Schneider: Some muddles in the models

This work is licensed under the Creative Commons | © David M. Schneider.

Some muddles in the models:

or, how the system really works1

David M. Schneider, University of Chicago


Part one. Alliance

I. The phrase ‘alliance theory’ and its opposition to what has been called ‘descent theory’ was first suggested by Dumont (1961a).

Alliance theory, with roots clearly in Durkheim and Mauss, has specifically arisen out of Lévi-Strauss’s Structures élémentaires … (1949) and has been developed by Lévi-Strauss, Dumont, Leach, and Needham. Descent theory also has its roots in Durkheim and Mauss, but its development has been through Radcliffe-Brown to Fortes, Goody, Gough, Gluckman, and, in certain respects, Firth.

This is an oversimplified picture, of course, but one which provides a reasonable beginning. It would oversimplify matters, too, but also be useful to point out, that where Durkheim tried to bridge the gap between positivism and idealism and ended up as an idealist in the remnants of some positivist clothing, Needham’s version of alliance theory is, if anything, squarely on the side of the idealists. Lévi-Strauss and Dumont, on the other hand, go with Hegel (Murphy 1963). Descent theory has moved in the direction of positivism; some of its misunderstandings stem from its positivist premises; and the direction which the younger descent theory people (Goody, Gough) have taken seems to me to be consistent with this view.

The dilemma of positivism is exemplified by a statement of Lévi-Strauss. Replying to Maybury-Lewis’s criticisms, Lévi-Strauss says:

‘… Mr. M. L. remains, to some extent, the prisoner of the naturalistic misconceptions which have so long pervaded the British school … he is still a structuralist in Radcliffe-Brown’s terms, namely, he believes the structure to lie at the level of empirical reality, and to be a part of it. Therefore, when he is presented a structural model which departs from empirical reality, he feels cheated in some devious way. To him, social [25] structure is like a kind of jigsaw puzzle, and everything is achieved when one has discovered how the pieces fit together. But, if the pieces have been arbitrarily cut, there is no structure at all. On the other hand, if, as is sometimes done, the pieces were automatically cut in different shapes by a mechanical saw, the movements of which are regularly modified by a cam- shaft, the structure of the puzzle exists, not at the empirical level (since there are many ways of recognizing the pieces which fit together); its key lies in the mathematical formula expressing the shape of the cams and their speed of rotation; something very remote from the puzzle as it appears to the player, although it “explains” the puzzle in the one and only intelligible way’ (Lévi-Strauss, 1960, p. 52).

The contrast may be put more specifically. To a degree, both alliance and descent theory are concerned with social structure. But, for descent theory, social structure is considered one or another variant of the concept of (a) concrete relations or groupings, socially defined, which (b) endure over time. To Radcliffe-Brown, social structure is the network of ‘actual social relations’; for Evans-Pritchard in discussing the Nuer (Evans-Pritchard 1940) it is the enduring social groups, the concrete lineages.3

For alliance theory, the problem is not what the concrete patterns of social relations actually are, although these are not neglected; it is not the actual organization of any specific group like a lineage. It is, instead, that construct or model which is fabricated by the anthropologist and which is presumed to have, as its concrete expression, the norms for social relations and the rules governing the constitution of social groups and their inter-relations (Lévi-Strauss, 1953).

Alliance theory grows out of (1) Durkheim’s distinction between mechanical and organic solidarity; (2) his notion of collective representations and specifically his and Mauss’s insistence that the fundamental socio-cultural categories of the culture itself must be understood in its own terms; and (3) out of Mauss’s ideas that are expressed in his essay, The Gift. This is perhaps the minimal and most immediately relevant set of ideas. others more derivative or tangential are involved, but need not detain us at this point. [26]

First, organic solidarity is essentially a state of affairs in which functionally differentiated parts are each non-viable parts but, when joined together, become a single, coherent, and viable ‘organism’. Functional differentiation is in turn based on two kinds of condition. One is that the parts have different roles which get those things done which need to be done if the whole is to survive. The other is that for the parts to work properly they must be oriented toward common goals. if the goals of one part are disparate from the goals of the other, then the parts will not make up a single, coherent, viable whole. This second feature may be considered to be most intimately related to ‘coherence’. Each part by itself means nothing; the parts together make an integrated and viable unit. The simplest example is of two unilineal lineages, each being able to provide for its essential needs in all ways except one: the rule of exogamy requires that marriage partners come from outside the group. These two lineages, both having a common aim — to provide for wives — exchange wives. The two lineages together, then, make up one viable and coherent society. That they may also exchange differentiated economic products only elaborates and reinforces their differentiation and interdependence.

The simplest form of differentiation centers on the definition of opposed but interrelated parts. Again, exogamy is the pertinent example. From the point of view of either of the two lineages, one ‘needs’ wives, the other ‘has wives to give’. The definitions are plus and minus, have and have not, and so on. These are essentially opposed elements, defined in terms of their polar qualities. Just what particular form this differentiation will take in any society is an empirical question. Yet it is implied, i think, that all societies operate in fundamentally this way: at every point of differentiation, there is at rock-bottom a polar, opposite kind of differentiating definition which in any particular case may be elaborated into gradients, subdivided into qualities, but basically, and in some sense ‘originally’, the elements are oppositional, unitary, and polar. (Lévi-Strauss, 1945, 1949, 1955, 1962, 1963; Needham 1958a, 1960d.)

There remains always the substantive problem: how is this particular social structure differentiated? What are its constituent parts? How are they interrelated? Here the idealistic [27] foundation of the theory becomes apparent. On the one hand, as a fundamental category of human mentality, dualism is expressed in a wide variety of particular social forms; on the other hand, it is the particular ideational categories, as these are consciously or unconsciously conceived by the members of the society itself, which play a role in regulating their action.

For example, in a matrilateral cross-cousin marriage system, where a ‘positive marriage rule’ (Dumont’s phrase (Dumont, 1961); a prescriptive system in Needham’s terms) obtains, the system is conceived of, from the point of view of any given lineage in that system, as one which is made up of dual relations — ‘we’ versus ‘they’ as wife-givers to wife-takers; ‘they’ versus ‘we’ as wife-takers to wife-givers. The system as a system is triadic; yet the conceptualization of the system from the point of view of the given lineage is dyadic (Lévi-Strauss, 1949, 1956; Needham, 1958a, 1960d, 1962a.)

But one must go further; in any society having a positive marriage rule, that class of persons who are potential wives is categorized in some particular way. To call this, for instance, MoBrDa is to commit the error of imposing our way of thinking on their system when these may (though they may not) be radically different. It is thus necessary to find out how they conceptualize their system, what their categories of social objects and actors are, how they are constituted and how they are conceived to behave.

This raises a core problem: symbolism. In alliance theory, it is held that a given structural relationship in a very important sense cannot be seen or observed as such. It is, instead, ‘expressed’ in a wide variety of different ways, and the ‘expressions’ of the relationship tend to be reiterative, though seldom identical from one form of expression to another. The problem in analyzing social structure is thus to search out the various forms of expression in order to comprehend the basic relationship which is structurally operative, but no particular form of it, no particular expression of it, can be taken as ‘it’ (Lévi-Strauss, 1960).

The contrast here with Radcliffe-Brown’s, Murdock’s, and similar conceptions is very clear. For Murdock, it is the concrete fact of where specific people actually reside which, bringing them [28] in proximity with one another, creates unilineal descent groups (Murdock, 1949). For the intellectualist, where they live, how they are brought into proximity with one another, and how the unilineal descent group is organized are all equally symbols of, or expressions of, the same inherent structural principle. Where residence rules cause or determine descent groups for Murdock, it is structural principles which account for both residence rules and descent groups for alliance theorists (Lévi-Strauss, 1960, p. 54; see Murphy, 1963, in this connection).

II. Lévi-Strauss, Needham, and George Homans are all psychological reductionists - each in his own way, of course.

Lévi-Strauss is quite emphatic that ‘sentiments’ and ‘emotion’ explain very little, if anything. His discussion of ‘anxiety’ as an explanation (either in Malinowski’s or Freud’s terms) is, however, the only discussion provided in any detail in his book Totemism. Here he rests his case on Radcliffe-Brown’s ‘turn about’ argument, namely, that it is not because people feel anxiety that they make magic, but that they feel anxiety because they make magic. Further, he adds an apparent empirical exception to Malinowski’s supposed ‘rule’, saying, ‘The empirical relationship postulated by Malinowski is thus not verified’ (Lévi-Strauss, 1963, p. 67). He bases this on the case of the Ngindo beekeepers, but he does not cite Kroeber’s earlier example of the Eskimo (Kroeber, 1948, pp. 603-604).

But for the rest, his position against sentiment, emotion, or anxiety remains an assertion.

‘Contrary to what Freud maintained, social constraints, whether positive or negative, cannot be explained, either in their origin or in their persistence, as the effects of impulses or emotions which appear again and again, with the same characteristics and during the course of centuries and millennia, in different individuals. For if the recurrence of the sentiments explained the persistence of customs, the origin of the customs ought to coincide with the origin of the appearance of the sentiments, and Freud’s thesis would be unchanged even if the parricidal impulse corresponded to a typical situation instead of to a historical event. [29]

We do not know, and never shall know, anything about the first origin of beliefs and customs the roots of which plunge into a distant past but, as far as the present is concerned, it is certain that social behavior is not produced spontaneously by each individual, under the influence of emotions of the moment. Men do not act, as members of a group, in accordance with what each feels as an individual; each man feels as a function of the way in which he is permitted or obliged to act’ (Lévi-Strauss, 1963, pp. 69-70).

Lévi-Strauss sharply dissociates himself from Durkheim’s position.

‘… in the last analysis Durkheim derives social phenomena as well from affectivity. His theory of totemism starts with an urge, and ends with a recourse to sentiment’ (ibid., pp. 70-71). ‘Actually, impulses and emotions explain nothing: they are always results, either of the power of the body or of the impotence of the mind. In both cases they are consequences, never causes. The latter can be sought only in the organism, which is the exclusive concern of biology, or in the intellect, which is the sole way offered to psychology, and to anthropology as well’ (Lévi-Strauss, 1963. pp. 70-71).

And so the intellect, cognition, how people think and not their emotions are the explanatory conditions for social phenomena. But notice how close is the pattern to that which is explicitly repudiated when phrased in terms of sentiment.

‘The advent of culture thus coincides with the birth of the intellect. Furthermore, the opposition between the continuous and the discontinuous, which seems irreducible on the biological plane because it is expressed by the seriality of individuals within the species, and in the heterogeniety of the species among each other, is surmounted in culture, which is based on the aptitude of man to perfect himself …’ (ibid., p. 100).

Origins, in other words, are not recoverable if we try to locate their emotional beginnings, but they are recoverable if we consider them to be in the intellect. This, it would seem, is because [30] the intellect is the source, the beginning and the continuing source.

‘… Bergson and Rousseau … have succeeded in getting right to the psychological foundations of exotic institutions … by a process of internalization, i.e., by trying on themselves modes of thought taken from elsewhere or simply imagined. They thus demonstrate that every human mind is a locus of virtual experience where what goes on in the minds of men, however remote they may be, can be investigated’ (ibid., p. 103).

The mathematical formula expressing the shape of the cams and their speed of rotation now assumes the form of the human intellect. It is the inventiveness of the human intellect, the rules of operation of the human mind, the fundamental qualities of mind which are the explanations.

‘It is certainly the case that one consequence of modern structuralism (not, however, clearly enunciated) ought to be to rescue associational psychology from the discredit into which it has fallen. Associationism had the great merit of sketching the contours of this elementary logic, which is like the least common denominator of all thought, and its only failure was not to recognize that it was an original logic, a direct expression of the structure of the mind (and behind the mind, probably, of the brain), and not an inert product of the action of the environment on an amorphous consciousness.’ (ibid., p. 90).

Except for minor embellishments, Lévi-Strauss has returned to the position at the turn of this century which Durkheim, Lévy-Bruhl, Boas, and Kroeber held. This is simply the position that culture, custom, belief, or social constraints are products of the human mind, understandable only when the laws of the human mind are known.

It is disconcerting, therefore, to follow the discussion of Totemism (Lévi-Strauss, 1963), which seems to review the history of our understanding of that phenomenon. It moves with care from the illusion of an entity shattered by Boas and Goldenweiser’s cold analysis, on to Elkin’s fumbling with too many broken pieces, and then on through to functionalism’s confusions until it reaches the climax in Chapter 4, ‘Toward the Intellect’. [31]

Toward the intellect, away from emotion and sentiment, using structural linguistics as the vehicle to ride to the rescue of associational psychology! But one suddenly notices that the intellect toward which one is headed is in one case fresh from the eighteenth century, and in the other from the earliest past of the twentieth; that the confusions of functionalism are all historically later developments; that the devastation of the entity called totemism by Goldenweiser, following directions clearly stated by Boas, comes after the very time and the very position to which Lévi-Strauss has returned by such a circuitous route. It was Boas, of course, who first wrote The mind of primitive man, in 1911, and Kroeber who insisted that kinship terms could be understood only as linguistic and psychological phenomena in 1917, just as it is Lévi-Strauss in 1962 who says that totemism is understandable as a product of the human mind.

One can only suppose that the separation of mind into ‘intellect’ and ‘emotion’ is the end-product of the same series of events which have led Homans up the Skinnerian path. Skinnerian psychology is not really too far from structural linguistics and Lévi-Strauss’s ‘intellect’ (Homans, 1962).

Homans and Lévi-Strauss have converged on the same intellectual position, each from his own very different starting-point, each backing away from the other with more than passing vigor, each denying any validity to the other’s position. Homans along a path of positivism, Lévi-Strauss following the course of intellectualism, both as psychological reductionists.

III. In certain respects three very old ideas form the core of all cross-cousin marriage theories: the first is the idea that cross-cousin marriage is somehow an integrative device. The second, that different cross-cousin marriage rules with different rules of descent form different kinds of units, so that, for instance, bilateral cross-cousin marriage yields a very different kind of arrangement from unilateral forms. The implications of the different forms of unilateral rule for the kind of integration which ensues is the third old idea, and this one dates at least back to Fortune (1933).

Cross-cousin marriage theories differ in significant details, on the ‘meaning’ of these details, and on ‘how the system really [32] works’ (Needham, 1957, p. 168). The derivation of much of the analysis from Rivers (Rivers, 1914a and b), and particularly the terminological correlates of the different forms of cross-cousin marriage, should be noted.

What Lévi-Strauss, Dumont, Leach, and Needham have done is to set these fairly well-established points in the context of a general theory of society, and to develop from that combination certain specific and specifically useful (as well as debatable) ideas which were not generally available before.

The first such specific outcome to be taken up here is that the relationship between intermarrying units in a society practicing cross-cousin marriage is one of ‘alliance’. And the problematic aspect of this is the question of just what is meant by ‘alliance’.

There is a series of closely interrelated ideas at work here. First, of course, there is the notion of organic solidarity, that is, the interrelation of differentiated parts no one of which can by itself be autonomous. Second, there is the notion that every social definition must in the nature of human mentality and in the nature of society be stipulated in terms of opposition, complementary dualism, and so forth. Third, there is the notion that structure inheres not in the concrete constitution of any particular society, but rather in the ‘model’ or the construct which must be developed by the anthropologist in order to understand that society; yet the structural principles which are ‘at work’, so to speak (my phrase, my phrasing), are somehow ‘real’, ‘existent’, ‘substantive’ and are expressed in the social definitions, the social conventions, the social rules of a particular society (Lévi-Strauss, 1953, 1960).

Briefly, there are two enduring structural features of any society practicing cross-cousin marriage, or having any ‘positive marriage rule’ (sister exchange, asymmetrical marriage rules, etc.). First, the rule of descent which aligns (not allies) a group into one unit; second, that of ‘alliance’, the relationship between two or more such units of which marriage is an expression.

Alliance theory insists that in the nature of social definitions, arising as they do from the inherent features of human mentality, ‘consanguinity’ can be defined only in opposition to ‘affinity’, and ‘affinity’ in opposition to ‘consanguinity’. Hence it is incon- [33] ceivable that a society with a positive marriage rule could consist in one consanguineally related group of social persons. Radcliffe-Brown (1953, p. 169), in criticizing Dumont’s treatment of Dravidian kinship terms, brings this issue out clearly. Radcliffe-Brown asserts that for the Kariera, ‘the terms “nuba”, “kaga” and “toa”, which are applied to large numbers of persons, are not terms for relatives by marriage’. In short, he denies that affinity, as a principle, is applicable to this society, and by implication that a society can consist only in statuses socially defined by consanguinity

So far as cross-cousin marriage systems are concerned, this problem of the necessary existence of a principle of alliance opposed to one of descent brings into question both the interpretation of the terminological system and the nature of the integration of such systems.

The problem of integration focuses on the implications of the idea of general exchange as proposed by Mauss (1954). Specifically, Lévi-Strauss follows Mauss in suggesting that if marriage is one mode of exchange, and if exchange between differentiated parts is seen as the basic mode of integration, not one exchange, nor one kind of exchange, but rather a series of exchanges of various kinds will all occur, and will all reinforce and reiterate the integration between exchanging groups. Hence the exchange of bridewealth, of goods and services, and so on are all understandable as part of the total exchange system and are all equally fundamental expressions of the structural principle of alliance. All expressions of alliance, be they pigs, gongs, food and services, or women, are equally expressions of this structural principle. In this sense, each expresses the principle.

Yet it is difficult to distinguish in the writings of alliance theorists between marriage as an expression of alliance, and marriage as creating alliance. It is one thing to say that marriage is one among many expressions of the structural principle of alliance; it is quite another to say that marriage creates alliance. To go a step further and say that both statements are true is to move into yet a third and not at all compatible position.

IV. In an asymmetrical marriage system, a prescriptive system with a positive marriage rule, the system is essentially triadic. [34] Nevertheless, the symbolic system is dyadic. This is shown by both Lévi-Strauss and Needham, and Leach seems to concur (Leach, 1961). The reason for this is that from the point of view of any given lineage or ego there are only two relations: wife-giver or wife-taker.

Needham may help to explicate matters. Near the conclusion of his reiterated exposition of Purum he says:

‘We see here, as elsewhere with prescriptive alliance, a mode of classification by which things, individuals, groups, qualities, values, spatial notions, and other ideas of the most disparate kinds are identically ordered within one system of relations. In particular, I would draw attention to the remarkable concordance and interconnection of social and symbolic structure [emphasis supplied]. In spite of the fact that structurally there must be three cyclically-related lines in the alliance system, the basic scheme of Purum society is not triadic but dyadic. Any given alliance group is wife-taker and therefore inferior to another, but it is also wife-giver and therefore superior to another group in a different context. That is, alliance status is not absolute but relative. The distinction to be appreciated is that between the triadic system and its component dyadic relation. It is through this mode of relation that the social order concords with the symbolic order’ (Needham, 1962a, pp. 95-96).

But a closer inspection reveals that both triadic systems and dyadic systems concord with dyadic symbolic systems. Needham says,

‘… the Aimol scheme of symbolic classification, which accompanies a two-section (symmetric) system, exhibits the same principles and even embodies some of the same pairs of terms as the Purum scheme, which accompanies an asymmetric system … our task is to see how a common mode of symbolic classification can cohere with these different social forms.

‘The key to this question is that in spite of the necessarily triadic structure of an asymmetric system its fundamental relation is dyadic…, and that the fundamental relation of Aimol society, though differently defined, is also dyadic. Here lies the basic similarity between the two social systems, and the structural feature with which the dualistic symbolic [35] classification is in both cases concordant. We may now regard these two systems as different but related means of ordering social relations as well as other phenomena by the same (dualistic) mode of thought. if this is the ease, it is this mode of thought itself which demands our attention if we are to understand them in any radical fashion’ (Needham, 1960d, pp. 100-101).

One must distinguish, then, between dyadic and triadic systems the first a two-section system, and one of symmetrical cross-cousin marriage, and the other an asymmetric, perhaps matrilateral, cross-cousin marriage system. These, in turn, are distinct from the symbolic system, that is, ‘…a mode of classification by which things, individuals, groups, qualities, values, spatial notions, and other ideas of the most disperate kinds are identically ordered within one system of relations’ (Needham, 1962, p. 95). The symbolic system ‘…concords with…’ the social system.

But just what does this concordance mean? Needham explains it in this way: ‘…what one is really dealing with in such a society as this is a classification, a system of categories, which orders both social life and the cosmos. That is, Purum social organization is ideologically a part of a cosmological conceptual order and is governed identically by its ruling ideas’ (Needham, 1962, p. 96).

In other words, ‘the social order’ is itself a symbolic order and is on no account to be confused with actual living persons, concrete groups, actual numbers, or actual relations between actual persons or groups.

V. The descent theory which follows from Radcliffe-Brown deals with actual living persons, concrete groups, actual relations between actual persons or groups. Consistent with such a view, Livingstone (1959) attacked Needham’s Purum analysis precisely on the ground that the system, if it worked at all, could not work very well, since some of the units in the exchange system have too many women to dispose of, and some not enough. Livingstone’s point is that, at best, the rules are difficult for people to follow; at worst, the rules are unworkable in real life.

Needham’s reply stresses, as one might well expect, that the structure of the system does not lie in the concrete constitution [36] of any particular exchanging group, nor in any particular cycle of exchanges. ‘Particular alliances and alliance groups may be expected to change, but the essential point is that the rules (matrilateral prescription, patrilateral prohibition) have not changed’ (Needham, 1960h, p. 499). Needham continues:

‘A social system is an abstraction relating (in this context) to lineal descent groups which are also abstractions. There is no place in an abstraction for substantive “specific groups”. It would be an odd and profitless use of the notion of social system to so identify it with substantive reality that every change … would be said to constitute a “breakdown”’ (p. 499). ‘I do not claim, though, that it is utterly unknown for a marriage to take place within the clan in an asymmetric system …’ (p. 499-500). ‘That the definition of alliance groups can be changed, and that alliances can be reversed, are well known facts in the literature on such systems as the Purum’ (p. 501). ‘When I used the word “stable”, on the other hand, I meant precisely (as I wrote) that in spite of demographic changes and fission of segments the rules of the alliance system continue to maintain the same type of social order between whatever groups are pragmatically distinguished’ (p. 502).

Needham’s reply to Livingstone clarifies this matter. But to my knowledge, no one has published a reply to Leach’s paper, ‘Structural Implications …’ (Leach, 1962). One of the points which Leach makes in this paper is that it is not ‘social segments’ conceived of as diagram lines which actually undertake marriage-in-a-circle; it is the social segment which Leach calls ‘local descent lines’, concrete units ‘on the ground’, which undertakes such marriages. For Needham’s version of alliance theory, Leach’s distinction, however true, is quite irrelevant. It is the social order as a conceptual order, the social system as a system of relational rules, which is the concern of alliance theory. Alliance theory is precisely and explicitly diagram lines as models of a social order for Needham.

Not so for Lévi-Strauss (or Dumont), however. Maybury-Lewis (1960) takes Lévi-Strauss sternly to task for this same error. Indeed, one may say that the criticism is the Leach Criticism, and consists in accusing Lévi-Strauss of confusing models with [37] reality, of mixing phenomena from disparate levels of abstraction in the same analytic image. And Lévi-Strauss’s reply to Maybury-Lewis (1960) should apply to Leach’s paper as well: ‘…I cannot be reproached for confusing dual organization, dualist system and dualism. As a matter of fact, I was avowedly trying to override these classical distinctions, with the aim of finding out if they could be dealt with as open — and to some extent conflicting — expressions of a reality, to be looked for at a deeper level’ (Lévi-Strauss, 1960, p. 46. Emphasis supplied). Or again, in the same vein, later in the same paper: ‘Hence the conclusion that “actual social segments” and “symbolic representations” may not be as heterogeneous as it seems, but that, to some extent, they may correspond to permutable codes. Here lies one’s right to deal with social segments and symbolic representations as parts of an underlying system endowed with a better explanatory value, although — or rather, because — empirical observation will never apprehend it as such’ (Lévi-Strauss, 1960, p. 54).

This should also help to clarify Lévi-Strauss’s view of the nature of ‘reductionism’. Where, for some, the reduction of one order of phenomenon to another constitutes a breach of analytic rules, it constitutes a proper form of explanation for Lévi-Strauss, Needham, and Dumont at least.

VI. Alliance theory continues a phrase closely associated with Durkheim, but alters its meaning in some respects. Durkheim saw and described ‘the social order’ as a system of norms and rules and regulations, values and ethical elements, and so he described it as a moral system, because it constrained people’s action. The phrase ‘moral system’ and the notion of morality are notably missing from Lévi-Strauss, Dumont, Needham, and Leach.

In Durkheim’s treatment of the problem of suicide the intimate, close connection between norms, values, constructs, and actual behavior is the very essence of the analysis. Perhaps it had to be, since he had forced himself by the skill of his intellect to deal with a problem of rate, which he was among the very first to distinguish clearly from that of incidence. The actor’s commitment to the norms and not the constitution of the norm itself, decided his fate. [38]

The problem of social order for Durkheim was the question of explaining how constraints actually operated to define, guide, and channel the behavior — the actual behavior — of people: real, living people. Social order is the order that is imposed on a person’s action (Parsons, 1949).

But ‘social order’ for Needham is the congruence of symbolic systems seen in logical terms; it has almost nothing to do with ‘groups pragmatically distinguished’. The word pragmatic can only make sense if it is taken to mean ‘conceptually’, and this in turn can mean that some native will talk about it. A figment of a figment of a figment of a native’s imagination will do — and indeed this is precisely the susbstantive element in the social order. Needham does not ever raise or deal with the problem of social order in this Durkheimian sense, which is the problem which Livingstone put to him: the symbolic system cannot order the actual marriages of the Purum very well. Needham replied that the social order that he is discussing is not to be located in the ‘rate’, but instead occurs at the level of the symbols, constructs, ideas, categories, rules for relating these, and so on (Needham, 1960, p. 499).

VII. Is it at this point that Lévi-Strauss’s separation of mind into intellect, on the one hand, and emotions and sentiments, on the other, becomes a less arbitrary action than it first seemed? It is not at all clear from his assertions just why this separation is necessary. It is conceivable that if Lévi-Strauss is concerned with the social order as a symbolic order, with the organization and configuration of that symbolic order, with the relationships between sub-systems of symbols, that the enormous weight, and almost indisputable weight, of Freud’s work on symbolism would serve him well. But instead Lévi-Strauss tells us that, unlike Kroeber’s, his own attitude toward Totem and Taboo has hardened over the years (Lévi-Strauss, 1963, p. 70).

It is odd, too, that the non-logical and irrational character of the logic of the unconscious as Freud describes it should fail to appeal to Lévi-Strauss.

But the other side remains clear. Durkheim was concerned with the actor’s commitment; Lévi-Strauss is concerned with what the actor may be committed to. Durkheim was concerned [39] with the integration of the objects of the actor’s commitment in the actor, as the actor saw it and was impelled to act on it; Lévi-Strauss is concerned with the integration of the system of objects of commitment, and how they integrate one with another. Durkheim needed something like an ‘urge’ and a ‘sentiment’ to explain how actors got and remained committed to objects; Lévi-Strauss needs something like intellect to explain the logic of the order among symbols. And, of course, how can there be, either in nature or in theory, intellect without emotion or emotion without intellect? Perhaps this explains why Freud haunts the pages of Lévi-Strauss, from the Structures élémentaires through the recent Totemisme.

The social order is a symbolic order, an order of ideas and categories, concepts and rules all ‘ordered’ by their logical relation. What imposes this order? How are the different levels of order related? Needham brings it all together in a single, felicitous paragraph which may well be the brief outline for a series of detailed volumes:

‘The present analysis [Aimol], together with that of Purum society, permits the proposition that both of the major types of prescriptive alliance feature basically a dyadic social relationship, and that with this there concords in these cases a dualistic scheme of symbolic classification [footnote omitted]. Furthermore, wherever the ethnography is adequate to the purposes of this type of investigation, it can be seen that in societies practising prescriptive alliance it is a dualistic symbolic classification which orders both the social and the symbolic system, characterizing both by the same mode of thought. This type of classification is the analogical elaboration, in all spheres of social concern, of a structural principle of complementary dualism; and this itself is one manifestation of the logical principle of opposition. It is in this sense that societies founded on prescriptive alliance may be interpreted as the most direct forms known of the social expression of a fundamental feature of the human mind’ (Needham, 1960d, p. 106).

In the same place, Needham stipulates some of his differences with Lévi-Strauss: [40]

‘In trying to pin down Lévi-Strauss’ idea of what is fundamentally important in the analysis of the sort of system we are examining two conclusions seem me [sic] to emerge: (1) in the matter of features of the human mind he proposes a number of formulations, each rather different from the others, which he declines to analyse further on the ground that they are psychological data; (2) for him the radically and essentially important idea is that of reciprocity. Admittedly, he is concerned with marriage regulation as means of exchanging women, a context in which reciprocity has an especially important place, but even here I think he over-rates its importance; for he disregards the conceptual systems of the societies within which women are exchanged, systems which are solidary with the exchange systems and which cannot be explained by the notion of reciprocity. Durkheim and Mauss indeed tried to derive concepts from social organization … but Lévi-Strauss would surely not do so in this case, since the notions he isolates are far more general than the forms of affinal alliance he examines’ (Needham, 1960d, p. 102).

VIII. A rule is certainly not the same thing as its observance. The rule that good little children go to heaven while those who are naughty go elsewhere must certainly be distinguished from an actual census conducted in Heaven, although the latter might possibly provide a statistical model (Lévi-Strauss, 1953). And even if it could be shown that not a single good child actually got to heaven, it would hardly alter the fact that this remains a rule, at least in some circles.

How can such a rule be ‘expressed by’ or ‘contained in’ some terminological pattern about children, goodness, naughtiness, heaven, or hell? This is perhaps difficult. But the matter is quite different with kinship terms. Here are symbols which, in some aspect of their character, can ‘embody’ or ‘express’ marriage rules. So Dumont (1953, 1957) and Yalman (1962) affirm for Dravidian kin terms, though Epling (1961) denies this. (We will come to Radcliffe-Brown’s objections shortly.)

This brings us around again to the question raised earlier, is marriage an expression of alliance, or does it create alliance as well as express it? [41]

This problem can be seen best in the matter of the radical distinction between prescriptive and preferential systems which Needham has drawn.

It seems doubtful that Lévi-Strauss and Dumont follow this distinction, but we have Leach’s statement which implies that he does (Leach, 1961).

If a marriage rule is to be treated as an expression of a structural principle, and if a marriage rule such as MoBrDa is taken as one expression of that principle, then it would seem to follow that a preferential rule (as distinct from the prescriptive rule) is equally clearly an expression of that same structural principle.

One might say that the preferential rule is modified, as an expression of the principle (as in Lévi-Strauss, 1960) but that in effect the ‘type of social-order’ (Needham, 1960h, p. 502) is unmodified. If the marriage rule is an expression of the principle of alliance, is it any less so if the rule is not, for some set of good reasons, required? The principle is still there. It is still expressed in the marriage rule. Just as the concrete number of actual marriages which follow the rule has no bearing on the structural analysis, so too it might seem that the rule expressed as an obligation or the rule expressed as a preference is equally clearly an expression of one and the same structural principle.

The marriage rule, like the terminological pattern and the goods which are exchanged, all reflect the principle of alliance Hence it does not seem clear why Needham should insist that for any given society there must simultaneously be certain forms of expression, and that each should be perfectly internally consistent as a form of expression of the structural principle. Needham insists that the terminological pattern must be consistent, or that the ethnographer has made a mistake (Purum); that the marriage rule must be prescriptive; he further finds it very hard to accept the fact that certain marriages which are wrong marriages (according to the rule) occur, though he contents himself that they are very few in number. (Dumont seems to differ on this point, if his genealogical examples in Hierarchy and Marriage Alliance are to be taken as an expression of his own views in this matter.)

Let me restate this in another way. Needham makes the radical distinction between prescriptive systems and preferential [42] systems. He goes to considerable trouble to assert which of thirty-three societies are prescriptive and which preferential (i.e. Needham, 1962). He then proceeds to deal with the prescriptive systems alone. He has not dealt with preferential systems, neither has he stated in understandable terms why preferential systems cannot be treated as having the same structure as prescriptive systems. It is by no means self-evident that they are different from the point of view of alliance theory. Indeed, I suggest that it is an induction from alliance theory that they are not different at all and that Needham fails to understand his own theory when he says that they are.

Part two. Segments

IX. The suggestion that Needham does not understand the implications of the theory he is following is perhaps premature at this point. It is true that Needham does not discuss at length the distinction between prescriptive and preferential systems. Neither does he explain why he insists on the radical distinction which he makes. It is certainly true that he might have analyzed a society like the Yaroro of Venezuela to show ‘… how such a system really works’ (Needham, 1957, 1958c, p. 323), for he obviously considers the material on that society valid and reliable, although he describes the source as being ‘unimpressive’, a word he does not clarify (Needham, 1962, p. 62). It is easy to see that the analysis of Purum as a prescriptive matrilateral system would be immeasurably strengthened by the analysis of a preferential matrilateral system to show, by contrast and example, precisely where the differences between such systems do lie and just what is at issue here.

What is at issue here is the idea that is best expressed by the oft-repeated phrase ‘… the same type of social order…’ (Needham, 1960, p. 502), or ‘a certain type of society’ (Dumont, 1961, for instance, where the phrase is also found as ‘… different types of societies’). In its simplest form, this means that one type of society has ‘a positive marriage rule’ and the other has not (Dumont, like Lévi-Strauss, sometimes says ‘… a “marriage preference”’ Dumont, 1961). The type without a positive marriage rule is one in which individual choice is exercised not with respect to aesthetic or purely personal considerations, but [43] one in which there is no limitation in principle on the social category (kinship category is a particular social category) from which the spouse is drawn, provided that the incest prohibition is observed (Lévi-Strauss, 1949).

This basic distinction, developed in this particular form by Lévi-Strauss in 1949, has a long history and a wide group of supporters of one or another of its forms, including Leach (1962), Goody (1959, 1961), Fortes (1959), and Freeman (1961).

The distinction follows from Durkheim’s notion of a ‘segmental society’. The image is that of a concrete society formed of homogeneous, repetitive parts. The parts are clans, and clans grow out of territorially based hordes (Durkheim, 1947, pp. 175 ff). The link between the homogeneous parts, in the limiting case which Durkheim posed, is nothing more than similarity, a kind of ‘birds-of-a-feather-flock-together’ principle. Only when the parts become functionally differentiated does a society begin to move away from, and fall on a point of the scale away from, that of mechanical solidarity and segmental organization (Durkheim’s evolutionary view is retained here).

But the problem which Durkheim did not solve satisfactorily remains. How are the segments defined and in what do they consist? How are they formed or maintained? Durkheim swung back and forth between a physically discrete aggregate or group of real persons, and the ideas (‘collective representations’) in terms of which such a physically discrete category might be formed. As I have already suggested, Radcliffe-Brown wrote as if he had the physical segment in mind (‘actual social relations’), Lévi-Strauss, Dumont, and Needham confine themselves to the rules which flow from the structural principles.

But in either case, some kind of ‘rules’ yields this particular ‘kind’ of society. Any other kind of ‘rules’ should not produce such a society.

The contrast between unilineal descent and bilateral systems seems to have been one of the most enduring models for this problem. (See Leach, 1962b, p. 132, quoted in Section XVII below, for a clear statement of this position.) With any of the unilineal descent rules (matrilineal, patrilineal, double unilineal), discrete, exclusive, multifunctional units can be formed within a single society when the unilineal units are exogamic. Where there [44] is no unilineal rule, such units cannot form within the endogamic boundary. Any endogamic unit can be discrete, exclusive, multifunctional as well and, given that it has certain other minimal characteristics, can form a self-sufficient entity within a society. Thus a caste within a society can be discrete, exclusive, multifunctional. The endogamic-exogamic boundaries must be stated in this problem.

It is, therefore, the exclusive, discrete unit which can be multifunctional. A unit which is discrete and exclusive and multifunctional, which is a self-contained, self-consistent, and self- perpetuating unit is, of course, a society. If it lacks one function necessary to maintain itself as such a unit it can nevertheless be a major social segment in a ‘segmental’ system, and, of course, one could conceivably scale societies according to the degree to which they are segmental.

We can now return to the problem of the conditions which can yield such segments, and of how within-segment and between-segment relations are defined. And this is the problem about which ideas and phrases such as ‘a certain type of society’ center; it is the problem between Fortes and Goody on the one hand, and Lévi-Strauss, Leach, and Dumont on the other.

For alliance theory, and for Lévi-Strauss who developed this variant of the notion, ‘… a certain type of society’ means a society which is very close to or approaches the model of the segmental society first stated by Durkheim. The limiting case of mechanical solidarity occurring entirely in terms of similarity is dropped as a model. Instead the next simplest form of organic solidarity — exogamy with a marriage rule — is used. Hence ‘… a certain type of society’ has a positive marriage rule, not a state of free choice, which is its opposite.

But this is the crux of the difference with Fortes and Goody. They do not concern themselves with the problem of whether the simplest form of organic solidarity is formed by a marriage rule and exogamy. They are concerned with segmental societies because they have a special theory of their own about how segments are formed and in what they consist, and how within-segment relations are ordered and how between-segment relations are structured. Marriage enters into this theory, but in a very different way. [45]

X. on one point both alliance theory and descent theory are in agreement. If a segment is to remain discrete, conceptually or concretely, it cannot have overlapping membership with any other segment. It can have relationships, but not overlapping membership.

Freeman (1961, pp. 200, 202) and Goodenough (1955), to cite but two writers on this point, maintain that the kindred cannot form a ‘group’, or corporate group, but must remain a category. Freeman makes the very useful distinction between the kindred and a kindred-based group, for portions of a kindred can come together for some specific purpose, for some limited time and thus form an ad hoc group.

The point remains the same. The kindred itself cannot be a group because it is defined by reference to a specific ego, because the category has overlapping interlocking membership, because at best it can become a discrete physical unit and remain so only for a limited time, by permitting some members to drop out (the ‘optive’ condition in Freeman, 1961, pp. 209-211) and rallying the rest for a stipulated short-run goal.

The whole man has to be in one and only one group, so that the group can be a physical as well as a conceptual group. But why is it necessary to be able to separate physically all of the groups?

I think that the reason for this, one to which both alliance theory and descent theory subscribe, is not that segments have to be spatially and physically separate and identifiable as such, but that the whole person as an aggregate of different commitments must be able to provide unqualified solidarity with the unit to which he belongs. It is this which permits the segment under certain conditions to be physically identifiable as such. But if a single person’s solidarity is qualified by membership in two or more different units of like order, then his commitment to, his solidarity with, one of them is qualified by the claims of the other upon him. Conflicting claims on two different persons are easily soluble; one goes one way, the other his own way. But conflicting claims on one person have to be adjudicated; one wins, the other wins, or both are qualified and compromised in some way.

If a system of intersecting claims on persons requires the adjudication of those claims, then this in itself becomes a mode of integration in such situations and also in such societies. This [46] has been repeatedly shown for feuds and so forth. It should not require explicit demonstration for the social system as a whole.

The whole person is thus of significance to both descent and alliance theory, and this notion constitutes, I suggest, an unanalyzed and unspecified condition of the model which alliance theory has produced. It is the condition of unqualified commitment which is requisite to the constitution of the segment — conceptually for alliance theory, concretely for descent theory — and it is this condition which radically reduces the kinds of segmental system to the few which have been pointed out. These are always based on kinship, because it is only descent rules — more precisely, unilineal descent rules or rigid endogamy — which dispose the whole man to one or another segment of society.

(At this point the argument could proceed along one of two different lines. One line would be to turn to the problem of choice, for Rivers, Fortes, Leach, Lévi-Strauss, and Needham are all quite explicit that it is only under conditions of unilineal descent, where choice is excluded, that the kinds of segment with which they are concerned can be formed. Rather than continue with that problem here, I have delayed its consideration to Section XVII and proceeded to discuss the descent theory views of the nature and constitution of segments.)

XI. A segment is a discrete conceptual or concrete entity in a segmental society. It is the descent rule which allocates whole persons to one or another segment. It is the descent rule which is the mode of recruitment that replaces the dead by the living. Because a discrete, exclusive unit is formed, this unit can be a perpetual unit (conceptually though not concretely; its members die and are replaced by new members).

A unit is one thing; a corporate group quite another. A descent rule, rigidly followed, only creates a conceptually distinct category. Its organization as a corporate unit depends upon other factors.

As a minimal definition, a corporate unit is one which is treated as internally undifferentiated by the other unit or units with which it has a specific relationship. Such units may be a person in a particular role, a group acting as a whole, a group represented by a person, and so forth.4 [47]

A corporate unit has a relationship with some other unit. The other’s treatment of it is consistent with its conception of that unit. If the unit is conceived of by the other as having an undifferentiated, unitary identity, then for that purpose, or for that particular relationship, it is corporate.

But it may be argued that to be corporate for some purpose or other equally means that the unit must also be able to act in a corporate fashion. Not only must it be conceived of as corporate but it must be able to act as a corporate unit.

Both of these conditions are met when the members of the unit see their relationship to each other as being ‘the same’, and when that same relationship is seen as ‘the same’ for all outsiders. Thus a unit may be treated as consisting entirely of agnates, and the members may define their relationship to each other not as father-son, brother-brother, etc, but rather as one of agnates.

The degree to which members have solidary bonds with each other, then, may be held as a condition or at least as a related condition, to the unit’s capacity to act as a unit and be treated as a unit.

Following Durkheim, it is presumed that solidarity is increased when different kinds of bond replicate each other. Hence the group which raises its food together, processes it, distributes it, consumes it; the group that is interdependent for its labor; the group which worships its own ancestors and thus is the same ‘church’; the group which owns its property as a unit — such a group, with repetitive bonds, each reinforcing the other, is more strongly solidary than is the group which merely shares a single function — say, its name or its emblem. And the degree to which it is solidary is directly proportional to the number of different solidary bonds among its members, and this is inversely proportional to the strength of any particular member’s bonds outside that unit So that if the agnates within a descent unit all are tied with repetitive bonds, and none of them has any very strong bonds with persons outside the group, a strongly corporate unit results.

The multifunctional character of a social segment, then, is one of the conditions which helps to maintain a segment’s segmental character. A unifunctional unit can hardly constitute a social segment. [48]

The segmental character of a society, therefore, depends on the corporate character of its segments, and the corporate character of its segments depends on the strength of the bonds among the members of the segment. The strength of the bonds among the members is in turn dependent in part upon having no compromising allegiance for some members of a group outside the group, but is in part also dependent on the kind of functions which the members undertake in common, and it is from this that the intensity of their commitment, the strength of the bonds among them, the degree of the corporateness of the group derives.

What kind of function so unites a group? If a unilineal descent group holds property rights in love magic, does this so unite a group? If it controls a dance or a form of poetry? If it has the right to recite certain prayers? Hardly. If a unilineal descent unit owns and controls property, productive property, real estate or cattle, then and only then does its status as a true segment emerge.

It is property, productive property, an estate, which makes a. segment a segment for Goody (1961), and it is at least the jural and political functions for Fortes, if I read him correctly (1959, 1962).

And so one sees the descent group defined by certain of its functions for Goody and Fortes. A descent group is a property-holding group where the property is productive property; it is (along with its descent rule) thereby corporate, a jural and political unit

XII. Fortes and Leach agree that the term ‘descent’ should be confined to unilineal descent and continually invoke Rivers’s name in ways that suggest that in this context, at least, it is sacred (Fortes, 1959; Leach, 1962). Freeman (1961) performs feats of eloquence with the ethnography of All Souls and Wales to help maintain Rivers’s division between descent as unilineal and a personal kindred as purely ego-centered. This unexpected concurrence arises, I suggest, from the fact that all three are committed to a particular model for a type of society, a segmental type of society. Note that this is not merely a descent rule, nor a combination of types of descent rules, nor a classification of descent systems. It is a type of society as a whole that is at stake.

Both descent theory and alliance theory have a vested interest, [49] so to speak, in particular definitions of segments which go to make up somewhat different theoretical models of segmental societies. Where alliance theory defines its model of a segment strictly in terms of a unilineal descent rule and exogamy, descent theory defines its segment in terms of corporate character. The corporate character of the segment is seen as a consequence of a unilineal descent rule, exogamy, and a high degree of solidarity related to the fact that it is the holder of productive property and so also is a jural and political unit. If alliance theory depends on the image of a descent rule plus marriage rule, descent theory depends on the image of a descent rule plus the jural and political functions which are connected with property ownership.

The last piece in the model of the segmental society, then, is the relationship between segments. And here is where so much of the smoke seems to come from. For alliance theory, the segments are related to each other by alliance, and alliance is either created by or expressed by (or both) marriage. For descent theory, the segments can be related to each other in a variety of different ways, among them, the bonds of complementary filiation and levels of segmentation. Some alliance theorists have complained that complementary filiation is just another name for descent and that descent theorists do not recognize marriage as a structural principle.

XIII. ‘It is the whole nature of the concept of “descent” which is at issue’ (Leach, 1961, p. 121).

Fortes, Freeman, and Goody have all gone to great pains to confine the concept of ‘descent’ to unilineal rules, forming corporate groups which have jural, political, and/or property-holding functions. Fortes (1959, pp. 207-208; 1962) also suggests that descent is, in the first instance, a matter of title to member ship, citizenship, and that group membership, although of major importance, is not the only defining criterion.

Fortes agrees that affinity is important in any society but, according to Leach, he seems to deny that alliance constitutes a structural principle for certain of them. Fortes seems to say that for the Tallensi, the fundamental structural principle is that of descent, as well as the lineage principle which is so closely interwoven with descent; that alliance is not a structural principle for [50] certain of the African lineage systems. (NB: Fortes seems to say. He does not say it in so many words.) Leach is on the offensive, yet cautious. Leach has not to my knowledge boldly asserted that, for Tallensi, alliance is just as fundamental a structural principle as the lineage principle. Instead, he has picked away from sheltered ground; he asserts that ‘Fortes … disguises [affinity] under his expression “complementary filiation”…. For Fortes, marriage ties, as such, do not form part of the structural system’ (Leach, 1957a, p. 54).

But Leach makes a more positive point, in criticizing Fortes’s notion of ‘filiation’. His criticism, if I take it correctly, focuses on the fact that ‘filiation.’ consists in a kind of descent rule, a relationship of consanguinity, when in fact it would seem that that relationship is precisely affinal. Hence his statement that ‘It is the whole nature of the concept of “descent” which is at issue’ (Leach, 1962, p. 121).

One confusion arises because of the somewhat different ways in which alliance theory and descent theory treat kinship terms. For alliance theory, kinship terms can be analyzed and understood without direct reference to the constitution of descent groups. For descent theory, this is not possible.

This can be seen from Dumont’s treatment of Dravidian kinship terms. (That this is also part of the MoBr-SiSo problem will become evident.)

Dumont (1953) says that in Dravidian kinship terminology the MoBr is defined in opposition to the father as father’s brother-in-law, the FaSi as the Mo’s Si-in-Law, and the Mo as a consanguine.

Now, this looks like sheer nonsense in descent theory terms. If one takes mother alone for a moment, descent theorists treat her as father’s wife; given a patrilineal lineage, she cannot ever be completely incorporated into ego’s lineage since her lineage membership is gained through her father; hence her relationship to her husband, her children, and to the men of her husband’s lineage is simply as an affinal, an in-marrying woman. The whole understanding of the levirate, the sororate, marital stability, and bridewealth turn on this. Thus marriage is brittle for the Hopi precisely because a woman is never released by her natal matrilineal lineage, her ties to her lineage remain strong, and so too the man’s ties to his own natal matrilineal lineage are strong; hence the [51] strain which is put on the marital tie. Descent group strength, that is, the various kinds of descent group solidarity, are seen as inversely related to conjugal ties and as proportional to parent-child and sibling ties (see in this connection, Schneider, 1961).

Mother is a consanguine to ego for descent group theory, but her relationship is not the same as the relationship of co-members of the same descent group. The consanguineal tie of father to son in patriliny is not the same as mother to son, yet both are consanguineal ties. The father-son tie consists both in a tie of consanguinity and in common descent group membership where the tie of mother to son is that of consanguinity, is opposed to the common descent group membership tie of father-son, and is as an affinal to the son in his capacity as father’s descent group member.

Mother, therefore, has a kind of composite status; this status turns on two kinds of relationship, one of affinity with respect to ego’s descent group, if it is patrilineal, and one of consanguinity to ego. Hence she is terminologically ‘wife’ to her husband, ‘mother’ to her son.

As mother is affinal to ego’s descent group, father’s sister is a consanguineal and cannot ordinarily lose her descent group membership. Hence, under no conditions could she possibly be seen as an affinal to ego.

Conversely, as mother is a consanguineal relative for her son, and as her brother is a consanguineal relative to her, mother’s brother is, through that series of links, a consanguineal of ego’s. And this consanguineal line, since it is ‘complementary’ to the descent line, is the line of ‘complementary filiation’.

Let me clear out immediately the misunderstanding based on Dumont’s treatment of kinship terms apart from group structure or behavior.

Dumont’s position is not so startling as it may look. Dumont is emphatically concerned with kinship terminology, he is not dealing with descent groups as such. Further, kinship terminology has to do with social categories. In his reply to Radcliffe-Brown, Dumont (1953b) is explicit; he is not dealing with behavior, but with the social definition of categories. There is a difference between kinship terminology and the structure of descent groups or the relation between the prescribed forms of behavior to ways in which kin are categorized. Kinship terminology ‘… is a con- [52] stellation revolving around the Ego … what is here called kin has, of course, nothing to do with actual groups, being only an abstraction arising from the oppositions; this again centres in Ego…. The whole could be called “terminological kin” to avoid confusion, and opposed to “terminological affines”. This is only a framework which is used and shaped by each group according to its particular institutions’ (Dumont, 1953a, p. 37. See also Dumont’s (1961a) attack on Gough, where the same position is reiterated).

But this begs the issue. If, for Dumont, kinship terminology is not kinship behavior, and if it ‘has … nothing to do with actual groups’, and if it yields conclusions which seem contrary to those which follow if behavior and actual groups are to be considered, how are these differences to be explained or reconciled? Dumont does not face this issue. The problem is whether or not MoBr can be considered an affinal or a consanguineal relative, or both. The problem is whether FaSi is to be considered an affinal or consanguineal relative, or both. The problem is to explain the MoBr-SiSo relationship. The problem is whether these are terminologically affinal or consanguineal relatives. To insist, as Dumont does, that terminology is and must be an ego-centered system is only to raise the question of how the ego-centered system articulates with the system as a total system. one cannot take shelter behind the assertion that these are different. The very difference itself is problematic.

Leach in his essay Rethinking Anthropology (1962a) seems to be showing that the Trobriand father is symbolized as an affine and this may contradict what Dumont’s Dravidian analysis seems to show with reference to mother. Dumont may be able to show that the Trobriand father is terminologically consanguineal if taken from the point of view of his child as ego and that Leach’s symbolic designation of the father as an affine is stated from the point of view of the wife and her matrilineal lineage vis-à-vis the husband and his influence on the child. Dumont could have considered such a problem in replying to Radcliffe-Brown, but he chose instead to say that he was dealing neither with behavior nor with Australian aborigines, a reply which i find almost totally irrelevant to any significant issue whatever. [53]

XIV. What, then, is the relationship of MoBr-SiSo? Is this to be seen as one of affinity, expressing alliance, or as one of complementary filiation?

For Goody (1959), the problem is quite simple. He is able to dispose of Lévi-Strauss in a little over two pages, reduce Dumont to a pale imitation of Lévi-Strauss, dismiss Leach in a phrase or two, and proceed to deal with the MoBr-SiSo relationship ‘… in West Africa’ in terms of the holder-heir conflict. The snatching and ritual stealing of objects by the SiSo is seen as a residual claim which he gets from his mother’s membership in her natal patrilineal property-holding group. As a member of the group, she has rights in property. But as a woman she can exercise only a few of these rights, and her rights in inheritance are ‘residual claims’. ‘…the legitimate sister’s son has nevertheless a shadowy claim upon the group by virtue of his mother’s position’ (p. 82). Leach’s disquiet over such terms as ‘shadowy claim’, ‘submerged rights’ and ‘residual sibling’ is not entirely misplaced, in my view. His criticism that descent, in the hands of Fortes, Goody, Gluckman, Gough, becomes a sliding scale, the boundaries of which can never be clearly stipulated, is not without merit. The point is that if the rights of the sister’s son over bits of property held by the mother’s brother, including perhaps the mother’s brother’s wife, whom sister’s son may ‘inherit’, are rights which are based on the consanguineal tie through his mother, and rights which depend on descent group membership of his mother, then these rights are transmitted exactly as is descent group membership. These are rights, that is, that are based on descent. If this is so, the problem is whether matrilateral filiation is not in this sense a ‘descent rule’ and so all patrilineal descent systems with matrilateral filiation are by definition double unilineal descent systems. Goody (1961) takes up this problem specifically and tries to break the impasse by saying that time double descent should be defined as a situation in which both matrilineal and patrilineal groups have property-holding functions. That when one group is a property-holding group and the other is not, this should be regarded as ‘a unilineal system with complementary descent group’ (Goody, 1961).

The outcome is clear for Goody and for Fortes. Descent has to do with group membership. When membership in a corporate [54] group is defined by one or other parent’s affiliation, then we can speak of ‘unilineal’ descent (Fortes, 1959). Goody insists that the group must be a property-holding group, and that the property, to qualify as proper or true double descent, must be real property. Goody’s position seems to depend on the assumption that in the long run it is things like food and wealth and power which really count, and if he criticizes Malinowski for reducing extra-familial kinship to intra-familial usages, and Lévi-Strauss for reducing kinship to marriage, it might be said with some grain of truth that property and power are what Goody prefers to reduce things to. (Goody’s position seems to have been modified somewhat in his 1962 book.) But the issue is clear. For Fortes, descent means unilineal descent groups. All other systems are systems of filiation, and filiation is universally bilateral. Fortes returns to Rivers’s definition of descent as meaning unilineal rules and group membership.

Briefly, then, Goody, following Fortes and Radcliffe-Brown, sees the MoBr-SiSo relationship as one in which residual sibling’s rights are transmitted in attenuated form and, therefore, the right of the SiSo to snatch or steal property as a right based on filiation — ‘… a shadowy claim upon the group by virtue of his mother’s position’. Leach criticizes the notions of shadowy claim, residual siblings, variable group strength, as basically making the notion of descent meaningless. He, following Lévi-Strauss, sees the claim of the mother’s brother on the bridewealth of the sister’s daughter as part of a continuing series of affinal exchanges of which the first bridewealth payment for the sister was but an initial act in the exchange. He would see the SiSo rights in the bits and pieces of the MoBr property as the other side of that exchange between allied, affinal groups. This view certainly avoids the involved and uncertain judgements of how shadowy a shadowy claim can be, or how residual a residual sibling may be, or how submerged a submerged right may be. Hence, for him, MoBr-SiSo is an affinal relationship, not a consanguineal one at all. But if it is an affinal relationship one must pose the question, can there be an affinal relationship without alliance? Can there be an affinal relationship between SiSo and MoBr without the structural principle of alliance being ‘there’ as one equal, if not more significant, structural feature of the system? [55]

The problem with respect to bridewealth and marital stability is very much the same. There is no room here to spell it out in particular detail. Briefly, if marital stability depends, in Gluckman’s formulation, on variable descent group strength, then the principle of descent is so elastic and so difficult to stipulate as to be virtually useless — in Leach’s view. The distinction between consanguinity and descent which Fortes draws, reserving descent for ‘group membership’ and confining it only to unilineal groups, is such as to make much oceanic and Southeast Asian material impossible to deal with in the framework of that theory. Leach’s position seems to be the orthodox one we have encountered thus far. Bridewealth and marital stability depend on the particular expression of alliance which obtains in the particular society.

XV. We have come a long way from whether Needham does or does not understand the implications of the theory he is following. There is some reason to believe that he does not, but we are not yet in a position to decide. First we must ask what is the role which marriage plays in that theory. Does it merely symbolize alliance, as the exchange of pigs or gongs may do?

For both alliance and descent theory, it is a type of society which is at issue. The segments are constituted by a unilineal descent rule for descent theory as well as for alliance theory. For both alliance and descent theory, the segments are internally differentiated in terms of kinship relations. For alliance theory, the segment is maintained (given its definition in terms of a unilineal descent rule) by its systematic relationship in opposition to other like units and its systematic relationship of marriage with a particular category of kinsmen in the segment to which it is related. This is a scheme, it is a conceptual system which relies on the logical development of a series of dichotomous elements. it is the conceptual. definition of the segments and the conceptual stipulation of how they are related which are the nub, the core, of alliance theory.

There are, then, two closely connected issues which create more unnecessary confusion than there need be. one is the issue of how the segments are conceptually defined, and of their ‘formal’ relationship to each other. The conceptual definition has a ‘function’, but it is not the function of a ‘degree of corporateness’ [56] as a concrete state of affairs. its function consists in its logical implications at a conceptual level. The second issue is the issue of how marriage relates two segments. It. ‘relates’ two segments in that it is a kind of relationship, not an act undertaken by two passionate persons. As a relationship, it is defined in terms precisely opposite to those of descent, the principle in terms of which the solidarity of the descent unit is expressed. And it is one more thing: it is that kind of relationship which explicitly and systematically entails particular categories within each segment and at the same time the entire segment. The relationship between segments is one of exchange; the categories within the segments which represent the systematic nexuses of the linkages are MoBrDa-FaSiSo or perhaps simply cross-cousin. But it is this linkage between categories which are at once categories within the descent unit and anchors between descent units which constitutes ‘marriage’.

Marriage, in this type of segmental system, is not the same thing as it is in a system constituted by unilineal descent units where the same kinship category is not the ‘anchor point’ between descent units. Where a man marries a woman who is until then in no sense a kinswoman, one may perhaps by stretching the term, speak of marriage creating an alliance. if the man represents one descent unit, and the woman represents another, and these two units are not already in a set or formal relationship to each other, then that particular marriage may create a bond and reiterate it in another way by the offspring of that union. But the classification of kinsmen within and between the descent units is not such that any particular union merely gives particular value to a relationship which existed beforehand in the classification of the kinship categories and their relationship.

In such a system, the definition of marriage is one which exists only with reference to and depends logically on the definition of the descent units, which in turn is dependent on the classification of kinship categories and aggregations of categories (units). These elements all depend on each other for their logical definition. To take marriage apart from the classification of kinsmen, or the mode of descent apart from the relationship between the segments, or the ‘corporateness’ of the segments apart from the way in which they are related to each other is simply to distort [57] and to deny their meaning. Marriage in such a system is not exactly the same thing as marriage in another system.

It is this, of course, that Dumont means when he says that Dravidian kinship terminology is an ‘expression of marriage’, and ‘embodies a marriage rule’ and so on. The terminology, as a system of classification, is so closely and inextricably articulated with the particular rule, that together they make a kind of unit to which each, apart, simply fails to add up.

What is marriage for a system without ‘a positive marriage rule’? This is the innovative, inaugurative relationship which ‘creates’. It does, as Fortes has so repeatedly stressed, constitute the core of a system of individuation, relating the child of the marriage to the class of mother’s brothers to and through those who are brother to his mother. The exchange between the descent units and the relationship between the descent units is a very different thing. In one system, descent units are already related; in the other, they become related. In one system they are necessarily related in a particular way, in the other they are not necessarily related until they become so.

In one kind of system, therefore, marriage plays one kind of role in the maintenance of the segment, in its definition, and in the definition of within-unit and between-unit relations. In the other kind of system — and this is the crux of one of the matters — it is not marriage so much as exogamy which maintains the definition of the segment, and of its within-unit and between-unit relations. For such a segment it is only exogamy which is necessary; for segments where a marriage rule obtains, it is both exogamy and marriage which are crucial elements in the system.

XVI. Two different kinds of system, each made up of identically structured segments, are really at issue. In one system, the segments are articulated into a logically interrelated system by the descent rule, the mode of classification of kinsmen, and the relationship of perpetual alliance between segments. In the other, the segments are defined by the descent rule, exogamy, and the variable bounding of the segments in terms of specific functions (domestic, jural, political, residential, territorial, and so on). This second kind of system consists in the proliferation of segments along genealogical lines. [58]

The first type of system can be called ‘segmental’, the second ‘segmentary’. Marriage, as we have seen, plays a very different structural role in each. But it is not marriage alone which decisively separates the two types.

The segmental and the segmentary are two types of a whole society or a whole system. Fortes has defended the segmentary type with skill and intellectual force, but he is defending a type. He may sometimes use different names, but he clearly distinguishes this type from all others. In his most recent paper, he uses the term ‘tribal society’, and he remains concerned with ‘residual rights’. ‘A man has unrestricted rights over his wife’s reproductive capacity in virtue of the bride-price he has paid to her father. But she never wholly forfeits her status as her father’s daughter This gives her residual claims on her father’s protection and him a lien on her’ (Fortes, 1962, p. 63). And Gluckman, in defending his sliding scale of corporateness, protests that he is only talking about a particular type of society.

Needham, in his turn, is defending a type; a whole closely articulated construct, but a type as a whole. He calls this ‘prescriptive’ and names the residual category, non-prescriptive types, ‘preferential’ and declines to discuss preferential systems. If one compares, as we have, for instance, what has been called ‘marriage’ in the two different types of society, it is evident that these are hardly the same phenomena, and perhaps should not be called by the same name and certainly should not be the bone of contention which Leach, Gough, and the others make of it. In a segmental system, a condition of perpetual alliance obtains conceptually; in a segmentary system, marriage is but the residuum of exogamy, the mark of the legitimate application of a descent rule, the condition providing for complementary filiation. To call these two things ‘marriage’ is to say that apples and eggs are both fruit because one is the fruit of a tree, the other of a hen.

In sum, the point of the discussion of this section is to show that the ‘segmental type’ and the ‘segmentary type’ do not refer to societies but to models. Any resemblance between these models (the models of Fortes, Leach, Lévi-Strauss, and Needham) and actual societies is wholly irrelevant to the point of this discussion, which is to show that Fortes’s model of a ‘segmentary’ society is a [59] type which is but one part of a typology; that Lévi-Strauss’s model of a ‘segmental’ society is also one type of a typology; that Needham’s ‘prescriptive’ and ‘preferential’ types are but two types of a typology; and that most of the heat of the discussion among these gentlemen is generated by the lively defense of a particular typology. Only occasionally (Livingstone is an example, Maybury-Lewis is another) has anyone said that the applicability of the model to a real society was gravely at fault. In the main, one author has argued that the other’s model has features which are inconsistent with the model of the first. Or they have argued that the names for the parts of the model are not used correctly. (‘Fortes … disguises [affinity] under his expression “complementary filiation” …’ Leach, 1957a, p. 54).

My distinction between ‘segmental’ and ‘segmentary’ thus applies to models, not to societies. But it should not be concluded from this that the models under no circumstances relate or are applicable to real societies. This is perhaps, for some of us, the ultimate concern and the ultimate test.

Part three. Choice and Needham’s typology

XVII. Following Lévi-Strauss, Needham formulates the basic difference between his prescriptive and preferential types in terms of whether ego has any choice in marriage. In a prescriptive system, ‘… the emphasis is on the very lack of choice; the category or type of person to be married is precisely determined, and this marriage is obligatory’ (Needham, 1962b, p. 9). Leach takes substantially the same position, even to the point of apparently accepting Needham’s typology of prescriptive and preferential where it might pay a few short-run polemic dividends (Leach, 1961). Later he elaborates at some length on the significance of choice (Leach, 1962b) and this elaboration is worth considering in detail to see just how the idea of choice operates in these different models.

Leach starts the discussion of double descent with the explicit declaration that he will follow Rivers’s definition so that ‘descent’ means unilineal descent units which are not such that ‘… membership itself is at all times ambiguous’ (Leach, 1962b, p. 132). The latter condition he exemplifies in terms of his ownership of shares [60] in the British-American Tobacco Company and the Samoan kin groups called aiga sa and fatelama.

‘… in such groups, not only is it the case that membership derives from choice rather than from descent, but the membership itself is at all times ambiguous. In a structure of this kind the potential groupings overlap, so that constantly each individual must ask himself: “Is it more to my advantage to fulfil my obligations towards group A or my very similar obligations towards group B?” It is because this ambiguity and choice does not automatically arise in true (that is, in unilineal) descent systems that Fortes and others have found it satisfactory to analyze unilineal descent systems as structures of jural obligation. In contrast, the analysis of any kind of cognatic kinship structure invariably ends by throwing the emphasis upon mechanisms of individual choice …’ (Leach, 1962b, p. 132).

Leach then proceeds to argue that

‘… with a simple unilineal descent … each individual is a member by birth of one exclusive segment of the total society and is linked by ties of filiation or of affinity (or perhaps both) with various other analogous segments. While Ego’s freedom of action with respect to membership of his own descent group is very severely restricted, his freedom of action with regard to his affinal and filiatory links with other descent groups is ordinarily wide’ (Leach, 1962b, p. 132).

Let us begin with Leach’s assertion that if ego is permitted to choose between being a member of one group or another, then by virtue of that mode of recruitment — actor’s choice as against ascription — the membership of those groups will overlap and be ambiguous.

There is a series of different elements involved which should each be distinguished. First, there is the question of the rules governing membership. Can one belong to one and only one such group, or can membership be multiple?

Second, if multiple membership is possible, can these member ships be simultaneous or must they be successive?

Third, if multiple membership is possible, are the units com- [61] plementary or competitive? That is, from the dialogue of soul-searching which Leach describes for his ego, it would seem that groups A and B both compete for ego’s commitment, and each requires the same token of commitment from ego. Ego is trying to resolve this conflict of competing claims by calling on his enlightened self-interest to guide him.

Fourth, if multiple membership is possible, and the units are competitive, do they call for the same or different tokens of commitment? Thus in two groups of like order, each claiming ego’s commitment against the claims of the other, one claims ego’s strength as a warrior, the other his skill as a farmer or the two groups could both call upon ego to contribute his labor.

Finally, there is the time dimension. For what period is membership required; is it single or multiple; for a year, for life, or for no stipulated period?

If the groups are single membership groups, and if, once choice is exercised, it cannot be changed but endures for life, then it is hard to see how the condition of choice makes membership in any sense ‘ambiguous’. Choice and ascription yield equally unambiguous groups under this set of conditions, so far as I can see.

If the groups are multiple membership groups, and if these are sequential rather than simultaneous (ego belongs in series, first to group A, then B. then C, etc.), again I find it difficult to understand how Leach can claim that the membership of such groups is by virtue of the matter of choice, overlapping and ambiguous. It may be that ego’s membership in a group is kept secret and so an outsider may not know who belongs and who does not, and an insider may be equally uncertain. Such a condition would surely make the membership of such a set of groups ambiguous. This is not a consequence of choice, but of secrecy.

If membership is multiple, and by choice, and the units are complementary, as in the case of double descent which Leach describes, then again there need be no ambiguity about membership as a consequence of choice, for it is quite clear from the case of double descent (in the pure sense in which Rivers defines descent) that if membership is by choice, or if it is by ascription, the consequences are precisely the same so far as ambiguity of membership is concerned. Within the units of like order — the patrilineal units, for instance — there is no overlapping, of course. [62] But those who are related to each other patrilineally, may not be related to each other matrilineally, and there is thus considerable overlapping, though there need be no ambiguity, in double descent systems. But if we once start considering double descent we must consider the problem of overlapping from the point of view of every kind of group and classificatory category within the society. It is obvious that there is always overlapping, there are always intersecting categories, and that some system of role segregation is required to integrate and articulate every social system.

It should be quite clear by now that what makes for ambiguous and overlapping membership is not necessarily the mode of recruitment at all. Even if recruitment is prescriptive and without choice, membership can be ambiguous if the criteria in terms of which membership is ascribed are ambiguous. Thus if recruitment is prescriptive on the basis of patrilineal descent and the criteria for establishing paternity are ambiguous, then multiple claims might well obfuscate membership lines.

In the particular form in which he presents it, Leach’s argument that the mode of recruitment is closely related to the ambiguity of the membership unit does not stand up. Nevertheless, it seems obvious that in general the mode of recruitment must relate to certain aspects of the structure of the unit and the system within which it occurs.

But here we return to our major theme. The model which Leach uses implicitly here is the model of the segmentary system, and the muddled part of that model is the notion that somehow the segment is not only a conceptual segment, but also in some way a physically distinct and concrete segment. For it is only with a segment so conceived that choice of membership, and frequent changes of membership, and multiple membership may create the kind of ambiguity which Leach describes.

If we turn now to Needham’s typology of prescriptive versus preferential, it seems clear why he uses choice as the essential discriminating element between these two types. The problem is that the segmental model requires that units be in an asymmetrical exchange relationship so that if A gives to B, A cannot receive that same kind of item from B, but must receive that kind of item from another unit. Thus, if A is a wife-giving unit, and B [63] a wife-taking unit, then A must take wives from a unit other than B — perhaps unit C. Since the act of ‘giving’ is constituted by a woman who marries, and the act of ‘taking’ consists in a man’s marriage, then the specification of obligation on the act of the man and the woman is simply another statement of the kind of relationship which obtains between the two units. If a man is required to take his wife from a wife-giving unit, there can be no mistake about the kind of relationship which obtains between those two units

This seems perfectly straightforward and simple. It is true in yet another sense. If units A and B are in a wife-giving to wife-taking relationship, then the rule that a man of B must take his wife from unit A is really nothing more than an injunction that he obey the rules of the relationship between the two units. For him to go elsewhere for a wife would be to repudiate that alliance. And this too, seems perfectly straightforward and simple. The relationship of alliance between units is the essence of the prescriptive type; it is the fact that the exchange between the units is asymmetric, that what one receives, it cannot return directly.

Indeed, if one can argue that the obligation to marry a particular category is not the precise correlate, at the level of individual action, of the structural relationship of the exchanging units, one may easily argue that the essence of the affiance model is corrupted by using the element of choice as the differentiating characteristic of the two major types. But such an argument seems unwarranted. It seems unwarranted because the guide to action, the imperative by which the men of such a system live, is only the necessary translation into action of the structural relationship of exchange. The requirement to marry MoBrDa is the statement at the level of action of the relationship of asymmetrical alliance between exchanging units.

Or so it would seem, except for two things. The first is that we are comparing the wrong things and so come to the wrong conclusion.

When we say that the system is prescriptive and mean by this ‘… the emphasis is on the very lack of choice; the category or type of person to be married is precisely determined, and this marriage is obligatory’ (Needham, 1962, p. 9), we can understand such a statement in contrast to and in contradistinction to some [64] category to which it is opposed. Here prescription is contrasted to choice. We know what a prescription is. It is the obligation to marry MoBrDa in an asymmetrical alliance system. What is ‘choice’ then? Choice obviously is where ego can marry MoBrDa or FaSiDa; or MoMoBrDaDa, or anyone else who pleases him. Look at the array of categories! One against an almost unlimited number.

But is this the proper contrast? Are these indeed the correctly opposed categories? Is MoBrDa in one system in any sense the comparable contrast with MoBrDa, FaSiDa, MoMoBrDaDa, and so on in the other system? By some stretch of the imagination it might be, if one had not learned by now the pitfalls of genealogy as against category (Needham, 1962b).

We early learned that in cross-cousin marriage systems the own real daughter of the own real brother of the own real mother in fact was not often married. We then learned that anyone who was ‘classified’ as mother’s brother’s daughter was an acceptable bride according to the rules in such cases, and this often included kin types such as mother’s mother’s brother’s daughter’s daughter, and so on. ‘And so on’ turned out, on analysis, to include women from certain lines, many of whom could not even trace a genealogical connection to ego, nor bothered to try. Moreover, if their actual genealogical connections prohibited them from being married, a rite could re-classify them into the marriageable category, as for instance is done in some societies like Purum. We learned too from Lévi-Strauss and Needham that although diagrams showed three lines in a circle, actually any given line might take wives from two, three, or four lines and might in turn give wives to two, three, or four lines, so that the chances of actually getting a genealogical category to coincide with marriage or alliance category were fairly slim indeed. And Lévi-Strauss showed how the essence of such a system was that of indirect exchange as opposed to direct exchange. A wife came from a wife-giving unit; a wife was given to a wife-taking unit; the unit which gave must take from elsewhere. Hence the better gloss for this category was ‘marriageable woman, i.e. a woman from a wife-giving unit’.

Two very important points now follow. First, for us to think of such a system as a MoBrDa system is wrong, for the category [65] is much more properly translated not by an irrelevant genealogical specification, but rather by such phrases as ‘marriageable woman’. Second, the essential element in the system is the asymmetrical alliance.

Having found the correct translations and discarded the inappropriate and inaccurate ones, we may now return to the formula and do some simple substitution. A prescriptive system is one in which ego has no choice, but must marry a ‘marriageable woman’. A ‘marriageable woman’ is a woman from a wife-giving unit. If this is a prescriptive system, what then is the opposite type? What is the contrast?

Is it a system in which ego has a ‘choice’? If so, what can that choice consist in? The opposite of a marriageable woman is an unmarriageable woman, or a prohibited woman. Is this the proper opposition? Can we say that a prescriptive system is one in which ego is obliged to marry a woman whom he is permitted to marry and a preferential system one in which ego is permitted to marry a woman who is prohibited? This seems sheer nonsense, but it is to such sheer nonsense that one is led if one starts with a structural problem and tries to define it in terms of individual action (Needham, 1962) on a choice versus no-choice basis.

In all models that we have heard anything about, a man is obliged to marry a marriageable woman and prohibited from marrying a woman of prohibited category. This applies to segmental and segmentary models alike. A woman who is not marriageable is one to whom the incest taboo applies. This too applies in segmentary as well as segmental models.

In alliance theory as well as in descent theory, then, ego must marry a permitted and not a prohibited woman. In alliance theory a permitted woman is defined as one who comes from or who is a member of a wife-giving unit; in descent theory she is one who is from any unit other than ego’s, since in that model units are not in an alliance relationship with each other.

To specify that in one system choice is permitted and in the other system it. is precluded is to confound two different levels of analysis and to fail to deal with the categories in terms of their proper conceptual definition. But it is one more thing, and this perhaps is its gravest error: it is essentially to nullify and con- [66] found most of the clarification which Lévi-Strauss, Needham, and Leach have all contributed to this question.

It is fair to say that what started out as the study of cross-cousin marriage has made sense in proportion to the degree to which we have gotten away from genealogical cross-cousins. The more our terminology has clung to the traditional names, like MoBrDa and FaSiDa, the more confused the problem has become.

This brings me to the final point in this connection. When the definition of the system as a system is stated in terms of the conditions under which an individual may act, it does so within the context of a given structure. The word ‘choice’ in English takes as its focus the problem of the individual’s action within the framework of a particular structure of the situation, and this structure is treated as given. Hence, it is a misplaced definition so far as the analysis of structure is concerned to discriminate two structures in terms of individual choice. The models are systemic models, they are not usable as models of situations within which actors choose among alternate courses. We are concerned with the question ‘How is this system structured?’ not with the question ‘Given this structure, how can a man pick his way through it?’ We are concerned with the question ‘What is the structure of the relationship between segments?’ not with the question ‘Is a man allowed to marry his cousin?’

Needham’s distinction between his prescriptive and preferential types is thus, from the point of view of alliance theory, untenable when it is stated in terms of choice. In both prescriptive and preferential systems, men are obliged to marry marriageable women and prohibited from marrying prohibited women. It is the definition of who is in the category of marriageable women which differs, not whether there is choice; it is how the lines are related, not whom the men are free to marry.

XVIII. The difficulties of dealing with types are clear, but they are easily forgotten. Just what is the ‘type’? Is it a Weberian ‘ideal type’? Is it an analytic construct which the anthropologist finds pedagogically useful? Is it a point on an evolutionary scale like ‘preliterate’, or ‘use of fossil fuels’? Is it a summary of the most frequently encountered characteristics in a finite array of items [67] that corresponds to no one concrete case? What function has the segmentary type in Fortes’s thinking or in his scheme of things? For Dumont, is it only a way of dividing those societies where affiance theory seems to hold from those in which it does not? Typologies have been used for more trivial purposes than these in the history of anthropology.

The difficulties of dealing with types are clear, but Needham’s work is an illuminating instance of some of the pitfalls.

In 1956 he said, ‘The greatest difficulty of all, however, is that we do not know what the facts are. There is no adequate account of unilateral cross-cousin marriage anywhere in the literature; and, until there is, there can be no great theoretical progress’ (Needham, 1956, p. 108).

As if to contradict this grave pronouncement, Needham proceeded to publish more than 16 papers and a short book on this subject, discussing Eastern Sumba, Purum, Vaiphei, Korn, Lamet, Chawte, Aimol, Siriono, and Wikmunkan at greater or lesser length. These have all been literary analyses (Needham, 1957) of material largely available when Needham wrote his pronouncement of 1956.

The remarkable thing about the ethnography available to Needham up to 1956 (and possibly even now) is that according to him it has all been plagued by a host of errors made by the ethnographers. The ethnographers have made mistakes and failed to provide crucial information. For instance, Needham feels that there is enough evidence to show that the Siriono are matrilineal, but that the ethnographer failed to note it (Needham, 1961, 1962, p. 64; Coult, 1962; Needham, 1963b). For the Wikmunkan, ‘I must say, too, that I should prefer to rely on his reports [those of Professor D. F. Thomson], for reasons which my analysis makes plain, than on those of the late Miss McConnell’ (Needham, 1963b, p. 145)… on this point Das’ ethnography is contradictory and that the categorized totals of marriages above are wrong’ (Needham, 1958a, p. 81). ‘These are consistent with a lineal descent system, and given the general patrilineal character of Kuki society we may safely assume that the Vaiphei are patrilineal also’ (Needham, 1959a, p. 401). ‘ZS (ws) cannot be interpreted in this way, and is apparently incorrect’ (Needham, 1960b, p. 103). Indeed, pages 102-104 of this paper are almost entirely devoted to [68] sorting out the presumed inconsistencies reported by the ethnographer.

Errors and inadequacies of ethnographers are one side of the coin. The other side shows up when others have re-analyzed some of the same data. Needham’s presentation can only then be appreciated for the true elegance and order which it imposes on the data. A closer look at the Purum, for instance, makes it appear to fit only very loosely into Needham’s type (White, 1963, pp. 130-145; Leaf, 1963). A closer look at Lamet does not yield so neat a picture as one might have expected from Needham’s 1960b paper. The material on the Pende is not so simply and decisively and undebatably ‘preferential’ as Needham claims, nor is the case of the Tismulan quite so clear (Needham, 1963, p. 68; Lane, 1962b).

The issue is simple. The inconsistencies, errors, inadequacies may in fact be inconsistencies, errors, inadequacies of the fieldwork and/or the reporting. Ethnographers do make mistakes. Informants do get their facts wrong. No one in the field can check every single bit of information he has collected. No ethnographer claims to infallibility.

The question, then, is this: Is this particular bit an inconsistency, an error in the data or not? Needham takes the ethnographic report and matches it against his model, his type. Every deviation of the ethnography from one or another element in the type suggests to Needham that the ethnography is wrong in one way or another. Needham never alters his type to accommodate the ethnography. Needham never changes his model to fit the data.

It seems, therefore, that Needham expects to find, free in nature, a concrete system which precisely replicates his type. If the type has characteristics X, Y, and Z in that order, then Needham expects to find that the Purum or the Lamet have characteristics X, Y, and Z in that order. Needham works with this type as if it were a kind of ‘missing link’, a real entity which a really good ethnographer who is a good hunter will be able to find — on Sumba perhaps, or among the Old Kuki. once it is found we will see ‘… how the system really works’ (Needham, 1958c, p. 323).

What else can we understand by such phrases as ‘… a pre- [69] scriptive alliance system based on exclusive patrilateral cross-cousin marriage not only does not exist, but on both formal and pragmatic grounds cannot exist — and this is the answer to the problem of unilateral “choice”’ (Needham, 1962a, p. 118). To our astonishment, we are now told that the system cannot exist as a social system if, in fact, it cannot work. Livingston’s demonstration that the Purum system could not actually work with the numbers of people reported now seems to be acknowledged to be true and just and the whole attack on it (Needham, 1960h) misconceived.

At what level does the system exist? For Needham, ‘the system’ seems to exist at the concrete level, at the manifest level, at the level of the explicit institutionalization of the categories and the rules relating those categories. There must be a prohibition on FaSiDa and a requirement to marry MoBrDa; these must be in unilineal descent units; MoBrDa must be terminologically distinguished from FaSiDa, etc. Any variation in the report is an error in the ethnography, not in the model.

For Needham, a society (Purum. Pende, Tismulan) is prescriptive or it is preferential. Needham can provide a list of those which are concretely and wholly in one or the other classification (Needham, 1962a, pp. 55-57).

Contrast this position with that of Lévi-Strauss. For the latter ‘the system’ is far more subtle. It exists and it is real. But its existence is only roughly manifest in the concrete categories and the socially institutionalized rules for relating these categories. Time, ecology, and a variety of factors affect the concrete manifestation of the principles.

Needham’s typology is one of whole, concrete entities. If they exist as such, Needham feels that they ought to be discoverable and when they are found, they ought to be ‘perfect crystals’, so to speak. It seems apparent that Needham who has been to Sumba, Sarawak, and Malaya has not even been able to find these perfect specimens himself.

Needham’s passionate dedication to his concrete types leads him into absurdities, into denying the existence of certain data which seem undeniable His repeated assertions that patrilateral prescriptive systems ‘cannot exist’ (Needham, 1962, p. 118) can only be maintained if he insists on seeing a perfect specimen [70] which, in its ethnographic report, is even more perfect than the matrilateral cases which have been so imperfectly reported. In addition to the Pende and Tismulan (Lane, 1962b), Salisbury’s Siane (Salisbury, 1962), the Busuma, Iatmul, and perhaps others seem fit to the models advanced, provided there is no insistence on the kind of perfection in their concrete manifestation which Needham demands. Needham’s effort to force the Siorono into a unilineal descent system because it has Crow-type terms is another example of the absurdity his typology requires of him. Needham feels that a proper alliance system cannot exist in the absence of unilineal descent groups. That Yalman’s Sinhalese manage without unilineal descent groups either proves that Yalman (like Holmberg) simply failed to see what must have been there, or that this is, in fact, like the Pende and Tismulan, really a case in the other category — preferential — and can therefore be dismissed from consideration.

There is, however, still another and perhaps more important error that Needham commits by the use of his concrete types: this is the implicit assumption that each social norm and rule has one and only one function and, because each norm and rule has only one function, its particular shape can be wholly determined or explained by that function. The rules which categorize kinsmen, Needham implies, function so to order the social universe as to constitute and maintain a system of asymmetrical alliance. For it is Needham’s assumption that kinship terms function only as the names by which social categories bearing a single mode of relationship with each other are designated.

Yet it seems clear that kinship terms have other functions than that of designating the categories to be related in an asymmetric alliance system. Among the Lamet or the Purum, for example, there may very well exist a system of asymmetric alliance, although the kinship terms may not be perfectly concordant with that system. Other loosely related systems occur too; sub-systems of variant modes may occur within the major framework; alternate modes of designation and categorization may occur. As I have tried to show elsewhere (Schneider & Homans, 1955), the alternate terms of American kinship are, on the one hand, consistent with a single, overall pattern, yet vary within the framework of that pattern, whereas the Zuni treatment of [71] FaSiDa as either FaSi or ESi varies across the lines of a basic framework (Schneider & Roberts, 1956). Alternate terms, as an order of data from a wide variety of different kinds of societies, from America to Zulu, from Zuni to Truk, constitute the decisive bit of evidence against any simple, unicausal view of the determination of a terminological pattern.

In short, Needham’s model differs in certain important respects from those of Leach and Lévi-Strauss. Needham expects that his model is a close approximation of the concrete shape and content. of concrete societies. He thus expects to find a society which approximates in the closest terms the model he has built. Like the descent theory model of the segmentary system, Needham’s prescriptive and preferential types are concrete types, they have exact counterparts in nature. So he has examined the literature and himself gone into the field. He has been able to show a large number of gaps and mistakes in the ethnographic reports because his model is perfect, but the work of the ethnographers is not. Though he speaks of principles at times, and though ideas like dualism and reciprocity are present in his work, he has moved, relentlessly almost, into the positivist position from which at first he had fled along with Lévi-Strauss. But he has backed into it at the end. His treatment of kinship terms is another example of this, for he acts as if the kin term system was a system determined entirely by the need to name categories whose only significance is their operation in a system of prescriptive alliance.

Part four. Conclusion

XIX. I have set up what might seem to be two types; alliance theory and descent theory. By alliance theory I mean only what Lévi-Strauss, Leach, Needham, or Dumont have said, and by descent theory what Fortes, Goody, Gough, or Gluckman have said.

I do not pretend that all is peace and harmony within the ranks of those I group together. Early in Section IV, I quoted Needham on his differences with Lévi-Strauss and I have pointed out other differences when they seem relevant. I do not believe that Fortes agrees with everything which Goody has written, or vice versa.

I do not think that there is a ‘thing’ or a ‘school’ or a specific [72] body of doctrine apart from what each person has said. I have used the names ‘alliance theory’ and ‘descent theory’ on Dumont’s suggestion and these words only mean that in certain carefully designated respects, Lévi-Strauss, Leach, Dumont, and Needham are in general agreement with each other and disagree with Fortes, Goody, Gough, and Gluckman, who are in closer agreement with each other on specific points than with the former.

My concern has been with the models these people use, with their internal consistency, and with certain of the implicit elements in the models. I have not tried to fit their models to the ethnographic facts. That is an entirely different problem from the one I have tried to deal with here.

The major criticism I have offered is that both the alliance theorists and the descent theorists have a tendency toward the development and propagation of whole-system, over-simple typologies. There has been a tendency to erect a typology and to defend it to the death against all corners; even against the facts where these prove stubborn.

This tendency toward the total-system typology is unevenly distributed among the theorists. Fortes is mildly addicted to his practice; Needham has a severe, and perhaps fatal case of it. Leach, despite his eloquent sermons (Leach, 1962a), is not entirely free of this vice himself (Leach, 1961).

I suspect that the addiction to such fruitless whole-system typologies is related to polemic tendencies. The greater the author’s commitment to polemic goals, the greater the exaggeration into which he is forced, the greater the extremes of one sort or another he finds himself using, the greater the oversimplification of idea and expression. The time when Leach seems to accept the prescriptive-preferential typology as a serious possibility is when he is angrily after the scalp of Barbara Lane (Leach, 1961). In his more reasonable moments, he is not too seriously taken with that particular device for the classification of ethnographic butterflies (Leach, 1962a). The fact that Lane made some fair, cogent and important points may or may not be related to the magnitude of Leach’s polemic explosion (also see Needham, 1963d).

One of the most serious difficulties with the descent theory [73] model, and a difficulty of which the alliance theory model is not entirely free, is the failure to distinguish the segment as a conceptual entity from the segment as a concrete, physical entity in the total system.

I have tried to suggest in Sections X and XVI that there may be a single limiting case where the segment as a conceptual as well as a concrete entity would seem to be a workable model, but in fact, of course, even that model makes no sense. Even if the segment is totally self-sufficient except for a single function — that is, it must marry outside and thus must be linked by marriage with other like segments — that one link turns out, on closer inspection, to consist in a veritable network of highly differentiated modes of relationship. For the marriage, if it does not entail co-residence (and if it entails co-residence, the segment is not physically distinct), must at least entail visiting husbands and/or visiting wives, and these in turn, related to their children as well as their agnates, set up an intricate network of filiative bonds, and these, in their turn, proliferate into gift exchanges, favors, claims and counter claims, agnates versus affines, and the warmth of the mother’s brother for his sister’s son and daughter. Leach’s down-to-earth discussion of local descent lines as the real things and not diagram lines, gives this good, hard feeling of a segment as a concrete entity (Leach, 1951). Evans-Pritchard’s classic treatment of the Nuer (Evans-Pritchard, 1940) is one of the early statements of this confusion, for it is evident when Evans-Pritchard is faced with the odd fact that, although the lineages are territorial units, and although they seem to be as patrilineal as patrilineal can be, descent is traced through women and many people live matrilocally. In order to reconcile these apparent contradictions, we were treated to those special gems of paradoxical obfuscation for which Evans-Pritchard is justly famous, such as the remark that ‘it is the clear, consistent, and deeply rooted lineage structure of the Nuer which permits persons and families to move about and attach themselves freely … to whatever community they choose by whatever cognatic or affinal tie they find it convenient to emphasize …’ (Evans-Pritchard, 1951, p. 28) or, to go it one better, ‘It would seem it may be partly just because the agnatic principle is unchallenged in Nuer society that the tracing of descent through women is so prominent and matrilocality so [74] prevalent’ (Evans-Pritchard, 1951, p. 28).

The failure clearly to distinguish the segment as a conceptual entity from its concrete counterpart as a group leads to the special requirement that its modes of recruitment, and the rules governing membership, must be such as to permit the segment to be a physically distinct entity — perhaps, even, to assemble in its full force for some ceremony or other. Therefore, that mode of recruitment to group membership which yields a distinct, concrete entity should not be confused with others. The unilineal descent rule is believed to be such a mode of recruitment par excellence, for this allocates a whole man to a group (see Section X above). In order, therefore, to protect the model of a segment as a physically distinct entity, unilineal descent is separated from all other modes of recruitment and we are left with a new and wholly odd definition of descent as only the unilineal form, everything else must be called by some other name.

I find it strange and perverse that Rivers, Fortes, and Leach insist that it is not even possible to say that there are two different kinds of descent rules, one unilineal, the other not; that the one yields unambiguous group boundaries, the other yields ambiguous group boundaries; the one yields what are regarded as ‘true social segments’, the other not. Rivers. Fortes, and Leach (Leach, 1962b) require that we must not even call that second kind of descent ‘descent’. We must banish its very name! As if calling it by another name would in fact banish it.

I have suggested, therefore, but perhaps too obliquely, that this is a perversion of the notion of descent, undertaken in the interest of protecting a typology of segmentary system, and of a model of a segment which is untenable to begin with. As an analytic scheme, it is misleading and thus harmful rather than useless.

Closely connected with this model of a segment is the notion that those forms of membership rule which permit any choice are different and yield different kinds of segment or group from those where no choice is permitted; that is, where the rule is ascriptive. Choice comes into the segmentary system model as an element opposed to ascription by ‘true’ unilineal descent, and it has a contrasting effect on group structure as well. Where choice is per- [75] mitted, it is believed that group boundaries are ambiguous and one must expect to encounter the entertaining dialogues which Leach offers us, of people patiently trying to unravel the tangled strands of their affiliation with different groups in terms of their enlightened self-interest (Leach, 1962b). I have tried to show that this is nonsense; that boundaries are ambiguous when the criteria for membership are ambiguous, not when choice is involved. I outlined a series of different kinds of choice condition, showing that the group boundary problem in each case is anything but ambiguous — indeed, quite as clear as in the purest case of ‘true’ (unilineal) descent.

In this section (XVII), I made the same case against Needham’s use of choice as the criterion for distinguishing his prescriptive from his preferential types. My argument here was that Needham had confused his oppositional categories. In both preferential and prescriptive systems ego is obliged to marry a marriageable woman; he has no choice about that in either type, and the word ‘choice’ is misapplied hero. In Needham’s prescriptive type, the marriageable woman comes from a wife-giving unit whereas in the preferential type she does not, and this is what distinguishes Needham’s two types, not the question of choice. (That the whole typology of preferential versus prescriptive is untenable for other reasons as well is beside the point here, which is that the use of choice only further confounds these issues.)

The point that choice is simply not a structurally relevant category applies to Lévi-Strauss’s use of it too, for choice is basically a statement of an actor’s course of action when the structure of the situation is taken as given. Lévi-Strauss, Needham, Leach, Fortes at least, and Gluckman (1950; see also Schneider, 1953a) in some of his writings, are primarily concerned with structure, not with individual action within the context of a structure. Choice does not say anything of significance or use about structure. This is what Parsons has properly called the confusion between the actor as the point of reference and the system as the object of reference.

I did not go into the matter in detail, but the problem of the ambiguity of boundaries, the rule of descent, the mode of recruitment are all closely linked with the problem of whether an alliance system requires unilineal descent. Yalman’s paper on the [76] Sinhalese is of importance in two respects. First, it is a case where the system of alliance is perfectly clear, and where it works perfectly well without having, what Fortes and Leach insist be called ‘true’ or unilineal descent. The opposition between affinity and consanguinity, in the form of the unilineal segment or descent group, does not stand up as a necessary condition to such systems if Yalman’s account is to be relied on. Ego-oriented systems do not lead to mires of confusion and boundary-ambiguity as Leach so vehemently claims. The second point of relevance is, of course, Needham’s desperate efforts to make unilineal systems out of the Siriono and the Vaiphei, for instance. His failure to deal with the Sinhalese case in this respect is significant.

And this brings us back to the first major criticism, instead of working with models made up of distinct pieces which are arranged and re-arranged into a variety of different permutations and combinations, Needham has saddled himself with a total-system model. Each little piece must be linked with every other little piece in a particular way to make a perfect constellation of a whole, crystal-clear system. The system must be lineal; the kin terms must be consistent with (patrilineal or matrilineal) lineal systems; the exchange units must be descent units, etc., etc., etc. So if the Siriono are not said to be matrilineal, then Holmberg must be wrong because as a total-system type, every piece must be in its proper place. Never once are we shown why a system, to be a prescriptive system, as a whole system, must be unilineal; why FaSiDa must be distinguished from MoBrDa.

Here Leach makes more sense than Needham. Where Needham is chained to his types. Leach is able to push alliance theory pieces as far as they can go without worrying about what is happening to a typology. Because of this, the structural role of marriage in segmentary systems now requires careful re-examination and restatement; the notion of filiation must be restated and redefined, though I expect it will survive; the sliding scale of corporateness, the residual rights and submerged rights, and imprecisely defined ‘strengths’ are now shown to be in dire need of repair. At the same time, by using Leach’s technique of pushing parts of the alliance theory model into new and untried areas, clear defects in that model itself have become apparent. I have already referred to Yalman’s Sinhalese kindred; Dumont [77] has already noted the problems raised by Barth’s FaBrDa marriage material and his Pathan work.

Let me be very clear on this problem. The model which is a total-system model, which yields a typology, and where there is no specific aim or purpose for which that typology is constructed is, I think, demonstrably a mistake. Too much time, effort, and energy are spent in mending the model, in protecting it from new data, in insuring its survival against attacks. It is too late in the history of the social sciences to think we can go out among societies and, by keeping our eyes open, sort them into their natural classes. It is not possible to operate like those in the story of the blind men and the elephant and hope that if only we can put enough blind men on the elephant we will get a good factual description of the beast — the total elephant. A typology is for a problem, it is not for sorting of concrete societies into unchangeable, inherent, inalienable categories. (I agree completely with Leach in this matter, as should be clear by now (Leach, 1962a).)

Instead of typologies we need a series of relevant elements, like descent, classification, exchange, residence, filiation, marriage, and so on; these need to be rigorously defined as analytic categories and then combined and recombined into various combinations and permutations, in different sizes, shapes, constellations.

The model of defined parts can be constructed with, or without Levi-Strauss’s kind of intellectualist or Hegelian assumptions, or the kind of positivism which Fortes requires. I have dwelt somewhat longer on some of the positivist difficulties than on the intellectualist problems, but each has its fair share of problems.

Finally, there is one point which needs stressing and which I only touched on. Alliance theory as a theory is capable of dealing with the symbol system as a system apart from, yet related to, the network of social relations. It has a way of dealing with problems of meaning which the descent theory of Radcliffe-Brown and Fortes does not have. Alliance theory, in the footsteps of Durkheim here as elsewhere, is cognizant of the importance of how the actor conceptualizes the structure (‘how the natives think’ perhaps) and the difference between this conceptualization and an outsider’s analytic construct of the system as a system. Where Radcliffe-Brown rejected culture in favor of what he was [78] pleased to call structure, and where Leach in an earlier work separated out cultural ornaments, alliance theorists have brought culture and social structure into an ordered relationship which even Needham’s gross manipulations of the Purum data cannot obscure.


Thanks are due to the following for permission to quote passagee from published works:

Beacon Press, New York, and Merlin Press, London, in respect of Totemism by C. Lévi-Strauss, translated by Rodney Needham; the Editor of Bijdragen tot de Taal-, Land-, en Volkenkunde in respect of ‘On Manipulated Sociological Models’ by C. Lévi-Strauss and ‘A Structural Analysis of Aimol Society’ by Rodney Needham; Dr E. R. Leach and the Council of the Royal Anthropological Institute of Great Britain and Ireland in respect of ‘On Certain Unconsidered Aspects of Double Descent Systems’, Man, 62, by E. R. Leach; The University of Chicago Press in respect of Structure and Sentiment: A Test Case in Social Anthropology by Rodney Needham, © 1962 by the University of Chicago. [79]


Names of journals are abbreviated as follows:
AA American Anthropologist
Bijdragen Bijdragen tot de taal-, land-, en volkenkunde
JPS Journal of the Polynesian Society
JRAI Journal of the Royal Anthropological Institute
SWJ Southwestern Journal of Anthropology

Beattie, J. H. M. 1957-58. Nyoro Kinship, Marriage and Affinity. Africa 27, 28.

Coult, A. D. 1962. An Analysis of Needham’s Critique of the Homans-Schneider Theory. SWJ 18.

De Josselin de Jong, J. P. B. 1952. Lévi-Strauss’ Theory on Kinship and Marriage. Mededelingen van het Rijksmuseum voor Volkenkunde.

Dumont, L. 1950. Kinship and Alliance among the Pramalai Kallar. Eastern Anthropologist 4.

———. 1953a. The Dravidian Kinship Terminology as an Expression of Marriage. Man 53, No. 54.

———. 1953b. Dravidian Kinship Terminology. Man 53, No. 224.

———. 1957a. Hierarchy and Marriage Alliance in South Indian Kinship. Occ. Pap. Roy. Anthrop. Inst. No. 12.

———. 1957b. Une Sous-caste de l’Inde du Sud. Paris: Mouton.

———. 1959. A Structural Definition of a Folk Deity of Tamil Nad: Aiyanar, the Lord. Contributions to Italian Sociology 3.

———. 1960. Le mariage secondaire dans l’Inde du Nord. 6th Int. Congress Anthropology, Vol. 2.

———. 1961a. Descent, Filiation and Affinity. Man 61, No. 11.

———. 1961b. Marriage in India: The Present State of the Question. Contributions to Indian Sociology 5.

———. 1961c. Caste, Racism. and ‘Stratification’: Reflections of a Social Anthropologist. Contributions to Indian Sociology 5.

———. 1961d. Les Mariages Nayar Comme Faits Indiens, L’Homme, 1, 1.

———. 1962. Le Vocabulaire de Parenté dans l’Inde du Nord. L’Homme, 2.

Durkheim, E. 1947. The Division of Labor in Society. Glencoe, Ill.: The Free Press.

———. 1951. Suicide. Glencoe, Ill.: The Free Press.

Epling, P. J. 1961. A Note on Njamal Kin-term Usage. Man 61, No. 184.

Evans-Pritchard, E. E. 1940. The Nuer. Oxford: Clarendon Press.

———. 1951. Kinship and Marriage Among the Nuer. Oxford: Clarendon Press.

Firth, R. 1957. A Note on Descent Groups in Polynesia. Man 57, No. 2.

Fortes, M. 1953. The Structure of Unilineal Descent Groups. AA 55.

———. 1959. Descent. Filiation, and Affinity. Man 59, Nos. 309, 331.

———. 1962. Ritual and Office in Tribal Society. In Gluckman, M. (ed.), Essays on the Ritual of Social Relations. Manchester: Manchester University Press.

Fortune, R. R. 1933. A Note on Some Forms of Kinship Structure. Oceania 4.

Freeman, J. D. 1961. On the Concept of the Kindred. JRAI 91.

Gluckman, M. 1950. Kinship and Marriage among the Lozi and Zulu. In Radcliffe-Brown. A. R. and Forde, C. D., African Systems of Kinship and Marriage. London: Oxford University Press.

Goodenough, W. 1955. A Problem in Malayo-Polynesian Social Organization. AA 57.

Goody, J. 1959. The Mother’s Brother and the Sister’s Son in West Africa. JRAI 89.

———. 1961. The Classification of Double Descent Systems. Current Anthrop. 2.

———. 1962. Death, Property, and the Ancestors. Stanford: Stanford University Press; London: Tavistock Publications.

Gough, E. K. 1952. Changing Kinship Usages in the Setting of Political and Economic Change Among the Nayars of Malabar. JRAI 82.

———. 1955. Female Initiation Rites on the Malabar Coast. JRAI 85.

———. 1956. Brahmin Kinship in a Tamil Village. AA 58.

———. 1959. The Nayars and the Definition of Marriage. JRAI 89.

Homans, G. C. 1962. Sentiments and Activities. New York: Free Press.

Kroeber, A. L. 1917. California Kinship Terminologies. Univ. Calif. Public. in Amer. Archaeol. and Ethnol. 12.

———. 1948. Anthropology. New York: Harcourt, Brace.

Lane, B. S. 1960. Varieties of Cross-Cousin Marriage and Incest Taboos: Structure and Causality. In Dole, G. & Carneiro, R., Essays in the Science of Society. New York: Crowell.

———. 1961. Structural Contrasts Between Symmetric and Asymmetric Marriage Systems: A Fallacy. SWJ 17.

Lane, R. & Lane, B. 1956. A Reinterpretation of the ‘Anomalous’ Six-Section Marriage System of Ambrym, New Hebrides. SWJ 12.

———. 1958. The Evolution of Ambrym Kinship. SWJ 14.

———. 1959. On the Development of Dakota-Iroquois and Crow-Omaha Kinship Terminologies. SWJ 15.

———. 1961. A Reconsideration of Malayo-Polynesian Social Organization. AA 63.

———. 1962a. Implicit Double Descent in South Australia and the Northeastern New Hebrides. Ethnology 1.

———. 1962b. Patrilateral Cross-Cousin Marriage. Ethnology 1.

Leach, E. R. 1951. The Structural Implications of Matrilateral Cross-Cousin Marriage. JRAI 81.

———. 1957a. Aspects of Bridewealth and Marriage Stability among the Kachin and Lakher. Man 57.

———. 1957b. On Asymmetrical Marriage Systems. AA 59.

———. 1958. Concerning Trobriand. Clans and the Kinship Category Tabu. In Goody, J. (ed.), The Developmental Cycle in Domestic Groups. Cambridge Papers in Social Anthropology, 1.

———. 1959. Social Change and Primitive Law. AA 61.

———. 1960. The Sinhalese of the Dry Zone of Northern Ceylon. In Murdock, G. P. (ed.), Social Structure of Southeast Asia. Viking Fund Pubs. 29. Chicago: Quadrangle.

———. 1961. Asymmetric Marriage Rules, Status Difference, and Direct Reciprocity. SWJ 17.

———. 1962a. Rethinking Anthropology. London: Athlone Press.

———. 1962b. On Certain Unconsidered Aspects of Double Descent Systems. Man 62, 214.

Leaf, M. 1963. ‘Age’ in Purum Social Structure. M.A. Thesis, Univ. of Chicago.

Lehman, F. K. 1963. The Structure of Chin Society. Illinois Studies in Anthropology No. 3.

Lévi-Strauss, C. 1945. L’Analyse structurale en linguistique et en anthropologie. Word 1 (also in Anthropologie Structurale, 1958).

———. 1949. Les Structures élémentaires de la parenté. Paris: Presses Universitaires de France.

———. 1953. Social Structure. In Kroeber, A. L. (ed.), Anthropology Today. Chicago: University of Chicago Press.

———. 1955. The Structural Study of Myth. J Am. Folklore 68.

———. 1956. Les Organisations dualistes existent-elles? Bijdragen 112.

———. 1960. On Manipulated Sociological Models. Bijdragen 116.

———. 1962a. Le Totémisme aujourd’hui. Paris: Presses Universitaires. (Trans. Needham, R., as Totemism. Boston: Beacon, 1963.)

———. 1962b. La pensée sauvage. Paris: Plan.

Livingstone, F. A. 1959a. A Formal Analysis of Prescriptive Marriage Systems Among the Australian Aborigines. SWJ 15.

———. 1959b. A Further Analysis of Purum Social Structure. AA 61.

Mauss, M. 1954. The Gift. Trs. I. Cunnison, London: Cohen & West.

Maybury-Lewis, D. 1960. The Analysis of Dual Organizations. Bijdragen 116.

Murdoch, G. P. 1949. Social Structure. New York: Macmillan.

———. 1960. Social Structure in S.E. Asia. Viking Fund Pubs. No. 29. Chicago: Quadrangle.

Murphy, R. F. 1963. On Zen Marxism: Filiation and Alliance. Man 63, 21.

Needham, R. 1954a. Reference to the Dead among the Penan. Man 54.

———. 1954b. The System of Teknonyms and Death-names of the Penan. SWJ 10.

———. 1954c. A Penan Mourning-Usage. Bijdragen 110.

———. 1955. A Note on some Murut Kinship Terms. J. Mal. Br. R. Asiatic Soc. 28.

———. 1956a. Ethnographic Notes on the Siwang of Central Malaya. J. Mal. Br. R. Asiatic Soc. 29.

———. 1956b. Review of Homans and Schneider. Am. J. Sociology 66.

———. 1956c. A Note on Kinship and Marriage on Pantara. Bijdragen 112.

———. 1957. Circulating Connubium in Eastern Sumba. A Literary Analysis. Bijdragen 113.

———. 1958a. A Structural Analysis of Purum Society. AA 60.

———. 1958b. The Formal Analysis of Prescriptive Patrilateral Cross-Cousin Marriage. SWJ 14.

———. 1958c. Review of Dumont ‘Une Sous-caste …’. Bijdragen 114.

———. 1959a. Vaiphei Social Structure. SWJ 15.

———. 1959b. An Analytical Note on the Kom of Manipur. Ethnos 3-4.

———. 1959c. Mourning-Terms. Bijdragen 115.

———. 1960a. Lineal Equations in a Two-section System. JPS 69.

———. 1960b. Alliance and Classification among the Lamet. Sociologus 10.

———. 1960c. Chawte Social Structure. AA 62.

———. 1960d. A Structural Analysis of Aimol Society. Bijdragen 116.

———. 1960e. Patrilateral Prescriptive Alliance and the Ungarinyin. SWJ 16.

———. 1960f. The Left Hand of the Mugwe. Africa 30.

———. 1960g. Jataka, Pancatantra and Kodi Fables. Bijdragen 116.

———. 1960h. Structure and Change in Asymmetric Alliance. AA 62.

———. 1961a. Notes on the Analysis of Asymmetric Alliance. Bijdragen 117.

———. 1961b. An Analytical Note on the Structure of Siriono Society. SWJ 17.

———. 1962a. Structure and Sentiment. Chicago: University of Chicago Press.

———. 1962b. Genealogy and Category in Wikmunkan Society. Ethnology I.

———. 1962c. Notes on Comparative Method and Prescriptive Alliance. Bijdragen 118.

———. 1963a. Prescriptive Alliance and the Pende. Man 63.

———. 1963b. Some Disputed Points in the Study of Prescriptive Alliance. SWJ 19.

———. 1963c. The Wikmunkan Mother’s Brother: Inference and Evidence. JPS 72.

———. 1963d. Symmetry and Asymmetry in Prescriptive Alliance. Bijdragen 119.

Parsons, T. 1949. The Structure of Social Action. Glencoe, Illinois: The Free Press.

Pospisil, L. 1958. Social Change and Primitive Law. AA 60.

Radclife-Brown, A. R. 1952. Structure and Function in Primitive Society. London: Cohen & West.

———. 1953. Dravidian Kinship Terminology. Man 53, No. 169.

Rivers, W. H. R. 1914a. Kinship and Social Organization. London: Constable.

———. 1914b. History of Melanesian Society. 2 vols. Cambridge University Press.

———. 1924. Social Organization. New York: Knopf.

Salisbury, R. F. 1956. Asymmetrical Marriage Systems. AA 58.

———. 1962a. Unilineal Descent Groups in the New Guinea Highlands. Man 56, No. 2.

———. 1962b. From Stone to Steel. Melbourne: Melbourne University Press.

Schneider, D. M. 1953a. A Note on Bridewealth and the Stability of Marriage. Man 53, No. 73.

Schneider, D. M. 1953b. Yap Kinship Terminology and Kin Groups. AA 55.

———. 1957. Political Organization, Supernatural Sanctions and the Punishment for Incest on Yap. AA 59.

———. 1961. The distinctive Features of Matrilineal Descent Groups. In Schneider, D. M. & Gough, K., Matrilineal Kinship. Berkeley: University of California Press.

———. 1962. Double Descent on Yap. JPS 71.

Schneider, D. M. & Homans, G. C. 1955. Kinship Terminology and the American Kinship System. AA 57.

Schneider, D. M. & Roberts, J. M. 1956. Zuñi Kin Terms. Notebook No. 3, Laboratory of Anthropology, University of Nebraska.

White, H. 1963. An Anatomy of Kinship. Englewood Cliffs, N.J.: Prentice-Hall.

Yalman, N. 1962. The Structure of the Sinhalese Kindred: A Re-Examination of the Dravidian Terminology. AA 64.

Publisher’s note: This is a reprint of D. M. Schneider. 1965. “Some muddles in the models: or, ‘how the system really works’.” In The Relevance of Models for Social Anthropology, edited by M. Banton, 25-86. ASA Monographs 1, London: Tavistock Publications; New York: Frederick A. Preager. We are very grateful to the ASA for granting us the right to reprint it. We remind the reader that we retain the style of the original, and indicate in the text the original page numbers in square brackets.


1. Since its presentation to the Cambridge Conference, I have added Sections XVII, XVIII, and XIX, clarified minor portions of the argument, and profited from the suggestions of Paul W. Friedrich, F. K. Lehman, Melford Spiro, Richard F. Salisbury, and Fred Eggan. Professor Claude Lévi-Strauss made some helpful suggestions, which led me to delete some of the material from the earlier draft which was incorrect or unclear. In particular, he pointed out that I had misunderstood him if I regarded him as an ‘idealist’ in the sense of treating ideas as such as fundamental to social life. Rather, if I understand his position correctly now, he regards himself as an ‘intellectualist’ in the sense that both ideas and action derive from qualities of mind, and that neither action nor ideas has any particular priority. Murphy (1963) has described this most accurately, I think. I had hoped to revise and greatly expand Section II, discussing in detail the problems raised by Homans’s and Lévi-Strauss’s psychological reductionism and their emphasis on the actor’s view of the system as distinct from the observer’s view. But all of this proved another very long essay for which space was not provided here.

3. In fact, of course, Evans-Pritchard seems to have implied very strongly that it was the idea of the lineage quite as much as the actual groups; for it becomes apparent on close inspection that the lineages as actual groups are not nearly so enduring, nor so concrete, nor so strongly localized as might be inferred from a hasty reading of his monograph.

4. I have used the term ‘corporate unit’, since the definition of the word ‘group’ and its associations and meanings are so problematic. A modern nation has a corporate identity. but it is not possible for it to act in unison and concert and as a physically corporate group in the sense, let us say, of a small village assembling in its entirety for a religious ritual (Schneider, 1961).