps that the antinomy between a principle of symmetry and a principle of asymmetry, as I have sought to study it here, corresponds to the Individualism/Holism and equalitarianism/ hierarchy oppositions, in which Dumont discerns a confrontation between the essential values of modern society and those of traditional civilisations. If one goes on to append a further distinction,

116

Comparison and conclusion

emphasised in the same studies by Dumont, between a valorisation of 'the relation to things' and, by contrast, a valorisation of 'the relation between men5, one is in a position to grasp that, to put it succinctly, there is an opposition between a substantivist world, where entities (including 'symbolic5 ones) are wholly distinct from each other, and a structural world, where relations alone are at work. When we say that a is 'greater than5 b, as a general rule we are directly comparing a and £, as two substances, without necessarily, or indeed exclusively, being concerned with their respective places in the set from which they derive. We refer a and b to a quantitative scale and we measure them. It follows that, when b is greater than a, nothing has changed, unless it is the measurable quantity of one or the other of them. Consequently, the figure that is opposed to either of these two (a>b, b>a) is the figure in which a is equal to b. This is how mathematics opposes relations of order to relations of equality or of equivalence. Division by two does not have an important part in it and it is simply said that such a division is possible or not, depending upon whether the number of elements in a given set is even or odd. If there are only two elements, dividing by two means that a is equal to b. Everybody knows that when someone says 'divide that into two!5 that, if no further instructions are added, he is being asked to divide it into two equal parts. In short, where binary sets are concerned, if a is equal to b, division by two is sufficient to define the elements; if it is not, one counts them up. One remains within the domain of substances. On the other hand, in hierarchical logic, dividing in two is necessarily an asymmetrical set. The definition of a whole is here a matter of relation rather than of substance. How is one to express a relation without attributing substance to the terms, except by expressing levels? But along a single line, if one maintains the condition of non-substantialisation of terms and if, as a consequence, one encounters no materialisation at specific points, the relation is necessarily an indeterminate one. To define a dual set is then to name two levels (at least), for example, that in which a>b and that in which a
Towards a comparative study

117

In short, it is the possibility of attributing substance which enables symmetry to occur and which justifies a one-level logic. Without substantialisation, one must envisage a system with several levels if one wishes to express relations. This is why the opposition between modern society and numerous traditional societies is also a contrast between the valorisation of substance and the valorisation of relation, between the valorisation of relations with things and that of relations with men, between the valorisation of the principles of equalitarianism and the valorisation of the principle of hierarchy. This same logic requires that we conceive of reversal as an operation rather than a contradiction. If, for us, to 'perform' is to go from a towards b, in the logic of levels, to perform an operation consists in going from a>b to a[b and a
118

Comparison and conclusion

involve the whole of society, one can go on to inspect the variations (reversals, totalisation) of the various dualist oppositions which emerge in the course of that examination, so as to identify the more restricted and less obvious hierarchies, in which the definition of the superior element is not easily attained through direct observation. THE SEARCH FOR VALUES

The binary/hierarchical, complementary/totalising and symmetrical/asymmetrical distinctions are absolute. Indeed, the discrepancy between the two sides of these oppositions is actually a condition of developing a comparative approach to non-modern societies. The importance of the values organising the symbolic system must therefore be emphasised. Hierarchy is treated as a quasi-universal principle, and in this regard I have quoted Dumont's observations about Parsons. But might this not be an over-statement in the opposite direction? Having criticised the binary method because it reflects the idea of symmetry which is characteristic of our mode of thought, am I not in danger now of invoking a perspective which is peculiar to those societies which resemble the hierarchical model of India? Would it not be preferable simply to say that man has always thought by means of distinctions and that a value may sometimes be added, at a second stage, to these distinctions? And do not the many studies included in Right and left (Needham 1973) provide a wealth of examples of distinctive opposition (binary opposition), even if they have allowed equal status too large a part in the systematisation of ideas? I hope that I have already shown, through the various examples presented and through the comparative background to my argument, that the problem here cannot simply be one of greater or lesser emphasis. The debate cannot be formulated thus, for two orders of fact - reality and the sociological treatment of it - are confused in it. It is a truism to assert that every society is faced with a number of distinctions which go to make up reality, and no human group confuses an animate being and a corpse, a man and a woman, or the male and the female principle. All human societies recognise and treat the alternation of births and deaths, and of day and night, and the changes in the seasons. But this treatment, which creates a particular symbolic system out of a universal matter, represents a combination of these distinctions in accord with a

Towards a comparative study

119

differential principle. The order of values and of ritual contexts is the expression of this principle. To disregard the above point is to overlook the fact that the only social mode of apprehension of the real is a differential one. Without mentioning classes, orders or status, I shall simply recall what Durkheim and Mauss said in 1903: 'society . . . is the unique whole to which everything is referred. Thus logical hierarchy is only another aspect of social hierarchy ...' (Durkheim and Mauss 1963, pp. 83-4). Even if the 'concentric circles' which these writers mentioned when schematising 'hierarchy' represented for them 'interlockings' according to the clan model, whose static quality thus allows us to distinguish between them and what I have termed hierarchy here, and even if the term 'social hierarchy' is an ambiguous one, they had still made the crucial point that 'classifications . . . are systems of hierarchised notions' because they are referred to a whole (p. 81). Which is also to say that there should be no distinction made in sociology between the analysis of ideas and of values, for society itself does not separate idea from value. A society's symbolism is not a domain apart, which would involve direct access to the universal, and which would simply sustain specific relations with 'ideology'. Classificatory structure is an integral part of ideology. How have I treated the Nyamwezi case? In studying right/left symbolism, I have highlighted a part of the ideological configurations which define Nyamwezi identity, namely, reference to the dead, royal authority, the conception of (annual) time etc. If one is therefore prepared to accept that right/left opposition does not lead to the development of a universal taxonomic structure, and that the model governing its application (a strictly equalitarian one, a complementary one or a hierarchical one; a purely diametric or a concentric one; a binary one, a triadic one, etc.) depends entirely on the type of society that is involved, one is bound to acknowledge that the study of classifications is no different, in its relation to the global analysis of a society, to those studies which are concerned with ceremonial exchanges, for example, or with expressions of authority. We are automatically involved with the analysis of values. Furthermore, in speaking of a 'hierarchical logic' and in opposing it to binary logic, I did not mean to compare two analytic logics, one of which would be more adequate than the other for thinking the universality of the human mind. My opposition was in fact concerned with sociological method, and I have contrasted an approach

120

Comparison and conclusion

which allows a special role (to scientific and modern) logic as such and one which seeks to free itself from such a logic and to give a special emphasis to the values of the particular society. One may draw much the same conclusions from a consideration of the systems of prohibitions. On the one hand, ' . . . taboos are organised into a system and refer to classifications which enable one to give a meaning to everything ...' (Coppet 1973, p. 704). On the other hand, the 'logical hierarchy' refers to the 'social hierarchy5 (cf. Durkheim and Mauss, above; I use the term here in the sense of a hierarchy of values). More generally, 'the order of the universe and the social order are one. Nature and society obey the same law (Coppet 1973, p. 704). It is therefore clear that the agent of ritual and magical action who places a given object on the right or on the left, who anoints one or two hands, or who looks at the Sky or at the Earth, is not simply locating himself in relation to a logical order which would be sufficient in itself inasmuch as it offers a meaning, i.e. inasmuch as it is a means to distinguish between things. On the other hand,'... thinking like that, the agent attributes an ethical value to this ordering of the universe which is entirely composed of injunctions and prohibitions. Classifications extend . . . their ramifications as if they were the extensions of a knowledge whose foundation is the commandment of the Law. And the force of this Law is the greater the more it seems to attribute to man an omnipotence over the phenomena of nature' (p. 704 - my emphasis). In non-modern societies, man does not only distinguish, he valorises; in the very act of classifying, he hierarchises. His apprehension of the distinctions that the real imposes cannot be separated from the hierarchical movement through which he relates them to the whole of the society. We know how different western philosophical and scientific thinking, because of its 'rational' and individualist character, is from that of non-modern societies. Let me quote again from Durkheim and Mauss's 1903 essay on classification: The centre of the first schemes of nature is not the individual; it is society ... the creative force of the universe and everything in it wasfirstconceived as a mythical ancestor, the generator of the society. This is how it is that the idea of a [scientific] classification was so hard to form ... a concept is the notion of a clearly determined group of things ... Emotion, on the contrary, is something esentially fluid ... Its contagious influence ... extends to everything about it Thus the history of scientific classification is, in the last analysis, the history of the stages by

Towards a comparative study

121

which this element of social affectivity has progressively weakened, leaving more and more room for the reflective thought of individuals' (Durkheim and Mauss 1963, pp. 86-8; the italics are my own). Just because one has learnt to stop treating this 'history5 in evolutionist terms and to grasp it through a structural comparison instead, one should not therefore forget that such a comparison requires that one retrace the road 'backwards', in a metaphorical sense. In our observation of non-modern societies, we should set aside 'the reflective thought of individuals' if we &re to apprehend value which, like an encompassing element, 'extends to everything about it'. I want now to try and give a more exact account of the contrast between thought which valorises and 'reflective thought' by returning to the formal plane. I will begin by quoting Dumont's definition of hierarchy again: 'I will therefore define hierarchy as the principle by which the elements of a whole are ranked in relation to that whole,

it being understood that in the majority of societies it is religion which provides a perspective on the whole . . . ' (Dumont 1970, p. 66). But this definition is mainly concerned with the hierarchy of pure and impure which underlies the hierarchy of status in the Indian caste system. In other contexts, the analysis of the elements of a symbolic corpus, as drawn from a ritual whole, and the analysis of a dualist type of classification in particular, has caused me to place particular emphasis upon totalisation and therefore to stress a hierarchy of planes affecting the whole, which as such is situated on a superior plane, and the element. It was in fact problems of this order which recently led Dumont to propose a more wide-ranging definition of the hierarchical relation, which ' . . . is very generally, that between a whole (or a set) and an element of this whole (or set): the element belongs to the set and is in this sense consubstantial or identical with it; at the same time the element is distinct from the set or stands in opposition to it. This is what I mean by the expression "the encompassing of the contrary".' (Dumont 1980, p. 240); 'I think the clearest formulation is gained by separating and combining two levels. At the superior level there is unity; at the inferior level there is distinction.' (Dumont 1980, p. 242). Hierarchical opposition involves levels. This is the crucial point about the opposition, on the plane of logic and of logical systems, between the mode of modern thought and that of traditional societies.1 Modern thought has only one level of logic at its disposal, whereas traditional societies are

122

Comparison and conclusion

capable of developing an infinite number of them. We are concerned here with a crucial question in the history of thought. Dumont (1978, p. 99) has recently suggested that the opposition between the 'breadth' of the logic of myths and the rationality of modern thought is rooted in the contrast between the two types of thought whose formal opposition I have emphasised here. Indeed, 'mythical' thought and 'rational' thought each derive from a different kind of coherence, involving two opposed valorisations. Modern discourse, which only unfolds on a single plane, is forced to aim for an 'unbroadened' totality, cut into slices which are placed end to end. The progression is linear and discursive, and the logic is an instrumental one, involving the 'relation of means to ends'. Hierarchical discourse, on the other hand, is able to conserve the full breadth of the totality, since it is itself multi-dimensional, and since it has several levels at its disposal. I emphasised in my introduction how the comparison between 'holistic' and 'individualist' societies forced me to consider the question of hierarchy. It is significant that a consideration of the formal characteristics of hierarchy - the oppositions of levels brings one to this comparison, by way of a contrast between one-dimensional modern rationality ('reflective thought') and that hierarchical logic which has several levels at its disposal (where the order of values unfolds). Let me elaborate a little. If it is nowadays clear that mythos was not a faltering intimation of logos, a stage prior to the era of'logic', and if one admits that the world of mythos possesses its own coherence, one should perhaps add that mythos is not a poor cousin oUogos or even its equal, but its senior. It is not senior to it in terms of time (save in those Greek democracies in which the idea of symmetry did in fact take root, twenty-five centuries ago, when people took it upon themselves to place 'power' 'in the middle' {en to meso)), but in terms of a diagrammatic comparison of societies, in which a one-dimensional logic shows up as a special case by contrast with the configurations in levels which feature in hierarchical systems of thought, as if space, which numbers three dimensions, had lost one of them and was no longer anything but a monotonous extension. THE CHANGE IN LEVEL

The essential purpose of the present work amounts to this. Binary and hierarchical perspectives are radically opposed to each other,

Towards a comparative study

123

and observation of non-modern societies may well leave us with no option but to gravitate towards a logic other than that of symmetry. But once one has begun to consider hierarchy, other perspectives open up, and the contrast described above between the 'breadth' of hierarchy and discursive rationality is further developed. The study of symbolic wholes has led us to highlight the hierarchy of levels. But, in positing a social logic of the symbolic we are implicitly positing laws regarding its efficacy also (without which there would be no ceremonial system). Faced with the general opposition between hierarchy and rationality, one would therefore expect to find a contrast between two types of efficacy. Furthermore, it is clear that the opposition between levels and the formal model which derives from it does not yet provide an account, or at any rate not an exhaustive one, of how a society conceives of itself. For such a conception, which refers to a cosmology, is always made of space and time. The order of the world is constructed on the basis of hierarchical distinctions, but it attains to meaning through the (ceremonial) movement of the system. These three points together give us cause to reject the notion that ritual practices are incoherent. The opposition between a system on a single plane and one on multiple planes, the rule regarding efficacy, and the time-dimension, may together be held to account for the existence of rituals of 'reversal' and for their corollaries, rituals which involve derision, expulsion (of a 'scapegoat') or destruction (sacrifice), and which are sometimes too readily consigned to a domain of non-meaning. If the hierarchy of levels is opposed to linear discourse, it is clear that the efficacy of each must be differently defined. Where linear discourse is concerned, efficacy, i.e. rationality, is, as I have noted, the 'relation of means to ends'. In short, since efficacy is defined exclusively in terms of rectilinear movement, it is only assessed in terms of the means employed to arrive as directly as possible at the designated end, and the 'shortest distance between two points' is a straight line. Where a hierarchy of levels is concerned, 'rationality', as it were, is assessed in terms of the possibility of establishing a relation between several positions or between several levels. Efficacy is multiple in the sense that there is no longer a notion of the 'shortest possible distance'. At the same time, once one has abandoned linearity, one has by the same token distanced oneself from a rectilinear conception of time and drawn closer to a conception of time as repetitive. Efficacy is then measured, on the very general logical plane under consideration

124

Comparison and conclusion

here, in terms of the relations between levels, and in terms of changes in level, which would be the only possible means of measuring a movement.2 In a hierarchical system, movement is characterised by changes in level. I have already used this expression when speaking of 'passing' from, for example, the level (a OR b) to the level {a AND b). This is not simply a metaphor, employed in the description of a model. The sociological change in level affects the social actor and the ritual object in several of the stages of their ceremonial journey. It is therefore to be expected that these journeys would be as frequently punctuated as they are by rites of reversal, sacrificial rites etc., in short by rites which manifest - and realise - a change in level. I want now to give a brief but more detailed account of the place of these rites, simply in order to delineate an area of study. The idea of movement is constituted by a succession of presences and absences. If one opts for a linear approach and a subs tan tialist perspective, one quickly recovers the object which was lost from view, for it was simply 'further on'; to observe a movement is here to follow a material body which shifts. For hierarchical logic, in which it is not a matter of following in a continuous fashion the trajectory of a body, the alternation between presence and absence is termed 'reversal' or 'expulsion', or even exchange (when, at any rate, it presides over a change in value), and 'sacrifice' also. This is because all these rites share the characteristic of operating a change in level. I would maintain that, generally speaking, it is the absence, not the presence, of such rites that should occasion surprise and a sense of there being a 'contradiction'. If a holistic system did not have them, movement itself would disappear, and the society would be frozen within a timelessness, since the (ceremonial) movement of these societies may only be defined 'from one level to another', through a sequence of transformations in which a new form - a reversed one, for instance - appears, which is thus seen to be affected by a new value, which is a new value (this 'form' not being that of a substance but that of a relation). In short, since one is no longer in a universe of substances and material movements that one may trace along a continuum, the ritual norm is the staging of modifications (of 'changes in level') in the order of values. The rite of'reversal' is thus to be expected, when one also knows - and the present work has demonstrated this - that reversal is one of the figures which marks a difference of level in the taxonomic model. Reversal does not constitute a contradiction. This is also the case

Towards a comparative study

125

with rituals of expulsion (of a 'scapegoat'), if one takes the totality of the circuit into account there also. Expulsion is not necessarily a casting out of something into an annihilated outside; it may instead be part of a circular movement constructed upon the basis of an inside/outside hierarchical opposition. Centripetal movements, which guarantee a completed circuit, correspond to centrifugal ones, but with changes of level at points at which passage occurs. That what it true of reversal is also true, mutatis mutandis, of expulsion, is something that I have already suggested in my analysis of the relation between diametric dualism (the 'not pure' type; type 2) and 'concentric dualism'. In short, expulsion is to the second figure what reversal is to the first, and there is no need to have recourse to a psychology of 'catharsis' to account for the phenomenon. Consider what was said above regarding colours. I insisted that the black of witchcraft (inferior level) was not different from the black of the rain (superior level), so that one could speak of reversal and of a change in level. The blackness of the rainclouds is a secondary rationalisation; the royal black is first of all the reversed valorisation of the black of the inferior level. I would even go so far as to say that the black of witchcraft is not only the inferior pole (of the inferior level), but the negative and the excluded one, and the arrival of the rain is in a certain sense the return, this time with a value that is superior, of this excluded element. The witchcraft relation is the exact opposite of the relation with the ancestors, and the divination which allows one to accuse a witch results in the expulsion of this latter (whilst the ancestor is 'installed' amongst his own people through sacrifice; cf. Tcherkezoff.1985). The witch is banished or killed, and his body is thrown out into the bush (instead of being buried in the village). Now, the exterior with which the witch comes into contact again is not devoid of meaning, even if from the point of view of the village world it is nothingness. The bush is the locus of universality for the Nyamwezi. It is the world of the lions, representatives of the royal principle in general, symbols of the community composed of all the Nyamwezi kingships. It is the world from which the founding kings of each kingdom came. It is the world of the rain, for rain always comes from beyond the horizon. It is also the place where the brotherhoods meet, associations of an individualist character which are spread throughout the whole country, transcending the local dimension of the kingdoms. These brotherhoods, which are many but which resemble each other, are

126

Comparison and conclusion

also the places where a diviner's career is established. The bafumu diviners are the necessary intermediaries between men and the ancestors. Even the king cannot do without them, and only the ritual association between king and diviners enables the rain to come each year. Now, it is apparent that belief in witchcraft is predicated upon the existence of a form of brotherhood. The witches gather and organise in much the same way as the genuine brotherhoods do. There is a continuity between the black of witchcraft and that of rain, between the exclusion of the witches and the annual return of the rain, and other facts reinforce this. Witches who are killed are devoured by hyenas; the witches enter villages at night in the form of shadows clinging to the backs of these same hyenas. Among the wild beasts which the Nyamwezi strongly oppose to cattle, the hyena occupy the last rank. In the same category, but in the first rank (united by the proximity enjoyed by extremes) are the lions, images of the devouring kings (the original conquering kings; the kings which 'eat' the guilty and which 'eat' the fines; the witch who kills, 'eats' his victim). They are brought to court, following an encounter with a hunter, so that their skulls and their skins may be exposed so as to form a kind of ritual tree. At the foot of this tree are piled up the riches 'eaten' by the king (cattle, hoes) which will feed the court and will enable the king to carry out kuhoja expulsions (bloody sacrifices) which will ensure the regularity of the rain. If the rain is to be controlled, this savage world must be integrated. Driving something out towards it thus involves a system, and the hierarchical system of colours does not describe a model that is different from it (black ( —) —> expulsion-integration —> black (+)). One could go on to note how it is that twins born of commoners are first of all excluded from society. The appearance of duality outside the court, and without being processed by ritual, is a monstrous thing, and a birth of this sort is an intimation of drought. The twins go on to reintegrate the society, but in 'being reborn' at the court. They no longer threaten drought, but serve rather to guarantee the cycle of cultivation. In considering this ritual of exclusion and reintegration, we also traverse the levels that we have traced in our study of number symbolism (2 ( —) —» expulsionintegration —» 2 ( + )). It is necessary to go still further and, in the wake of these remarks on rituals of reversal or of expulsion, say a word regarding other rites which, though they have long had a place in anthropological theory,

Towards a comparative study

127

are often seen too much in terms of a universalism which is simplistic. I have in mind ceremonial exchanges, and particularly those which follow a murder or a declaration of marriage. One's perspective on exchange, in its generality, is profoundly affected by the existence of hierarchised levels. Exchange which is unequal, unbalanced and asymmetrical (to use a negative description first of all) becomes the norm; exchange involving equality of status only features in a subordinate position. In the case of exchange involving equal status, movement remains on the same level; one is brought back to the example of linear movement (a single level). Thus, in some societies it is characteristic for revenge for a murder, for the exchange of murders (a death for a death), to be a subordinate sequence, and a weak or even ineffectual response when compared with a sequence of murder —> 'blood price', where a payment follows on from a murderous attack (and this payment would seem to differ from any notion of a 'fine' or of'compensation' which would seem to be equivalent to the death - cf. Tcherkezoff, 1986). Some societies regard what in our eyes would constitute a return to equilibrium, the 'law of talion', as a sign of incompletion. The Nyamwezi example shows that revenge involving a murder is rare by comparison with a sequence of 'prices'. Furthermore, such a sequence involves a great number of different variants. Even in the case of payment, it is clear that what might still pass for an equivalent response (which in fact it is not) is completed by another prestation, which completely rules out any interpretation in terms of equilibrium. The 'blood price', njigu (the 'fifty') symbolically reconstructs the body of the murdered person; but in addition a goat which must be cut in two is required. Only then is the 'peace' truly established, through the setting to work, in the lethal act itself, of the superior 'head' + 'back' figure. There are other particularly illuminating examples, which serve to show how it is that after a perfectly symmetrical exchange of murders and even of money, a final large-scale payment unbalances the whole series and then, and only then, enables peace to be established. This is true of the 'Are'are of Malaita, who develop a system which is all the more striking for the fact that, in certain cases, the payment which is made 'in addition' is made by the original victim's group. 3 Now, analysis shows that in these examples the money paid represents all of the dead. Being then the image of the whole* it constitutes in my opinion a referent at the superior

128

Comparison and conclusion

level. To pass from revenge to a response involving money is to pass from an egalitarian response5 to a hierarchical one, and is thus to pass to the superior level; the efficacy of the 'blood price5 is thus coherent. There is a sense in which the 'bridewealth' is also affected. Consider the African lobola, which is regarded as a kind of extension of the model of generalised exchange. Undoubtedly, the lack of a positive marriage rule is counterbalanced by this obligation to give livestock. Thus, in each group, the livestock received for a sister will return with the brother, thereby establishing a circulation which ideally will be completed when the livestock given at one point or another of the circuit, 're-enters the corral' (Levi-Strauss 1968a, p. 00). But these payments, insofar as they actually make a total circuit out of the various marriages, though at the level of livestock and not at that of women, represent a value superior to what would be involved in the 'purchase' or the 'compensation' given to the group for 'the loss of a woman' (cf. Radcliffe-Brown). This is certainly true for the Nyamwezi, as it is for the Swazi, to whose language the term lobola belongs. Upon closer inspection, kulobola in some sense means the buying of life, and the buying of it in order to pay the dead (with livestock-money, images of the dead). I will close this parenthesis on exchange here, in order to return to the other rites which render still more doubtful the claims of our logic to give a coherent description of them. This is the case with sacrifice, as I shall try to show by presenting several examples encountered whilst considering Meru, Nyamwezi, Dogon and Chinese symbolism. I have repeatedly pointed out that a ritual act or a mythical event 'renders asymmetry explicit'. In creating warriors, whose bodies are composed of two halves of different origin, the Meru Mugwe establishes an asymmetrical totalisation. But, in order to ensure that the operation is effective, he inserts one of his horns, which is meant to effect a kind of soldering. In marrying two Nyamwezi fiances, the kanumba crosses their hands and reverses the 'hand of the bow'/'hand of the back' distribution. But their hands must hold a bugali (sorghum paste) ball, as in the case of a sacrifice, and duality is also expressed by the balls themselves, with the first being cast to the ground and only the second being eaten. In the sacrifice of the goat 'cut in two', which accompanies the 'blood price' prestation, its efficacy depends upon the 'head/back' configuration, which the sacrificial cutting up affirms in the clearest possible manner. In Nyamwezi sacrifice also, the red colour of the blood is the chromatic expression of the threshold

Towards a comparative study

129

which articulates the black/white hierarchy. The purificatory sacrifices, in which anointing with blood is essential, enable an impure act, a 'white' one, to be transformed and its author to be 'blackened', i.e. purified (the victim, as we have already learnt, is black and/or white; a coat composed of brown, of 'red' spots, means that the animal is unsuited for sacrifice). In Dogon myth, it would not have been sufficient for mixed twinship to have replaced the single grain in order to have instilled life in the system; it would also have been necessary for one birth to be premature and for the twin couple to be thus itself unbalanced. From then on there are true twins and there are false twins. Finally, in the Chinese myth, the Sky/Earth opposition only becomes significant when the column which links the two poles is shattered. Sky and Earth then lean slightly, and the contrast between their reciprocal orientations become manifest. I cannot pretend that the above list is a properly comparative one, for what I have done is to group together isolated facts rather than compare social totalities. But it is meant to illustrate what an extended analysis of sacrifice - or at any rate a certain type of ritual putting to death - might serve to show in holistic societies, namely, that sacrifice can effect a change in level. Its efficacy, its operative character, derives from the fact that it manipulates a class of particular objects, which participate in both of the two levels on which sacrifice is situated. 6 The change in level is no longer manifested through hierarchical succession, such that a > b —> a < b (reversal) or inside—* outside (expulsion), or individual—> money (exchange), but through the transformation of an undifferentiated object into one whole explicitly composed of hierarchised parts. The horn of the mythical Mugwe which links and solders the two moieties of the Meru warriors is clearly the guarantor for the Mugwe's 'tail' or 'left hand', in that it is both a part of the body and the whole (since it is the container of Ugwe, the essence of the Mugwe's authority). It is characteristic in this respect that these horns are said to be 'hollowed out' and 'nobody could tell what was inside them'; likewise, the left hand of the Mugwe must always 'be hidden'. Finally, the horns are two in number; one enables one to reconstruct the first warrior, the other enables one to reconstruct the second. They used to be together on the Mugwe's head, but now they are separate, each in the body of a warrior. The ball ofbugali for its part is what passes from left to right; it is also what is in part thrown away, in part eaten (as in sacrifices to the

130

Comparison and conclusion

familial ancestors). In addition, the livestock sacrificed in the putting to death of the dual type ('in twos') is defined both as a local 'good' (nsavo) and as a global good, an image of the local group and of the close ancestors and an image of royalty and of the collective ancestors. Red, the colour of blood, has an essentially ambiguous definition, for it both favours the coming of rain and, through other aspects of nature, stops the rain. It is an image of death, but of a death which is at the same time the source of renewal of life (the gushing out of the kuhoja blood is an image of the gushing out of life; every birth is a kuhoja going out, a kufuma). I have already mentioned the ambiguous status of left-handed persons and albinos in Dogon society; it stems from the situation created by the mythical premature birth (twinhood/singularity; perfect twinhood/failed twinhood). Finally, the Chinese column is by definition both 'of the Sky' and 'of the Earth'. In rites and in myths, the change of level may assume the form of a destruction of an object which is suited to constitute a hierarchical threshold. This operation would seem to define a certain kind of ritual destruction, in which a hierarchical transformation is effected, the unitary notion of sacrifice thereby being dissolved and replaced by the alternative options of a change of level or not, and of an equalitarian or a hierarchical movement. If one connects these remarks with my suggestions regarding rites of reversal or of expulsion and regarding exchanges, one will grasp that a hierarchical perspective allows a new approach to ritual, whereby it is conceived as a succession of transformations in an oriented space. It is clear to what extent egalitarian logic can create illusions. As far as it is concerned, reversal is 'contradiction', exclusion is the liberatory expulsion of an excessive fullness which re-establishes the prior harmony, exchange is closed when it is 'balanced', sacrifice is a 'communication' between profane and sacred or a 'communion' between the various participants in the ritual. In short, the inequality which dominates the movements of the ritual has to be reduced, for otherwise it remains contradictory. According to a hierarchical perspective, on the other hand, reversal is to be expected, non-reciprocity in exchange is the norm, and sacrifice and expulsion may serve as the means of a transformation. This latter notion becomes an operative concept, for the orientation of a society's symbolic space (the hierarchy of values) turns certain transformations into changes of level by contrasting them with other relations, which remain equalitarian. One thus avoids modes of

Towards a comparative study

131

interpretation which are ethnocentric (assumptions about equalitarianism, symmetry, egalitarian exchange, which are all 'modern' values) or psychological, with the animal victim replacing the man in the sacrifice, the blood price money replacing the corpse which would have corresponded to the first murder, in the sense that these substitutes would allow one to avoid having recourse to sacrificing a man and would 'eliminate' violence. One also avoids a binary division by means of which the rites would be classified: culture/nature, power/magic, power/reversal (in the sense of'rites of reversal'), structure/marginality, etc. One is then also in a position to transcend the alternatives that anthropological theory comes up against when it reflects upon the nature of symbolic efficacy (is the symbolic system a system of signs which refers to a universal meaning or does it function so as to conserve the structure, i.e. to reduce tensions whilst at the same time maintaining the differences?) (cf. Auge 1978). These remarks suggest a possible direction for anthropological research. Whilst a universal theory of'sacrifice' (or of'exchange') as such would represent a turning away from what is essential, one can investigate how societies organise equalitarian and hierarchical relations and which group they place in a subordinate position. This direction stems from the highlighting of the individualism/holism opposition (with its corollaries symmetry/asymmetry, 'flat' rationality/hierarchy of levels). Similar perspectives for the study of ritual will, I hope, emerge, as other anthropological oppositions between modern and traditional social values become available. The emergence of further comparative axes - in addition to that of holism/individualism - should each time offer new possibilities for the analysis of non-modern societies.

Notes

CHAPTER I RIGHT/LEFT OPPOSITIONS

1 There is also some difficulty in dispensing with the third pole, the red one. Bernardi has not presented any pointers as to this earlier ternary scheme. He does however note that the Meru class as 'red' the colonisers who made the Meru 'similar to women'. This valorisation of red, in a context involving war, refers to the victory of the 'red' clans over the 'white' ones. This 'violent' aspect of'red' is not an aberrant or unfamiliar feature of this region. We also find the Nyamwezi describing the colonisers as 'red bulls', in a system which is also based upon the black/white opposition. The violent aspect of the red, and the need to take account of this colour and of its ambiguity in studying the black-white contrast, has been strongly emphasised by V. W. Turner in his article on colour classification in east and central Africa (1966). It is however significant that Turner comes up against the contradictory aspects of the system elsewhere. Situating his article quite explicitly in a line of studies that run from Durkheim, Hertz, Levi-Strauss, Needham (on the Meru), as well as Leach, Evans-Pritchard, and Beidelman (on the Kaguru), he remains in this respect on the same terrain as his predecessors, drawing no definite conclusions from the contradictions, 'ambivalences' and 'ambiguities' that he notes. In a monograph that I am drafting on the Nyamwezi, I will present an account of the red-black-white triad. Here I will only tackle the black/white pair. In fact, in the Nyamwezi case, and doubtless also in certain comparable cases described by Turner, red is in a certain sense outside of the black/white system (cf. Luc de Heusch's remark to this effect, regarding Turner's article, in his introduction to Bradbury, Geertz et al., 1966, pp. 14-15). It is not merely on another level from that on which black and white are contrasted, which might lead one to believe that it is a pole encompassing the two others; it is also outside the hierarchy of levels constituted by the taxonomy of colours. It is not a colour-value, but a sign of the sort of ritual operation (sacrifice and, by extension, war and the destruction by fire of royal emblems) which is necessary in ceremonial practice for there to be a change of level (cf. chapter 6). Red is the operator of the reversals occurring in the case of black and white. [For a study of operators in a change of level, see Tcherkezoff forthcoming.] 2 'the Mbere and the Kamba were famous for their medicines, it was assumed by the Meru elders that those medicines had been stolen by mischievous treachery from their own Mugwe. This story might have been the reason for the opinion formed by some Tharaka elders that there was also a Mugwe among the Kamba 132

Notes to pages 25-30

133

and the Mbere: an opinion that did not hold good when I extended my research to those tribes' (Bernardi 1959, p. 69). 3 Nor have I quoted from the article on Nyoro symbolism, in which a far richer material is presented, and in which Needham goes some way towards posing the problem of reversal. But, on the one hand, the basis of his method remains unchanged and, on the other, it seemed important to analyse the study which served as a point of departure and as a reference for the subsequent analyses of several researchers. Some of this article's conclusions have on occasion been treated as proven fact (cf. J.-F. Vincent 1978, p. 504; and below, chapter 5 note 2; cf. Needham 1980). CHAPTER 2 NYAMWEZI CLASSIFICATION

1 The great diviners are given royal funerals, organised by the court. Their tombs, like those of kings and of the arch-priest mgabe, remain for a long time a place of veneration and of sacrifices, in contrast to the tombs of ordinary men (whether commoners or nobles), where such things are not held, and where it is deemed sufficient to honour the calabashes, jewels and maholero domestic animals which, through being consecrated, are turned into the ephemeral or lasting receptacles of the soul of the dead person. 2 In distinguishing the mizirnu from the badugu I do not mean to suggest that there is, strictly speaking, a distinctive opposition involved. First of all, badugu is a term designating a kinship relation. It is applied to all living bilateral kin and to those who are dead, going back three generations. Mizimu refers to the quality of being an ancestor, which is acquired by the 'dead' (vafwa) when, about six months after death, when the 'soul' (myuye) had left the dead person, leaving only a 'dead piece of wood' (igogo), this 'soul' appears to his descendants, obliging them to sacrifice to him. Thus, the badugu ancestors are mizimu. However, this general term is not employed in the context of familial sacrifice, when the Nyamwezi can sacrifice to close ancestors, whom they are able to identify, and whom they term their badugu (kuhoja badugu, to sacrifice to one's badugu (ancestors)). The mizimu are thus, in relation to the living, the ancestors to whom one can no longer sacrifice, for they are from then on too distant, and with whom one must envisage another type of relation through: —

the royal sacrifices, fed by the livestock coming from each group (where it served as an image of the badugu ancestors) and carried out at the court and addressed to the royal ancestors, badugu ('kin') of the king and by extension badugu of all (the king is the 'father' of all; but the people speak of them as of bahanya, of'grown-ups'); the memory of the royal ancestors is perpetuated endlessly, and they thus subsume all the mizimu of the kingdom in a vast kinship system; — the ceremonial exchanges involving a 'price', in which the treatment of these 'prices' (in livestock) is linked to a sacrifice (sometimes these 'prices' culminate in the court, thereby bringing us back to the preceding circuit). In short, the badugu are, strictly speaking, mizimu. One should speak of (mizimu) badugu and of (non-badugu) mizimu. More simply, one could follow Nyamwezi usage and speak of badugu ancestors and of mizimu ancestors, of close (familial)

134

Notes to pages 30-2

ancestors and of collective ones. Among the royal ancestors (mizimu wa batemi 'ancestors of the kings', or bahanya, 'great' ancestors), one should differentiate between royal ancestors in the strict sense (the dead kings) and 'royal' ancestors (who have become such by assimilation), and, as regards these latter, one should further differentiate between those who are explicitly such: (a) twins, reversed-born, and untouchable criminals and (b) in a certain sense all the mizimu of each person beyond three generations, to the extent that the Nyamwezi also employ the term bahanya in their prayers to designate those ancestors who are beyond nameable categories and that one sometimes invokes alongside the principal named ancestor who is the intended recipient of the sacrifice. 3 The latter are not generally distinguished from the Nyamwezi and the Sukuma (cf. Cory 1953, 1954; Bosch 1930). 4 As is frequently the case in this part of Africa, the ruling dynasties are made out to be invaders. In the traditional ethnography these are seen as instances of the 'Hamitic' invasions which brought the civilising element of the State, or of an embryonic form of it, to the bands of Bantu hunters or agriculturalists. It is easier for us today to assess this confusion between 'races' and linguistic groups and to see through the special value accorded by Europeans to the 'Ethiopoid' physical type, such that they could not help but see it as the one appropriate to 'nobles' and conquerors. Even if we set aside the question of the problematic existence of these recent invasions, historians in fact opt for a more distant origin for the system of butemi chiefship (Cf. Austen 1969; Gottberg 1967). On the other hand, Nyamwezi-Sukuma society has no distinctions between the different orders of the tutsi I hutu I twa type, which exists in Burundi, and which interweaves noble and commoner statuses or pastoral and agricultural occupations (distinctions which, at any rate, even in Burundi, may not be directly superimposed). The 'external' character of the batemi (plural of ntemi, 'king'), which is undoubtedly connected in some way to the numerous migrations which criss-crossed the whole of this region prior to the nineteenth century, is, during the period in which I have set this study, a structural fact. As opposed to all the others, the king is 'the master of the meat' and the 'great devourer' who is always marked by duality. He is thus the one who comes from the outside to set up the sacrifice of the bull, which enables the creation of the world to be finished; but he is also the devourer of the inside, the one whose court, at the centre of the society, is the place where one comes to 'have oneself be eaten', i.e. to die symbolically in certain kinds of breach of prohibitions (cf. note 8). On the other hand, beyond this duality involving a 'devourer of the outside and of the inside', the 'outside' character of the kings is opposed to the autochthonous character of the authority of the arch-priest mgabe, the 'master of the mhojct (cf. Chapter 4, n. 12).

5 These acts include the defacement of a lion's hide, where 'the king is a lion'; bringing twins into the world, where 'the kings are like twins', followed by the destruction of the entrance gate to the royal court; the murder of a man, followed by the destruction of the cult objects of the royal ancestors. All of these acts render the subject both criminal and untouchable. It is among the whole set of 'nobles' (king's 'councillors', 'royal kin', 'kin by earth', criminals and untouchables) that are to be found the dignitaries who, according to the traditional

Notes to pages 32-7

135

ethnography, provide the core of the 'administrative and territorial hierarchy': royal councillors, province chiefs, village chiefs, along with the 'religious' figures, such as the ngabe or ntemi wa mhoja, the high-priest who is 'master of prosperity', and the great diviners (bafumu bamanga), whose proper place, in this functionalist perspective, cannot be properly defined. 6 The handful of men who opt for the path of transgressive destruction in the manner indicated in note 5 above should undoubtedly be excluded. Their symbolic death means that they take leave of commoner status forever and therefore even go so far as to abandon their wives, their children and their ancestors, so as to become forever 'the king's children', living at the king's court, and serving as ritual agents without however having the kind of cultic responsibilities that a father of a family will hold. 7 The royal ancestors (in the strict sense of the word) that one is supposed to honour are not actually all members of the one lineage or clan but, as I have said, just the dead kings, who succeed each other according to a matrilineal principle ('sister's son'). 8 This work clearly differs from the 'corvee' on account of its religious context and on account of the symbolic time that each adult devotes to it (two days a year). 9 These levies are low (two out of fifteen); they are deducted from certain 'bride prices' and 'blood prices' or, in exceptional cases, directly from the owner. 10 For further information on these points, as well as on the ceremonial circulation which, in sacrifice and exchange, articulates the hierarchy of the two references (local and global), cf. TcherkezofT, 1981. 11 I shall not repeat the term 'Nyamwezi-Sukuma' upon every occasion, but I shall always be referring to the whole of the group, which was formerly designated by the first term alone (cf. the German ethnography: Blohm 1933, Bosch 1930) and is nowadays, as a consequence of the English colonial administration, divided up into two smaller entities. 12 This animal may not from then on be sacrificed in any other ceremony, nor may it die a natural death. One waits until it is old and is close to death. Only then is it killed, and by a specific allied group, who afterwards make up a bracelet of hide (these are the bapugo, the 'joking kin', who 'have given a woman to the first ancestor'). As far as the collectivity is concerned, the same procedure is followed, and the same fate is suffered by the royal bull (the 'elder king', ntemi nkuru), who is at the head of the royal herd and by the king (ntemi) himself. Each of them, from the very first day that they are taken in hand, are consecrated to the collectivity. Each of them is untouchable and immortal, and is killed by suffocation, 'so that death may not be seen', thereby outstripping their natural death. 13 My information regarding the number of fibres derives from Zinza country. 14 The division of the sacrifical animal 'into two' is indisputably a 'head'/'back' division, as is demonstrated by an analysis of the 'blood price' rites, in the course of which this particular sacrifice, amongst others, is performed (cf. p. 51, and what is said there about the 'fifty'). It is also said that the 'back' is the maternal side because 'the child clings to the back of his mother'. This is undoubtedly an a posteriori rationalisation, which would seem to derive its justification from the 'bow'/'back' pair; there, in fact, when one speaks of the

136

Notes to pages 37-41

'bow side' instead of the 'head side', the opposite term 'back' causes astonishment. One should perhaps see this as the result of an evolution. In the more recent English ethnography, the term 'bow' dominates and the opposition between the two 'sides' is always presented as 'bow'/'back' (buta/mgongo). On the other hand, in the Nyamwezi texts collected by Blohm in the years 1890-1910, which constitute the most valuable but hitherto the least exploited source (it contains over two hundred commentaries and descriptions of rituals in the original language), one continually comes across the mutwe/mugongo ('head'/'back') opposition. However, in both these two sorts of material, one encounters the buta/midinho ('bow'/'spoon') opposition, which expresses the contrast between the sexes. A man is denoted by the 'bow' which he always holds in his left hand, a woman by the wooden spoon which is in continual use in her kitchen (and which she holds in her left hand when she is a child and in her right hand when she is an adult and married) (cf. Blohm 1933, text 54; Bosch 1930, p. 158). There are numerous rites in which boys and girls, men and women each hold their respective emblem. There is thus reason to suppose that there has been a partial superimposition between the two pairs, 'head'/'back' and 'bow'/'spoon', the man's authority in the conjugal relation (expressed by the 'bow', cf. below) having been compared with the superiority of the 'paternal side', ku mutwe, which is expressed in the transmission of individual prohibitions (migilo) and in inheritance, as also, to a certain extent, in parental authority (which has nothing to do with 'patrilinearity', a principle whose operation in this society has been exaggerated). If the system has evolved, and evolved recently (as the difference between Blohm's texts and the surveys of the colonial epoch would seem to suggest), the rapprochement between the authority of the 'bow' (belonging to the order of status) and the superiority of the 'head'/'side' (belonging to the hierarchic order and linked to the 'head'/'back system), and the rapprochement between the maternal 'back' side and the feminine 'spoon' sex should be treated as a weakening of hierarchy as a principle organising a holistic symbolic system, to the advantage of (modern) individualism, which tends to abolish distinctions of levels and to confuse the status of the individual with the level of value. 15 Night is the time for witchraft; the star of the night, the moon, is the one which brings epidemics. 16 The 'anomaly' of man being placed 'on the left' is not new. Vincent's recent (1978) article describes a similar case and clearly demonstrates that it involves neither an anomaly nor 'universality of pre-eminence' (see chapter 5 note 2). 17 The parents become quasi-members of the royal family; they are buried with the same honours as the latter. 18 This in fact accords with the matrilineal succession of kings (see chapter 4 note 3)19 This does not entail a concept of'belonging' in the juridical sense, still less the presence of unification. It is a problem of ceremonial circulation: the father will have privileged access to the 'bride prices' which will arrive for his daughters or for the daughters of his sons. For the following facts, cf. Blohm 1933, text 54L

Notes to pages 43-6

13 7

CHAPTER 3 RIGHT AND LEFT

1 One could also consider the bracelets consecrated in honour of the ancestors and worn, as I noted above, on the left or on the right. There exist two different kinds (we lack information as to a possible difference in their use), bead bracelets and copper ones. There are black beads, which are worn on the right, and white beads, which, as we already know, are worn on the left. Likewise, one can distinguish between a red copper bracelet, which is worn on the left in honour of paternal ancestors, and a yellow copper one, which is worn on the right in honour of maternal ancestors. The two categories of bead bracelet are always consecrated by a woman, no matter which ancestor the bracelet in question is dedicated to. There is however a distinction between the different kinds of copper bracelet, in that the red copper ones are consecrated by a man, and the yellow copper ones by a woman. But both of them will be clasped, on the wrist of the sacrificer, by a man (Bosch 1930, p. 103). This last fact undoubtedly involves another opposition which I have not mentioned so far; the woman is the 'opening', the man is that which fastens and recloses, as far as the conjugal context is concerned. The hearth is only complete when the feminine triangle of the three stones is centred on the fire, which is male, and, most crucially, when it is 'covered' by a piece of pottery (modelled by the woman but fired by the man, Gas-s et at. 1973, p. 349). Further on in this book we again come across the theme of completion represented by the 'four', as formed by 3 + 1. When there is a birth, which causes a couple to cross a new threshold, there is reversal. In order that the child may come into the world undamaged, and in order that the woman may survive, the passage must be opened. Everything that is tied and knotted must be undone (strings, hanging objects), everything that is taut must be relaxed (one must unstring the bowstring), and all the pot lids must be raised (Bosch 1930, pp. 63 and 88; Blohm 1933, p. 11 and texts 54 b-d). 2 The union between 'head' and 'back' reiterates, as I have said, the division of the sacrificial animal, which is cut 'in two' in order to 'create unity', and it is said with respect to the twins that this duality, which becomes a unity of a superior order, is the defining feature of royal reference (totalising duality). The other nomenclature, 'bow' and 'back', rests on an opposition between the man, who carries a bow in order to hunt and 'protect his house', and the woman, whose child 'clings' to her back (cf. chapter 2 note 14). If'back' and 'wooden spoon' are not therefore superimposable, it is interesting to note that the symbol of the bow, which characterises the boy as much as it does the husband or the father, may possibly be an invariant one. One of the phases of the marriage ceremony shows how 'the bow' must be the occasion of battles if it is to perpetuate itself: ' . . . two or three of the most robust of the young people are obliged to fix the hook in the fiancee's house, the hook upon which her suitor's bow will hang; that done, the house is stormed. But the fiancee has taken her own precautions: robust guards are posted all around her h u t . . . every intrusion is repulsed, by force if necessary. At nightfall, our adventurers approach . . . the important thing is to surprise the guardians and to overcome them swiftly . . . once the hook is fixed, they sing of their victory: we have bought back the house!' Bosch 1930, p. 376). If this ruse fails, one must actually buy back the right to

138

Notes to page 46

hang the hook: a hoe will suffice. The bow is thus perpetuated, whether by ruse or by the payment of a hoe, for the man/woman hierarchy, on the symbolic plane that is in question here, itself assumes the form of the permanence/change opposition: from childhood to marriage, the man remains con the left', the woman passes from 'left' to 'right'. In a family in which marriage is effected by means of a livestock price, the children are 'those of the father' and hold the bow or the wooden spoon on the left. In marriage, if the reference changes (father —» society, the king, the kanumba) (cf. note 3 below), the man for his part remains on the left of this new reference. 3 The marriage lasts several days and is divided into six main stages. The two parties gather for two successive banquets, first at the house of the fiancee, and the second time at the house of the fiance. There follows the discussion of the 'price', then the taking possession of one of the huts of the fiancee's father by means of 'the bow'. There is then the actual ceremony, in which the newly married people reverse their symbolic emblems, which in this case are two flour or meat balls. Finally, there is the blessing of the parents. For the initial two meals each of the fiances arrive surrounded with the objects that symbolise them. The fiance arrives at his fiancee's house in the company of his sisters and his other kin. One carries the bow, another an arrow, whilst others carry a spear or an axe or a gourd. The bow, which is the main symbol, carried by the fiance's youngest sister, serves, with the arrow, as a guarantee of the qualities of a hunter and a fighter that he as the future husband will need to have. The spear represents his authority in his new house. With his axe he clears land and gathers wood for the fire. The gourd will protect him in his travels. At the second meal, the fiancee comes with her sisters, who carry her pearl apron and the vase of unguents, which guarantee an untroubled conjugal life and the fertility of the spouse. They also carry the cooking paddle, the huge midinho wooden spoon, which will enable the woman to feed her whole family well. It is interesting to compare these ceremonies with the origin myths, in which the man does not so to speak create anything, and in which the woman brings with her the definition of the human condition (the obligation to die and to till the earth), and the foundation of society (seasoning and fires, the Nyamwezi feminine and masculine principles respectively, whose combination enables cooking to occur Bosch 1930, p. 372). In a part of the ensuing ceremonies, each of the fiances will hold either the bow or the spoon in their left hand. And the relatives' benedictions also call this principle to mind as they anoint the married pair on their left hand and foot. But at the same time another principle is asserted, which will clearly become dominant when the newly wed couple have their first two children. And the right/left opposition then takes as its point of reference a body which is the new union, a body represented by a kind of marriage priest, who is external to kinship and is a symbol of the society. This, as we have already said, is an old woman, the kanumba, who is 'the great stove of the marriage', and the 'co-director' of the age class of married adults (Malcolm 1953, p. 202). It is worth noting that, on another plane (but from a homologous perspective), the king is the 'master of the meat' of the society. In the two final phases, after the central ceremony — to which I shall return below - we see the husband coming up and placing himself on the left of his wife. Immediately afterwards, the

Notes to pages 46-50

139

kanumba sets out to look for them. She places herself between them, a little in front of them, and extends the small finger of her right hand to the wife, who grasps it, gets up and follows her. At the same time the kanumba extends the left finger to the husband. She then leads them both, the husband on her left, the wife on her right, towards their hut. It is she now who is theframe of reference. But prior to this the exchange of sorghum flour balls had already taken place, which, through a reversal, marked the emergence of the new principle. 'When the young people have left the hut, its door is shut and things are fixed from the inside in such a way as to prevent anyone entering. The kanumba has remained there with all the married persons. The kanumba places two seats side by side and has the fiances sit down on them, the man on the left, the woman on the right. She herself places herself in front of them, on her knees on the ground, and turning her back to them; she has placed two plates before them, containing polenta and sauce. At this moment all chatting stops and there is a solemn silence. With each of her hands, the kanumba breaks off a small piece of polenta, and makes a ball with it by rolling it in her hands, then she pricks the little finger of each hand. It is on his little finger that the king wears the ring made of the hide of the sacrificed bull, at the time of the great festivals marking the passage between the "white" and the "black" periods. This done, she crosses her arms on her shoulders, so that the ball adhering to the left little finger - called ku buta ("of the bow") - passes over to the right side and, conversely, the one adhering to the right little finger passes over to the left. Thus, the left hand symbolizing the man presents the ball which it is carrying to the woman, and the right hand symbolizing the woman presents its ball to the man' (Bosch 1930, p.382). Each gesture is repeated once, for the two fiances must let the first ball presented to them by the kanumba fall, and they can only eat the second one which she rolls. We shall see below that the number 1 is incomplete and that we must always go up to two, at any rate when the context indicates a passage to the superior level. I also want to draw the reader's attention to the following fact. After the wedding night, one kills 'the animal of the young girls' house'; the married woman has left this house. Then the man and the woman enter the village, each wearing an article of clothing belonging to the other. Once again, reversal

marks the new state (Cory 1953, p. 44). 4 There are many instances of this. I shall mention two of them. When the king's councillors attend apluviator diviner, in the sowing season, they remain with him for three days; on the fourth day, they return with the requisite magical preparation (Millroth 1965, p. 132). A king's son who has broken the prohibition against eating the 'elder king' (the royal bull which one suffocates when it becomes too old, outside the royal court) may not re-enter the court for three days (Millroth 1965, p. 159). 5 The kind of exchange that I have in mind here is the sort which is always accompanied by the payment of a 'price' (nsavo), which is only concerned with murder and with marriage, in all their variant forms and with all their consequences, and which is defined by the estabishment of a 'debt' which 'never ends', by contrast with immediate and egalitarian exchange, which does not involve a 'price', and which is employed in cases of'mutual help' and only has to

140

6

7

8

9

10

Notes to pages 50-6

do with neighbourly relations. Only the reference for the first type of exchange is situated at the 'superior level'. One should likewise distinguish between the incomplete one and the 'royal' one, where the latter is a symbol of Unity. At their coronation, the king and queen should sit down on a single stool, with only one leg, for this latter 'represents the unity of the kingdom' (Cory 1951). The complex symbolic structure here discernible in numbers, and for which binary oppositions such as even/odd do not provide a satisfactory explanation, also occurs elsewhere. There are telling instances of it in Melanesia, where only a distinction in terms of levels enables one to understand the ceremonial structures and the movements which animate them. Amongst the 'Are'are of Malaita, the sequence of the first nine numbers also assumes a particular form, which one sees at work in marriage, in murder, in funerals, in festivals of power and even in musical forms : 1, 4, 8; 9, 7 (cf. Coppet 1970; Coppet and Zemp •978). There is more than one kind of'prohibition', and it too is a notion that is open to hierarchisation. The Nyamwezi distinguish between mgilo and mwiko. The former is predominantly concerned with alimentary practices; it concerns the individual and the sanction remains individual (the leper); the reference is the father. The latter varies a great deal as far as the domain of prohibition is concerned, but the sanction employed is more precisely defined and is almost invariably concerned with the fertility of woman or of the soil. Furthermore, the sanction is collective to the extent that the whole community feels itself to be 'sullied'. Prohibitions which, in a general sense, concern death, and those which are declared and sanctioned by the king, obviously belong to the second category. The hunter receives this cloth if the lion's hide is complete, if no claws are missing on either paw. Where this is not the case, the hunter is regarded as a 'murderer' and becomes a member of the royal court, both a courtier and a prisoner for life, a kind of ritual servant who may however be saddled with serious responsibilities (ban 'ikulu) (Bosch 1930, pp. 88-9; cf. Blohm 1933, text 152; cf. Tcherkezoff forthcoming). These examples, like the previous analyses, involve an opposition between the royal and the 'ordinary' domain, between global and local reference, and between 'royal' value and that of the familial ancestors. This is obviously not in any direct sense a question of status. On the one hand, there are individuals and groups, commoners and nobles, as locally defined. On the other hand, there are the royal rites, which are performed for all and which the king, 'the father of all', directs, and which involve the global plane and a superior reference. What then does the difference in statuses involve? To take one example, what do the great diviners (bamanga) do when they want to invoke all their ancestors, on both sides? In structural terms, some 'positions' are occupied: 'head'-white + 'black'-black, perfect royal totality; black coat and a white spot on the forehead, the favoured animal for a royal sacrifice (and one which the nobles sometimes use); we have also observed the reversed figure of a black head and a white back, which a commoner may use when addressing himself to both of his two categories of ancestor. The diviner is not usually a member of the court. He is, however, different from a commoner. He has ancestors whose powers of

Notes to pages 36-g

141

'clairvoyance' were great, and he himself is invested with a particular 'force' which enables him to invoke them better. On the merely logical plane of colour associations, there would be many 'positions' that the diviner might still envisage, and many choices of colour combinations would be possible which would mark him off from the king, the nobles and the commoners. But, where hierarchy is involved, there are more restrictions, and this twofold need for demarcation is answered by a double reversal: the pole will be univocal, a single colour, but the colour in question will be white. The diviner will choose a white goat for this sacrifice (cf. Tanner 1958, p. 226). It should be compared with the isumbi nolo solo mentioned by Bosch (1930), p. 101) in the list of objects consecrated as maholero, the emblem of the bamanga diviners, consecrated in honour of the ancestors of the two lineages, and it is a small white, oval stone, threaded on to a leather thong and worn around the neck on a string of mhoja (the bark of the tree of'peace'). While we are on the subject of practices having to do with status, it is worth recalling that the 'nobles' by preference sacrifice black bovines with white spots on their foreheads (Bosch 1938-9, p. 89). 11 This may sometimes be a witch-ancestor, i.e. a dead relative who was (unjustly) executed after an accusation and who returns to haunt. Another of Bosch's observations (1930, p. 229) may be pertinent here: someone who has been accused of witchcraft and who survives the ordeal by vomiting the poison that he has been forced to drink is declared 'white' (kwera); he is innocent and it is his accuser who is condemned. The example of the chicken, particularly on account of Cory's text (quoted above) is the best known (or the only one known) of the data in Nyamwezi society respecting dualist oppositions. Beidelman, in his 1961 article (pp. 250 and ff.), which he wrote in response to Needham's stimulating article, 'the left hand of the Mugwe' (which had been published the year before), notes that, in spite of the lack of any studies of this topic in Africa, the enthnographic literature contains unsuspected riches. Amongst other examples, he refers to Cory as reporting 'an especially fine example of dualistic example with reference to social organisation' (p. 251). These words imply that Beidelman was predominantly concerned with the lateral opposition rather than with the colour opposition. In fact, divination does not only consist in specifying the author of it, by noting the place that the spot is on the animal's body. The chicken's head represents the client who has consulted the diviner, and if there is a spot upon it, that indicates that this person has sought to bewitch someone, but that the ill-doing has rebounded upon himself. The head thus serving as a point of reference, an opposition is naturally implied between the two wings, each of which is associated with a side of the person's family. One wing refers to the relatives (and to the ancestors) 'of the back', the other to the relatives and ancestors 'of the head'. The most telling point here is that the place occupied by the affines depends upon the sex of the client. A man's affines will be on the wing which is associated with his maternal relatives ('of the back'); a woman's affines (and her husband would certainly be included in this category) will be on the wing associated with her paternal relatives ('of the head'). But the left—right division, which had undoubtedly attracted Beidelman, cannot be employed. In fact, Cory, whose work is later taken up in much the same terms by Millroth (1965), simply says that the chicken is laid upon the ground with its

142

12

13 14

15

16

17

Notes to pages 59-64

wings spread, and that the left wing refers to the 'back' side and the right wing to the 'head5 side, without specifying whether the animal is placed upon its stomach or upon its back, and without stating whether the observer (Cory himself) is speaking of the left wing in relation to himself, the chicken being, for example, on its back, or whether he is speaking of the chicken's wing as seen if it was placed on its stomach. For want of such specifications, one risks developing a left-maternal / right-paternal dichotomy which would be in accord with the classical theory but which would be discrepant with the rest of the material. This is why I have not directly used this datum here. Note that Bosch's description (1930, pp. 220-1) is, by contrast, one that we are familiar with, the right wing being 'the mother' and the left wing 'the father'. He appears to be referring to the right wing of the chicken. Blohm's description (1933 text 146), on the other hand, returns us to the head/back opposition: the paternal relatives are on the animal's head, and the maternal relatives are on the back and paws. So as to forestall any confusion, let me remind the reader that I distinguish between 1. symbolic levels, levels of inferior and superior, encompassed and encompassing value; 2. the context of the 'ordinary' rites, as performed by the ordinary run of mortals, i.e. by everyone except the king, which is opposed to the context of the royal rites, organised by the court in the interest of and, sometimes, with the participation of all: the plane and domain of ordinary, familial ritual and the plane of royal ritual; 3. commoner status, that of people who are neither nobles, nor diviners, nor twins, nor murderers enjoying the king's protection, which is opposed to various other statusses mentioned above, and which is further divided into elder of the house and junior, man and woman, celibate and married, married with a complete 'price' or not, etc.; 4. local and global, familial and royal references: badugu ancestors/collective-royal ancestors. The reader should bear in mind that the witches' domain is the night and darkness. This relation is only concerned with murder. If a death is imputed to the action of a witch, he is killed by a surprise attack; one then pays the 'five' to the court (because one has breached the royal prohibition against shedding blood), where one is congratulated by the king. Second marriage seems to be associated with remarriage after widowhood, thus placing in question the life of the first wife and her children (Bosch 1930, p. 436; Cory 1953, p. 10). The cultivated/wild opposition, which I cannot go into here, plays an important part in defining all rituals, since it is mhoja work which culminates in the domestication of the dead (see Tcherkezoff 1985). The latter are in fact primarily wild animals; the work of ritual entails turning them into domestic ones. In other contexts, this opposition underlies the inside/outside, close ancestor/first ancestor, familial ancestors/royal ancestors distinction. Distinctions which are internal to the representations of the ancestors as wild (lion/warthog, snake of two colours/snake of one colour - as mentioned here) correspond to the bovine—ovine/caprine ordering, which hierarchises sacrificial victims in terms of the distinction between royal and commoner sacrifices (for the mla, cf. Bosch, 1930, pp. 93, 142). A dialectic of the Hegelian type is directed at an ultimate totality which does not

Notes to pages 64-ji

143

pre-exist the process, by contrast with the hierarchical totality in relation to which levels are constructed (cf. Dumont 1971, pp. 77-8). 18 Initiation consists essentially in transforming each person's relation to the ancestors by making it possible for the initiand to master the art of sacrifice. CHAPTER 4 HIERARCHY AND POWER

1 The association between the ngabe and the lateral and numerical oppositions (as noted above, p. 40) should be handled with care, for it does not derive from rituals which have actually been observed - by contrast with all the other ethnographic data — but from narratives concerning a legendary figure, Luhinda. This figure is of crucial importance for the study of the relation between king and priest, but the cycle of legends concerning him requires an analysis extending over the whole interlacustrine region (cf. note 12 below). For the facts respecting the shaven/unshaven opposition, cf. Bosch 1930, P-5IO2 The Nyamwezi do not emphasise mixed twinship and, instead of being concerned with the opposition between the sexes, refer to the relative seniority of the twins. They are always called Kulwa and Doto, the 'big' and the 'little' (Gass et at., 1973, p. 364). Among the Dogon, by contrast, mixed twinship is patently hierarchised. It is at any rate important that the twins should not be identical (boy-girl) or that they should be differentiated as they emerge from their birth ritual (senior/junior), so that the definition of a twin is preserved, i.e. the most intimate union of the most opposed principles. This, at any rate, is how twins are defined in a hierarchical system, and it is what accounts for the importance accorded this spectacular form of birth, in that it represents the simultaneous affirmation of principles which are normally distributed over time and hierarchised in space. This is why, among the Nyamwezi, we find that twins, before the purificatory rite held at their birth, threaten kingship and the whole country with death, and after the rite, which brings everybody to the court, are a sacred image of kingship (they also become 'royal ancestors' and are a sign of prosperity for the country). Indeed, their birth represents a production of the 'two', a 'royal' production and a raw image of kingship. Which is what is implied by the very fact that their birth concerns kingship and necessitates a rite held at court. But it is a priori a royal image of death, because it is not oriented and is timeless, and is in short non-hierarchised. Which is why the king must perform a 'purification', force the relatives to 'pay' and, as the rite shows, integrate these two children symbolically into the court. They can then 'be' kings' as the Nyamwezi put it; they are the royal, hierarchised Unity of a couple (here senior/junior, there male/female). It would seem to be significant that, by contrast with definitions of this sort, which are appropriate to a system of hierarchical thought, a birth is defined as spectacular in our own society when there is a repetition of the same, and it is only the quantitative aspect of this repetition which is emphasised. You only have to look at the headlines in the tabloid newspapers. A twin birth is of no interest, triplets merit a photograph, whilst quadruplets deserve a commentary. Beyond this point, with quintuplets and sextuplets, everyone's interest is aroused, and the authorities will even go to witness the event. In a 'single level' logic, an

144

3

4

5

6

7

Notes to pages 71-5

order can only be defined in quantitative terms: the 'two' is merely larger than the 'one5, but inferior to the 'three', which is itself inferior to the 'four' and so on. I would add that, as regards the opposition between the sexes, another highly productive 'totalising' figure may be the image of a 'brother-sister couple', which also displays at once the greatest proximity and the greatest distance. In some societies, it is this figure which is the bearer of value and which seems to dominate the kinship system and, more generally, the exchange system (cf. Iteanu 1983; for the senior/junior pair, cf. Barraud 1979). Cory tells us that it was the right arm in cases of uterine succession (which, it should be remembered, was the norm), and the left arm in cases of agnatic succession. In most kingdoms the agnatic rule emerged in the colonial period, and we therefore know this feature to be a recent one, with the shift from the old (uterine) rule to the new (agnatic) one being a direct consequence of the foreign presence, which was first Arab and then European. Bosch and Cory relate that a pair were sacrificed, and I have accepted their testimony here. Blohm's text, however, informs us that the children sacrificed were three in number. I do not regard this as a contradiction and would maintain that it is a variant. Indeed, Blohm's text describes a similar arrangement, with a girl at the king's head and the boy at his feet, in the position described above. The second girl supported the king's legs on her shoulders. One might suppose that this triadic sacrifice also represented an image of the totality. In fact, a man and a woman constitute a totalising image, as I have said; but, apart from the image of the conjunction of the opposed sexes, I have referred to that of the 'five', which was 2 + 3, the basic figure of the series associated with man and with woman. A boy and two girls may also constitute the 'five'. People in fact say that in exchanges, in a general sense, the female animal is worth two male ones and that, correlatively, a woman 'is worth' two men, in 'blood prices' particularly. Thus, a bull is the 'five' and a heifer is worth 'ten'. If a man is worth 'one', a woman must be payed as 'two', and 2 X 2 + 1 = 5. If the 'man's price is 50, and that of a woman is therefore 100, two women and a man correspond to: 250 = 5 X 5 (X 10). In fact, the remarks concerning marriage, and those concerning the value of the five in exchanges, are themselves a part of this 'sacrificial' context of the kuhoja (cf. the analysis of exchange as 'sacrifice' in Tcherkezoff 1981, vol. 2). This fact has important consequences for the manner in which one considers ceremonial exchanges with the court and the system oimasumulafineswhich are imposed as a consequence of transgressing royal prohibitions. The system assumes the form of a 'sacrificial' circuit and not that of a penal system. The kind of royal duality that has been a recurrent feature of the present analysis would seem to have much in common with the royal symbolism that has been investigated in a number of earlier studies, at any rate where their authors have aspired to a totalising vision. They have taken binary oppositions into account but, over and above them, have emphasised the fact that the royal figure is 'outside of all classes', 'central' and a 'mediator'. I ought at this point to refer to Masao Yamaguchi's study (1974), which is meant as a further development of Hocart's thought on this subject, and which did much to awaken my interest in this kind of approach to kingship (D. de Coppet drew this text to my notice and arranged for me to meet its author). Masao Yamaguchi's

Notes to pages 75-8

145

exhaustively documented study advances the study of royal symbolism. Faced with a dualist classification, the author begins by presenting a lengthy analysis of the bipolar configurations which feature in all Jukun representations. Although he does not restrict them to an exclusively binary choice, he does not however go so far as to pose the question of hierarchical opposition. But he does go on to give a powerful demonstration of the 'extra-class' character of the royal figure who 'mediates' or rather 'abolishes' the dualist classification in order to situate himself 'beyond the ordered limit of time and space'. The king, an androgynous figure, is 'more remarkable' than the order established by the classification, for he recreates a totalising beyond, which is the society's identity. This is only one aspect of the question, however, as we saw above, and an observation of this sort must be treated as a starting-point only. As several examples have shown, there are not two radically separate planes, even if they are stated to be hierarchised, such that there would be an absolute distinction between a domain of binary opposition (which would thereby be too cut off from its reference) and a domain of'mediation' or of the abolition of oppositions. The 'superior level' and the 'inferior level', to introduce yet again the terms I have employed here, are not two irreducible modes of thought but an encompassing and an encompassed. Both of them are therefore perpetually in relation to each other, and according to a hierarchical mode. 8 Cf. Blohm's translation of text 55a: anormal geborene Kinder dem Hduptlinge gehoren, for 'the twins and the kasindje are the nsavo of the king' (= they entail a 'price' relation to the king); cf. Cory 1953, p. 88 and Spellig 1927, p. 218). 9 I have employed Dumont's formula here. In the case of India, it expresses an absolute disjunction. The hierarchy of the pure and the impure which underlies the hierarchy of the castes is absolutely independent of royal power. Furthermore, the domain of artha, involving the exercise of 'force' and the pursuit of 'self-interest', is subordinated to the domain of dharma, the Brahmanic value which is the supreme value. When artha dominates, we are on an inferior level, and there and only there is dharma inferior: 'Hierarchy thus offers the possibility of reversal: that which at a superior level was superior may become inferior at an inferior level. The left can become the right in what might be called a "left situation" . . . ' (Dumont 1980, p. 244). Even when we go on to consider other examples apart from the Indian one, a general fact still holds, namely, that in a holistic society, power is not identified with the hierarchy of values, with the global ideology, and may even go so far as to eliminate it: 'The moment hierarchy is eliminated, subordination has to be explained as the mechanical result of interaction between individuals, and authority degrades itself into power, power into influence and so on.' (Dumont 1977, p. 10). The fundamental problem is not a theory of 'power' but 'the relation between "power" and values or ideology' (p. 10). In India, power is absolutely encompassed. In Nyamwezi society, the relation is more complex, as I have implied here, and in the pages that follow. But on the general plane, 'power' - which is not simply 'kingship' - is still encompassed by hierarchy. 10 It is of interest to note that the founding hero Luhinda was buried in the two kingdoms that he had conquered and unified, with the head being buried in the one kingdom and the body in the other (cf. Mworoha 1977, pp. 86-7 and chapter 5 note 9).

146

Notes to pages 79-82

11 No restriction was placed upon emigration and it was common. People would set themselves up in a neighbouring kingdom, for a drought might in fact be very localised. No land shortage ever occurred, for ownership of land was unknown. This freedom of movement was a constant threat to those kings who did not know how to bring about rain and, if the inhabitants were all fleeing the country, the courtiers would kill or depose the king. 12 A study of the exchanges and representations concerning the animal world also reveals a construction involving levels. At the first level, inside and outside are opposed in the same way as domestic and wild animals are; goats and cattle serve as an image of the ancestors 'fixed' near to where one lives, whilst snakes and warthogs serve as an image of the wandering ancestors who may come in violence to claim something off their descendants. On the royal plane, the same opposition concerns the bull and the lion. But, like bulls, lions are in the court, whether dead (an animal encountered in the bush is killed and brought into the court after a special ceremony) or 'living' (the king 'is a lion'). Parts of their bodies play a role similar to the calabashes, the Iwanga and the goats consecrated as maholero play for the commoners (see Tcherkezoff forthcoming). The court is thus the outside of the inside. The outside/inside relation changes level. The outside becomes dominant, at the same time as it becomes conjunct with the inside and no longer disjunct with it (the outside then transcends the plane of the particular kingdom; it is the place 'of lions', that of all the Nyamwezi kingships and even of the entire interlacustrine region). A legendary figure called Luhinda has much to do with the relation to the outside and with the relation, on this plane, between king and priest. This figure is an ambiguous one. (cf. note 1). He is believed to have established the royal kingdom on the north-west fringe of the country, in Buzinza. Luhinda has, on the one hand, the features of a conquering king; but his conquest is a peaceful one, dependent upon magical ruses and upon a justification in terms of autochthony (he is the son of an emigre from Buzinza who had established himself in Buganda): 'Luhinda travelled unceasingly in the country that he had subdued and everywhere he had himself proclaimed Mgabe or great chief rather than king.' (Gass et al. 1973, pp. 340-45). This title is reminiscent both of the religious, pacific and autochthonous figure of the great mgabe priest who is to be found in numerous Nyamwezi kingdoms and of another figure, one that is manifestly royal in character. Indeed, the mugabe {Mgabe) is the supreme king of Nkore, a country to the north of Rwanda and to the east of Buganda and of the Haya states, one of which, in the northern part of this group, is Kiziba. And another set of myths has it that Ruhinda (= Luhinda) was the founder of this royalty. Ruhinda's origin does not therefore lie to the south of the interlacustrine group, in Zinza country, but to the north (Nyoro) in the Cwezi dynasty. A variant of this second set of myths establishes a link between the narratives 'from the north' and those 'from the south', by treating Ruhinda, heir of the Cwezi kings, as a conqueror who only passed through Nkore, and then went down to Kiziba and to Buzinza, retracing in reverse the path taken by his emigre father, such as it is represented in the narratives from the south. His arrival at Buzinza is clearly a stranger's arrival and he is thus contrasted with the Nyamwezi mgabe. It is interesting, however, to note that in these narratives 'from the north', the

Notes to page 82

147

royal origin of Ruhinda, by way of the Cwezi kings, is itself ambiguous. He is sometimes the natural son of the king and of a female servant whom the high-priest of the court had caused to sleep beside the king one evening, when the king was drunk (because this servant had saved the high-priest from a trap into which the king had sought to make him fall). There is thus a sense in which Ruhinda is as much the son of the priest as of the king. Sometimes Ruhinda is the 'delegate' (mugabe) of the Nyoro king (mukama), who has entrusted him with the task of guarding the livestock at Nkore; consequently, Ruhinda declares himself independent. We should note here that the guardian of the royal livestock is actually the repository of the tradition and continuity of kingship, the livestock representing the dead kings and the perpetuation of the 'spirit' of kingship. It is a role of this very sort that is entrusted to the Nyamwezi mgabe. He is responsible for watching over the royal tombs and, in the rites which are performed there, he always precedes the king. In front of the royal bull, he recites the genealogy of the kings. I would also add that, as far as the Nyamwezi are concerned, the mgabe is often presented as the descendant of the former autochthonous kings who were conquered by the new dynasties and that, on the other hand, the king and the mgabe are said to form a bupugo relation (cf. Varkevisser 1973; an inaugural marriage brought a period of war to an end and established formalised 'joking' relationships). Mgabe (Blohm, Bosch) = ngabe (Cory, Gass). The image of Luhinda as 'outside' king also features in the habitual forms of invocation preceding a sacrifice. One turns to all four points of the compass, to the firmament and to infernum: 'it is thou, Likube, the Creator who gives us the white and the black nhendeko beads (the beads of the bracelets consecrated as 'head' and 'back'; cf. Bosch 1930, p. 101), thou, the morning sun ...; west, evening sun, country of Luhinda whence the hoes come, thou who art our king . . . ; north, whence come the waterspouts and the livestock, the water which we are given for cultivation . . . ; south, whence comes the drought, the soldier ants . . . ; Heaven, which divides up everything, which gives us a fertile earth, children and health; Earth, which is bad, which is ever gaping, which never gives us anything, which covers over men in their tombs, accept our sacrifice, allow us to be at peace with our dead . . . ' (this is a precis of Bosch 1938-9, p. 90, whose transcript is not in Kinyamwezi; cf. Millroth 1965, p. 167). Here Luhinda is the king of the outside, the 'country whence the hoes come'; opposite is the east, the side of the Creator; it may also be the side of the priest. One should reserve judgement on this last point, given how heterogeneous the contexts once again are. But it is said that the Creator is on the side of the Balataru diviners (Gass et al. 1973, p. 388), great and mythical diviners responsible for handing down the art of divination. They are part of a larger set of myths concerning the migration of the Balaturu population (Nyaturu, Taturu). The latter is also held to have handed down the art of domesticating animals and of sacrifice, before it withdrew in a catastrophic and 'reserved' manner (the cows began to go into their corrals backwards and to give blood instead of milk; all the Balaturu departed leaving red traces on the rocks; this fall and this departure evidently refer to certain themes in the Cwezi epic, which is known throughout the interlacustrine area) (cf. Gass et al 1973, pp. 337—8, 413). Two linked observations are appropriate here. The opposition of east/west «•»

148

Notes to pages 82-92

inside/outside <-* the Creator and (?) the priests-diviners/the king, the historical origin - as opposed to the cosmological origin - is not identical with the rain/drought opposition (which here = north/south), reminding us yet again that the priest/king contrast may not be superimposed upon the one that distinguishes between hierarchy and power and, even less obviously, upon that which would distinguish between 'religious powers' and 'political powers'. It is of interest, on the other hand, to note the encompassing role of the east, which is the side of the Creator and of the two kinds of pearls, the black and the white. This point calls to mind other representations, in which the morning sun, which is 'red' (cf. dawn in the initiatory hut), is both the accomplice of the sun and its adversary. It burns the earth; any red-coloured object will stop the rain falling. But another kind of manipulation of red will attract the thunder, which is definitely murderous and is sent by the sun, but will be followed by the redeeming rain (Gass et al. 1973). It is also worth noting that the Earth/Sky opposition, in which the valorisation of the second pole is quite explicit, lends further emphasis to the orientation of values that, I would maintain, organise the inside of the initiary hut, with the devalorised image being below and the superior configuration above. 13 Here the margin between the two is slender, at any rate before the emergence of the warrior-kings of the epoch of Mirambo. I would point out, however, that, in cases of murder for example, conflict could arise between royal authority, which aimed to have the protagonists accept the formula of the 'price' paid in the presence of the court, and the group of victims, who would demand vengeance and sometimes would obtain it, with the king having to 'yield' to their pressures (cf. Bosch 1930, p. 511). CHAPTER 5 THE SAME QUESTIONS ASKED OF OTHER EXAMPLES

1 It would have been a dauntingly large task to have represented the left/right, masculine/feminine etc, oppositions in their entirety, but study of CalameGriaule's books (1965, pp. 29off.) will serve to convince the reader of the richness of the material in this domain and of the manner in which it can be handled. I have chosen instead to limit myself to the special case of twins, which is of special significance for the Dogon, which does in itself require very extensive treatment and which may lead on to a study of the whole of their symbolic system. 2 This 'African complementarity', as Dieterlen terms it, does as we have seen, imply a hierarchy. In a recent analysis of the system of the Mofu of the Cameroons (1978), based upon a detailed ethnographic study, Vincent suggests a distinction between classifying and hierarchy which calls Dieterlen's study to mind. On a first plane, left and right classify and 'serve simply to emphasise irreducible difference', that between man and woman in particular. This distinction becomes a classificatory grill into which oppositions between numbers or points of the compass etc. may also enter. The problem of pre-eminence then disappears, and the numerous systems of alternation may then be understood in terms of their context: for a particular act defined as impure, a man will use one hand, a woman the other. When, however, a pure act is involved, the opposite hand will in either case be used. As Vincent stresses,

Notes to pages 92-103

149

this is a farther illustration of what Granet had shown and one should, he writes, 'speak, as Germaine Dieterlen has done, of complementarity'. The hierarchy appears on another plane. 'In certain cases, however, the superior status enjoyed by men causes the concepts associated with being a man to topple over on to the beneficient side, so that this symbolic system comes to seem a hierarchised one . . . ' (p. 498). The male series, which is characterised as right or left depending upon the context (and upon the society), will be superior. This difference between domain of distinction and that of hierarchy seems to me an important one, so long as one inscribes it within the hierarchical principle: there is a hierarchical mode/distinctive or equalitarian mode encompassing. This is not to argue that all the oppositions deriving from the man/woman classificatory grid soould be placed in the 'distinctive' domain, thereby distinguishing them from all the other oppositions, which belong to hierarchy. One can undoubtedly analyse certain Mofu examples as belonging to two levels, which are neither two distinct models produced by the observer nor two distinct principles of the indigenous taxonomy. On the one hand, some binary oppositions involve more than just a distinctive man/woman opposition (a lefthanded person is invincible in war, the left hand side characterises mourning and sacrifice, pp. 488-92). On the other hand, there are reversals which demonstrate the change of level involved in shifting between the principle of alternation and the hierarchical principle. In divination with pebbles, an odd number favours men, whilst an even one favours women. But if it is a whole community that is the object of a diviner's enquiry (as with the choice of date of a harvest festival for a 'massif), the odd numbers will be interpreted as a sign of the ancestors' assent (p. 496). East and west also form a system. For a man, the east is the lucky direction, and he must sleep with his head pointing towards the east; for a woman, it is the west. 'Some informants, however, say that . . . everyone (men and women) turn their face towards the west. By contrast, the chief of the 'massif', and this is much emphasised, is buried facing east, 'because he alone is capable of tolerating this side' (p. 497). If the Mofu chief, considered as a symbolic person, is like the 'massif itself, a symbolic totality placed at a superior level, it too is an instance of a figure to be encountered among the Nyamwezi, where the king, even when a woman, is 'always male'. I had also stated that this Unity/alternation hierarchical opposition was homologous with the one which seemed to be dominant, namely, Duality/singularity. At the inferior level, reversal takes the form of alternation, for here the perspective is 'arranged', so to speak, around the main gap between element and set (man-woman/ 'massif or chief of the 'massif; odd-even/always odd; west—east/east). It may then be possible to give a direct expression of the whole construction in the following hierarchical form (east encompasses west; superior level: E > W —» Chief/others, inferior level: W > E —> man/woman), rather than first bringing up the distinctions (man-woman), then value (on the man's side), in order finally to analyse the symbolism of the 'massif or of the chief. 3 The bear, the animal associated with the Earth sub-moiety, indicates the number to those belonging to the Sky (Tsizhu) moiety and to the Water (Wazhazhe) sub-moiety. The beaver, associated with the Water sub-moiety, indicates the number to those belonging to the Sky moiety and to the Earth sub-moiety.

150

Notes to pages iog-2g

4 The fact that the bororo only encompasses them on the Cera side, that of the 'superior' moiety, is of secondary importance here (cf. pp. 157-62). CHAPTER 6 TOWARDS A COMPARATIVE STUDY

1 It is not modern ideology as a whole that I have in mind here for it, like any society's ideology, involves different levels, but rather the mode of thought which is valorised by modern society, i.e. scientific rationality, which tends to obscure the hierarchical dimension of all ideology. 2 'Rational' efficacy is not ignored but is subordinated, since it only serves to define facts which stay on a single level. The hierarchical mode does not exclude its contrary, and thereby it differs from the 'rational' mode; it subordinates and it encompasses. 3 Daniel de Coppet has used this example to demonstrate the importance of disequilibrium in ceremonial exchanges of this type (cf. Coppet 1970). The reader is thus invited to redress the error in anthropological theory involved in considering such ceremonial exchanges in terms of equality, equalitarianism or of drawing up a balance for each partner (for a more recent reconsideration of the problem, Coppet 1981). 4 For the Nyamwezi, it is sacrifice which transforms the recent dead into money or livestock. The image of the whole set of ancestors is the 'image of the whole' of the society to the extent that we are here concerned with societies narrowly enclosed within a single law, that of the ancestors, with the latter having sole rights over life and death. 5 Such an answer may be strictly reciprocal (a death for a death) or reciprocal in the larger sense (a death for the theft of lifestock, where such theft is treated as 'murder', as it is among the Nyamwezi). In either case, it is still egalitarian. 6 This observation derives from some remarks made by Raymond Jamous (cf. Jamous 1980).

Bibliography

Abrahams, R. G. 1967a. The political organisation of Unyamwezi - Cambridge: Cambridge University Press 1967b. The peoples of greater Unyamwezi, Ethnographic survey of Africa XVII. London: International African Institute Auge, M. 1978. Vers un refus de Talternative sens-fonction, L'Homme, vol. 18, no 3-4, pp. 139-54 Austen, R. A. 1969. Ntemiship, trade and state building, pp. 133-47 in D. F. McCall and N. R. Bennett (eds), Eastern African History. New York: Boston University Papers on Africa, vol. 3 Barraud, C. 1979. Tanebar-Evav: une societe de maisons tournee vers le large. Cambridge: Cambridge University Press; Paris: Editions de la Maison des sciences de l'homme Beidelman, T. O. 1961. Right and left hand among the Kaguru: A note on symbolic classification, Africa, vol. 31 no 3, pp. 250-7 Bennett, N. R. 1971. Mirambo of Tanzania. Oxford: Oxford University Press Bernardi, B. 1959. The Mugwe, a failing prophet. London: Oxford University Press Blohm, W. 1933. Die Nyamwezi. Berlin: De Gruyter. Vols 2 and 3 Bosch, F. 1930. Les Banyamwezi, Bibliotheque Anthropos, Vol. 3, fascicule 2 1938-9. Contribution a l'etudes des sacrifices, Grands Lacs, vol. 55 Calame-Griaule, G. 1965. Ethnologie et langage, la parole chez les Dogons. Paris: Gallimard Coppet, D. de 1970, 1, 4, 8; 9, 7. La monnaie: presence des morts et mesure du temps, UHomme, vol. 10, no 1, pp. 17—39 1973. Tabou, pp. 702—5 in Encyclopedia universalis, vol. 15, Paris: Encyclopedia universalis France 1981. The life-giving death, in S.Humphreys & H.King (eds), Mortality and immortality: The anthropology and archaeology of death. New York: Academic Press Coppet, D. de, and H. Zemp. 1978. 'Are'are. Unpeuple melanesien etsa musique. Paris: Editions du Seuil Cory, H. 1944. Sukuma twin ceremonies, Tanganyika notes and records, no 17; pp. 34-44 1946. The Buyeye, a secret society of snake-charmers, Africa, vol. 16, no 3, pp.160-78 1951. The Ntemi. London: Macmillan 1953. Sukuma law and custom. London: Oxford University Press 1954. The indigenous political system of the Sukuma and proposals for political reform, Eastern African studies, no 2

152

Bibliography

i960. Religious beliefs and practices of the Sukuma/Nyamwezi tribal group, Tanganyika notes and records', no 54 Dieterlen, G. 1968. Norme et lateralite en Afrique occidentale, in R. Kourilsky and P. Grapin (eds), Main droite et main gauche, norme et lateralite. Paris: Presses universitaires de France Dumont, L. 1967. Homo hierarchicus. Paris: Gallimard 1970 Homo hierarchicus, tr. by M. Sainsbury, Chicago, London. Chicago University Press 1971. On putative hierarchy and some allergies to it, Contributions to Indian sociology, new series 5, pp. 61-81 1977a. Homo aequalis. Paris: Gallimard 1977b. From Mandeville to Marx. Chicago, London. University of Chicago Press 1978. La communaute anthropologique et l'ideologie, L'Homme, vol. 18, no 3-4, pp. 83-110 (in Essais sur Vindividualisme, Paris, Seuil 1983) 1979. Homo hierarchicus. Paris: Gallimard, collection «Tel» 1980 Homo hierarchicus, tr. by M. Sainsbury, revised edition. Chicago, London. University of Chicago Press Durkheim, E. 1975. Textes. Paris: Editions de Minuit. vol. 2 Durkheim, E. and Mauss M. 1963. Primitive classification, ed. and tr. by R. Needham. Chicago. University of Chicago Press 1968. De quelques formes primitives de classifications pp. 13-89 in M. Mauss, Oeuvres. Paris: Editions de Minuit. vol. 2 Gass, P. M. etal. 1973. Le Bugwe, le Buzinza. Croyances, magie et superstitions des Bagwe ou Basukuma etablis au sud du lac Victoria-Nyanza, Anno pontificio mus. Mission etnol. Roma, 37, pp. 325-83 et 385-459 Gottberg, A. 1967. On the historical importance of the crisis of the Fundirika Empire, £tudes qfricaines, pp. 60-83 Granet, M. 1953. La droite et la gauche en Chine, Bulletin de Ulnstitut francais de sociologie, vol. 3, no 3; in £tudes sociologique sur la Chine. Paris: Presses universitaires de France 1973. 'Right and left in China', pp. 43—59, in R. Needham (ed.), Right and left. Essays on dual symbolic classification. Chicago. University of Chicago Press Hertz R. i960. Death and the right hand. tr. by R. and C. Needham. Aberdeen. Cohen and West 1970. La preeminence de la main droite. fitude sur la polarite religeuse, pp. 84—109 in R. Hertz, Sociologie religieuse et folklore. Paris: Presses universataires de France Heusch, L. de 1972. Introduction, in R. E. Bradbury, C.Geertz et al (eds), Essais d'anthropologie religieuse. Paris: Gallimard Holmes, C. F. and J . M . Kadase Mange. 1971. Mfumo ya Bukwimba, African historical studies, vol. 4 no 1 Ikombe. 1943. The legend of Nkanda (Holmes ed.), Tanganyika notes and records, no 15

Iteanu, A. 1983. La ronde des echanges. De la circulation aux valeurs chez les Orokaiva. Cambridge: Cambridge University Press; Paris: fiditions de la maison des sciences de l'homme Jamous, R. 1980. Honneur et baraka. Les structures sociales traditionnelles dans le Rif.

Bibliography

153

Cambridge: Cambridge University Press; Paris: Editions de la Maison des sciences de l'homme La Flesche, F. 1973. Right and left in Osage ceremonies, pp. 32-42 in R. Needham (ed), Right and left. Essays on dual symbolic classification. Chicago: University of

Chicago Press Levi-Strauss, Cl. 1958. Anthropologie structural. Paris: Plon. 1962. Lapensee sauvage. Paris: Plon

1966 The savage mind. London. Weidenfeld and Nicholson 1968a. Les structures elementaires de la parente, 2e ed. Paris, la Haye: Mouton

1968b. Structural anthropology, tr. by C.Jacobson and B.G. Schoepf. Harmondsworth, Allen Lane 1969. The elementary structures of kinship, tr. by J . H . Bell, J.R. von Sturmer and R. Needham, ed. by R. Needham. London. Eyre and Spottiswoode Malcolm, D. W. 1953. Sukumaland. An African people and their country. London: Oxford

University Press Mauss, M. 1968. Oeuvres. Paris: Editions de Minuit. vol. 2 Millroth, B. 1965. Lyuba: traditional religion of the Sukuma, Studiae ethnograficae

Uppsaliensae, vol. 13 Mworoha, E. 1977. Peuples et rois de VAfrique des Lacs. Dakar, Abidjan: Les nouvelles editions africaines Needham, R. i960. The left hand of the Mugwe: An analytical note on the structure of Meru symbolism, Africa, vol. 30, no 1, pp. 20-33 1967. Right and left in Nyoro symbolic classification, Africa, vol. 37, no 4, pp.425-51 (ed). 1973. Right and left. Essays on dual symbolic classification. Chicago: University of

Chicago Press 1980. Reconnaissances. Toronto: University of Toronto Press. Spellig, F. 1927. Die Banyamwezi, Zeitschriftftir Ethnologie, vol. 50, no 3-6, pp. 62-6 and 201-57 Tanner, R. E. S. 1957. The installation of Sukuma chiefs in Mwanza district, African Studies, vol. 16, no 4, pp. 197-209 1958. Ancestor propitiation ceremonies in Sukumaland, Africa, vol.28, no 3, pp.225-31 Tcherkezoff, S. 1981. Le rituel dans l'ordre des valeurs. Hierarchie et sacrifice dans les royautes nyamwezi a la fin du XIXe siecle. These de 3e cycle, EHESS 1985. The expulsion of illness or the domestication of the dead - A case study of the Nyamwazi of Tanzania, in Interpreting Illness, M.Auge, ed., History and anthropology, vol. 2, no. 1, 1985 1986. Vengeance et hierarchie, ou comment un roi doit etre nourri, in R. Verdier (ed), La vengeance. Paris; Editions Cujas forthcoming'Les jumeaux nyamwezi. Efficacite rituelle etchangement de niveau' Turner, V. W. 1966. 'Colour classification in Ndembu ritual', in R. E. Bradbury, C. Geertz et al., eds, Anthropological Approaches to the Study of Religion, London,

Tavistock Varkevisser, C. 1973. Some functions and characteristics of the Sukuma Luganda (clan), Tropical man

Vincent, J.-F. 1978. Main gauche, main de l'homme: essai sur le symbolisme de la gauche et de la droite chez les Mofu, pp. 485-508 in Systemes de signes. Hommage a G. Dieterlen. Paris: Hermann

154

Bibliography

Wright, A. C. A. 1954. The magical importance of pangolins among Basukuma, Tanganyika notes and records, no 36, pp. 71—2

Yamaguchi, M. 1974. La royaute et le symbolisme dualiste chez les Jukun du Nigeria, Journal of Asian and African studies, no 8

Index

Bradbury, R. E., 132m Abrahams, R., 35, 69 alternation of pre-eminence, see Calame-Griaule, G., 148m reversal change of level, 18, 24, 68, 90, i n , analogy, 65, 67 ancestors, 29-34 possim, 41-5 passim, 52 124-5, I29> i 3 2 n I 55> 59, 62, 63, 65, 72-3, 125, Chinese, the, 7, 68, 92-8, 115, 128 I33n2, I34n2, i4onio, 141ml, classification, 3-7 passim, 29-42, 57, 142m, I42ni6, I5on4 100, 106, 119, 120, 132m, 14^12 'back', 65, 72 and values, 57, 100, 120 familial, 34,41,44,45, 71 see dualism 'head', 65, 72 complementarity, 7, 9, 17, 21, 89, 93, matrilateral, 34, 45, 52, 55, 63 112, 113, 118, 128-9, H^112 patrilateral, 34, 45, 52, 55, 63 contradiction, 5, 9, 68, 124; see also royal, 34, 41,44, 45, 73 dialectics Coppet de, D., vii, 120, I4on7, I44n7, 'Are'are, the, I4on7 asymmetry, 9, 13, 19, 20, 41, 58, i5 o n 3 8 9 - 9 0 , 108, 113, 115, 128 Cory, H., 20, 38, 39, 40, 48, 50, 56, 57, Auge, M., vii, 131 62, 69, 70, 78, 81, I34n3, Austen, R. A., I4on6, 141ml, 144113, I45n8, 147m 2 Bambara, the, 90 dialectics, 64, 115, 142m 7 Barraud, C , I44n2 Dieterlen, G., 7, 89-92, I48n2, Beidelman, T. O., 132m, 141ml Bennet, N. R., 32 Dogon, the, 7, 89-92, 128, 129, 130, Bernardi, B., 15-25 passim, I33n2 148m binary opposition, 5—8, 15, 18, 19, 21, dualism, 4-6, 8, 10, 14, 75, 89, 92-3, 26,31,39,67,69,73,93,94, 99, 106-12, 119, 125 'concentric', 107-9, m - 1 2 , 119, 125 'diametric', 107—8, n o , i n , 119, 125 Blohm, W., 20, 35, 37, 38, 39, 40, 50, 'pure diametric', 111-12, 119 60, 62, 65, 75, 80, 136m9, 137m, duality, totalising, 20, 23, 49, 72, 90, 142ml, 144114, blood price, 25, 30, 51, 127-8, 131, Bosch, F., 37-40 passim, 43, 44, 46, 47, 5<>> 52, 53, 57, 59, 62, 63, 65, 75, 81, I34n3, 137m, I37n2, 138^, I39n3, Hong, i4inio, 141ml, 142ml, I42ni6, 143m, I47ni2, I48ni2 Bororo, the, 108-9

of powers, 21, 69, 83-4, 148m 2 Dumont, L., vii, 8-12 passim, 22, 77, 92, 115-16, 121, I43ni7, I45n9 Durkheim, E., 15, 119-21, 132m iZcole Sociologique, 4

encompassing (relation of), 23, 41, 48, 57, 67, 71, 83, 85, 96-7,

156

Index

equalitarianism, 9, 21, 22, 41, 93, 115, 117, 127, 131, 149112, 150113 and hierarchy, 115, 130, 149112 Evans-Pritchard, E. E., 5, 132m exchange, ceremonial, vii, 29-30, 50, 119, 127, 130-1, I33n2, i35nio,

expulsion, see ritual Gass, P. M., 35, 37, 39, 47, 50, 59, 63, 64, I46ni2, I47ni2, I48ni2 Geertz, C , 132m Gottberg, A., I34n4 Granet, M., 7, 13, 92-8, I49n2 and the alternation of pre-eminence, Hertz, R., 3, 4, 5, 6, 12, 13, 14, 15, 21, 92, 132m Heusch, L. de, 132m hierarchy, 5-14 passim, 26, 29, 30, 38, 39, 40, 43-68, 77, 92, 93, 94, 115, 116, 117, 119-24, 131, I45n7, I45n9, I48n2, 150m of values, 26, 77, 120 of levels, 30, 40, 43-68, 77, 94, 123, holism, 10,29,77, 115, 131, see individualism Holmes, C. F., 50 idea-value, 9, 73, 112 ideology, see value individualism, 10-12, 22, 77, 115, 122, 131, I36n4 see holism initiation, 52, 53, 62-5 Iteanu, A., 84, 144^ Jamous, R., I5on6 Jukon, the, vii, Kaguru, the, 132m Kikuyu, the, 76 kingship, 10, 33, 49, 69-85, 93, 125, 136m 8, I39n4, i4omo, 144^ and priesthood, 69-76 La Flesche, F., 7, 92-3, 98-106 Leach, E. R., 132m levels, 6, 9, 14, 17, 26, 30, 31, 89, 104,

i n , 117, 122, 124, 125, 129, 132m, I36ni4 change in, 18, 24, 68, 90, 111, 124, 125, 129, 132m hierarchy of, 30, 43-68, 132m Levi-Strauss, C., vii, 4, 10, 93, 99, 101, 106, 108, 117, 128, 132m logic, European, 9, 68, 115-16, 119, 121-2, 150m hierarchical, 64, 68, 115, 119, 121-2, 124 Malcolm, D. W., 138^ Malinowski, B., 108 Maori, the, 14 Mauss, M., 94, 98, 119-21 Meru, the, 4-6, 13-21, 40, 65, 76, 128-9, 132m, I32n2 and the Mugwe, 13—22, 40, 76, 128-9, Melanesia, 84, Millroth, B., 20, 38, 39, 49, 53, 54, 56, 59,62, 76,81, I39n4, 141ml, Mofu, the, I48n2, Mworoha, E., i45nio myth, 81, 89, 122, 129 Needham, R., 4, 6, 13, 15-23, 25-6, 35>69>76> i3 2 n l > i33n2> H l n l Nyamwezi, the, 6, 7, 9, 29-76, 111, 114-15, 119, 125, 128 Nyoro, the, 4, Omaha, the, 99 opposition, binary, 15, 31,69, 73, i n , 113, centre/periphery, 108, no-11 complete/incomplete, 48-9 even/odd, 36-7, 38, 46-51, 1 4 0 ^ hierarchical, 73, 89, 94, 111, 113—14, H5 n 7 inside/outside, 125 man/woman, 35, 45-6, 48, 72, I37n2, 148m patrilateral/matrilateral, 35, 38 right/left, 14, 38-42, 43-68, 92-8 sacred/profane, 108 twin birth/single birth, 90 Osage, the, 7, 10, 92-3, 97-106, 108, i n , 114-16

Index Parsons, Talcott, 11, 118 Polynesia, 5 power, 5, 11,22,29,74, 76 and the sacred, 5,17 duality of, 21-2, 40, 69, 76 powers, theory of, 15, 17, 69 priest, see kingship prohibition, 56-7, 79, I4on8 Radcliffe-Brown, A. R., 128 reference, 9, 17, 26, 30, 92, 104, 106, 108, i4onio and classification, 106 duality of, 30 reversal, 6-9, 13, 18, 41, 43, 46, 56, 58-62, 67-8, 71-3, 75, 92-3, 94, 96, 105, 117, 123-4, 130 right hand, the pre-eminence of, 3—6, 8, 12, 13, 22, 93-4; see also opposition, right/left Right

157

royal, 52-4, 58, 60, 72, 74 Shilluk, the, 78 Sioux, the, 7, 106 Spellig, F., 81, I45n8 symbolism, 5, 15, 21, 22, 25, 51, 93, i n , 113-31 symmetry, 8, 9, 13, 21, 25, 113, 116, 123,131 Swazi, the, 128 Tanner, R. E., 48, 14m 10 totalisation, 10, 41, 54, 58, 61, 67-8, 7i-5>93, IOO, 105, 117, 118, 128 totemism, 99, 100 Trobriand Islands, 108 twinship, 7, 30, 39, 43, 46-8, 50, 54, 71, 74, 75, 89-92, 126, 130, values, 5-10, 74, 76-7, 84, 89, 100, i n , 115, 118-22, 124, I4omo,

and left, 4 , 5 , 7 , 1 1 8

ritual, 8, 34, 59, 91, 123-4, I3 0 " 1 context, 9, 19, 21,31, I42ni2 of destruction, 123, 130 of expulsion, 123, 124, 125 of rebellion, 8

Varkevisser, C , 50

sacred, the, 4, 10, 13-15, 21 sacrifice, 20, 30, 43, 50, 52-4, 56, 58-60, 69, 72, 74, 81, 124, 128, 129, 130, 131 familial, 52-4, 58, 60, 72, 74

Wright, A. C. A., 48

Winnebago, the, 93, 99, 107-8, 111 witchcraft, 32, 34, 57, 59, 60, 62, 125-6, I36ni5, 142ml,

Yamaguchi, M., vii, 1 4 4 ^ Zemp, H.,