Rethinking Anthropology – E. R. Leach

RETHINKING ANTHROPOLOGY LONDON SCHOOL OF ECONOMICS MONOGRAPHS ON SOCLL ANTHROPOLOGY Managing Editor: Anthony For^e The Monographs in on Social Anthropology were established modem The by 1940 and aim to publish results of anthropological research of primary interest to specialists. continuation of the series was made possible from the Wenner-Gren Foundation for Anthropological Research, and more recently by a further grant from the Governors of the London a grant in aid School of Economics and Political Science. re under the direction of an Board associated with the Department of Anthropology of the London School of Economics and Political Science. Editorial The Monographs LONDON SCHOOL OF ECONOMICS MONOGRAPHS ON SOCIAL ANTHROPOLOGY No. 22 Rethinking Anthropology by E. R. LEACH UNIVERSITY OF LONDON THE ATHLONE PRESS NEW YORK: HUMANITIES PRESS INC Published by THE ATHLONE PRESS UNIVERSITY OF LONDON at 2 Gotcer Street, Distributed by Tiptree London wci Book Services Ltd Tiptree, Essex F irst edition, 1961 First paperback edition with corrections, 1966 Reprinted, 1968, 1971 E. R. Leach, 1961, 1971 U. K. U. K. sB N o 485 19522 4 cloth sB N B o 485 19622 o paperback U. S. A. s N 391 00146 9 paperback First printed in 1961 by ROBERT CUNNINGHAM AND SONS LTD ALVA Reprinted by photo-litho by JOHN DICKENS & CO LTD NORTHAMPTON 4- M75†² Preface The title of this collection properly belongs only to the first essay. On 3 December 1959 1 had the honour to deliver the first Malinovvski Memorial Lecture at the London School of Economics. The Editorial Board of the London School of Economics Monographs in Social Anthropology enerously offered to publish the text of my lecture but added the flattering suggestion that I should reprint a number of my other essays at the same time. I have accordingly appropriated the title of my Malinowski lecture for the whole collection. I do not pretend wholly consistent with that The essays extend over a period of fifteen years and is that th e viewpoint of the latest (Chapter i) of the earliest (Chapter 2) but there is, I think, a certain continuity of theme and method in all of them. When they were first written all these essays were attempts to ‘rethink anthropology'.All are concerned with problems of others, I ‘theory' and are based on ethnographic facts recorded by my own contribution being primarily that of analyst. In each case have tried to reassess the known facts in the light of unorthodox assumpSuch heresy seems to me to have merit for its own sake. Unconventional arguments often turn out to be wrong but provided they provoke discussion they may still have lasting value. By that criterion each of the essays in this book is a possible candidate for attention. tions. Among social anthropologists the is game f building new theories on the ruins of old ones almost an occupational disease. Contemporary arguments in social anthropology are built out of formulae concocted by Malinowski, Radcliffe-Brown an d Levi-Strauss who in turn were only ‘rethinking' Rivers, Durkheim and Mauss, who borrowed from Morgan, McLennan and Robertson- Smith the total outcome of all — and so on. Sceptics may think that despite all this ratiocination adds up to very little; our pedagogical subtleties, the diversities of human custom remain as bewildering as ever. But that we admit.The contemporary social anthropologist is all too well aware that he knows much less than Frazer imagined that he knew for certain. But that perhaps is the point. The contributions to anthropological pedantry collected in this book add little to the sum of human knowledge but if they provoke some readers to doubt their sense of certainty then they will have served their purpose. A note on the interconnections between the different papers draft of Chapter 2 may prove still helpful. The first was written in 1943 while I was on VI PREFACE and still in direct contact with Jinghpaw speakers. ppeared in the 1945 volume of the J. R. A. I, this was not actually published until 1950. These details of dating are relevant because they explain why my paper contains no reference to Chapters 15 and i6 of Levi-Strauss, Les structures elementaires de la parente (1949) and reciprocally why the latter work ignores the new information provided by my paper. Chapter 3, which was originally a Curl Prize Essay, was completed in the spring of 195 1 and seems to have been the first English language commentar)' on Levi-Strauss's magnum opus though, presumably, my paper and J.P. B. de Josselin de Jong's monograph Levi-Strauss's Theory on Kinship and Marriage (1952) were going through the press at the same time. Although I here criticized Levi-Strauss on the grounds of ethnographical inaccuracy my sympathy with his general theoretical point of view is very great. Professor Levi-Strauss has himself noted the similarity between the view of ‘social structure' implicit in my first Jinghpaw paper (Chapter 2) and his own (Levi-Strauss, 1953, p. 525 n), and in all my subsequent publications my debt to Levi-Strauss is obvious.The relationship of Chapter 4 to earlier literature will be apparent from the references in the text. Although it was not intended to be controversial it provoked Dr Kathleen Gough into a vigorous reply (Gough, 1959). The crucial part of my argument here is that I emphasize the need to distinguish between affinity regarded as an alliance between corporate kin groups and those individual affinal ties which bind a particular wife to a particular husband. This theme recurs in Chapter 5 and again in active military service t Although Chapter i. Chapter 5, as indicated in the text, is linked with a long correspondence which appeared in the pages of Man in 1953 and 1954 but the response which it evoked from my close academic colleagues is only marginally connected with this earlier discussion. Dr Goody has denounced my whole argument as grounded in fundamental error (Goody, 1959, p. 86 ) and Professor Fortes has taken up most of two issues of Man to expound my fallacies and confusions (Fortes, 1959b).Both these explosions of academic wrath were provoked by a single sentence in my essay, namely ‘Thus Fortes, while recognizing that ties of affinity have comparable importance to ties of descent, disguises the former under his expression â€Å"complementary filiation† (see below p. 122). The exact sense in which this statement is an ‘error' is still not clear to me for in the course of his denunciation Fortes reaffirms his view that ‘complementary filiation is a function of affinal relations' (Fortes, 1959b, p. 209) which is precisely the argument I sought to controvert. ^ Professor Fortes has called his article *a rejoinder to Leach', and — readers of Chapter i of this book need to appreciate that ‘a among other things in it is intended as rejoinder to Fortes'. Reference to a short note Man (i960. Art. 6) will perhaps help to mak e this clear. The two short papers on time symbolism reprinted in Chapter 6 do PREFACE influence of Professor Levi-Strauss Vll not form a series with the other chapters of the book though again the is pronounced. Although my ‘Cronus and Chronos' appeared in print in 1953 while Levi-Strauss's ‘The Structural Study of Myth' was only published in 1956, I had in fact already heard Professor Levi-Strauss's lecture on this topic before I wrote my essay.Explorations, the Toronto University publication in which my Chapter 6 was originally published, carried on its fly leaf the statement that it was ‘designed, not as a permanent reference journal that embalms truth for posterity, but as a publication that explores and searches and questions' and both my papers are correspondingly brief and tentative. Nevertheless a number of my friends have suggested that the arguments they contain are of more than ephemeral interest; hence the reissue here^ Chapter i contains a considerabl e amount of matter which was not included in the spoken text of my Malinowski lecture. The other essays^ appear as originally printed, except for the correction of misprints, and one or two very minor alterations intended to clarify the argument. The Introductory Notes at the beginning of Chapters 2-6 are new. Acknowledgements I am indebted to the Council of the Royal Anthropological Institute of Great Britain and Ireland for permission to reprint the essays published here as Chapters 2, 3, 4 and 5 and to Professor E. S. Carpenter and the University of Toronto for permission to reprint the two short essays included in Chapter 6.I am indebted to a personal grant in aid from the Behavioral Sciences Division of the Ford Foundation for facilities employed while preparing } * j] : -^ these papers for publication. E. R. L. Contents 1. RETHINKING ANTHROPOLOGY I 2. JINGHPAW KINSHIP TERMINOLOGY THE STRUCTURAL IMPLICATIONS OF MATRILATERAL CROSS-COUSIN MARRIAGE 28 3. 54 4. POLYANDRY, INHERITAN CE AND THE DEFINITION OF marriage: with PARTICULAR REFERENCE TO SINHALESE CUSTOMARY LAW ASPECTS OF BRIDEWEALTH AND MARRIAGE STABILITY IO5 5. AMONG THE KACHIN AND LAKHER 6. II4 TWOESSAYS CONCERNING THE SYMBOLIC REPRESENTATION OF TIME (i) 124 Cronus and Chronos, 124 (ii) Time and False Noses, 132 Rethinking Anthropology my arrogant title. Since 1930 British Anthropology has embodied a well defined set of ideas and -^objectives which derive directly from the teaching of Malinowski and Radcliffe-Brown this unity of aim is summed up in the statement that British social anthropology is functionalist and concerned with the comparative analysis of social structures. But during the last year or so it has begun to look as if this particular aim had worked itself out.Most of my colleagues are giving up the attempt to make comparative generalizations instead they have begun to write impeccably detailed historical ethno- tET Social me begin by explaining — graphies of particular peoples. I regret this new tendency for I still believe that the findings of anthro- pologists have general as well as particular implications, but functionalist doctrine ceased to carry conviction? why has the understand what is happening in social anthropology I believe we need to go right back to the beginning and rethink basic issues really elementary matters such as To — hat we mean by marriage or descent or the unity of difficult siblings, and that is — for basic concepts are basic; The the ideas one has about them are deeply entrenched and firmly held. One bias of the things we need to recognize is the strength of the empirical which Malinowski introduced into social anthropology and which essential core of social anthropology has stayed with us ever since. is understanding of the way of life of a single particular people. This fieldwork is an extremely personal traumatic kind of experience and the personal involvement of the anthropologist in his work is reflected in wh at he produces.When we read Malinowski we get the impression that he is stating something which is of general importance. Yet how can this be? He is simply writing about Trobriand Islanders. Somehow he has so assimilated himself into the Trobriand situation that he is able to make the Trobriands fieldwork a microcosm of the whole primitive world. successors; for Firth, Primitive citizen of —the And the same is true of his is Man is a Tikopian, for Fortes, he a Ghana. The existence of this prejudice has long been recognized / but we have paid inadequate attention to its consequences.The difficulty of achieving comparative generalizations is directly linked with the problem of escaping from ethnocentric bias. 2 RETHINKING ANTHROPOLOGY As is appropriate to an occasion I when we honour the memory of Bronislaw MaUnowski, am going to be thoroughly egotistical. I shall imply there my own is merit by condemning the work of in my closest friends. But purpose is to distinguish between two rather similar varieties of comparative generalization, both of which turn up from time to time in contemporary British social anthropology.One of these, which I dislike, derives from the work of Radcliffe-Brown; the other, which I admire, derives from the work of Levi-Strauss. It is important that the differences between these two approaches be properly understood, so I shall draw my illustrations in sharp contrast, all black and all white. In this harsh and exaggerated form Professor Levi-Strauss method my malice. My might well repudiate the authorship of the ideas which I am trying to convey. Hence my egotism; let the blame be wholly mine. My problem is simple.How can a modern social anthropologist, with all the work of Malinowski and Radcliffc-Brown and their successors at his elbow, embark upon generalization with any hope of arriving at a satisfying conclusion? My answer is quite simple too; it is this: By thinking of the organizational ideas that are present in any societ y as constituting a mathematical pattern. The rest of what I have to say that is simply an elaboration of this cryptic statement. concern is with generalization, not with maintained that the objective of social anthropology was the ‘comparison of social structures'.In explaining this he asserted that when we distinguish and compare different types of social structure we are doing the same kind of thing as when we distinguish different kinds of sea shell according to their structural type (RadcliffeBrown, 1953, p. 109). Generalization is quite a different kind of mental First let me emphasize my comparison. Radcliffe-Brown operation. Let me illustrate this point. two points can be joined by a straight line and you can represent this straight line mathematically by a sm^G. first order algebraic equation.Any three points can be joined by a circle and you can represent this circle by a quadratic or second order algebraic equation. It would be a generalization to go straight on fro m there and say any : Any n points in a plane can be joined by a curve which can be represented by an equation of order n-i. This would be just a guess, but it would be true, and it is a kind of truth which no amount of comparison can ever reveal. Comparison and generalization are both forms of scientific activity, but different. Comparison is a matter of butterfly collecting —of classification, of the rrangement of things according to their types and subtypes. The followers of Radcliffe-Brown are anthropological butterfly collectors and their approach to their data has certain consequences. For example, according to RadclifTe- Brown's principles we ought to think of Trobriand society : RETHINKING ANTHROPOLOGY as 3 classification a society of a particular structural type. The might proceed thus: Main Type Sub-type: Sub-sub-type : societies societies societies composed of unilineal descent groups. composed of matrilineal descent groups. composed of matrilineal descent groups i n which he married males of the matrilineage live together in one place and apart from the females of the matrilineage, and so on. In this procedure each class preceding it is a sub-type of the class immediately in the tabulation. its uses, but it has very serious has no logical limits. Ultimately discriminated in this way as a sub-type Now I every just agree that analysis of this kind has is limitations. One major defect known society can be that it from any other, and since anthropologists are notably vague about what they mean by ‘a society', this will lead them to distinguish more and more ocieties, almost ad infinitum. This is not just hypothesis. My colleague Dr Goody has gone to great pains to distinguish as types two adjacent societies in the Northern Gold Coast which he calls LoWiili and LoDagaba. A careful reader of Dr Goody's works will discover, however, that these two ‘societies' are distinct simply the way that field Dr Goody notes from two has chosen to de scribe the fact that his neighbouring communities show some curious discrepancies. If limit Dr Goody's methods of analysis were pushed to the we should be able to show that every village community throughout is he world constitutes a distinct society which distinguishable as a type from any other (Goody, 1956b). Another serious objection is that the typology makers never explain why they choose one frame of reference rather than another. RadcliffeBrovsTi's instructions were simply that ‘it is necessary to compare societies the economic system, the with reference to one particular aspect . . . political system, or the kinship system' . . . this is equivalent to saying that you can arrange your butterflies according to their colour, or their size, or the shape of their wings according to the him of the moment, but no matter what you do this will be science. Well perhaps, in a sense, it is; but you must realize that your prior arrangement creates an initial bias from which it is later extremely difficult to escape (Radcliffe-Brown, 1940, p. xii). Social anthropology is packed with frustrations of it this kind. An obvious Ever since example is the category opposition patrilineal/matrilineal. has been customary for anthropologists to distinguish unilineal from non-unilineal descent systems, and writing of the Iroquois, Morgan began among that it the former to distinguish patrilineal societies from atrilineal societies. is These categories now seem to us so rudimentary and obvious extremely difficult to break out of the straitjacket of thought which the categories themselves impose. 4 RETHINKING ANTHROPOLOGY Yet if our approach is to be genuinely unbiased we must be prepared to consider the possibihty that these type categories have no sociological significance whatsoever. It may be that to create a class labelled matrtis as irrelevant for our understanding of social structure as the creation of a class blue butterflies is irrelevant for the understanding of t he anatomical structure of lepidoptera.I don't say it is so, but it may be; it is lineal societies time that we considered the possibility. J I warn you, the rethinking of basic category assumptions can be very disconcerting. But Let me cite a case. Dr Audrey Richards's well-known contribution to African Systems of Kinship and Marriage is an essay in Radcliffe-Brownian typology making which is rightly regarded as one of the ‘musts' of undergraduate reading (Richards, 1950). In this essay Dr Richards asserts that societies is ‘the problem' of matrilineal the difficulty of combining recognition of descent through the oman with the rule of exogamous marriage, and she classifies a variety of matrilineal societies according to the way this ‘problem' is solved. In effect her classification turns on the fact that a woman's husband the two men. jointly possess rights in the woman's brother and a woman's children but that matrilineal systems differ in the way these rights a re allocated between is the prior category assumptions. Men have kinds of society, so why should it be assumed from the start that brothers-in-law in matrilineal societies have special ‘prob- What I object to in this ll brothers-in-law in lems' which are absent in patrilineal or bilateral structures? really What has lay a matrilineal society, she has decided to restrict her comparative obser-ations to matrilineal systems. Then, having selected a group of societies which have nothing in common except that they are matrilineal, she is naturally led to conclude that matrilineal descent is the major factor to which all the other items of cultural behaviour which she happened here with the Bemba, is that, because Dr Richards's own special knowledge describes are functionally adjusted.Her argument I am afraid is a tautology; her system of classification already implies the truth of what she claims to be demonstrating. This illustrates how Radcliffe-Brown's taxonomic assumptions fit in with the ethnocentric bias which I mentioned earlier. Because the typefinding social anthropologist conducts his whole argument in terms of tempted particular instances rather than of generalized patterns, he is constantly to attach exaggerated significance to those features of social organization which happen to be prominent in the societies of which he himself has first hand experience. The ase of Professor Fortes illustrates this is same point in rather a different way. His quest not so much for types as for prototypes. It so happens that the two societies of which he has made a close study have certain similarities of structural pattern for, while the Tallensi are patri- RETHINKING ANTHROPOLOGY lineal 5 and the Ashanti matrilineal, both Tallensi and Ashanti come unfiliation', usually close to having a system of double unilineal descent. Professor Fortes has devised a special concept, ‘complementary w^hich helps him to describe this double unilineal element in the Tallen si/Ashanti pattern w^hile rejecting the notion that these societies actually possess double unilineal systems (Fortes, 1953, p. 33; 1959b). It is interesting to note the circumstances which led to the development of this concept. From one point of view ‘complementary filiation' is simply an inverse form of Malinowski's notion of ‘sociological paternity' as applied in the matrilineal context of Trobriand society. But Fortes has done more than invent a new name for an old idea; he has made it the corner stone of a substantial body of theory and this theory arises logically special circumstances of his own field experience.In his earlier writings the Tallensi are often represented as having a somewhat extreme form of patrilineal ideology. Later, in contrast to from the Rattray, Fortes placed an unambiguously matrilineal label upon the Ashanti. view, is The that merit of ‘complementary it is filiation', from Fortes's point of a concept which applies equally well to bot h of these contrasted societies but does not conflict with his thesis that both the Tallensi and the Ashanti have systems of unilineal descent. The concept ecame necessary to him precisely because he had decided at the start that the more familiar and more obvious notion of double unilineal descent was inappropriate. In retrospect Fortes seems to have decided that double unilineal descent is a special development of ‘complementary filiation', the latter being a feature of all unilineal descent structures. That such category distinctions are contrived rather than natural is evident from Goody's additional discrimination. Goody asserts that the LoWiili have ‘complementary descent rather than a dual descent system'.Since the concept of ‘complementary filiation' was first introduced so as to help in the distinction between ‘filiation' and ‘descent' and since the adjective ‘complementary' cannot here be given meaning except by reference to the word â⠂¬Ëœdescent', the total argument is clearly tautologous (Fortes, 1945, pp. 134, 20of; 1950, p. 287; 1953, p. 34; 1959; Goody, 1956b, p. 77). Now I do not claim that Professor Fortes is mistaken, but I think he is misled by his prior suppositions. If making and from enthnocentric science. we are to bias we must let escape both from typology turn to a different kind ofInstead of comparison repeat. Generalization us have generalization; instead of inductive; it butterfly collecting let us have inspired guesswork. Let me is consists in perceiving it is possible general laws in the circumstances of special cases; guesswork, a gamble, you may be wrong or you may be right, but if you happen to be right you have learnt something altogether new. In contrast, arranging butterflies according to their types and sub-types is tautology. It merely reasserts something you know already in a slightly different form. 6 RETHINKING ANTHROPOLOGY But if you are going is o start guessing, you need I to kno w how to guess. . d this wliat I am getting at when say that the form of thinking should be mathematical. Functional ism in a mathematical sense is not concerned with the interconnections between parts of a whole but with the principles of operation of partial systems. There is a direct conflict here with the dogmas of Malinowski and Malinowski's functionalism required us to think of each Society (or Culture, as Malinowski would have put it) as a totality Radcliffe-Brown. of a made up kinds — number of discrete empirical ‘things', of rather diverse institutions', e. g. groups of people, customs. These ‘things' are functionally interconnected to form a delicately balanced mechanism rather like the various parts of a wrist watch. cliff†e- The functionalism of Rad- Brown was equally mechanical though the focus of interest was different. RadclifTe-Brown was concerned, as it were, to distinguish wrist watches clocks, whereas Malinowski was interested in the gener al attributes of clockwork. But hath masters took as their starting point the notion that a culture or a society is an empirical whole made up rom grandfather of a limited two societies number of readily identifiable parts and that when we compare we are concerned to see whether or not the same kinds of is parts are present in both cases. This approach a mechanic but appropriate for a zoologist or for a botanist or for it is not the approach of a mathematician nor of an engineer and, in gineer. my view, the anthropologist has much in common with the en- But that is my private bias. I was originally trained as an engineer. The entities which we call societies are not naturally existing species, neither re they man-made mechanisms. But the analogy of a mechanism has quite as much relevance as the analogy of an organism. This is not the place to discuss the history of the organic analogy as a model for Society, but its arbitrariness is often forgotten. Hobbes, who developed his notion of a social organism in a very systematic way, discusses in his preface whether a mechanical or an organic analogy might be the more appropriate for his purpose. He opts for an organism only because he wants to include in his model a metaphysical prime mover (i. . God Life Force) (Hobbes, 1957, p. 5). In contrast RadcHffe-Brown employed the organic analogy as a matter of dogma rather than of choice (e. g. Radcliffe-Brown, 1957, pp. 82-86; 1940a, pp. 3, lo) and his butterfly collecting followers have accepted the appropriateness of the phrase ‘social organism' without serious discussion. Against this complacency I — must protest. It is certainly the case that social scientists must often resort all to analogy but eternity. we are not committed to one type of model making for Our task societies s to understand and explain what goes on in society, how work. If an engineer tries to explain to you how a digital computer RETHINKING ANTHROPOLOGY bolts. 7 works he doesn't spend his time classifying different kinds of nuts and He concerns himself with principles, not with things. He writes out argument as a mathematical equation of the utmost simplicity, somewhat on the lines of o + i = i i + i = 10. No doubt this example is frivolous; such computers embody their information in a code which is transmitted in positive and negative impulses denoted by the digital symbols o and i.The essential point is that although the information which can be embodied in such codes may be enormously complex, the basic principles on which the computing machines work is very simple. Likewise I would maintain that quite simple mechanical models can have relevance for social anthropology despite the acknowledged fact that the detailed empirical facts of social life display the utmost complexity. I don't want to turn anthropology into a branch of mathematics but I believe we can learn a lot by starting to think about society in a mathehis : ; matical way.Considered mathematicall y society is not an assemblage of things but an assemblage of variables. A good analogy would be with that branch of mathematics known as topology, which may crudely be described as the geometry of elastic rubber sheeting. If I have a piece of rubber sheet and draw a series of lines on it to symbolize the functional interconnections of some set of social phenomena and I then start stretching the rubber about, I can change the manifest shape of my original geometrical figure out of all recognition and yet clearly there is a sense in which it is the same figure all the time.The constancy of pattern is not manifest as an objective empirical fact but it is there as a mathematical generalization. By analogy, generalized structural patterns in anthropology are not restricted to societies of any one manifest structural type. you will tell me that topology is one of those which mere sociologists had best avoid, but I am not in fact proposing anything original. A very good simple account of the nature of topology appears in an article under that title in the current edition of the Encyclopaedia Britannica.The author himself makes the point that because topology is a non-metrical form of mathematics it deserves especial attention from social scientists. I Now know that a lot of alarming scientific mysteries The fundamental Any closed curve is arc of a circle is ‘the variable in topology ‘the is the degree of connectedness. same as' any other regardless of its shape; the same as' a straight line because each is open ended. Contrariwise, a closed curve has a greater degree of connectedness than an arc. If of pattern case if we apply these ideas to sociology particular relationships e cease to be interested in and concern ourselves instead with the regularities relationships. is among neighbouring In the simplest possible there be a relationship p which intimately associated with another relationship q then in a topological study we shall not concern ourselves 8 RETHINKING ANTHROPOLOGY with the particular characteristics of/) and q but with their mutual characteristics, i. e. with the algebraic ratio p'^q. But it must be understood that the relationships and sets of relationships which are symbolized in this way cannot properly be given specific numerical values.The reader should bear this point in mind when he encounters the specimens of pseudo-mathematics which occur later in this paper. All propositions in topology can also be expressed as propositions in symbolic logic (see Carnap, 1958, chapter G) and it was probably a consideration of this fact which led Nadel to introduce symbolic logic into own view is that while the consideration book (Xadel, 1957). of mathematical and logical models may help the anthropologist to order his last My his theoretical arguments in an all this intelligent way, his actual procedure s should be non-mathematical. The pattern relevance of to my main theme that the saTne structural may turn up in any kind of society patrilineal —a mathematical approach matrilineal makes no prior assumption that from non-unilineal systems or structures. all unilincal systems are basically different structures from the contrary, the principle of parity leads us to discount rigid category distinctions of this kind. On Let me try to illustrate I for the occasion shall take my point with an example. To be my example from Malinowski. Malinowski reported, as a ppropriate Most of you will know that fact of empirical ethnography, that the Trobrianders profess ignorance of the connection between copulation and pregnancy and that this ignorance serves as a rational justification for their system of matrilineal descent. From the Trobriand point of view ‘my father' (tama) is not a blood relative at all but a kind of affine, *my mother's husband' (Malinowski, 1932a, p. 5). However, alongside their dogmatic ignorance of the facts of life, Trobrianders also maintain that every child should resemble its mother's husband (i. . its father) but that no child could ever resemble a member of its own matrilineal kin. Malinowski seems to have thought it paradoxical that Trobrianders should hold both these doctrines at the same time. He was apparently bemused by the same kind of ethnocentric assumptions as later led a Tallensi informant to tell Professor Fortes that ‘both parents transmit their blood to their offspring, as can be seen from the fact that Tallensi children may resemble either parent in looks' (Fortes, 1949, p. 35; my italics). This is mixing up sociology and genetics.We know, and apparently the Tallensi assume, that physical appearance is genetically based, but there is no reason why primitive people in general should associate ideas of genetic inheritance with ideas about physical resemblance between persons. The explanation which the Trobrianders gave to Malinowski was that a father impresses his appearance on his son by cohabiting repeatedly with the mother and ther eby ‘moulding' (kuli) the child in her womb (Malinowski, 1932a, p. 176) which is reminiscent of the Ashanti . RETHINKING ANTHROPOLOGY view that the father shapes the body of his child as might a potter (Rattray, 1929, p. 9). This Trobriand theory is quite consistent with the view that the father is related to the son only as mother's husband that is, an affine and not as a kinsman. There are other Trobriand doctrines which fall into line with this. The father's sister is ‘the prototype of the lawful woman' (Malinowski, 1932a, p. 450) which seems to be more or less the equivalent of saying that — the father (tama) is much the same sort of relation as a brother-in-law.Again, although, as Powell has shown (Powell, 1956, p. 314), marriage with the father's sister's daughter is rare, the Trobrianders constantly assured Malinowski that this was a very right and proper marriage. Evidently in their view the category tama (which includes both father and father's sister's s on) is very close to that of lubou (brother-in-law) (Mal- inowski, 1932a, pp. 86, 451). The similarity is asserted not only in verbal expression but also in the pattern of economic obligation, for the harvest gift (urignbu) paid by a married man is due both to his mother's husband tama) and to his sister's husband (lubou) (Malinowski, 1935, I, pp. 386, 413-18). From my point of view this cluster of Trobriand beliefs and attitudes is a ‘pattern of organizational ideas'— it specifies a series of categories, in a particular relationship and places them with one another as in an was biased by his down to earth empiricism, by European prejudices and by his interest in psycho-analysis, and he refused to accept tlie Trobriand doctrine at its face value. Instead he refurbished his concept of ‘sociological paternity' which he had originalgebraic equation.But Malinowski ally devised to fit a quite different context, that of patrilineal organization among On to the Australia n Aborigines (Malinowski, 19 13, p. 170-83). this earlier occasion Malinowski had used ‘sociological paternity' relations show how between parents and children and between spouses derive from customary rules and not from any universal facts of biology or psychology, but in the later application of these ideas to Trobriand circumstances he shifts his ground and the argument becomes confused by the introduction of naive psychological considerations. On the face of t ‘sociological paternity', as used in The Sexual Life of attitudes Savages, seems to mean that even in a society which, like the Trobriands, sociological still denies the facts of ‘biological paternity', pertain to paternity, as zve understand it, which far, may be found. So so good. But Malinowski goes further than this. Instead of arguing, as in the Australian case, that kinship attitudes have a purely social origin, he now insists that social attitudes to kinship arc facts. rooted in universal psycholog ical The paternal relationship contains elements which are necessarily resent in the father/child relationship of all societies, no matter what the circumstances of custom and social structure confusing. may be. This is all very On the one hand the reader is is told quite plainly that the Trobriand child taught to think of his father as a non-relative, as an lO RETHINKING ANTHROPOLOGY individual with the special non-kinship status of mother's husband. But on the other hand the reader is forced to conclude that this ‘IVobriand mother's husband is related to the mother's child ‘as a sociological father', that is to say by ties of kinship as well as by tics of affinity.The argument, as a whole, is self-contradictory. is You may about. well think that this a yery hairsplitting point to make a fuss How can it possibly make any difference whether I think of a parti- cular male as my father or as is my mother's husband? Well, all I can say that anthropologists do Professor Fort es, Dr Goody and Dr Kathleen Gough on this subject that worry about such things. are so disturbed by my heretical yiews oflF time to try to bruise my owski's argument (Fortes, 1959)-